vvEPA
United States
Environmental Protection
Agency
Office of Research and
Development
Washington DC 20460
EPA/600/4-90/001
March 1990
Direct/Delayed Response
Project: Quality
Assurance Report for
Physical and Chemical
Analyses of
Soils from the
Mid-Appalachian
Region of the
United States
rf-ch ?«
-------
EPA/600/4-90/001
March 1990
Direct/Delayed Response Project:
Quality Assurance Report for Physical and
Chemical Analyses of Soils from the
Mid-Appalachian Region
of the United States
by
G.E. Byers, R.D. Van Remortel, M.J. Mian,
J.E. Teberg, M.L. Papp, B.A. Schumacher,
B.L. Conkling, D.L Cassell, and P.W. Shaffer
A Contribution to the
National Acid Precipitation Assessment Program
U.S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
Environmental Monitoring Systems Laboratory, Las Vegas, Nevada 89193
Environmental Research Laboratory, Corvallis, Oregon 97333
-------
Notice
The information in this document has been funded wholly or in part by the United States
Environmental Protection Agency under Contract Number 68-03-3249 to Lockheed Engineering &
Sciences Company and Cooperative Agreement CR 814701-01 to the Environmental Research Center
of the University of Nevada at Las Vegas. Additional cooperation has been provided under Contract
Number 68-03-3246 to NSI Technology Services Corporation. It has been subject to the Agency's
peer and administrative review, and it has been approved for publication as an EPA document.
Mention of corporation names, trade names, or commercial products does not constitute
endorsement or recommendation for use.
This document is one volume of a set which fully describes the Direct/Delayed Response
Project, Mid-Appalachian Soil Survey. The complete document set includes the major data report,
quality assurance plan, laboratory analysis handbook, field and preparation laboratory operations
and quality assurance reports, and analytical laboratory quality assurance reports. Similar sets are
being produced for each Aquatic Effects Research Program component project. Colored covers,
artwork, and the use of the project name in the document title serve to identify each companion
document set.
The correct citation of this document is:
Byers, G. E.1, R. D. Van Remortel1, M. J. Miah1, J. E. Teberg1, M. L Papp1, B. A. Schumacher1, B. L
Conkling2, D. L Cassell3, and P. W. Shaffer3. 1990. Direct/Delayed Response Project: Quality
Assurance Report for Physical and Chemical Analyses of Soils from the Mid-Appalachian
Region of the United States. EPA/600/4-90/001. U.S. Environmental Protection Agency, Las
Vegas, Nevada.
1 Lockheed Engineering & Sciences Company, Las Vegas, Nevada 89119.
2 Environmental Research Center, University of Nevada, Las Vegas, Nevada 89154.
3 NSI Technology Services Corporation, Corvallis, Oregon 97333.
-------
Abstract
The Direct/Delayed Response Project was designed to address the concern over potential
acidification of surface waters by atmospheric sulfur deposition in the United States. The Mid-
Appalachian Soil Survey was the last of three DDRP surveys conducted from 1985 through 1989.
The purpose of these synoptic soil physical and chemical surveys was to characterize watersheds
in regions of the United States believed to be susceptible to the effects of acidic deposition. This
document describes the implementation of a quality assurance program and the verification of the
analytical data base for the Mid-Appalachian Soil Survey. It is directed primarily towards the users
of the data base who will be analyzing the data and making various assessments and conclusions
relating to the effects of acidic deposition on watersheds of the region.
The quality assurance program was designed in order to satisfy predefined measurement
quality objectives and to assess the measurement uncertainty of field sampling, sample preparation,
and laboratory analysis. The objectives were directed towards the attributes of detectability,
precision, accuracy, representativeness, completeness, and comparability. The quality assurance
program uses quality evaluation and quality control samples and is described and evaluated for its
ability to identify and resolve potential deficiencies during the survey that would otherwise have had
an adverse effect on the data quality. The computerized analytical data entry system "LEVIS" is
also evaluated. Summary estimates of measurement uncertainty in relation to overall data
uncertainty are provided.
The fifty analytical laboratory parameters included in the analytical data base are segregated
into nine groups to allow comparison among similar parameters. Detectability is assessed in terms
of detection limits for instrumentation, contract requirements, and the total system. The
assessment of precision relied strongly on the analysis of measurement quality samples as a
function of intralaboratory within-batch objectives. Two-tiered precision objectives were established
for most of the parameters. A step function statistical method was used to estimate the
components and magnitude of measurement uncertainty. Overall variability from measurement and
population sources was estimated from routine data. Accuracy was investigated in terms of
analytical bias, laboratory differences, and laboratory trends in relation to known reference values.
Completeness objectives were defined for all measurement phases. Comparability was approached
as a complex issue encompassing three levels of detail. Four types of uncertainty estimates, called
delta values, were calculated for each parameter.
The results show that the measurement quality objectives for detectability, precision,
accuracy, representativeness, and completeness were generally satisfied. Measurement uncertainty
was generally low in relation to overall data uncertainty. A series of conclusions and
recommendations are provided at the end of the report. The recommendations will be useful in the
planning of future projects of this nature.
This report is submitted in partial fulfillment of Contract Number 68-03-3249 by Lockheed
Engineering & Sciences Company, Las Vegas, Nevada, under sponsorship of the U.S. Environmental
Protection Agency. The report covers survey activities that occurred from September, 1988, to July,
1989.
HI
-------
Contents
Section Page
Notice ii
Abstract iii
Figures viii
Tables xii
Acknowledgments xiii
Acronyms and Abbreviations xiv
Executive Summary xvii
1. Introduction 1
Project Overview 1
Mid-Appalachian Soil Survey 3
Project Staff Responsibilities 4
Primary Cooperators 4
Organization of the Report 5
Description of Parameter Groups 5
Description of Parameters 6
2. Quality Assurance Program 10
Quality Assurance Optimization 10
Consolidated Preparation Laboratory Facility 10
Separation of Mineral and Organic Samples 12
Changes in Parameters and Methods 12
Consolidation of Total Elemental Analysis 13
Sample Design 13
Statistical Design 14
Computerized Data Entry and Verification 14
Acceptance Criteria 14
Selection of Analytical Laboratories 15
Statements of Work 15
Performance Evaluations 16
Award of Analytical Contracts 16
Distribution of Batches 17
Analytical Laboratory Quality Control 17
Sample Receipt and Storage 17
Laboratory Protocols 18
Sample Homogenization 18
Instrument Calibration 18
Data Entry 18
Data Reporting 19
Preventive Maintenance 19
Technical Systems Audits 19
(continued)
-------
Contents (continued)
Section page
Special Studies 20
Elemental Analysis Study 20
Homogenization Study 21
Filter Pulp Study 22
Independent Analysis of Calcium and Magnesium 22
Assessment Procedures 22
Quality Attributes 22
Measurement Quality Objectives 23
Detectability 24
Precision 26
Accuracy 28
Representativeness 30
Completeness 31
Comparability 31
Uncertainty Estimation 32
Statistical Model 32
Sampling Class/Horizon Groups 33
Measurement Quality Samples 35
Quality Evaluation Samples 35
Quality Control Samples 38
Data Verification 41
Design and Implementation of LEVIS 42
Verification Checks 43
Confirmation and Reanalysis Requests 45
Internal Consistency Checks 46
Overview of the Data Bases 47
3. Results and Discussion 48
Assessment of Detectability 48
Assessment of Precision 50
Moisture and Particle Size Analyses 51
Soil pH 57
Exchangeable Cations in Ammonium Chloride 61
Exchangeable Cations in Ammonium Acetate 68
Cation Exchange Capacity and Exchangeable Acidity 73
Extractable Cations in Calcium Chloride 77
Extractable Iron, Aluminum, and Silicon 84
Extractable Sulfate and Sulfate Adsorption Isotherms 93
Total Carbon, Nitrogen, and Sulfur 103
Assessment of Accuracy 107
Bias 107
Laboratory Differences 108
Laboratory Trends 109
Assessment of Representativeness 111
Assessment of Completeness 112
(continued)
VI
-------
Contents (continued)
Section Page
Assessment of Comparability 112
Comparison of Statistical Methods 112
Comparison of Measurement Quality Samples 112
Comparison of Sampling, Preparation, and Analytical Methods 113
Comparison of User-Defined Acceptability Criteria 115
Quantitative Comparison of the DDRP Surveys 115
Interlaboratory Comparison Study 116
Assessment of Data Uncertainty Components 117
Additive Components of Uncertainty 117
Effect of Measurement System on Data Uncertainty 118
4. Conclusions and Recommendations 122
Quality Assurance Optimization 122
Sample Flow 122
Sample Design 122
Data Verification 123
Data Quality Attributes 124
Detectability 124
Precision 125
Accuracy 126
Representativeness 128
Completeness 128
Comparability 128
Uncertainty Estimation 129
References 130
Glossary 134
Appendices
A. Analytical Data Verification and Validation Flags 140
B. Quality Assurance Reanalysis Templates 146
C. Statistics for Step Function Uncertainty Estimates 164
D. Inordinate Data Points Influencing the Uncertainty Estimates 178
E. Precision Plots for Particle Size Fractions 186
F. General Statistics for the Laboratory Audit Samples 194
G. Range and Frequency Histograms of the Data Sets 200
H. Technical Systems Audit Reports 251
I. Plots of Moving Averages for Laboratory Audit Samples 269
J. Step Function Approach for Estimating Data Uncertainties 320
K. Data Tracking and Verification Forms 333
VII
-------
Figures
Number Page
1-1 Regional surveys of the Direct/Delayed Response Project 2
2-1 Flowchart of verification activities in the Mid-Appalachian Soil Survey 11
2-2 Example of two-tiered precision objective 26
2-3 Example of an accuracy window with upper, lower, and reference values 29
2-4 Quality evaluation and quality control soil samples for mineral and organic batches . . 37
3-1 Range and frequency distribution of MOIST for precision 53
3-2 Range and frequency distribution of SAND for precision 54
3-3 Range and frequency distribution of SILT for precision 55
3-4 Range and frequency distribution of CLAY for precision 56
3-5 Range and frequency distribution of PH_H2O for precision 58
3-6 Range and frequency distribution of PH~002M for precision 59
3-7 Range and frequency distribution of PH~01M for precision 60
3-8 Range and frequency distribution of CA~CL for precision 63
3-9 Range and frequency distribution of MG^CL for precision 64
3-10 Range and frequency distribution of K_CL for precision 65
3-11 Range and frequency distribution of NA_CL for precision 66
3-12 Range and frequency distribution of AL_CL for precision 67
3-13 Range and frequency distribution of CAjOAC for precision 69
3-14 Range and frequency distribution of MG_OAC for precision 70
3-15 Range and frequency distribution of K_OAC for precision 71
3-16 Range and frequency distribution of NA_OAC for precision 72
3-17 Range and frequency distribution of CEC_CL for precision 74
3-18 Range and frequency distribution of CEC~OAC for precision 75
3-19 Range and frequency distribution of AC_BACL for precision 76
3-20 Range and frequency distribution of CA~CL2 for precision 78
3-21 Range and frequency distribution of MG~lCL2 for precision 79
3-22 Range and frequency distribution of K_CL2 for precision 80
3-23 Range and frequency distribution of NA_CL2 for precision 81
3-24 Range and frequency distribution of FE_CL2 for precision 82
3-25 Range and frequency distribution of AL~CL2 for precision 83
3-26 Range and frequency distribution of FE_PYP for precision 86
3-27 Range and frequency distribution of AL~PYP for precision 87
3-28 Range and frequency distribution of FE_AO for precision 88
3-29 Range and frequency distribution of AL_AO for precision 89
3-30 Range and frequency distribution of SI~AO for precision 90
3-31 Range and frequency distribution of FE_CD for precision 91
3-32 Range and frequency distribution of ALjCD for precision 92
3-33 Range and frequency distribution of S04_H2O for precision 95
3-34 Range and frequency distribution of SO4~PO4 for precision 96
3-35 Range and frequency distribution of SO4~0 for precision 97
3-36 Range and frequency distribution of SO4_2 for precision 98
3-37 Range and frequency distribution of SO4~4 for precision 99
(continued)
viii
-------
Figures (continued)
Number Page
3-38 Range and frequency distribution of SO4_8 for precision 100
3-39 Range and frequency distribution of SO4J6 for precision 101
3-40 Range and frequency distribution of SO4_32 for precision 102
3-41 Range and frequency distribution of C_TOT for precision 104
3-42 Range and frequency distribution of NJTOT for precision 105
3-43 Range and frequency distribution of SJTOT for precision 106
B-1 Quality assurance reanalysis template for SAND, SILT 147
B-2 Quality assurance reanalysis template for CLAY 148
B-3 Quality assurance reanalysis template for PH H2O, PH 002M, PH 01M 149
B-4 Quality assurance reanalysis template for CA,~MG, K, NA_CL; CA,~MG, K, NA OAC ... 150
B-5 Quality assurance reanalysis template for AL_CL ~ 151
B-6 Quality assurance reanalysis template for CEC_CL, CEC_OAC 152
B-7 Quality assurance reanalysis template for AC_BACL . . 7 153
B-8 Quality assurance reanalysis template for CA~CL2 154
B-9 Quality assurance reanalysis template for MG~lCL2, K_CL2, NA_CL2 155
B-10 Quality assurance reanalysis template for FE_CL2 . . T ~. 156
B-11 Quality assurance reanalysis template for AL~CL2 157
B-12 Quality assurance reanalysis template for FE7AL_PYP; FE, AL, SI AO; FE, AL CD 158
B-13 Quality assurance reanalysis template for SO4_H2O, SO4_PO4 ~. ~. 159
B-14 Quality assurance reanalysis template for SO4_0-32 160
B-15 Quality assurance reanalysis template for C_TOT 161
B-16 Quality assurance reanalysis template for N_TOT 162
B-17 Quality assurance reanalysis template for S~TOT 163
E-1 Range and frequency distribution of VCOS for precision 187
E-2 Range and frequency distribution of COS for precision 188
E-3 Range and frequency distribution of MS for precision 189
E-4 Range and frequency distribution of FS for precision 190
E-5 Range and frequency distribution of VFS for precision 191
E-6 Range and frequency distribution of COSI for precision 192
E-7 Range and frequency distribution of FSI for precision 193
G-1 Range and frequency distributions for MOIST 201
G-2 Range and frequency distributions for SAND 202
G-3 Range and frequency distributions for VCOS 203
G-4 Range and frequency distributions for COS 204
G-5 Range and frequency distributions for MS 205
G-6 Range and frequency distributions for FS 206
G-7 Range and frequency distributions for VFS 207
G-8 Range and frequency distributions for SILT 208
G-9 Range and frequency distributions for COSI 209
G-10 Range and frequency distributions for FSI 210
G-11 Range and frequency distributions for CLAY 211
G-12 Range and frequency distributions for PHJH20 212
G-13 Range and frequency distributions for PH~002M 213
G-14 Range and frequency distributions for PH~01M 214
(continued)
IX
-------
Figures (continued)
Number Page
G-15 Range and frequency distributions for CA_CL 215
G-16 Range and frequency distributions for MG^_CL 216
G-17 Range and frequency distributions for K_CL 217
G-18 Range and frequency distributions for NA_CL 218
G-19 Range and frequency distributions for AL~CL 219
G-20 Range and frequency distributions for CA_OAC 220
G-21 Range and frequency distributions for MG_OAC 221
G-22 Range and frequency distributions for K_OAC 222
G-23 Range and frequency distributions for NA_OAC 223
G-24 Range and frequency distributions for CEC_CL 224
G-25 Range and frequency distributions for CEC~OAC 225
G-26 Range and frequency distributions for ACJ3ACL 226
G-27 Range and frequency distributions for CA~CL2 227
G-28 Range and frequency distributions for MG_CL2 228
G-29 Range and frequency distributions for K_CL2 229
G-30 Range and frequency distributions for NA_CL2 230
G-31 Range and frequency distributions for FE_CL2 231
G-32 Range and frequency distributions for AL_CL2 232
G-33 Range and frequency distributions for FE~PYP 233
G-34 Range and frequency distributions for AL~PYP 234
G-35 Range and frequency distributions for FE_AO 235
G-36 Range and frequency distributions for AL_AO 236
G-37 Range and frequency distributions for SI_AO 237
G-38 Range and frequency distributions for FE^_CD 238
G-39 Range and frequency distributions for AL_CD 239
G-40 Range and frequency distributions for SO4_H2O 240
G-41 Range and frequency distributions for SO4_PO4 241
G-42 Range and frequency distributions for SO4_0 242
G-43 Range and frequency distributions for SO4^2 243
G-44 Range and frequency distributions for SO4_4 244
G-45 Range and frequency distributions for SO4J3 245
G-46 Range and frequency distributions for SO4J6 246
G-47 Range and frequency distributions for SO4_32 247
G-48 Range and frequency distributions for CJTOT 248
G-49 Range and frequency distributions for N_TOT 249
G-50 Range and frequency distributions for S_TOT 250
1-1 Range and frequency distributions for MOIST 270
1-2 Range and frequency distributions for SAND 271
1-3 Range and frequency distributions for VCOS 272
1-4 Range and frequency distributions for COS 273
1-5 Range and frequency distributions for MS 274
1-6 Range and frequency distributions for FS 275
1-7 Range and frequency distributions for VFS 276
1-8 Range and frequency distributions for SILT 277
1-9 Range and frequency distributions for COSI 278
1-10 Range and frequency distributions for FSI 279
1-11 Range and frequency distributions for CLAY 280
1-12 Range and frequency distributions for PH_H2O 281
(continued)
-------
Figures (continued)
Number
1-13 Moving average plots of laboratory trends for PH_002M ....................... 282
-14 Moving average plots of laboratory trends for PH_01M ........................ 283
-15 Moving average plots of laboratory trends for CA_CL ......................... 284
-16 Moving average plots of laboratory trends for MG_CL ................ . ....... 285
-17 Moving average plots of laboratory trends for K_CL .......................... 286
-18 Moving average plots of laboratory trends for NA_CL ......................... 287
-19 Moving average plots of laboratory trends for AL_CL ......................... 288
-20 Moving average plots of laboratory trends for CA~OAC ....................... 289
-21 Moving average plots of laboratory trends for MG_OAC ....................... 290
-22 Moving average plots of laboratory trends for K_OAC ........................ 291
-23 Moving average plots of laboratory trends for NA_OAC ....................... 292
-24 Moving average plots of laboratory trends for CEC_CL ....................... 293
-25 Moving average plots of laboratory trends for CEC_OAC ...................... 294
-26 Moving average plots of laboratory trends for AC_BACL ....................... 295
-27 Moving average plots of laboratory trends for CA_CL2 ........................ 296
-28 Moving average plots of laboratory trends for MG_CL2 ....................... 297
-29 Moving average plots of laboratory trends for K_CL2 ......................... 298
-30 Moving average plots of laboratory trends for NA_CL2 ........................ 299
-31 Moving average plots of laboratory trends for FE_CL2 ........................ 300
-32 Moving average plots of laboratory trends for AL_CL2 ........................ 301
-33 Moving average plots of laboratory trends for FE_PYP ........................ 302
-34 Moving average plots of laboratory trends for AL_PYP ........................ 303
-35 Moving average plots of laboratory trends for FE_AO ......................... 304
-36 Moving average plots of laboratory trends for AL_AO ......................... 305
-37 Moving average plots of laboratory trends for SI ~AO ......................... 306
1-38 Moving average plots of laboratory trends for FE_CD ......................... 307
1-39 Moving average plots of laboratory trends for ALjCD ......................... 308
1-40 Moving average plots of laboratory trends for SO4_H2O ...................... 309
1-41 Moving average plots of laboratory trends for SO4_PO4 ....................... 310
1-42 Moving average plots of laboratory trends for SO4~0 ......................... 311
1-43 Moving average plots of laboratory trends for SO4_2 ......................... 312
1-44 Moving average plots of laboratory trends for SO4_4 ......................... 313
1-45 Moving average plots of laboratory trends for SO4_8 ......................... 314
1-46 Moving average plots of laboratory trends for SO4J6 ........................ 315
1-47 Moving average plots of laboratory trends for SO4~32 ........................ 316
1-48 Moving average plots of laboratory trends for C_TOT ......................... 317
1-49 Moving average plots of laboratory trends for N_TOT ......................... 318
1-50 Moving average plots of laboratory trends for S~TOT ......................... 319
K-1 DDRP Form 102 (shipping form) ........................................ 334
K-2 DDRP Form 500 (data confirmation/reanalysis reanalysis form) ................. 335
K-3 DDRP Form 600A (internal consistency confirmation form) .................... 336
K-4 DDRP Form 600B (internal consistency confirmation form) ..................... 337
xi
-------
Tables
Number Pago
1-1 Analytical Parameters Measured in the Mid-Appalachian Soil Survey 7
2-1 Distribution of Batches for General Soil Analysis 17
2-2 Contract-Required Detection Limits for the Analytical Laboratories 25
2-3 Within-Batch Precision Objectives for the Analytical Measurement Quality Samples . . 27
2-4 Sampling Classes Used to Group Different Soil Types 34
2-5 Primary Horizon Types for Sampling Class/Horizon Groups 34
2-6 Distribution of Field Measurement Quality Samples and Routine Samples 36
2-7 Status and Assessment Purpose of Measurement Quality Soil Samples 36
2-8 Number and Percentage of Quality Evaluation and Routine Samples 36
2-9 Quality Control Requirements for the Analytical Sample Batches 39
2-10 Soil Chemistry Relationships and Delimiters 45
2-11 Internal Consistency Checks Performed for Parameter Correlations 46
3-1 Detection Limits for Contractual Compliance and Independent Assessment 49
3-2 Achievement of MQOs for Precision of Moisture and Particle Size Analysis 51
3-3 Achievement of MQOs for Precision of Soil pH 57
3-4 Achievement of MQOs for Precision of Cations in Ammonium Chloride 61
3-5 Achievement of MQOs for Precision of Cations in Ammonium Acetate 68
3-6 Achievement of MQOs for Precision of Cation Exchange Capacity and Acidity 73
3-7 Achievement of MQOs for Precision of Cations in Calcium Chloride 77
3-8 Achievement of MQOs for Precision of Extractable Iron, Aluminum, and Silicon .... 84
3-9 Achievement of MQOs for Precision of Extractable Sulfate and Isotherm Parameters 93
3-10 Achievement ot MQOs for Precision of Total Carbon, Nitrogen, and Sulfur 103
3-11 Analytical Bias Estimates 107
3-12 Contribution of Outlying Audit Samples to Overall Analytical Bias 108
3-13 Analytical Laboratory Differences Pcoled Across Laboratory Audit Samples 110
3-14 Significant Differences in Distribution of Field Duplicates and Routine Samples 113
3-15 Completeness of Routine Data from Verified and Validated Data Bases 114
3-16 Multiple Comparison of Interlaboratory Mean Values 116
3-17 Comparison of Measurement Uncertainty Components 119
3-18 Delta Values and Ratios for Assessment of Uncertainty Components 120
4-1 Precision Indices for Parameter Groups 126
4-2 Accuracy Indices for Parameter Groups 126
4-3 Laboratory Difference Indices for Parameter Groups 127
A-1 Analytical Data Verification Flags 141
A-2 Analytical Data Validation Flags 144
C-1 Table of Statistics for Step Function Uncertainty Estimates 165
D-1 Inordinate Data Points Having a High Degree of Influence on Precision Estimates . . 179
F-1 Summary Statistics of Analytical Parameters for Laboratory Audit Samples 195
XII
-------
Ackno wledgments
External peer reviews of this report by the following individuals are gratefully acknowledged:
Dr. Mark B. David, Department of Forestry, University of Illinois, Urbana, Illinois; Ms. Lora S.
Johnson, Technology Applications Inc., Cincinnati, Ohio; and Mr. James R. Webb, Department of
Environmental Sciences, University of Virginia, Chariottesville, Virginia. We also appreciate the
special interest peer reviews performed by Dr. John K. Taylor (retired), National Institute of
Standards & Technology, Gaithersburg, Maryland, and Mr. David V. Peck, Lockheed Engineering &
Sciences Company, Las Vegas, Nevada.
The authors wish to acknowledge the following individuals for their technical assistance
during the development of this document: D. S. Coffey, M. G. Johnson, J. S. Kern, and C. J. Palmer,
NSI Technology Services Corporation, Corvallis, Oregon; C. C. Brandt and R. S. Turner, Oak Ridge
National Laboratory, Oak Ridge, Tennessee; L K. Fenstermaker, University of Nevada, Las Vegas,
Nevada; and R. L Tidwell, K. C. Shines, J. V. Burton, R. L. Slagle, M. H. Bartling, W. H. Cole, D. W.
Sutton, J. M. Boyd, and B. N. Cordova, Lockheed Engineering & Sciences Company, Las Vegas,
Nevada. We appreciate the contribution of the DDRP Technical Director, M. R. Church,
Environmental Research Laboratory, U.S. Environmental Protection Agency, Corvallis, Oregon, in the
development of the introduction section of this report.
The following individuals provided computer graphics or logistical support and are gratefully
acknowledged: B. N. Cordova, L. A. Stanley, and A. M. Tippett, Lockheed Engineering & Sciences
Company, Las Vegas, Nevada. In addition, B. N. Cordova is acknowledged for his exceptional
support in computer graphics and word processing.
Finally, we appreciate the support of our technical monitors, L. J. Blume and D. T. Heggem,
U.S. Environmental Protection Agency, Las Vegas, Nevada, during the course of this survey.
XIII
-------
Acronyms and Abbreviations
AAS atomic absorption spectrometry
AC_BACL exchangeable acidity in barium chloride triethanolamine
AC_KCL exchangeable acidity in potassium chloride
AD analytical duplicate sample
AERP Aquatic Effects Research Program
AL_AO extractable aluminum in acid oxalate
AL_CD extractable aluminum in citrate dithionite
AL_CL exchangeable aluminum in ammonium chloride
AL_CL2 extractable aluminum in calcium chloride
AL_KCL exchangeable aluminum in potassium chloride
AL_PYP extractable aluminum in sodium pyrophosphate
ALSOW analytical laboratory statement of work
ANC acid neutralizing capacity
ANOVA analysis of variance
CA_CL exchangeable calcium in ammonium chloride
CA_CL2 extractable calcium in calcium chloride
CA_OAC exchangeable calcium in ammonium acetate
CEC cation exchange capacity
CEC_CL cation exchange capacity by ammonium chloride
CEC_OAC cation exchange capacity by ammonium acetate
CLAY total clay fraction
C:N carbon:nitrogen
COS coarse sand fraction
COSI coarse silt fraction
CRDL contract-required detection limit
C:S carbon: sulfur
C_TOT total carbon
DDRP Direct/Delayed Response Project
df degrees of freedom
DL-QCCS detection limit quality control check sample
DQO data quality objective
EGME ethylene glycol monoethyl ether
EMSL-LV Environmental Monitoring Systems Laboratory at Las Vegas, Nevada
EPA U.S. Environmental Protection Agency
ERL-C Environmental Research Laboratory at Corvallis, Oregon
FAL low-range field audit sample
FAO organic field audit sample
FAP mineral field audit sample
FD field duplicate sample
FE_AO extractable iron in acid oxalate
FE_CD extractable iron in citrate dithionite
FE_CL2 extractable iron in calcium chloride
FE_PYP extractable iron in pyrophosphate
FIA flow injection analysis
(continued)
xiv
-------
Acronyms and Abbreviations (continued)
FES flame emission spectrometry
FSI fine silt fraction
FS fine sand fraction
ICP inductively coupled argon plasma
IDL instrument detection limit
IIDL independent instrument detection limit
IFB invitation for bid
K_CL exchangeable potassium in ammonium chloride
K_CL2 extractable potassium in calcium chloride
K~OAC exchangeable potassium in ammonium acetate
LAL low-range laboratory audit sample
LAO organic laboratory audit sample
LAP mineral laboratory audit sample
LEVIS Laboratory Entry and Verification Information System
MASS Mid-Appalachian Soil Survey
MG_CL exchangeable magnesium in ammonium chloride
MG_CL2 extractable magnesium in calcium chloride
MG_OAC exchangeable magnesium in ammonium acetate
MOIST percent air-dry soil moisture content
MS medium sand fraction
MQO measurement quality objective
NA_CL exchangeable sodium in ammonium chloride
NA~CL2 extractable sodium in calcium chloride
NA_OAC exchangeable sodium in ammonium acetate
NAPAP National Acid Precipitation Assessment Program
NCSS National Cooperative Soil Survey
NE Northeastern
NSWS National Surface Water Survey
N_TOT total nitrogen
ORNL Oak Ridge National Laboratory
P percentile
PD preparation duplicate sample
PE performance evaluation
PHJH2O pH in deionized water
PH_002M pH in 0.002M calcium chloride
PH_01M pH in 0.01 M calcium chloride
QA quality assurance
QART quality assurance reanalysis template
QC quality control
QCAS quality control audit sample
QCCS quality control check sample
QE quality evaluation
R reference value
RS routine sample
RSD relative standard deviation
SAND total sand fraction
SBRP Southern Blue Ridge Province
SCR soil chemistry relationship
SCS Soil Conservation Service
SD standard deviation
SDL system detection limit
(continued)
xv
-------
Acronyms and Abbreviations (continued)
S/H sampling class/horizon group
SI_AO extractable silicon in acid oxalate
SILT total silt fraction
SO4J3 zero mg S/L sulfate isotherm parameter
SO4_2 two mg S/L sulfate isotherm parameter
SO4~4 four mg S/L sulfate isotherm parameter
SO4_8 eight mg S/L sulfate isotherm parameter
SO4J6 sixteen mg S/L sulfate isotherm parameter
SO4J32 thirty-two mg S/L sulfate isotherm parameter
SO4JH2O extractable sulfate in deionized water
SO4~PO4 extractable sulfate in sodium phosphate
SP SUR specific surface
S_TOT total sulfur
USDA U.S. Department of Agriculture
VCOS very coarse sand fraction
VFS very fine sand fraction
XVI
-------
Executive Summary
Introduction
The U.S. Environmental Protection Agency (EPA), as a participant in the National Acid
Precipitation Assessment Program, has designed and implemented a research program to predict
the long-term response of watersheds and surface waters in the United States to acidic deposition.
Based on this research, a sample of watershed systems will be classified according to the time
scale in which each system will reach an acidic steady state, assuming current levels of acidic
deposition. The Direct/Delayed Response Project (DDRP) was designed as the terrestrial com-
ponent of the EPA Aquatic Effects Research Program.
The mapping for the DDRP Mid-Appalachian Soil Survey was conducted in portions of
Pennsylvania, Virginia, and West Virginia during the spring and summer of 1988 and the sampling
took place the during the fall of 1988. These activities initiated the third full-scale regional survey
of the DDRP. The physical and chemical properties that were measured in the soil samples are
listed below.
Physical and Chemical Parameters Measured In the Direct/Delayed Response Protect Mid-Appalachian Soil Survey
Air-dry moisture content
Total sand
Very coarse sand
Coarse sand
Medium sand
Fine sand
Very fine sand
Total silt
Coarse silt
Fine silt
Total clay
pH in deionized water
pH in 0.002M calcium chloride
pH in 0.01M calcium chloride
Ca in 1.0M ammonium chloride
Mg in 1.0M ammonium chloride
K in 1.0M ammonium chloride
Na in 1.0M ammonium chloride
Al in 1.0M ammonium chloride
Ca in 1.0M ammonium acetate
Mg in 1.0M amonium acetate
K in 1.0M ammonium acetate
Na in 1.0M ammonium acetate
CEC in 1.0M ammonium chloride
CEC in 1.0M ammonium acetate
Ex. Acidity by barium chloride-TEA
Ca in 0.002M calcium chloride
Mg in 0.002M calcium chloride
K in 0.002M calcium chloride
Na in 0.002M calcium chloride
Fe in 0.002M calcium chloride
Al in 0.002M calcium chloride
Ext. Fe in pyrophosphate
Ext. Al in pyrophosphate
Ext. Fe in acid oxalate
Ext. Al in acid oxalate
Ext. Si in acid oxalate
Ext. Fe in citrate dithionite
Ext. Al in citrate dithionite
Ext. sulfate in deionized water
Ext. sulfate in sodium phosphate
Sulfate isotherm 0 mg sulfur
Sulfate isotherm 2 mg sulfur
Sulfate isotherm 4 mg sulfur
Sulfate isotherm 8 mg sulfur
Sulfate isotherm 16 mg sulfur
Sulfate isotherm 32 mg sulfur
Total carbon
Total nitrogen
Total sulfur
Quality Assurance Optimization
As a result of conclusions and recommendations from the DDRP Northeastern region and
Southern Blue Ridge Province soil surveys, a number of improvements were made in the quality
assurance program for the Mid-Appalachian soil survey. A single preparation laboratory was
established in Las Vegas, Nevada, in close proximity to the EPA Environmental Monitoring Systems
Laboratory - Las Vegas (EMSL-LV) quality assurance (QA) staff. This allowed the development of,
and adherence to, strictly defined sample preparation protocols and the ability to track and control
progress at the laboratory on a real-time basis.
Mineral and organic samples were placed in separate batches because of differences in
analyte concentrations and in the soil:solution ratios required for analysis. As a result, the
analytical laboratories were able to perform instrument calibration and sample analyses within
XVII
-------
narrower linear dynamic ranges, allowing the QA staff to make more reliable assessments of the
resulting data quality.
Several changes in analytical methods and procedures were initiated in the Mid-Appalachian
survey. For example, the exchangeable aluminum in 1M potassium chloride parameter was replaced
by exchangeable aluminum in 1M ammonium chloride as part of the measurement of exchangeable
cations. This change in extraction solution reduced the analysis time for exchangeable aluminum
while retaining similar experimental conditions. The quantity of mineral soil used in the
exchangeable cation analyses was increased relative to the extraction solution which facilitated
instrumental analysis. The amount of cellulose filter pulp used in extractions was decreased and
a pulp prewash step was added to reduce contamination by exchangeable cations in the pulp
material. The determination of extractable cations in 0.002M calcium chloride using a mechanical
extractor was changed to an overnight extraction using a mechanical shaker. Also, pH was
measured on this extract rather than on a separate extract as used in previous surveys. Total
carbon, nitrogen, and sulfur were analyzed by a single laboratory in response to the Mid-
Appalachian survey QA requirements for tighter measurement quality and the specialized nature of
the analytical instrumentation required for these analyses.
An effort was made to use measurement quality samples in the field, at the preparation
laboratory, and at the analytical laboratories in such a way as to provide optimal real-time control
and evaluation of data quality. Of particular importance was the addition of field audit samples
at the sampling phase to allow the estimation of sampling and system-wide measurement
uncertainties. The statistical approach in the Mid-Appalachian survey made use of a balanced
hierarchical design that allowed the various components of measurement uncertainty to be
estimated with respect to the larger population uncertainty. A step function technique was
developed by the QA staff to evaluate data quality in terms of predefined objectives and in relation
to the routine sample data.
A Laboratory Entry and Verification Information System (LEVIS) was developed for use by the
QA staff and the analytical laboratories. The LEVIS program facilitated the entry, edit, and review
of raw data and the calculation of final data values. The program also performed verification
checks for the measurement quality samples and produced QC summary reports.
Many significant changes in the batch acceptance criteria were initiated. Among them were
the tightening of contract-required detection limits and precision requirements. Several new
measurement quality samples were introduced to check both precision and accuracy. Acceptable
accuracy windows for laboratory audit samples were established and used as contractual
requirements for the laboratories. A template was also developed by the QA staff to assist in
setting major and minor flags, and helped to eliminate the subjectivity present in the previous DDRP
surveys in regard to reanalysis requests.
Data Quality Assessment
The quality assurance (QA) program for soil sampling, sample preparation, and sample
analysis were designed to satisfy measurement quality objectives (MQOs) for the resulting data and
to assess the variability of sampling, preparation, and analytical performance. The MQOs for this
survey were primarily directed toward the attributes of detectability, precision, accuracy, and
completeness. Representativeness and comparability of the data were also assessed, although
quantitative MQOs were not imposed.
Detectability
The instrument detection limit is the lowest value that an analytical instrument can reliably
detect above instrument background concentrations. Acceptable initial instrument detection limits
were established prior to any sample analysis and subsequent values were determined and
xviii
-------
reported on a batch-by-batch basis. Contracts with the analytical laboratories specified maximum
allowable instrument detection limits and, if a reported batch detection limit was invalid, the batch
was reanalyzed for that parameter. Secondary checks on the reliability of the instrument detection
limits were made using independently determined values.
System detection limits were estimated using low concentration field audit samples which
reflected the uncertainty introduced during soil sampling, preparation, extraction, and analysis.
These limits allowed data users to identify soil samples which had a measured concentration that
was statistically different from the reagent or calibration blanks.
Precision
Precision is the level of agreement among multiple measurements of the same soil
characteristic. Measurement imprecision is distinct from the overall variability in the population
itself. Determination of measurement imprecision and its sources in the Mid-Appalachian soil
survey relied strongly on analysis of the measurement quality samples and was a function of the
intralaboratory within-batch precision MQOs defined in the QA Plan. Overall variability stemming
from measurement and population sources was estimated from the routine data.
The precision MQOs were characterized by a two-tiered assessment system. Below a specific
concentration, called the "knot", precision was defined as an absolute standard deviation in
parameter reporting units; above the knot, precision was defined as a percent relative standard
deviation. To address the issue of concentration-dependent variance, the range of soil analyte
concentrations was partitioned into appropriate intervals within which the error variance was
relatively constant. A step function was fitted across the intervals to represent the error variance
for the entire concentration range. Different step functions were used to assess variability in
selected measurement quality samples, and variability in the population of routine soil samples
collected was also estimated.
Accuracy
Accuracy is the level of agreement between an observed value and the "true" value of a soil
characteristic. Data from the laboratory natural audit samples were used to estimate analytical
accuracy and data from the field natural audit samples were used to assess accuracy from a
system-wide measurement perspective. Each audit sample type was assigned a range of
acceptable values for each parameter in the form of an accuracy window, which was derived from
an MQO-based confidence interval placed around a weighted estimate of the mean calculated using
analytical data from the previous DDRP surveys.
The three aspects of accuracy investigated were bias, laboratory differences, and laboratory
trends. Analytical bias was considered to be the quantitative measure of accuracy used in the
estimation of measurement uncertainty. Laboratory differences were assessed in relation to known
reference values and, in conjunction with laboratory trends, served as quantitative and qualitative
evaluations of analytical laboratory performance.
Representativeness
The representativeness objectives of the survey were qualitative and quantitative in nature.
The general objectives were that: (1) the soil pedons sampled by the field sampling crews be
representative of the soil sampling class characteristics, (2) the samples that were collected be
homogenized and subsampled properly by the preparation laboratory personnel, and (3) the field
duplicate samples adequately represent the range and frequency distribution of analyte
concentrations found in the routine samples.
XIX
-------
Completeness
The completeness objectives of the survey were to ensure that: (1) all pedons designated for
sampling were actually sampled, (2) all samples received by the preparation laboratory were
processed, and (3) all samples received by the analytical laboratories were analyzed and that 90
percent or more of the required measurements were made on all of the samples.
Comparability
Comparability of data from the three DDRP surveys was treated as a complex issue having
several levels of detail which should be considered. Level 1 comparability was established on the
basis of statistical evaluation methods, measurement quality samples, and the sample collection,
preparation, and analysis methods used. Level 2 comparability was established by the acceptability
and useability of the verified data bases as defined by the data users. Level 3 comparability
allowed the direct quantitative comparison of data for each parameter of interest.
Uncertainty Estimates
The term "uncertainty" was used to describe the sum of all quantifiable sources of error
associated with a given phase of the measurement system or in the sample population.
Uncertainty estimates, or delta values, were calculated for each parameter using the square root
of the sum of the within- and between-batch variances and squared bias term. Four delta values
were calculated for each parameter. The delta, values represent analytical laboratory uncertainty
and were estimated using the low-range laboratory audit samples. The delta2 values represent
the confounded uncertainty of sample preparation and analysis, and were estimated using the
preparation duplicates. The delta3 values represent the confounded overall measurement uncertainty
of field sampling, sample preparation, and sample analysis, and were estimated using the field
duplicates. The delta4 values represent uncertainty due to the spatial heterogeneity of the routine
sample population confounded with the overall measurement uncertainty of field sampling, sample
preparation, and sample analysis, and were estimated using the sampling class/horizon groups of
routine samples. The sampling class/horizon groups refer to configurations of the samples that
were believed to have similar physical and chemical properties in relation to soil responses to acid
deposition.
Quality Evaluation and Quality Control Samples
Quality evaluation (QE) samples were used to assess overall measurement uncertainty and
to provide an independent check on the quality control (QC) procedures. The QE samples were
known to the QA staff but were either blind or double-blind to the sampling crews and preparation
or analytical laboratory personnel. Six types of QE samples were used in the Mid-Appalachian soil
survey: (1) field duplicates (soil samples were collected by each sampling crew from one horizon
of every third pedon sampled and were placed randomly in the sample batch with the other
samples from the same pedon); (2) field audits (duplicate mineral soil samples or triplicate organic
soil samples were sent by the QA staff to the sampling crews for processing as if they were
routine samples); (3) low-range field audits (low concentration mineral soil samples were sent
with the field audits to the sampling crews by the QA staff); (4) preparation duplicates (a pair of
preparation duplicates, one split from the field duplicate sample and one split from its associated
routine sample, was created at the preparation laboratory and placed randomly in the sample
batch); (5) laboratory audits (duplicate mineral soil samples or triplicate organic soil samples,
identical to the field audits were sent by the QA staff to the preparation laboratory for inclusion in
each batch); and (6) low-range laboratory audits (low concentration mineral soil samples identical
to the low-range field audits were sent to the preparation laboratory by the QA staff for inclusion
with each mineral soil batch).
xx
-------
The composition of QC samples was known to the analyst and the analytical results from
each laboratory were required to satisfy the batch acceptance criteria as the samples were
analyzed. Immediate feedback on the functioning of the analytical system allowed sample
processing and analytical deficiencies to be resolved quickly, resulting in minimal error form these
sources. Nine types of QC samples were used in the Mid-Appalachian soil survey: (1) QC audit
samples were used to control bias and reduce between-laboratory and between-batch measurement
uncertainty; (2) analytical duplicates were used to control analytical within-batch precision; (3)
calibration blanks were used as a check for sample contamination, analytical carryover effects, and
baseline drift in the analytical instrument immediately after calibration; (4) reagent blanks served
as a check for reagent contamination; (5) QC check samples served as a check on the accuracy
and consistency of the instrument calibration throughout the sample batch analysis; (6) detection
limit QC check samples eliminated the necessity of determining the detection limit every day and
allowed accuracy to be estimated at the low end of the calibration range; (7) matrix spikes were
used to determine the sample matrix effect on analytical measurements; (8) adsorption spike
solutions served as a check for reagent contamination; and (9) ion chromatography resolution
samples were used to provide evidence of acceptable peak separation for the sulfate analyses.
In addition to the use of QC samples for quality control, technical system audits,
or on-site evaluations were also conducted. Each analytical laboratory underwent an audit after
successfully analyzing a set of performance evaluation samples prior to receiving a contract.
Second-round audits were performed after each laboratory had completed most of the analyses
on two sample batches. During third-round auditing, two laboratories underwent two additional
audits. The third laboratory underwent only one additional audit.
Data Verification
The analytical laboratories and the QA staff used LEVIS for data verification. Phase one of
the LEVIS program included the data entry component and two verification components. The
analytical laboratories entered and evaluated data as it was produced using a QC summary report
of batch data characteristics and sixteen soil chemistry relationships. The data were then
transferred to the central computer at EMSL-LV. The QA staff evaluated preliminary and formally
submitted batches from the analytical laboratories using precision and accuracy windows which
were checked by LEVIS. The LEVIS program flagged data which did not meet the MQOs.
Following the initial data evaluation for a batch, the QA staff prepared a summary document of all
flagged parameters, indicating whether a flag originated from an unacceptable value for the QC,
chemistry, precision, or accuracy criteria. The number and severity of the flags for each parameter
were checked using the QA reanalysis template to determine if reanalysis was required.
After completion and receipt of all final batch data from the analytical laboratories, two
internal consistency checks were performed to check for possible outliers in the routine data. A
correlation approach was used to assess internal consistency in which the coefficients of
determination were obtained by performing weighted linear regressions. From the regressions,
studentized residuals and difference of fit statistics were calculated to identify extreme data values
that could be considered outliers. Approximately 3 percent of the data underwent confirmation
using this consistency check. The second internal consistency check used the data structured into
a pedon/horizon data set using the original soil profile sequence. Data for each pedon were visually
scanned by soil scientists for consistency of the parameter values from one horizon to the next.
Approximately 2 percent of the data underwent confirmation using this consistency check.
Data Management
The field sampling and sample preparation data were entered into SAS-AF raw data files
on personal computers at EMSL-LV. The analytical data were entered into LEVIS on personal
computers located at the laboratories. Data verification was accomplished by a systematic
evaluation of completeness, precision, internal consistency, and coding accuracy. Apparent
xxi
-------
discrepancies were appended with flags unless they could be corrected. After verification was
complete, the data bases were frozen and sent to the Oak Ridge National Laboratory in Tennessee
to undergo validation in cooperation with personnel at the EPA Environmental Research Laboratory
at Corvallis, Oregon, and at EMSL-LV. The validation procedures included a specific assessment
of outlying data points for inclusion or omission in validated data sets based on assigned
confidence levels.
Results and Discussion
Detectabiiity
The calculated instrument detection limits were less than the contract-required limits for every
parameter. Therefore, the MQOs for detectability in the Mid-Appalachian Soil Survey were
completely satisfied. System detection limits were used to assess system-wide detectability.
Using a criterion of 80 percent or more of the routine sample concentrations exceeding the system
detection limits as a basis for assessment, most of the parameters were suitable for all data uses
throughout the concentration range. The six exceptions were exchangeable calcium and sodium,
iron extracted in calcium chloride or acid oxalate, silicon extracted in acid oxalate, and total sulfur.
Data users should use caution when assessing these parameters, as a significant portion of the
routine sample concentrations were less than the corresponding system detection limits and may
be difficult to distinguish in regard to overall system detection.
Precision
The analytical within-batch precision objectives were satisfied for most of the parameters.
Occasionally, an objective was not satisfied for an upper or lower tier of a parameter. When the
two tiers were pooled over the total concentration range for each of the parameter groups, a
precision index showed that all parameter groups satisfied the "overall" precision objectives.
The precision of the preparation duplicates was about the same, on the average, as the
laboratory audit samples and satisfied the MQOs except for iron in acid oxalate. This indicates that
the preparation laboratory performed very well in subsampling the bulk soil samples. For the field
duplicates, only a few parameters exceeded the precision objectives. In addition, the relatively
low precision for the field duplicates for some parameter groups suggests that the component of
error from soil sampling, which includes spatial heterogeneity within the sampled horizons, is a
large portion of the data collection error.
Accuracy
Accuracy in the Mid-Appalachian soil survey was evaluated by estimating analytical bias with
respect to a reference value, defined as the mean of an accuracy window for a given parameter.
Laboratory differences and trends were assessed by comparing mean values for the laboratories,
combined across audit samples, to the pooled reference values.
Analytical bias was negligible when compared with the system detection limit for all
parameters for which system detection limits where established. The only case of significant bias
occurred for the cation exchange capacity in ammonium acetate for the two laboratories which
used the titration procedure. Bias values for the two newly-established parameters for the Mid-
Appalachian survey, i.e., aluminum in ammonium chloride and silicon in acid oxalate, were both
much less than their respective detection limits. The percentage of observations that were outside
the respective accuracy windows and the magnitude of their contribution to the overall bias
estimates were also calculated. The results show a very wide range in the ratios of bias for values
outside the window compared to the total analytical bias. Except for total sulfur, every parameter
had fewer than 20 percent of its values outside the accuracy windows. Although the percentage
of samples outside the window for total sulfur was high, the contribution to overall bias was
xxii
-------
relatively low. Only 22 percent of the parameters had values outside the window that contributed
more than one-third to the bias.
Approximately 54 percent of the parameters showed significant laboratory differences by
Scheffe's pairwise comparison test. Only four parameters (total silt, pH in 0.002M calcium chloride,
cation exchange capacity in ammonium acetate, and silicon in acid oxalate) showed all three
laboratories to be significantly different from each other. There were no general trends for any
specific laboratory for most of the parameter groups. For the cations in calcium chloride, however,
Laboratory 2 was significantly different from the other laboratories in all cases shown and
Laboratory 4 was significantly lower for the sulfate isotherm parameters than the other laboratories
in the four cases that showed differences.
Moving averages of the laboratory audit samples were plotted to identify situations when a
particular laboratory showed an upward or downward trend over time for a given parameter.
Generally, the trends did not show extreme divergence with respect to the accuracy window
acceptance criteria. However, certain data users may find the trends to be noteworthy for a
specific data analysis.
No single laboratory was consistently superior to the others for all parameters or parameter
groups regarding low differences. Each laboratory appears to have individual strengths for specific
analytical methods. This is probably a reflection of the combination of experience, instrumentation,
and laboratory management practices within each laboratory, resulting in a patchwork of
differences on a parameter group basis.
Representativeness
All pedons sampled were within the range of morphological characteristics outlined in their
respective sampling classes. The homogenization and subsampling procedures at the preparation
laboratory produced representative analytical soil samples of known and accepted quality. Overall
precision for the preparation duplicates was approximately equivalent to that of the laboratory audit
samples, hence, the procedures were shown to be suitable for creating representative subsamples.
The analyte concentrations in the field duplicates generally were representative of the range
and frequency distribution of routine sample concentrations. Analyte concentrations in the
preparation duplicates were generally representative of the corresponding field duplicates. The
audit samples, as expected, were usually representative only of the overall range of data from the
routine samples.
Completeness
Sampling of the specified pedons had a completeness level of 100 percent, and processing
was accomplished for 100 percent of the field samples satisfying the sample receipt criteria at the
preparation laboratory. Analytical completeness in the verified data base was 100 percent for all
parameters. Sufficient validated data were generated to make conclusions for each parameter in
the Mid-Appalachian survey data bases, with a completeness level of 98 percent or higher for all
parameters. Data qualifiers, or flags, for completeness were inserted in the data base to indicate
any missing values.
Comparability
The statistical methods were comparable among the three DDRP surveys. Additional audit
samples were used to optimize QE and QC activities in the Mid-Appalachian survey although they
did not directly affect the comparability of the data from the surveys.
In most cases the sampling, preparation and analytical methods and protocols for the three
surveys were comparable, although there were several parameter protocol modifications in the Mid-
xxiii
-------
Appalachian survey. The only modification which may affect comparability was the change of
aluminum extractant from 1M potassium chloride to 1M ammonium chloride.
As a result of more stringent MQOs in the Mid-Appalachian survey, overall measurement
quality generally was better in the Mid-Appalachian data bases. Therefore, it is possible that some
of the Northeastern region and Southern Blue Ridge Province data may not be suitable for the same
data analysis procedures performed on the Mid-Appalachian data without some form of caveat.
Two examples are the organic soil sulfate isotherm data and the specific surface data from the
Northeastern and Southern Blue Ridge Province surveys which are generally considered to be
suspect by data users.
Significant differences among the surveys were identified using the median-range, mineral soil
audit sample common to the three surveys. Only nine parameters showed differences among
surveys at the .01 significance level. Although 82 percent of the parameters common to all three
surveys were not significantly different at this level across surveys, data users should exercise
caution when using data from the nine parameters exhibiting significant differences.
As part of the DDRP, an interlaboratory methods comparison study was conducted which
compared soil analysis data from two laboratories using the DDRP methods to 16 statistically
randomly-chosen independent laboratories using their own methods which were similar, but not
identical to, the DDRP methods. The results of the study show the DDRP data to be comparable
with data from the independent laboratories.
Uncertainty Estimates
Within-batch imprecision estimates increased, as expected, from analytical (delta, values) to
sample preparation (delta2 values) to field sampling (delta3 values) to sampling class/horizon
groups (delta4 values). The between-batch precision estimates were generally low.
The overall measurement uncertainty in the routine samples was based on the delta3 values
which were estimated from the field duplicate samples. The delta3 values, in relation to the
associated de!ta4 values, were used to provide the data users with a basis for assessing the
contribution of the measurement system to the data uncertainty. Using this procedure,
measurement uncertainty was negligible for 90 percent of the parameters measured) of the data.
Conclusions and Recommendations
As a result of previous conclusions and recommendations from the DDRP Northeastern and
Southern Blue Ridge Province surveys, a number of improvements were made in the quality
assurance program for the Mid-Appalachian survey. The quality assurance data are presented in
a manner considered to be the most appropriate for use by the primary data users. The data
quality evaluation procedures and the report format resulted from regular interactions with the data
users, and the assessment by several external reviewers. Each user has a subjective concept of
data quality as well as a knowledge of the specific level of data quality required for his/her own
use. The users are therefore encouraged to become familiar with the text, figures, and tables to
facilitate the identification of data satisfying their specific requirements.
The consolidation of sample preparation facilities at a single laboratory facilitated quality
assurance of the samples from the field sampling through the sample analysis phase. The
separation into different batches of mineral and organic samples should be continued. In addition,
measurement quality samples should continue to be distributed among batches and analytical
laboratories in such a way as to provide a balanced design for assessment purposes.
The use of a computerized data entry and verification system which allowed the calculation
of final data values and produced a list of flags and data entry errors for each sample batch
xxiv
-------
greatly facilitated the verification and reanalysis decision-making process. A similar program should
be tailored for each future survey, with the addition of a method that will identify and confirm
outlying data points for real-time control purposes.
Considerable effort was expended throughout the three surveys to improve detectability of
various parameters and significant improvement was made for the exchangeable cations and sulfur
parameters. Additional methods research is needed to improve detectability further. In addition,
both instrument and system-wide detectability should be defined in the MQOs. As part of the
implementation of such MQOs, it is recommended that low concentration audit samples, entered
into the system during the sampling phase, continue to be utilized as substitutes for soil blank
samples.
The precision results indicate that the analytical within-batch precision objectives were
satisfied in most cases. Occasionally an objective was not satisfied for an upper or tower tier of
a parameter. It is recommended that the lower and upper tier precision objectives for exchangeable
aluminum in ammonium chloride and extractable silicon in acid oxalate be modified as described
in the report. The preparation and field duplicate samples generally satisfied the precision
objectives.
Audit sample accuracy windows were developed for use in the Mid-Appalachian survey from
data collected in the previous two surveys. The use of a quality control audit sample should be
continued in future surveys. In addition liquid audit samples should be incorporated into the quality
assurance program and be used to differentiate soil extraction error from instrument error.
Emphasis should be placed on the use of audit sample control charts by the QA staff to identify
abnormal scatter outside the accuracy windows during the batch analysis.
No single laboratory was consistently superior to the others for all parameters or parameter
groups. To control interlaboratory differences in future surveys, it is recommended to continue the
selection of a specific laboratory to perform analysis on a parameter basis for those parameters
or parameter groups that reveal inherently high differences of where specialized instrumentation is
used, e.g., total elemental analysis. Also, a stringent performance evaluation process should be
continued to select quality contract analytical laboratories.
Data comparability across the three surveys was generally good. It is recommended,
however, that a methods comparison be performed for the two soil extraction methods used for
exchangeable aluminum, i.e., 1M ammonium chloride and 1M potassium chloride.
The step function statistical approach has been shown to be an effective procedure for
evaluating measurement quality issues in environmental data spanning a wide concentration range.
It is recommended that additional research and development be undertaken to identify an optimal
step function procedure that is fully compatible with the measurement quality sample design.
xxv
-------
-------
Section 1
Introduction
This report serves as documentation of
quality assurance (QA) activities that were
undertaken in the physical and chemical analy-
sis of soils collected during a major soil char-
acterization effort conducted in the Mid-Appa-
lachian region of the United States. Section 1
provides overviews of the project and the
survey work that was performed. The primary
participants are outlined and the analytical
parameters measured in the survey are de-
scribed.
Project Overview
The effects of acidic deposition on
freshwater ecosystems have been the subject
of intensive research during the 1980s. The
U.S. Environmental Protection Agency (EPA)
recently completed the National Surface Water
Survey (NSWS) of the lakes and streams of
the United States considered to be most
vulnerable to acidic deposition, i.e., those with
the lowest acid neutralizing capacity (ANC), as
described in Linthurst et al. (1986) and Kauf-
mann et al. (1988). Analysis of the water
chemistry data from the NSWS and other
sources in conjunction with temporal and
spatial atmospheric deposition data indicates
that long-term deposition of sulfur-containing
compounds originating from the combustion of
fossil fuels has decreased the ANC of some
surface waters in eastern North America
(Altshuller and Linthurst, 1984; NAS, 1986). The
movement of sulfate and other mobile anions
through watershed soils and associated cation
leaching are the most widely accepted mecha-
nisms of this acidification process (Galloway
et al., 1983; Driscoll and Newton, 1985).
Given that acidification of some surface
waters has occurred, critical scientific and
policy questions focus on whether acidification
is continuing in the regions of concern, whe-
ther it is just beginning in other regions, how
extensive the effects might become, and over
what time scales effects might occur. The
EPA is addressing these questions through the
activities of the Direct/Delayed Response
Project (DDRP). The DDRP is being conducted
as part of the EPA Aquatic Effects Research
Program (AERP) under the congressionally
mandated National Acid Precipitation Assess-
ment Program (NAPAP) as described in Church
et al. (1989). The overall purpose of the DDRP
is to characterize geographic regions of the
United States by predicting the long-term
response of watersheds and surface waters to
acidic deposition. The DDRP draws its name
from the consideration of whether acidification
might be immediate, i.e., "direct", or would lag
in time, i.e., "delayed", because of soil buffering
characteristics.
Recent trend analyses have indicated
that the rate of sulfur deposition is slowly
declining in the northeastern United States
but is increasing in the southeastern United
States. If a "direct" response exists between
sulfur deposition and surface water alkalinity,
the extent of current effects on surface water
probably would not change much at current
levels of deposition, and conditions would
improve as the levels of deposition decline. If
surface water chemistry changes in a "delayed"
manner, e.g., due to chemical changes in the
watershed, future changes in water chemistry
(even with current or declining rates of deposi-
tion) become difficult to predict. This range
of potential effects has clear and significant
implications for public policy decisions on
sulfur emissions control strategies.
The DDRP was not established to identi-
fy the exact mechanisms and processes of
-------
surface water acidification; rather, the principal
mandate was to make regional projections of
future effects of sulfur deposition on long-term
surface water chemistry based on the best
available data and most widely accepted
hypotheses of the acidification process.
Detailed investigations into processes of soil
and surface water acidification are being
conducted as part of other projects within the
AERP and NAPAP (USEPA, 1989).
Although surface waters can be affected
by acidic deposition originating from emissions
many miles distant, the concept of the water-
shed as a unit is critical in understanding
current and future aquatic effects. Indeed, for
drainage lake and reservoir systems in the
Northeast, Upper Midwest, and Southern Blue
Ridge regions, most ANC production occurs as
a result of biogeochemical processes within
the surrounding watershed (Shaffer et al.,
1988; Shaffer and Church, 1989). Initially, the
use of existing regional soils data was consi-
dered for the DORP analyses. Existing soils
data bases, however, were limited with respect
to their application to surface water acidifica-
tion issues. First, such data are available
primarily from lowland agricultural regions,
whereas surface water acidification occurs
principally in relatively undisturbed upland
systems. Second, such data bases generally
do not include a number of key variables
relevant to soil chemical interactions with
acidic deposition. Therefore, three major soil
surveys were specified to aid in the creation of
new regional soils data bases for the DDRP
(Lee et al., 1989). The surveys were designed
to allow specific soil types to be linked with
the existing NSWS data bases that describe
the chemistry of low ANC lakes and streams.
The three DDRP surveys shown in Figure
1-1 focused on regions of the United States
within which sensitive surface water systems
were known to exist. Nonsensitive systems
were also known to exist in these regions.
The Northeastern (NE) Soil Survey was con-
ducted in 1985 in the states of Maine, New
Hampshire, Vermont, Massachusetts, Connec-
ticut, Rhode Island, New York, and portions of
Pennsylvania. The Southern Blue Ridge Pro-
vince (SBRP) Soil Survey was conducted in
1986 in portions of Virginia, Tennessee, North
Carolina, South Carolina, and Georgia. This
report focuses on the Mid-Appalachian Soil
Survey (MASS) which was conducted in 1988 in
portions of Pennsylvania, Virginia, and West
Virginia. Surface water acidification in these
regions was studied during the Eastern Lakes
Survey in 1984 and in the National Stream
Survey, Phase I, in 1985.
Northeast
Region
Mid-Appalachian
Region
Southern Blue Ridge
Province
Figure 1-1. Regional Surveys of the Direct/Delayed
Response Project (from Lee et al., 1989)
The four specific goals of the DDRP are
to: (1) characterize the variability of soil and
watershed attributes across these regions, (2)
determine which soil and watershed charac-
teristics are most strongly related to surface
water chemistry, (3) estimate the relative
importance of key watershed processes in
controlling surface water chemistry across the
regions of concern, and (4) classify the sample
of watersheds with regard to their response
characteristics and extrapolate the results
from the sample of watersheds to the regions
of concern (Church et al., 1989). A wide variety
of data sources and methods of analysis are
used to address the DDRP objectives. In
addition to the data collected during the sur-
veys, other sources of information include the
following data bases:
National Surface Water Survey (NSWS)
[water chemistry data],
Acid Deposition Data Network (ADDNET),
including GEOECOLOGY [atmospheric
precipitation chemistry data],
Soil Conservation Service (SCS) Soils-5
[soil physical and chemical data],
-------
Topographic and Acid Deposition System
(ADS) [total sulfur deposition data], and
U.S. Geological Survey [runoff data].
Data from EPA long-term monitoring sites,
episodic event monitoring sites, and intensively
studied watersheds are used in the data
analysis. Uncertainty and error propagation
estimates are an important part of all levels of
analysis (USEPA, 1989).
The DDRP watersheds were selected as
a subset of lake and stream systems surveyed
in the NSWS (Church, 1989) and were charac-
terized as probability samples to ensure that
results could be extrapolated to a specified
target population. A series of maps of soils,
vegetation, land use, and depth to bedrock
were prepared for each DDRP watershed by
the United States Department of Agriculture
(USDA) Soil Conservation Service (SCS). Soil
sampling classes were defined for each DDRP
region, and soils selected from these classes
were sampled and analyzed for physical and
chemical characteristics. Soils were aggre-
gated within sampling classes to develop
class means and variances used in represent-
ing characteristics, e.g., by mass or area
weighting, of the watersheds of interest.
A variety of complementary data analy-
ses are being performed as part of the DDRP,
including the statistical evaluation of inter-
relationships among atmospheric deposition,
mapped watershed characteristics, soil chemi-
stry, and current surface water chemistry. The
principal goal of these analyses is to verify
that the processes and relationships incor-
porated in the subsequent modeling analyses
reasonably represent the systems under study.
Watershed retention of atmospherically deposi-
ted sulfur is an important consideration, as are
the dynamics of sulfur retention via sulfate
adsorption by the soils. Also considered are
"single-factor" models (Bloom and Grigal, 1985;
Reuss and Johnson, 1986) of the influence of
acidic deposition on the supply and transport
of base cations from soils to surface waters.
The purpose of this modeling is to evaluate
the potential relative importance of cation
exchange as a mediation process for surface
water acidification.
Watershed models are used in the DDRP
to project future integrated effects of atmos-
pheric sulfur deposition on surface water
chemistry. Three models specifically developed
to investigate the effects of acidic deposition
on watersheds and surface waters are being
applied: The Model of Acidification of Ground-
water in Catchments (MAGIC) as described by
Cosby et al. (1986); the Enhanced Trickle Down
(ETD) model as described by Nikolaidis et al.
(1988) and Schnoor et al. (1986); and the
Integrated Lake-Watershed Acidification Study
(ILWAS) model as described by Chen et al.
(1983) and Gherini et al. (1985). Projections of
changes in annual average surface water
chemistry are being made for each region for
at least 50 years using two different scenarios
of atmospheric sulfur deposition (Church et al.,
1989).
Mid-Appalachian Soil Survey
The DDRP technical director had overall
responsibility for the MASS. There were five
sampling crews, each consisting of three to
four soil scientists, that were assigned the
task of sampling selected soil pedons. A
pedon is a three-dimensional body of soil
having lateral dimensions large enough, i.e., 1
to 10 square meters in area, to permit the
identification and sampling of its soil horizons,
or horizontal soil layers. In each of the 150
soil pedons selected for sampling, all soil
horizons were described on a coded field data
form. A total of 844 soil horizons from these
pedons were sampled in accordance with
documented protocols (Appendix A in Kern and
Lee, in press). In addition to collecting 5-
kilogram routine soil samples, each sampling
crew collected one duplicate sample for QA
purposes from every third pedon sampled.
Details of the soil mapping and sampling
activities are contained in a separate QA
report (Kern and Lee, in press).
As part of the MASS, a preparation
laboratory was established at Las Vegas,
Nevada, to facilitate processing of the field-
moist bulk soil samples collected by the samp-
ling crews and to perform preliminary analyses
on these samples. The samples were pre-
pared in accordance with standard operating
procedures (Appendix A in Papp and Van
Remortel, in press). The samples were air-
dried, disaggregated, and sieved at the labora-
tory. Field-moist pH, organic matter by loss
on ignition, air-dry moisture, and rock frag-
ments in the soil samples were determined,
and the samples were subsequently homoge-
nized and subsampled to produce analytical
-------
samples. In addition, the bulk density of repli-
cate soil clods and known volume samples
collected by the sampling crews was mea-
sured. The analytical samples were grouped
into batches of approximately 40 samples per
batch. Measurement quality samples, includ-
ing field duplicates, natural audit samples, and
preparation duplicates, were placed in each
batch for QA purposes. The samples were
added to each batch in a random manner and
the batches were subsequently distributed
among three analytical laboratories contracted
by EPA. A detailed account of the preparation
laboratory operations is provided in a separate
QA report (Papp and Van Remortel, in press).
Soil physical and chemical analyses were
conducted according to documented protocols
(Blume et al., in press) by three contracted
laboratories. These protocols ensured that the
soil sample analyses would yield analytical
data of known and acceptable quality. The
sampling crews and laboratory personnel
received training in the protocols for their
respective activities, and on-site technical
systems audits were conducted of all partici-
pants. Frequent communication between QA
personnel and laboratory personnel was estab-
lished to identify, discuss, and resolve issues
raised during the survey.
The QA protocols specified for the MASS
were used to maintain consistency in the field
sampling, sample preparation, and sample
analysis. The design of the analytical QA
program has an integral effect on the resulting
data quality derived from the physical and
chemical analyses of the soil samples. Know-
ledge of the data quality enables potential
users to determine if the data satisfy their
specific requirements for the data analysis
models or other assessments of the physical
and chemical data. In addition, the QA pro-
gram was designed to ensure that the DDRP
data are comparable within and across the
regions of concern.
Project Staff Responsibilities
A summary of the different DDRP staff
responsibilities at each cooperating EPA labo-
ratory is provided below.
Environmental Research Laboratory,
Corvallis, Oregon (ERL-C)
The DDRP staff at ERL-C is active in all
aspects of the DDRP soil surveys. Respon-
sibilities of the personnel for the MASS include:
developing the experimental design and QA
oversight for soil mapping and sampling,
analyzing, validating, and interpreting data,
preparing progress reports (periodic and
final reports with contributions from other
laboratories relative to their responsibili-
ties),
assessing and resolving all science-
related issues (jointly with other laborato-
ries),
preparing QA data reports for soil map-
ping, sampling operations, and data analy-
sis, and
coordinating survey activities with NAPAP
management staff.
Environmental Monitoring Systems
Laboratory, Las Vegas, Nevada (EMSL-LV)
The DDRP staff at EMSL-LV has exper-
tise in matters relating to QA implementation,
logistics, and analytical services. The respon-
sibilities of the personnel at this laboratory in-
clude:
developing and implementing the QA
project plan, methodology, and standard
operating procedures for soil sample
preparation and analysis,
coordinating logistical support and equip-
ment disbursement for field and prepara-
tion laboratory operations,
developing and implementing the data
verification and management procedures
for soil sampling, preparation, and analy-
sis (jointly with other laboratories),
analyzing and interpreting data (jointly with
ERL-C),
preparing QA data reports for preparation
laboratory and analytical laboratory opera-
tions, and
resolving issues pertaining to laboratory
QA, logistics, and analytical services.
Primary Cooperators
The use of federal interagency agree-
ments by EPA allowed the DDRP to engage
the support of several organizations that have
expertise in areas of importance to the project,
as outlined below.
-------
Oak Ridge National Laboratory, Oak Ridge,
Tennessee (ORNL)
The DDRP staff at ORNL has expertise in
managing, manipulating, and restructuring
large data bases to satisfy data analysis
needs. The ERL-C staff oversees the activities
of ORNL for the following responsibilities:
developing and maintaining a data base
management system (jointly with other
laboratories),
preparing computer-generated summary
tables, statistics, and graphics for reports,
and
assisting with data validation and inter-
pretation activities.
Soil Conservation Service (SCS)
The SCS has expertise in soil mapping
and characterization on a national scale.
Under the guidance of the DDRP staff at ERL-
C, the SCS has the responsibility for:
soil mapping of the MASS watersheds,
field site selection and characterization,
and
soil sampling of selected pedons.
Organization of the Report
This document has been organized into
four main sections. Section 1 provides an
overview of the MASS objectives and analytical
parameters. Section 2 describes the imple-
mentation of the overall QA program, its rela-
tion to data quality assessment, and the use
of measurement quality samples during the
various stages of data collection and verifica-
tion. Section 3 provides the data quality
results and discussion for the analytical para-
meters measured in the survey. Data quality
achievements in relation to objectives estab-
lished at the beginning of the survey are
evaluated. Section 4 addresses the conclu-
sions and recommendations that have been
generated from these findings, particularly
in regard to issues of concern for sample
analysis, improvement in QA design, and
suggestions for QA efforts in future surveys.
Throughout the report, data quality is de-
scribed in terms of detectability, precision,
accuracy, representativeness, completeness,
and comparability.
Description of Parameter Groups
The DDRP QA staff organized the 50
analytical parameters into nine groups and
subsequently evaluated each group indepen-
dently according to the objectives specified for
the survey. The nine parameter groups are
briefly summarized below:
(1) Moisture and Particle Size Analysis
(eleven parameters) - The air-dry soil
moisture content is determined in order
to define all subsequent aliquots on an
oven-dry weight basis. Particle size
analysis is performed on the less than 2-
millimeter size fractions of mineral soils
for characterization and classification
purposes. (Note: the 2-20 mm fraction
rock fragments were determined at the
preparation laboratory).
(2) Soil pH (three parameters) -- The pH is a
measure of free hydrogen ion activity ex-
pressed as a negative logarithm. The pH
measurements are determined in three
soil suspensions: deionized water and
two different strengths of calcium chlo-
ride solution. The deionized water sus-
pension is universally used in soil labo-
ratories, and the calcium chloride sus-
pension is widely used to address the
effects of naturally-occuring salts on pH.
(3) Exchangeable Cations in Ammonium
Chloride (five parameters) -- The exchan-
geable cations (calcium, magnesium,
potassium, sodium, and aluminum) are
extracted with an unbuffered ammonium
chloride solution during the cation ex-
change capacity (CEC) determinations.
These cations, with the exception of
aluminum, represent the soil base cat-
ions and can be used in the calculation
of percent base saturation of the soil
and in defining selectivity coefficients
and cation pools for the DDRP models.
(4) Exchangeable Cations in Ammonium
Acetate (four parameters) -- The exchan-
geable cations (calcium, magnesium,
potassium, and sodium) are extracted
with a pH 7.0 buffered ammonium ace-
tate solution during the CEC determi-
nations. These cations can be used in
the calculation of percent base satura-
tion of the soil and in defining selectivity
-------
coefficients and cation pools for the
DDRP models.
(5) Cation Exchange Capacity and Exchange-
able Acidity (three parameters) - The
CEC indicates the ability of a soil to
adsorb cations, especially the exchange-
able base cations mentioned above. The
CEC is highly correlated with the buffer-
ing capacity of the soil. Two saturating
solutions for the exchangeable cation
component of CEC are used: a pH 7.0
buffered ammonium acetate solution for
the cation component of "total" CEC, and
an unbuffered ammonium chloride solu-
tion as the cation component of "effec-
tive" CEC at the natural pH of the soil.
Exchangeable acidity measures the ex-
changeable hydrogen ions that are held
in the inorganic and organic complexes
on the soil particle surfaces, including
the hydrogen ions contributed from the
aluminum complexes as well as the
active acidity in solution. A buffered
barium chloride triethanolamine extrac-
tion indicates the "total" exchangeable
acidity.
(6) Extractable Cations in Calcium Chloride
(six parameters) - Lime potential [pH -
1/2 pCa] is used as an input for certain
predictive models. Aluminum potential
[3pH - pAI] is another important charac-
teristic for watershed modeling. The soil
is extracted with a calcium chloride
solution and analyzed for calcium and
aluminum concentrations. The magnesi-
um, potassium, sodium, and iron concen-
trations also are determined and are
compared to cation concentrations in
other extracts.
(7) Extractable Iron, Aluminum, and Silicon
(seven parameters) - Sulfate adsorption
is highly correlated with the presence of
iron and aluminum. Each of three ex-
tracts estimate a specific iron or alu-
minum fraction: sodium pyrophosphate
estimates organic iron and aluminum;
acid oxalate estimates organic iron and
aluminum plus sesquioxides; and citrate-
dithionite estimates nonsilicate iron and
aluminum. Silicon is also determined in
the acid oxalate extraction.
(8) Extractable Sulfate and Sulfate Adsorp-
tion Isotherms (eight parameters)
Extractable sulfate is determined in two
different extracts: deionized water,
which estimates loosely-bound sulfate;
and 500 mg P/L as sodium phosphate,
which estimates inorganic sulfate held
by electrostatic forces or by ligand ex-
change on the anion exchange sites.
The ability of soil to adsorb sulfate is
related to anion adsorption capacity.
Isotherms are developed by placing soil
samples in six magnesium sulfate solu-
tions of different concentrations: 0, 2,
4, 8, 16, and 32 mg S/L A determination
is made of the amount of sulfate remain-
ing in solution after one-hour contact
with the soil, and subtraction yields the
net sulfate sorption. The isotherms
describe sulfate partitioning between
solid and solution phases under labora-
tory conditions and are used to predict
changes in sorted and dissolved sulfate
in the soil as a result of altered sulfur
inputs.
(9) Total Carbon, Nitrogen, and Sulfur (three
parameters) - Total carbon and nitrogen
are closely related to the type and amou-
nt of soil organic matter. Total sulfur
measures both organic and inorganic
sulfur (mostly adsorbed sulfate) in the
soil. Ratios of total sulfur are used to
characterize sulfur retention in soils and
provide a benchmark to monitor future
inputs of anthropogenic sulfur.
Description of Parameters
Throughout this document, physical and
chemical soil parameters are referenced either
by a data-variable or descriptive parameter
name. A list of parameters and their corres-
ponding descriptions is given in Table 1-1. The
order of the parameters is consistent with
their order of presentation in this report.
-------
Table 1-1. Analytical Parameters Measured In the Mid-Appalachian Soil Survey
Parameter
Description of Parameter
MOIST Air-dry soil moisture measured at the analytical laboratory and expressed as a percentage on an oven-
dry weight basis; mineral soils are dried at 105°C, organic soils at 60°C.
SAND Total sand is the portion of the sample with particle diameter between 0.05 mm and 2.0 mm, and is
calculated as the summation of percentages for individual sand fractions: VCOS + COS + MS + FS + VFS.
VCOS Very coarse sand is the sand fraction between 1.0 mm and 2.0 mm, and is determined by sieving the sand
which has been separated from the silt and clay.
COS Coarse sand is the sand fraction between 0.5 mm and 1.0 mm, and is determined by sieving the sand which
has been separated from the silt and clay.
MS Medium sand is the sand fraction between 0.25 mm and 0.50 mm, and is determined by sieving the sand
which has been separated from the silt and clay.
FS Fine sand is the sand fraction between 0.10 mm and 0.25 mm, and is determined by sieving the sand which
has been separated from the silt and clay.
VFS Very fine sand is the sand fraction between 0.05 mm and 0.10 mm, and is determined by sieving the sand
which has been separated from the silt and clay.
SILT Total silt is the portion of the sample with particle diameter between 0.002 mm and 0.05 mm, and is
calculated by subtracting from 100 percent the sum of the total sand and clay.
COSI Coarse silt is the silt fraction between 0.02 mm and 0.05 mm, and is calculated by subtracting the fine silt
fraction from the total silt.
FSI Fine silt is the silt fraction between 0.002 mm and 0.02 mm; it is determined by pipetting and is calculated
by subtracting the clay fraction from the less than 0.02-mm fraction.
CLAY Total clay is the portion of the sample with particle diameter of less than 0.002 mm; it is determined by
pipetting.
PHJH20 pH determined in a deionized water extract using a 1:1 mineral soil to solution ratio or 1:5 organic soil to
solution ratio, measured with a pH meter and combination electrode.
PH_002M pH determined in a 0.002M calcium chloride extract using a 1:2 mineral soil to solution ratio or 1:10 organic
soil to solution ratio, measured with a pH meter and combination electrode.
PH_01M pH determined in a 0.01M calcium chloride extract using a 1:1 mineral soil to solution ratio or 1:5 organic
soil to solution ratio, measured with a pH meter and combination electrode.
CA_CL Exchangeable calcium determined with an unbuffered 1M ammonium chloride solution; approximately 1:13
mineral soil to solution ratio or 1:52 organic soil to solution ratio are used; atomic absorption spectrometry
or inductively coupled plasma atomic emission spectrometry is specified.
MG_CL Exchangeable magnesium determined with an unbuffered 1M ammonium chloride solution; approximately
1:13 mineral soil to solution ratio or 1:52 organic soil to solution ratio are used; atomic absorption
spectrometry or inductively coupled plasma atomic emission spectrometry is specified.
K_CL Exchangeable potassium determined with an unbuffered 1M ammonium chloride solution; approximately
1:13 mineral soil to solution ratio or 1:52 organic soil to solution ratio are used; atomic absorption
spectrometry or flame emission spectrometry is specified.
NA_CL Exchangeable sodium determined with an unbuffered 1M ammonium chloride solution; approximately 1:13
mineral soil to solution ratio or 1:52 organic soil to solution ratio are used; atomic absorption spectrometry,
inductively coupled plasma atomic emission spectrometry, or flame emission spectrometry is specified.
AL_CL Exchangeable aluminum determined with an unbuffered 1M ammonium chloride solution; approximately
1:13 mineral soil to solution ratio or 1:52 organic soil to solution ratio are used; inductively coupled plasma
atomic emission spectrometry is specified.
(Continued)
-------
Table 1-1. Continued
Parameter
Description of Parameter
CA_OAC Exchangeable calcium determined with 1M ammonium acetate solution buffered at pH 7.0; approximate-
ly 1:13 mineral soil to solution ratio or 1:52 organic soil to solution ratio are used; atomic absorption
spectrometry or inductively coupled plasma atomic emission spectrometry is specified.
MG_OAC Exchangeable magnesium determined with 1M ammonium acetate solution buffered at pH 7.0;
approximately 1:13 mineral soil to solution ratio or 1:52 organic soil to solution ratio are used; atomic
absorption spectrometry or inductively coupled plasma atomic emission spectrometry is specified.
K_OAC Exchangeable potassium determined with 1M ammonium acetate solution buffered at pH 7.0; approximately
1:13 mineral soil to solution ratio or 1:52 organic soil to solution ratio are used; atomic absorption
spectrometry or flame emission spectrometry is specified.
NA_OAC Exchangeable sodium determined with 1M ammonium acetate solution buffered at pH 7.0; approximately
1:13 mineral soil to solution ratio or 1:52 organic soil to solution ratio are used; atomic absorption
spectrometry, inductively coupled plasma atomic emission spectrometry, or flame emission spectrometry
is specified.
CEC_CL Cation exchange capacity determined with an unbuffered 1M ammonium chloride solution is the effective
CEC which occurs at approximately the field pH when combined with the acidity component; approximately
1:13 mineral soil to solution ratio or 1:52 organic soil to solution ratio are used; samples are analyzed for
ammonium content by one of three methods: automated distillation/titration; manual distillation/automated
titration; or ammonium displacement/flow injection analysis.
CEC_OAC Cation exchange capacity determined with 1M ammonium acetate solution buffered at pH 7.0 is the
theoretical estimate of the maximum potential CEC for a specific soil when combined with the acidity
component; approximately 1:13 mineral soil to solution ratio or 1:52 organic soil to solution ratio are used;
samples are analyzed for ammonium content by one of three methods: automated distillation/titration;
manual distillation/automated titration; or ammonium displacement/flow injection analysis.
AC_BACL Total exchangeable acidity determined by titration in a buffered (pH 8.2) barium chloride triethanolamine
extraction using an approximately 1:30 soil to solution ratio.
CA_CL2 Extractable calcium determined by a 0.002M calcium chloride extraction; a 1:2 mineral soil to solution
ratio or 1:10 organic soil to solution ratio are used; the calcium is used to calculate lime potential; atomic
absorption spectrometry or inductively coupled plasma atomic emission spectrometry is specified.
MG_CL2 Extractable magnesium determined by a 0.002M calcium chloride extraction; a 1:2 mineral soil to solution
ratio or 1:10 organic soil to solution ratio are used; atomic absorption spectrometry or inductively coupled
plasma atomic emission spectrometry is specified.
K_CL2 Extractable potassium determined by a 0.002M calcium chloride extraction; a 1:2 mineral soil to solution
ratio or 1:10 organic soil to solution ratio are used; atomic absorption spectrometry or flame emission
spectrometry is specified.
NA_CL2 Extractable sodium determined by a 0.002M calcium chloride extraction; a 1:2 mineral soil to solution ratio
or 1:10 organic soil to solution ratio are used; atomic absorption spectrometry, inductively coupled plasma
atomic emission spectrometry, or flame emission spectrometry is specified.
FE_CL2 Extractable iron determined by a 0.002M calcium chloride extraction; a 1:2 mineral soil to solution ratio
or 1:10 organic soil to solution ratio are used; inductively coupled plasma atomic emission spectrometry
is specified.
AL_CL2 Extractable aluminum determined by a 0.002M calcium chloride extraction; a 1:2 mineral soil to solution
ratio or 1:10 organic soil to solution ratio are used; the aluminum concentration obtained from this procedure
is used to calculate aluminum potential; inductively coupled plasma atomic emission spectrometry is
specified.
FE_PYP Extractable iron determined by a 0.1M sodium pyrophosphate extraction using a 1:100 soil to solution
ratio; the pyrophosphate extract estimates organically-bound iron; inductively coupled plasma atomic
emission spectrometry is specified.
(Continued)
-------
Table 1-1. Continued
Parameter Description of Parameter
AL_PYP Extractable aluminum determined by a 0.1M sodium pyrophosphate extraction using a 1:100 soil to solution
ratio; the pyrophosphate extract estimates organically-bound aluminum; inductively coupled plasma atomic
emission spectrometry is specified.
FE_AO Extractable iron determined by an ammonium oxalate-oxalic acid extraction using a 1:100 soil to solution
ratio; the acid oxalate extract estimates organic and amorphous iron oxides; inductively coupled plasma
atomic emission spectrometry is specified.
AL_AO Extractabie aluminum determined by an ammonium oxalate-oxalic acid extraction using a 1:100 soil to
solution ratio; the acid oxalate extract estimates organic and amorphous aluminum oxides; inductively
coupled plasma atomic emission spectrometry is specified.
SI_AO Extractable silicon determined by an ammonium oxalate-oxalic acid extraction using a 1:100 soil to solution
ratio; inductively coupled plasma atomic emission spectrometry is specified.
FE__CD Extractable iron determined by a sodium citrate-sodium dithionite extraction using a 1:30 soil to solution
ratio; the citrate dithionite extract estimates non-silicate iron; inductively coupled plasma atomic emission
spectrometry is specified.
AL_CD Extractable aluminum determined by a sodium citrate-sodium dithionite extraction using a 1:30 soil to
solution ratio; the citrate dithionite extract estimates non-silicate aluminum; inductively coupled plasma
atomic emission spectrometry is specified.
SO4_H20 Extractable sulfate determined with a double deionized water extract; this extraction approximates the
sulfate which will readily enter the soil solution and uses a 1:20 mineral soil to solution ratio or 1:40
organic soil to solution ratio; ion chromatography is specified.
SO4_PO4 Extractable sulfate determined with a 0.016M sodium phosphate (500 mg P/L) extract; this extraction
approximates the total amount of adsorbed sulfate and uses a 1:20 mineral soil to solution ratio or 1:40
organic soil to solution ratio; ion chromatography is specified.
SO4_0 Sulfate remaining in a 0 mg S/L solution following equilibration with a 1:5 mineral soil to solution ratio
or 1:20 organic soil to solution ratio; the data are used to develop sulfate isotherms; ion chromatography
is specified.
SO4_2 Sulfate remaining in a 2 mg S/L solution following equilibration with a 1:5 mineral soil to solution ratio
or 1:20 organic soil to solution ratio; the data are used to develop sulfate isotherms; ion chromatography
is specified.
SO4_4 Suifate remaining in a 4 mg S/L solution following equilibration with a 1:5 mineral soil to solution ratio
or 1:20 organic soil to solution ratio; the data are used to develop sulfate isotherms; ion chromatography
is specified.
SO4_8 Suifate remaining in a 8 mg S/L solution following equilibration with a 1:5 mineral soil to solution ratio
or 1:20 organic soil to solution ratio; the data are used to develop sulfate isotherms; ion chromatography
is specified.
S04_16 Sulfate remaining in a 16 mg S/L solution following equilibration with a 1:5 mineral soil to solution ratio
or 1:20 organic soil to solution ratio; the data are used to develop sulfate isotherms; ion chromatography
is specified.
S04_32 Sulfate remaining in a 32 mg S/L solution following equilibration with a 1:5 mineral soil to solution ratio
or 1:20 organic soil to solution ratio; the data are used to develop sulfate isotherms; ion chromatography
is specified.
C_TOT Total carbon determined by rapid oxidation followed by infrared detection or thermal conductivity detection
using an automated CHN analyzer; total carbon can be used to characterize the soil organic matter.
N_TOT Total nitrogen determined by rapid oxidation followed by infrared detection or thermal conductivity detection
using an automated CHN analyzer; total nitrogen can be used to characterize the soil organic matter.
S_TOT Total sulfur determined by automated sample combustion followed by infrared detection of evolved sulfur
dioxide.
-------
Section 2
Quality Assurance Program
Quality assurance may be defined as "a
system of activities whose purpose is to
provide to the producer or user of a product or
service the assurance that it meets defined
standards of quality with a stated level of
confidence" (Taylor, 1987). The QA program for
the MASS was designed to ensure that the
data collected were of the highest integrity and
that the data quality could be evaluated and
documented. These procedures included the
preparation of written protocols and manuals
describing: (1) soil sampling, preparation, and
analysis; (2) application of QA procedures
during field and laboratory activities; and (3)
verification of the descriptive and analytical
data (see Figure 2-1). Specific aspects of the
QA program are described in the following
subsections.
Quality Assurance Optimization
Due to overlapping verification activities
during the DDRP NE and SBRP surveys, there
was insufficient time to fully evaluate the NE
data in order to optimize the latter survey to
produce data of higher quality. However, there
was sufficient time available to evaluate the
NE and SBRP data before the MASS was
implemented. The two primary objectives
during the evaluation phase were to identify
procedures that could improve data quality and
to develop the QA staff operations in such a
way as to expedite the survey. An improve-
ment in data quality was necessary because
the data users had determined that the data
for certain parameters in the NE and SBRP
surveys contained more uncertainty than was
desirable for their intended use. In order for
the MASS data to be available for the 1990
NAPAP assessment, the EMSL-LV operations
had to be completed in a significantly shorter
timeframe than in the previous surveys.
The subsections below are intended to
highlight the major features of optimization
applied to the MASS sample preparation and
analysis. The discussion is intended only as
a summary of the features because additional
details are provided in other parts of this
section or in supporting documents as refer-
enced.
Consolidated Preparation Laboratory
Facility
The NE and SBRP surveys utilized multi-
ple laboratory facilities to perform the neces-
sary sample preparation tasks. These facili-
ties were also used as storage and distribu-
tion centers for consumable field sampling
supplies. The following difficulties were en-
countered with multiple preparation facilities:
the progress of the preparation labora-
tories was difficult to track, especially
concerning sample batch shipment to the
analytical laboratories;
batches were not always sent within a
regular timeframe, leading to difficulties in
tracking the progress of batch analysis;
control of possible contamination sources
at the different facilities was inconsistent;
the distribution of samples among sam-
pling crews and laboratories was irregular
and unbalanced;
significant time and expense was required
to perform technical systems audits at
multiple facilities;
real-time control of data quality at the
laboratories was inconsistent;
variable air-dry moisture contents in soils
occurred due to different geographic
locations of laboratories;
subtle differences in preparation techni-
ques and interpretation of protocols oc-
curred; and
10
-------
tSWS DAIADASC
AttHSHtO SUBSCT j
MASS
DDRP SOIL SURVEY
FLOWCHART
^ OE/OC
Figure 2-1. Flowchart of verification activities In the Mid-Appalachian Soil Survey.
11
-------
« some sampling crews acquired an inor-
dinate amount of supplies at the expense
of other crews because of unsatisfactory
disbursement and control of field equip-
ment and consumable supplies.
The DDRP management team determined
that more real-time control of the preparation
facilities was necessary. This determination
ted to the establishment of a single MASS
preparation laboratory located in Las Vegas,
Nevada, in close proximity to the EMSL-LV QA
staff. The consolidation of sample preparation
activities allowed:
« rapid and consistent soil drying due to the
low relative humidity;
establishment and consistent adherence to
strictly-defined sample preparation proto-
cols;
the ability to track and control progress at
the laboratory on a real-time basis;
a central control and distribution point for
field sampling supplies, thereby alleviating
shortages;
elimination of confounded multi-laboratory
measurement uncertainties at the prepara-
tion phase;
better distribution of samples among
sampling crews and laboratories;
advanced controls against sample con-
tamination; and
minimal time and expense for conducting
technical systems audits.
Separation of Mineral and Organic
Samples
Previously in the DDRP surveys, organic
soil samples were randomly intermingled with
mineral soil samples in each batch. For a
number of reasons, this practice was modified
for the MASS. Organic soil horizons generally
have much higher analyte concentrations than
mineral soil horizons within a pedon. Organic
soil to solution ratios are often quite different
than mineral soil to solution ratios for the
laboratory analysis of a given parameter. The
fluctuation between sequential high and low
sample readings on an analytical instrument
can produce erroneous data because of a
"carryover" effect from the instrument. Also,
certain parameters were not required to be
analyzed for organic soils.
For the MASS, mineral and organic
samples were placed in separate batches.
Samples were defined as organic if their
organic matter content was 20 percent or
more by weight as measured at the prepara-
tion laboratory. This enabled the laboratories
to perform the instrument calibration and
sample analysis within narrower linear dynam-
ic ranges and allowed the QA staff to better
assess the quality of data generated by the
analyses. As a result of this development,
different precision objectives were defined for
the mineral and organic batches.
Changes in Parameters and Methods
Several changes in analytical methods
and procedures were initiated for the MASS as
a result of data quality limitations and recom-
mendations from the NE and SBRP surveys,
and from special studies initiated to address
concerns raised during those surveys. The
changes included:
Analysis of exchangeable acidity in 1M
potassium chloride (AC_KCL) was exclud-
ed from the analytical parameters because
it was not used in the summation of
cations for the calculation of CEC or
base saturation in the previous surveys;
as a result, the data users recommended
replacing the determination of exchange-
able aluminum in 1M potassium chloride
(AL_KCL) with a determination of exchan-
geable aluminum in 1M ammonium chloride
(AL_CL) as part of the measurement of
exchangeable cations; the change reduced
the analytical time required to determine
exchangeable aluminum while retaining
very similar extraction conditions.
The determination of specific surface was
excluded for the MASS sample analysis
and no substitute physical determination
was recommended; the specified method
in the two earlier surveys was ethylene
glycol monoethyl ether (EGME), which was
a time-consuming method that resulted in
widely varying data for the high surface
area of the clay fraction in the soil samp-
les.
Silicon was added to the list of analytical
parameters; silicon extracted by acid
oxalate (SI_AO) is a measure of amor-
phous silica, which is a component of the
total silicon in the soil; in the previous
surveys, X-ray fluorescence was used to
measure total silicon in a 10 percent
representative subsample of the popula-
tion of routine soil samples.
12
-------
The analytical procedures for the exchan-
geable cations in both ammonium chloride
and ammonium acetate were revised; in
the first two surveys it was found that
several of the cations were at very low
concentrations in many mineral soil samp-
les; in order to increase the concentration
of the cations in the extract prior to in-
strumental analysis, it was recommended
that the quantity of mineral soil used in
the extraction be increased from 2.5 to 5.0
9
The results of a special study included a
recommendation that the amount of cel-
lulose filter pulp used in extractions be
reduced from 1.0 to 0.5 g; the pulp was
found to contain low levels of exchange-
able cations and could possibly contribute
to the measured cation exchange capacity;
the study tested several other materials
for possible substitution, but none were
found to be superior to the cellulose for all
purposes; to minimize possible contamina-
tion, however, the procedure for the prep-
aration of the cellulose pulp was revised
to include a washing step using hydro-
chloric acid.
The determination of extractable cations in
0.002M calcium chloride using a mechani-
cal extractor was changed to an overnight
extraction using a mechanical shaker; in
addition, PH_002M was determined on this
extract rather than on a separate extract
used for soil pH in the previous surveys.
The strength of the triethanolamine solu-
tion used for total exchangeable acidity
(AC_BACL) was reduced from 0.2M to
0.1M; in addition, the titer strength was
reduced from 0.1M to 0.05M hydrochloric
acid; these changes were recommended in
order to more accurately determine the
total acidity in soils having low acidity
content.
The use of vanadium pentoxide as a
combustion accelerator was recommended
for the determination of C_TOT and N_TO-
T; this accelerator is used to increase the
temperature and, therefore, decrease the
reaction time and ensure a more complete
combustion; it replaced the formerly used
accelerator which was composed of iron
chips, granular tin, and copper.
Consolidation of Total Elemental
Analysis
For the NE and SBRP surveys, all con-
tract laboratories analyzed total carbon, nitro-
gen, and sulfur, and reported the results along
with the other parameters. Due to the MASS
requirements for tighter measurement quality
and the specialized nature of the analytical
instrumentation, it was recommended by the
data users that these three elemental para-
meters be analyzed by only one carefully
selected laboratory. Performance evaluation
protocols were written specifically for these
analyses, as well as a separate statement of
work and special contractual considerations.
Sample Design
One of the conclusions drawn from the
two previous DDRP surveys was that the
sample design should be more robust and
enable more information to be derived from the
analysis of measurement quality samples. It
was also noted that there were insufficient
samples in the batches to assess the effects
of the various phases of sample measurement,
i.e., sampling, preparation, and analysis, in
relation to the system-wide measurement
uncertainty. The possibility of contamination
introduced through the measurement system
was a common concern voiced by reviewers of
the DDRP QA program.
For the MASS, an effort was made to
use measurement quality samples in the field,
at the preparation laboratory, and at the
analytical laboratories in such a way as to
provide optimal real-time control and assess-
ment of data quality. Of particular importance
was the addition of field audit samples at the
sampling phase to allow the estimation of
sampling and system-wide measurement
uncertainties. Data from previous surveys
were used to develop ranges of acceptable
analyte concentration for the assessment of
accuracy. The hierarchical sample design
described later in this section and in the QA
Plan (Papp et al., 1989) utilized blind and
double-blind samples to provide a complete
evaluation of the data quality issues.
13
-------
Statistical Design
A number of design characteristics were
investigated during the two previous DDRP
surveys, particularly in regard to statistical
analysis issues. During the data quality as-
sessment of the NE survey data base, issues
were raised regarding concentration dependen-
cy of the error variance, mathematical transfor-
mation of data, and data reporting criteria. It
was eventually decided that the logarithmic
transformation technique originally applied to
address concentration dependency was not
suitable for the requirements of the data
users, and that a statistical measure more
closely linked to the quality objectives and the
routine sample data was needed.
As a result of these findings, the QA
staff developed a step function technique
(Miah et al., 1989), described later in this
section, to evaluate data quality in terms of
predefined objectives and in relation to the
routine sample data. This approach made use
of a balanced hierarchical design that allowed
the various components of measurement
uncertainty to be estimated with respect to the
overall uncertainty in the population of data
collected. The information was used to cau-
tion data users about those situations whore
measurement uncertainty is a significant
contributor to overall data uncertainty.
Computerized Data Entry and
Verification
The most labor-intensive and lengthy
operation in the DDRP QA program at EMSL-
LV was the verification of analytical data. For
the NE and SBRP surveys, the QA staff used
hardcopy batch data packages for verification
while the data were being entered on personal
computers by ORNL staff. Lack of an auto-
mated verification program meant that preci-
sion estimates and the majority of verification
checks were calculated by hand, with a single
batch evaluation requiring approximately 100
hours to complete. Therefore, laboratories
would not receive affirmation of batch accep-
tance for a number of weeks after submission.
This delay often affected the reanalysis proce-
dure at the analytical laboratories because
they did not always have the option to reana-
lyze the initial extraction solutions and oc-
casionally had to re-extract the soils at the
expense of both time and money.
Due to the critical timeframe for delivery
of the verified MASS data, a Laboratory Entry
and Verification Information System (LEVIS)
was developed that could be utilized at the
laboratories (see Data Verification). The LEVIS
program facilitated the entry, editing, and
review of intermediate data and the calculation
of final data values for the MASS. The pro-
gram also performed verification checks for
the measurement quality samples and pro-
duced summary reports of data quality.
A modem-linked laboratory personal
computer was available to facilitate the trans-
fer of preliminary data which were used to
evaluate the current status of laboratory
operations and data entry. The laboratories
delivered the official analytical data to the QA
staff in hardcopy and on floppy disks in the
LEVIS data base structure. The QA staff
performed additional verification checks on a
personal computer and the data were reviewed
for confirmation/reanalysis requirements. The
data generated by batch reanalysis were
entered and verified in the same manner.
The LEVIS program greatly improved the
timeliness of the verification operation. Com-
plete verification of all quality control (QC),
precision, accuracy, and soil chemistry relation-
ship data for a batch of samples was ac-
complished in approximately four hours. The
increased efficiency of the verification process
aided in sample reanalysis because the analy-
tical laboratories were notified in time to
reanalyze the existing extracts.
Acceptance Criteria
During the early stages of the DDRP, the
QA staff lacked a definitive basis for establish-
ing acceptance criteria because previous
studies of this type had not instituted such an
intensive QA program. The batch acceptance
criteria for the NE survey were based on the
abilities of the analytical laboratories to satisfy
specific QC requirements, e.g., for reagent
blanks, and precision objectives for individual
pairs of audit samples and duplicate samples.
Nearly identical criteria were used for the
SBRP survey. In the absence of established
accuracy windows, a subjective "visual" as-
sessment of accuracy was made by the data
reviewer using data from previous sample
batches. Allowances for deviations from the
contractual requirements were somewhat
14
-------
subjective, with few firm protocols established
for corrective action. Precision was checked
only if sample concentrations were greater
than or equal to 10 times the contract-required
detection limit (CRDL), and a percent relative
standard deviation was the only statistical
criterion applied. Failure to satisfy the preci-
sion objectives for the laboratory audit samp-
les justified a request for batch reanalysis,
while requests for reanalysis on the basis of
duplicate sample variability were subjective.
The ongoing precision of the duplicate samples
was tabulated and if the variability for a spe-
cific batch were extremely high, in the judge-
ment of the data reviewer, reanalysis was
requested. A series of chemistry relationships
were checked and used to encourage the
laboratories to check for data reporting errors
such as transposed digits. In several instan-
ces, batch reanalysis was requested due to
the failure of a large, unfixed number of rela-
tionship failures in the judgement of the data
reviewer. The QA staff began to use reanaly-
sis templates late in the survey to aid in
standardizing the basis for requesting reanaly-
sis. For the SBRP survey, an additional labor-
atory audit sample pair was incorporated into
each batch to provide a better check of the
analytical laboratory precision.
For the MASS, many significant changes
in the acceptance criteria were initiated. The
CRDLs and precision requirements of the QC
samples, described later in this section, were
tightened. Additionally, precision was checked
on all pairs of measurement quality samples
without regard to low concentration values
approaching the CRDL. A two-tiered scale
was used for precision, where the lower range
of concentration was assessed using an
absolute standard deviation and the upper
range of concentration was assessed using a
relative standard deviation.
Several new measurement quality samp-
les were introduced to check both precision
and accuracy. Field audit samples were
included in each batch to measure system-
wide precision components. Accuracy win-
dows for laboratory audit samples were es-
tablished and used as contractual require-
ments for the laboratories. Detailed reanalysis
templates were developed by the QA staff to
assist in setting major and minor flags and to
eliminate the subjectivity present in the NE and
SBRP surveys in regard to reanalysis requests.
The analytical laboratories were also contrac-
tually required to satisfy a larger group of
chemistry relationships, with batch reanalysis
requested if six or more relationships failed in
a single batch. Upon completion of analysis,
two types of internal consistency checks were
implemented to identify possible outliers and
request confirmation of these data or reanaly-
sis of the affected samples.
Selection of Analytical
Laboratories
It was determined before the onset of
the DDRP surveys that EPA laboratories and
the analytical laboratories participating in the
EPA Contract Laboratory Program did not have
the specific instrumentation nor expertise to
conduct the analysis of all required DDRP
parameters. Therefore, analytical services
were solicited from commercial analytical
laboratories, and a laboratory selection pro-
cess was developed that would objectively
evaluate analytical performance to satisfy the
data requirements for the DDRP. These re-
quirements had to be clearly and concisely
documented in statements of work and the
data consistently and objectively evaluated for
each laboratory in order to identify those most
qualified to perform the analyses.
Statements of Work
The contracting process included the
preparation of detailed analytical laboratory
statements of work (ALSOWs) which defined
the analytical and QA requirements in a con-
tractual format. Two ALSOWs were created,
one for general soil analysis (W802498D1) and
the other for total elemental soil analysis of
carbon, nitrogen, and sulfur (W802621D1).
Although the MASS analytical protocols and
initial draft of the QA Plan were prepared in
the early phases of the planning process, the
laboratory methods and data quality require-
ments had to be presented in a contractual
format in order to obtain support services.
This process involved careful review of the
analytical and logistical requirements, i.e.,
reporting and QC stipulations, to ensure their
clarity in the ALSOW and their ability to be
satisfied according to the contract specifica-
tions. The analytical laboratories were re-
quired to follow the methods exactly as speci-
fied in the ALSOW. The project officer was
15
-------
authorized to provide technical clarifications for
the contract laboratory, but contractual chan-
ges could be made only with the written
approval of the EPA contract officer.
An invitation for bid (IFB) was prepared
for each ALSOW and subsequently advertised
in the Commerce Business Daily to solicit
analytical support from interested laboratories.
It was anticipated that approximately 800 to
1,000 soil samples would be collected in the
MASS. A contractor could bid on the analysis
of one or more bid lots (40 to 800 soil samp-
les per bid lot for general analysis, 80 to 1,600
samples per bid lot for elemental analysis).
Bid lots were specified for delivery to the
analytical laboratories at a maximum rate of
one batch per week, with each batch contain-
ing approximately 40 samples. Delivery of the
completed batch data package by the contrac-
tor was required within 60 days of sample
receipt. Reanalysis was requested of suspect
data identified through data quality manage-
ment activities, and full payment was withheld
pending compliance with any such requests.
Initial responses to the IFBs were received
from 54 laboratories for general analysis and
9 laboratories for elemental analysis.
Performance Evaluations
All interested laboratories received a set
of five pre-award performance evaluation (PE)
samples. As part of the bid submission
package for qualification, these laboratories
were required to analyze the PE samples and
to report the results within 25 days following
sample receipt. The PE samples consisted of
four mineral audit samples and one organic
audit sample, and were intended to represent
the low, medium, and high analyte concentra-
tions expected in the soil samples collected
during the survey. Data packages received
from each competing laboratory were evaluat-
ed and graded on the QC, precision, accuracy,
and chemistry relationships of analytical data
as well as the quality and completeness of the
data package, using the scoring sheet provi-
ded in Appendix E of the MASS QA Plan (Papp
et al, 1989). This evaluation identified those
laboratories that could successfully perform
the analyses at the level of quality required for
the DDRP. During this stage, 49 laboratories
withdrew from the competition for general
analysis because of the high number of para-
meters to be analyzed or the stringent instru-
ment specifications. Also, five laboratories
withdrew from the competition for elemental
analysis.
Three of the remaining five laboratories
that submitted bids for general analysis suc-
cessfully passed the PE sample evaluation and
were visited by EPA technical and contracting
representatives who verified the qualifications
and capabilities of these laboratories to meet
or exceed the contractual requirements. The
EPA team determined whether each analytical
laboratory had adequate facilities, equipment,
personnel, and technical capabilities to analyze
large numbers of samples in accordance with
the ALSOWs and within the required timeframe
of the survey. These visits also provided an
opportunity to clarify contractual specifications
with laboratory personnel and to identify
deficiencies that were observed during the PE
phase.
Three of the remaining four laboratories
that submitted bids for elemental analysis
successfully passed the PE sample evaluation.
The laboratory that submitted the most com-
petitive bid was visited by EPA representatives
who verified the qualifications and capabilities
of this laboratory to meet or exceed the con-
tractual requirements. Had this laboratory not
passed the on-site evaluation, the EPA team
would have visited the laboratory which sub-
mitted the next most competitive bid until a
fully qualified laboratory could be identified.
Award of Analytical Contracts
The laboratories that successfully pas-
sed both the performance and on-site evalua-
tions were awarded contracts to provide
analytical services for the MASS general soil
analysis. However, because only one bid lot
for the elemental analysis contract was to be
awarded, the qualifying laboratory that sub-
mitted the most competitive bid and success-
fully passed the on-site evaluation was subse-
quently awarded the elemental analysis con-
tract. Coincidentally, this laboratory was also
one of the three laboratories that was award-
ed a contract for general analysis.
In the previous two DDRP surveys, there
were four participating analytical laboratories
that were described as Laboratories 1, 2, 3,
and 4 in the QA reports (Van Remortel et al.,
1988; Byers et al., 1989). All four laboratories
16
-------
analyzed samples in the NE survey, while only
Laboratories 1, 2, and 3 analyzed samples in
the SBRP survey. Three of these same four
laboratories were selected for the MASS
sample analysis, namely, Laboratories 1, 2,
and 4. Laboratory 3 did not submit a bid for
either the MASS general or elemental analyses,
although the NE and SBRP data from this
laboratory are used in many of the compara-
bility discussions in Section 3 of this QA
report,
Distribution of Batches
Prior to initiating the routine sample
analysis, the analytical methodology was
employed in the analysis of three batches
containing several replicates of various types
of audit samples. One batch was sent to
each of the selected analytical laboratories
and the resulting data were used only to
examine the utility of the batch acceptance
criteria and other operations components of
the MASS. Subsequently, 27 mineral soil
batches and 2 organic soil batches were
analyzed among the three laboratories, as
outlined in Table 2-1. Organic and mineral
soils were placed in separate batches because
of the expected wide variability in analyte
concentrations.
Table 2-1. Distribution of Batches for General Soil
Analysis
Number of
Laboratory batches Batch code*
L1
L2
L4
9
9
11
30104* 30107, 30110, 30113, 30116,
30119, 30122, 301256, 30128
30106, 30109, 30112, 30115, 30118,
30121, 30124, 30127, 30130
30105, 30108, 30111, 30114, 30117,
30120. 30123, 30126, 30129,
30131C, 30132C
" L2 also performed the elemental soil analyses (C TOT,
N_TOT, S_TOT) for all 29 batches.
* Batch later retrieved from L1 and sent to L4 for
reanalysis of exchangeable calcium and magnesium
(coded as batches 39104 and 39125, respectively).
c Batch contained only organic samples.
Analytical Laboratory Quality
Control
Each analytical laboratory was required
to comply with stringent QC requirements
during the MASS sample analysis. Specific
requirements were stated in the QA Plan (Papp
et al., 1989) for sample receipt and storage,
use of laboratory protocols, sample homogeni-
zation, instrument calibration, data entry, data
reporting, and preventive maintenance. These
subjects are described in the following sec-
tions.
Sample Receipt and Storage
The routine bulk samples were collected
from each of the selected soil pedons by five
sampling crews. These bulk samples were
delivered to a single preparation laboratory
located at EMSL-LV, where the samples under-
went processing. The laboratory staff pre-
pared two homogenized analytical samples
from each air-dried bulk sample: a general
analysis sample weighing approximately 450 g
and an elemental analysis sample weighing
approximately 50 g. All leftover soil was
archived in long-term cold storage.
All samples received at each analytical
laboratory were checked for sample integrity
by a receiving clerk who recorded on the
shipping form, DDRP Form 102 (see Appendix
K), the date that the samples were received.
The sample labels were checked against the
form to identify discrepancies. A copy of the
form was then sent to the QA manager. If
there were any discrepancies or problems,
such as sample leakage or insufficient sample
volume, the QA manager was notified immedi-
ately for instructions. The samples were
refrigerated at 4°C as soon as possible after
receipt by the laboratory and were kept refrig-
erated when not undergoing routine analysis.
After all analyses were completed, the samp-
les were placed in long-term cold storage to
ensure sample integrity in the event that
reanalysis was requested.
17
-------
Laboratory Protocols
Instrument Calibration
The analytical methods and associated
protocols used in the MASS were selected to
ensure that the resulting data would be com-
parable with other similar data bases, e.g., the
DDRP NE and SBRP data bases. General
laboratory QC protocols included the use of
trained laboratory personnel, suitable labora-
tory facilities, appropriate instrumentation with
known performance characteristics, and labor-
atory supplies of sufficient quality and quantity
for the specific purpose. The laboratories
were occasionally required to use specific
manufacturers and models of instruments.
Written standard operating procedures, a
laboratory QA plan, and a record of in-house
samples were required. Control charts, with
99 and 95 percent control limits, were plotted
for the QC soil samples and liquid check
samples, and were updated daily for each
parameter undergoing analysis. The charts
were submitted to EMSL-LV upon request at
any time during, or immediately following, the
survey. On-site technical systems audits
ensured that the laboratory protocols and
procedures were followed correctly (see Ap-
pendix H).
Sample Homogenization
The bulk soil samples processed by the
preparation laboratory were air-dried, disag-
gregated, and sieved to retain the less than
2-mm fraction (Papp and Van Remortel, in
press). Analytical samples were prepared by
homogenization and subsampling using a
closed-bin riffle splitter. Immediately after
receiving these samples from the preparation
laboratory, the analytical laboratories homo-
genized each sample by passing the entire
sample through a closed-bin riffle splitter five
times in succession. After this step, approx-
imately half of the contents of each sample
was stored and the other half was used for
the analyses. Sample movement in the labora-
tory, e.g., transport or shaking, was restricted
in order to minimize sample segregation by
particle size and density. Sample aliquot
removal was accomplished by random inser-
tion of a sampling device, e.g., spatula or
scoop, into the sample bottle.
For most methods, a calibration curve
was established using a minimum of three
points within the linear dynamic range of the
instrument unless the manufacturer's recom-
mendations specified the use of more than
three points within the range. The linear
dynamic range was determined by a least
squares regression analysis (Steel and Torrie,
1960), and the correlation of concentration and
instrument response was required to be 0.99
or higher. The lowest concentration of the
calibration standards was required to be not
more than 10 times the CRDL, as described
later in this section.
The concentrations of standards brack-
eted the expected range of sample concentra-
tions without exceeding the linear dynamic
range, with one exception. Laboratory 4 was
allowed to use an upper calibration standard
of 2 mg/L Ca2+ for calcium extracted during
the analysis of CA_CL2, even though sample
concentrations may have exceeded the upper
calibration limit. This exception was made
because this laboratory had been able to
demonstrate instrument linearity throughout
the range of calcium concentration. However,
Laboratory 4 was required to verify linearity by
using a 100 mg/L Ca2+ (± 5%) standard for
each batch. In all other cases, the labora-
tories diluted and reanalyzed samples if their
concentrations exceeded the linear dynamic
range.
Data Entry
After all samples in a batch were ana-
lyzed for a given parameter, the results were
entered on raw data forms and subsequently
entered into a specialized computer system
using a personal computer supplied to each
analytical laboratory by the EPA (see Data
Verification). A single-entry system was em-
ployed in conjunction with a complete visual
data check performed by the laboratory mana-
ger prior to formal data submission. The
computer system had the ability to produce
screens of standardized data forms that
allowed entry of sample weights, solution
volumes, dilution factors, instrument readings,
titrant volumes, and titer acid normalities for
routine sample analysis, all with the appro-
priate predefined number of significant decimal
18
-------
places. Separate data forms or spaces were
provided for the QC sample data, and a page
was also provided to allow the documentation
of any comments concerning the analyses or
samples by the laboratory analyst or manager.
Data Reporting
Analytical data were reported according
to the protocols specified in the ALSOW (USE-
PA, 1988). For formal submission of a batch
of data, each laboratory was required to
provide: (1) a signed cover letter that indi-
cated the date, laboratory name, and batch
number being formally submitted; (2) a copy
of DDRP Form 102; (3) a floppy disk contain-
ing all data files generated; and (4) hard
copies of all appropriate data forms. The data
contained on the floppy disk and the hard
copies were required to be identical or a
resubmission was requested following iden-
tification and correction of any deficiencies.
Each laboratory manager signed a completed
data form to indicate that the batch data had
been reviewed and that the samples were
analyzed exactly as described in the ALSOW.
Any deviations from the protocols were re-
quired to be documented on the form. Copies
of the raw data were submitted upon request
of the QA manager; otherwise, all original raw
data were retained at the analytical labora-
tories. The raw data include data system
printouts, chromatograms, notebooks, indi-
vidual data sheets, and control charts.
If data confirmation was requested, i.e.,
a recheck of suspect data values, the analyti-
cal laboratories submitted a completed DDRP
Form 500 (see Appendix K) indicating whether
the questioned data values were correct. If a
correction was made in the data, the labora-
tory manager made note of the change, ini-
tialed the form, and provided an explanation
for the change. The correction was made in
the data base and an updated data reporting
form was generated. Raw data supporting the
change were provided upon request of the QA
manager. There were no laboratory qualifiers,
or tags, reported by the laboratories during the
MASS, whereas the reporting of tags was
prevalent in the previous DDRP surveys.
Instead, any comments relating to the tag
definitions in the MASS QA Plan (Papp et al.,
1989) were entered as free-form notes on the
data reporting forms.
Preventive Maintenance
The analytical laboratories were required
to maintain their instruments in good operating
condition. Instrument breakdowns or failures
were required to be repaired within a reasona-
ble time period and documented in an instru-
ment logbook. Each logbook contained infor-
mation pertaining to routine maintenance and
cleaning, calibration problems, and repairs. If
an instrument failed during analysis of a
batch, all samples in that batch were reana-
lyzed following instrument repair and recalibra-
tion.
Analytical balances were calibrated at
the beginning of each day of use. A balance
was considered to be in need of maintenance
if the calibration readings exceeded the range
of the respective Class S weight at the level of
sensitivity required for a particular analysis. A
check of the logbooks during the technical
systems audits showed that the balances
were adequately calibrated during the sample
analysis period.
Technical Systems Audits
An exercise which was extremely useful
in controlling data quality in the MASS was the
technical systems audit, or on-site evaluation,
which was conducted to ensure that all labora-
tories adhered to the protocols in a consistent
manner. Each analytical laboratory underwent
a minimum of three technical systems audits
during the survey. Between audits, frequent
communications were maintained between the
QA staff and designated laboratory personnel
to ensure that each laboratory was satisfying
the QC requirements and to make a prelimi-
nary evaluation of data quality and laboratory
performance. This contact enabled the QA
audit team to become familiar with analytical
difficulties at the laboratories that were subse-
quently discussed during audit visits.
The first on-site technical systems audits
were performed in September 1988, after each
laboratory had successfully analyzed the set of
pre-award PE samples. The QA manager and
the QA soil chemist evaluated each of the
laboratory functions that were pertinent to the
analyses. A questionnaire included in the QA
Plan (Appendix F in Papp et al., 1989) was
used to guide this evaluation as well as to
19
-------
document each laboratory's performance in a
consistent manner. The lead auditor sum-
marized all observations in an audit report (see
Appendix H) and brought any discrepancies to
the attention of the laboratory manager.
Second-round technical systems audits
were conducted in November 1988, after each
laboratory had received two batches of soil
samples and had completed analysis of the
majority of parameters. By this time, prelimi-
nary data had been received by the QA staff
via the modem-linked computer data entry
system. The audit allowed an opportunity to
discuss all problems associated with the data
and the analytical methods. Of major concern
to all laboratories during this audit were the
new extraction procedures for CEC and subse-
quent cation analyses. The method specified
that a polyester pre-plug and an Acrodisc be
used in the extraction device rather than the
cellulose filter pulp used in the earlier DORP
surveys. The laboratory managers suspected
that the new procedures, which specified a
larger quantity of soil, were not producing an
efficient extraction because the mechanical
extractor had a tendency to clog when extract-
ing soils of high clay content. The procedures
were subsequently modified to return to the
use of cellulose filter pulp, but at a reduced
quantity. During this audit, operation of the
computer data entry and verification system
(see Data Verification) was discussed and any
questions pertaining to its use were answered.
In general, the laboratories were functioning
within the protocols of the contracts.
Third-round technical systems audits for
Laboratories 1 and 2 were conducted in Janua-
ry 1989, while Laboratory 4 was visited again
in March 1989. All preliminary data were
reviewed, focusing on parameters not satisfy-
ing QE/QC criteria in a consistent manner. The
sample reanalysis criteria and the procedures
used by the QA staff to evaluate the data,
including the DDRP measurement quality
sample design, were discussed. The auditors
toured each laboratory's facilities and held
individual discussions with the laboratory
technicians. The modified LEVIS data system
used by Laboratory 2 for elemental analysis
data was evaluated during the audit and
problems were discussed and rectified.
Laboratories 1 and 2 underwent a fourth-
round technical systems audit during May
1989. Laboratory 4 was not visited because
the previous audits had shown this laboratory
to be in control of all analysis and reporting
tasks. Laboratory 1 had completed analysis
on all of its batches, so this audit was con-
ducted to review the final data, archive the
computer hardware and analytical data base,
discuss contractual issues, and specify proce-
dures for returning the hardware and remain-
ing soil samples to EMSL-LV. Laboratory 2
had been experiencing difficulty in satisfying
the carbon ratio relationships for elemental
analysis samples exhibiting low C_TOT con-
centrations; therefore, subsamples of the soils
in Batch 30114 were sent to three other labora-
tories to verify the values reported by this
laboratory. It was concluded that the Labora-
tory 2 data were valid, but that the theoretical
chemistry relationships may not have been
valid for elemental analysis at these low
concentrations (Schumacher and Shaffer, in
review). An appropriate delimiter was applied
to each carbon relationship. This laboratory
was also experiencing difficulty in satisfying
the accuracy criteria in the extractable iron and
aluminum analyses. The median-range audit
samples for the latter batches in the survey
showed a trend toward high bias and were
outside the control limits when plotted on the
QC charts. The problem appeared to be
associated with higher than usual tempera-
tures in the laboratory during the extraction
phase of the analysis, which was conducted
in an enclosed room with minimal ventilation.
Once this situation was rectified, the samples
were reanalyzed and the resulting data were of
acceptable quality.
Special Studies
Before the MASS was implemented, the
DDRP data users and QA staff raised some
questions regarding soil preparation and
analytical procedures. As a result, a series of
small-scale research projects were undertaken
to address the issues that were raised.
Descriptions of the studies and their findings
are provided in the following subsections.
Elemental Analysis Study
It was noted early in the MASS sample
analysis that a relatively high number of car-
bon to nitrogen (C:N) and carbon to sulfur
(C:S) ratios failed to satisfy the expected
acceptance criteria for the total elemental
20
-------
analysis performed by Laboratory 2, especially
for those samples having low C_TOT con-
centrations. The boundary conditions for the
ratios were based upon theoretical and report-
ed values for these ratios in the soil science
literature. The C:N and C:S ratios were ex-
pected to be from 7 to 50 and 40 to 400,
respectively, with delimiters applied to ac-
comodate instrument variability at low con-
centrations. A study was conducted to deter-
mine whether: (1) the data generated by the
laboratory were of acceptable quality, (2) bias
was present in the data submitted by the
laboratory, (3) the expected ranges established
for the C:N and C:S ratios accurately repre-
sented the actual soil data, and (4) there was
a particular source of error causing certain soil
samples to exceed the acceptance ranges.
In the study, two independent labora-
tories and Laboratories 2 and 4 analyzed
aliquots of the samples associated with Batch
30114 and reported the results in a specially
coded Batch 39114. Laboratory 4 analyzed the
samples using the MASS elemental analysis
methods, while the independent laboratories
and Laboratory 2 employed other commonly
used methods for total elemental analysis.
The results showed that a failure of six or
more chemistry relationships in the batch
occurred at all four laboratories using the
established criteria. It was noted that most of
the relationships failed on soils containing
C_TOT of less than 0.3 percent. An evaluation
of intralaboratory precision showed that the
replication of C_TOT and N_TOT data was
excellent, with only one observation above the
acceptable variability limit. The S_TOT values
were reproducible among the analytical labora-
tories, although not to the same degree as
C_TOT and N_TOT (Schumacher and Shaffer,
in review).
The study concluded that the data gener-
ated by Laboratory 2 for elemental analysis
were of acceptable quality to justify their
incorporation into the final MASS data base.
The cause for the failure of the ratios ap-
peared to be based within the empirical rela-
tionships and not in the data generated by the
laboratory. It was determined that the rela-
tionships were originally developed for agricul-
tural soils that were generally higher in total
carbon, nitrogen, and sulfur than forest soils
in the Mid-Appalachian region. Appropriate
delimiters were used for applying the C:N and
C:S ratios to low concentration samples.
Homogenization Study
In the previous DDRP surveys, a statisti-
cal logarithmic transformation of soil analytical
data suggested that the soil sample homogen-
ization procedure was possibly contributing an
inordinate measure of uncertainty to the data.
An extensive literature search revealed a lack
of research specifically focusing on the preci-
sion and accuracy of various homogenization
methods. Therefore, a study was conducted
at EMSL-LV in the summer of 1988 to examine
the various homogenization techniques used to
prepare soil samples for analysis. Three of
the most commonly used homogenization
techniques were examined further: (1) cone-
and-quartering, (2) riffle splitting using both
open- and closed-bin riffle splitters, and (3)
random sampling after disaggregating and
sieving. A method evaluation was initiated
using four soils of widely varying textures.
Several subsampling approaches were used to
evaluate the homogenization techniques,
including the collection of subsampies after a
series of passes during homogenization and a
timed experiment to determine the efficiency of
each technique. The analyses chosen to
determine the effectiveness of each technique
and the loss of fines from the system were
particle size distribution, organic matter by
loss on ignition, and pH in 0.01M calcium
chloride.
The study concluded that use of either
the closed- or open-bin riffle splitter is prefera-
ble to the cone-and-quartering or random
sampling homogenization techniques. Cone-
and-quartering required as much as six times
longer for homogenization than riffle splitting
and exhibited a greater loss of fines. Random
sampling was the most time-effecient techni-
que but was the least effective from a variabil-
ity standpoint. The closed-bin riffle splitter is
recommended over the open-bin riffle splitter
due to its greater ability to retain the fines
that could be lost as dust during homogeniza-
tion. It was determined that five passes
through the splitter are optimal to provide the
most homogeneous sample, i.e., lowest varia-
bility, in terms of particle size distribution and
organic matter, and additional passes ap-
parently resulted in greater particle segregation
(Schumacher et al., in review; Papp and Van
21
-------
Remortel, in press). It was also determined
by EMSL-LV statisticians that other statistical
techniques, e.g., step function, applied to the
data did not indicate the same type of inor-
dinate homogenization uncertainty as the
logarithmic transformation technique.
Filter Pulp Study
A study was initiated to examine the
precision and physical advantages of several
types of filter material utilized in the mechani-
cal extraction of base cations from soils as
specified in the MASS analytical methods
(Blume et al., in press). An evaluation of
alternatives to the continued use of Schleicher
& Schuell No. 289 cellulose filter pulp was
warranted because of cation contamination
and its effect on CEC data observed in the
previous surveys. Also, there were difficulties
in using the standardized method for prewash-
ing the pulp prior to use.
Three bulk fiber materials (cellulose pulp,
polyester fiber, and rayon fiber) used for
filtration, three types of disposable filter cas-
settes (teflon, nylon, and polysulfone), and five
kinds of disposable membranes (two mixed
esters, glass, polypropylene, and teflon) with
reuseabie holders were evaluated as alterna-
tives to fre cellulose pulp. The filter materials
were evaluated for their physical suitability,
contamination by six major soil elements, and
the time required in extraction and equipment
preparation between analyses.
The results of the study showed that the
cellulose pulp was the most hydrophilic and
possessed better packing qualities than either
the polyester or rayon fibers. Chemical con-
tamination was highest using the rayon fiber,
especially for exchangeable sodium. However,
each type of material exhibited distinct con-
tamination characteristics for the six primary
cations measured in the DDRP. The dispos-
able membranes allowed soil to exude around
the edges, and rupturing of the membranes
occurred under vacuum. The nylon and poly-
sulfone cassettes were much more reliable
than the teflon cassettes which exhibited
hydrophobic properties. In general, the cas-
settes were costly and difficult to set up, and
accounted for some soil loss.
The study concluded that a polyester
fiber plug placed in the syringe tip, used in
conjunction with a cassette, greatly improved
the extraction process and reduced soil loss
and clogging of the filters. This apparatus
was initially adopted for the MASS extractions.
During the survey, however, it was determined
that the apparatus was not suitable for soils
high in clay due to ongoing clogging problems.
As the cellulose fiber was easily available,
inexpensive, easy to prepare, and demonstated
reproducible extractions without soil loss, it
was recommended that the use of cellulose
filter pulp be continued for the MASS soil
extraction methods. To minimize the potential
for contamination, the size of the filter plug
was reduced from 1.0 g to 0.5 g.
Independent Analysis of Calcium and
Magnesium
During the survey, Laboratory 1 experi-
enced difficulty in the analysis of calcium and
magnesium (CA_OAC, CA_CL, MG_OAC,
MG_CL, and MG_CL2) in Batch 30104 and
magnesium (MG_CL2) in Batch 30125. Al-
though two reanalyses of the batches were
performed by this laboratory, the data contin-
ued to fail the acceptance criteria. Therefore,
the two batches of samples were retrieved
from Laboratory 1 and were receded as Batch-
es 39104 and 39125, respectively. Laboratory
4 was contracted to analyze the relabeled
samples for the five parameters, with the
anticipation that better data quality could be
achieved for the two batches. The resulting
data were of acceptable quality; therefore, the
QA staff substituted these data in the analyti-
cal data base as replacements for the previ-
ous calcium and magnesium data from Labor-
atory 1.
Assessment Procedures
The QA staff at EMSL-LV was respon-
sible for evaluating the results of the MASS
sample analyses and reporting to the DDRP
technical director on the quality of the verified
data bases. The characteristics of data quali-
ty and their methods of assessment are
described in the following subsections.
Quality Attributes
The attributes of data quality are qualita-
tive and quantitative characteristics that are
designed to provide an overall assessment of
22
-------
quality during the measurement phases of the
MASS data collection activities. Measurement
quality was defined in terms of the following
six quality attributes:
Detectability the determination of the
low-range critical value of a soil charac-
teristic that a method-specific procedure
can reliably discern,
Precision - the level of agreement among
multiple measurements of the same soil
characteristic,
Accuracy - the level of agreement between
an observed value and the "true" value of
a soil characteristic,
Representativeness - the degree to which
the soil data collected accurately represent
the population of interest,
Completeness - the quantity of soil data
that is successfully collected with respect
to the amount intended in the experimental
design, and
Comparability - the similarity of soil data
from different sources included in a given
set of data, and the similarity of methodol-
ogies from related projects across the
regions of interest.
Measurement Quality Objectives
To address the overall DDRP objectives,
conclusions must be based on scientifically
sound interpretations of the data base. Accor-
dingly, the EPA requires all monitoring and
measurement programs to have established
objectives for data quality based on the antici-
pated end uses of the data (Stanley and Vern-
er, 1985; USEPA, 1986). The utility of the data
is defined by the ability to confirm, reject, or
discriminate among hypotheses formulated in
the various DDRP computer simulation models.
The primary purpose of the QA program is to
increase the likelihood that the resulting data
base meets or exceeds overall data quality
objectives (DQOs). The quality of data can be
quantified in relation to the DQOs, thereby
allowing the data user to evaluate the hypothe-
ses with a known level of confidence.
The methodology for establishment of
DQOs has evolved over the course of the
DDRP soil surveys. In contemporary usage,
DQOs are statements of the levels of uncer-
tainty that data users are willing to accept in
the data. Uncertainty that exceeds the DQOs
is considered to render the data unreliable in
the judgement of the data users. The DQOs
encompass all components of uncertainty
resulting from sample measurement, operation-
al activities, e.g., soil mapping or sample
aggregation into sampling class/horizon con-
figurations, and regional uncertainties, e.g.,
spatial variability. For the MASS QA program,
measurement quality objectives (MQOs) were
identified as a component of user-defined
DQOs. The MQOs are specific goals defined
by the data users that clearly describe the
data quality that is sought for each of the
measurement phases. They are defined quan-
titatively, where possible, according to quality
attributes for each phase and are comparable
to the "DQOs" specified in previous DDRP
surveys.
The MQOs for the soil characterization,
preparation, and analysis phases were speci-
fied in the MASS QA Plan (Papp et al., 1989).
The survey objectives were reviewed by scien-
tists familiar with analytical methods and soil
characterization techniques, including soil
chemists and laboratory personnel outside of
the program. The MQOs were defined using
information gained from the DDRP NE and
SBRP surveys and from peer review comments
in accordance with the analytical methods,
instruments, and procedures selected.
Soil sampling in the MASS included the
physical removal of soil samples from an
excavated pit as well as the characterization
of the soil pedon and sampling site (Kern and
Lee, in press). The primary goal of the MASS
sampling was to collect soil samples from
representative pedons of 15 established sam-
pling classes (Lee et al., 1989). The sampling
classes were intended to represent the range
of soil types encountered in the Mid-Appala-
chian region. The MQOs developed for sample
analysis were not applicable to the field sam-
pling activities, hence, specific sampling objec-
tives were developed to ensure that field
operations were conducted in a consistent
manner and that an estimate of variability
among sampling crews could be provided. The
field sampling effort produced both qualitative
and quantitative soil data from pedon charac-
terization, e.g., soil color and horizon thickness,
and quantitative data from soil analysis, e.g.,
field measured pH.
To assess the analytical MQOs, a series
of measurement quality samples were ana-
lyzed together with the routine samples in a
23
-------
manner that was statistically relevant and
which would allow conclusions to be drawn
concerning the data quality. There are a
number of assumptions and considerations
upon which the batch sample design was
based:
The analytical MQOs could be attained if
each batch satisfied the batch acceptance
criteria set for the quality attributes.
A two-tiered approach for measuring ana-
lytical precision would allow accurate
assessment of the low concentration
samples apart from the median and high
concentration samples.
Approximately 25 percent of the overall soil
analysis effort would be applied to the
analysis of measurement quality samples
required for assessment purposes. There
were approximately 40 samples in a batch,
hence, about 10 of these were measure-
ment quality samples.
« Measurement uncertainty at any given
phase, e.g., analytical within-batch pre-
cision, could be evaluated with data from
a minimum of 20 measurement quality
samples of a given type (Barth et al.,
1989). Since 27 mineral soil batches were
analyzed during the MASS, a sufficient
number of samples were present to make
reliable across-batch statistical estimates.
The primary sources of measurement
uncertainty to be identified, controlled, and
assessed were produced by sampling,
preparation, and analysis activities.
« Each of the primary sources of uncertainty
could be considered to be a combination
of several smaller sources of uncertainty,
e.g., the analytical component could in-
clude within-laboratory, within-batch,
between-laboratory, and between-batch
uncertainties.
Accuracy includes the characteristics of
bias in relation to predetermined audit
sample accuracy windows, and laboratory
differences and trends in relation to an
accepted reference value within each
window.
Details regarding the implementation and
assessment of the analytical laboratory MQOs
in the MASS are described in the following
subsections. [NOTE: Some small changes to
the QA program were implemented subsequent
to the final submission of the MASS QA Plan
(Papp et al., 1989). These included raising the
required detection limit for AC_BACL from 0.25
to 1.25 meq/100g in organic soils. Also, an
error was made in Appendix C of the QA Plan
with regard to the soihsolution ratios used in
the sample analyses. Table 1-1 of this report
shows the actual ratios that were used by the
analytical laboratories.]
Detectability
An important factor to consider in the
measurement quality evaluation is the detec-
tion limit, which is defined as the lowest value
of a characteristic that a method-specific
procedure can reliably discern. The primary
consideration in the evaluation of detectability
is whether a measured sample value can be
considered significantly different from the
measured value of a sample blank (Keith et al.,
1983). The probability that an analytical signal
is not simply a random fluctuation is depen-
dent on how many standard deviations the
analytical signal varies from the mean value of
blank responses (Long and Winefordner, 1983).
The assessment of detectability issues in the
MASS required an investigation of previously
obtained precision data for low concentration
samples.
A commonly recognized value for the
detection limit is three times the standard
deviation of a set of blank samples. A signal
measured at this level or greater would have
less than a 1-in-100 chance of being the result
of a random fluctuation of the blank, assuming
a one-tailed normal distribution of the blank
samples. In the absence of "soil blank" samp-
les, laboratory stock solutions and low con-
centration replicate samples have previously
been used in the DDRP to estimate the varia-
bility expected of blank samples (Van Remortel
et al., 1988; Byers et al., 1989). Although liquid
blank samples were used extensively in the
AERP aquatics surveys, development of a
suitable synthetic soil blank for system-wide
use in the DDRP has not been feasible.
With these limitations in mind, the DDRP
data users and QA staff defined contract-
required detection limits (CRDLs) based on the
variability expected from a series of five cali-
bration blanks analyzed in each batch of
samples. The CRDLs were used in assessing
contractual compliance by the analytical labor-
atories with regard to instrument detectability.
Each laboratory was required to satisfy the
CRDLs for specified parameters, as presented
24
-------
in Table 2-2. It was not appropriate to es-
tablish CRDLs for the particle size, pH, or
sulfate isotherm parameters due to the nature
of their measurement procedure, i.e., by gravi-
metric analysis, potentiometric readout, and
sulfur solution, respectively; each of these
parameters had expected ranges of concentra-
tion that were well above the critical detection
limits of the relevant measuring devices.
The actual instrument detection limits
(IDLs) measured throughout the survey by the
laboratories were calculated from the pooled
variability data of the calibration blanks. The
IDLs identified the lowest analyte concen-
trations, averaged across batches, that could
be reliably detected by the laboratories using
the specified analytical instruments and proce-
dures, During the actual sample analysis,
IDLs were defined as three times the pooled
standard deviation of at least 15 nonconsecu-
tive calibration blanks run on three separate
days. The IDLs were required to have values
less than or equal to the CRDL In some
analyses a blank signal was not obtained; in
these cases, the IDL was defined as three
times the pooled standard deviation of 15
nonconsecutive replicate analyses of a stan-
dard whose concentration was four times the
actual detection limit or the CRDL, whichever
concentration was less. Acceptable initial
IDLs were established prior to any sample
analysis and subsequent IDLs were deter-
mined arid reported on a batch-by-batch basis
upon submission. Initial IDLs were valid for
one month after their establishment; if more
than one month had elapsed since establish-
ment, the laboratories were required to es-
tablish new IDLs.
The IDL determination for AC_BACL
followed the same method used in the es-
tablishment of the initial IDL throughout the
survey due to the strong dependence of the
calibration blank upon the individual batch of
extracting solution. For all other parameters,
after the validity of the initial IDLs were es-
tablished, the first batch of samples was
analyzed and the IDLs reported for this batch
were determined as composites of the calibra-
tion blanks from the batch and the calibration
blanks run on the second and third day ot the
initial IDL determination. Similarly, the IDLs
for other batches were established as com-
posites of the calibration blanks from the
batch being analyzed and the calibration
Table 2-2. Contract-Required Detection Limits for the
Analytical Laboratories
Parameter
CA CL
MG CL
K CL
NA CL
AL_CL
CA OAC
MG OAC
K OAC
NA_OAC
CEC_CL
II
CEC_OAC
it
AC_BACL
CA CL2
MG CL2
K CL2
NA CL2
FE CL2
AL_CL2
FE PYP
AL PYP
FE AO
AL AO
SI AO
FE CD
AL_CD
SO4 H2O
SO4 PO4
SO4_0
C TOT
N TOT
S~TOT
Reporting
units
meq/100g
II
it
n
II
meq/100g
n
it
it
meq/100g
it
n
n
'
ii
meq/100g
il
n
11
H
"
wt %
II
II
II
II
II
II
mg S/kg
II
mg S/L
wt %
11
n
CRDL
Units
0.003
0.005
0.002
0.003
0.015
0.003
0.005
0.002
0.003
0.15*
0.075C
0.15*
0.075C
0.25tf
1.25rf
0.0050
0.0008
0.0003
0.0004
0.0005
0.0011
0.005
0.005
0.005
0.005
0.005
0.002
0.002
0.50
0.50
0.025
0.010
0.005
0.001
a
mg/L
0.05
0.05
0.05
0.05
0.10
0.05
0.05
0.05
0.05
0.0075*
1.0517
0.0075*
1.05C
0.005*
0.025"
0.50
0.05
0.05
0.05
0.10
0.10
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.025
0.025
0.025
" Contract-required detection limit in reporting units and
instrument units (mg/L or ppm), respectively.
* For titration method, in meq/100g and meq,
respectively.
c For flow injection analysis (FIA) method, in meq/100g
and meq/L, respectively.
d For mineral samples and organic samples, in meq/100g
and meq, respectively.
NOTE: Detection limits not applicable for the physical
parameters, soil pH, and the SO4_2-32 parameters.
blanks from the two previous batches or initial
days, such that the calibration blanks were
determined over three separate days and
expressed as an "ongoing" IDL It was essen-
tial that the analytical laboratories analyze the
batches sequentially to establish the proper
IDL determinations. If a given batch IDL was
25
-------
invalid, the batch was reanalyzed for that
parameter.
At the end of the survey, independent
IDLs (IlDLs) were calculated by the QA staff
using the pooled variability data of the low
concentration DL-QCCS samples across all
batches. The IIDLs served as independent
checks on the reliability of the IDLs because it
was assumed that the variabilities associated
with the calibration blanks and DL-QCCS
samples wou!d be similar.
Laboratory analysis, of course, is only
one of many steps in the overall process of
generating raw data for a soil sample collec-
ted from tho field. For the .MASS, it was
possible to route low concentration audit
samples through the field sampling phase and
ai! subsequent phases of tha measurement
system, thereby allowing a system detection
limit (SDL) to be estimated. The SDLs were
calculated as three times the pooled standard
deviation of the low-range field audit (FAL)
samples from ai! mineral soil batches (see
Measurement Quality .Samples). Variability in
the FAL samplos encompassed uncertainty
introduced during sampling, preparation, ex-
traction, and analysis, and could include sam-
ple contamination incurred during any or all of
these phases. This type of detection limit
allowed a data user to identify when a given
soil sample had a measured concentration that
could be considered as statistically different
from a reagent blank or calibration blank that
had passed through the measurement system.
For the MASS, the FAL samples exhibit many
of the features that might be expected in a
hypothetical sci! blank sample; therefore, the
resulting variability would be expected to
parallel that of system-wide blanks.
Another important factor to consider in
the evaluation of detectability is the implication
of the calculated detection limits for assessing
the quality of the routine dataset. By estima-
ting the percentage of routine sample data
values that were greater than their correspond-
ing SDLs, specific parameters wera identified
that might not have been measured precisely
enough to satisfy the requirements of data
users. Use of the SDL information in this
manner is very similar to the establishment of
a "decision limit" that is of value to the
data modeling staffs. A high percentage of
samples tjreeHos" ;.'!,-;:; ;
sartly the rosuii of tt v
possible ill..i t!x-. ,.;;,>
the prima-".- 60'ji-. :; r-
FAL samp'ofr- UrfoJ /:
each para.?,.,-.
Precision
among r<:- ^ ..-. :;.:./*:...
soil chara';fr;, \,-'\c .,; p.-
feature of the offtne-'o;? <
tiered syst:--!;. >'-.;,
assessing v
the precisio :bi,-
precision , ''..::
multiplyin-j LI\ *.
batch yrer:;-,:c n
each >'ji ;:;,: \.-.
traticn ran," ;?:,. ,
of the da: - -'( -.
limits a:'& >, -
dard ddv:- ;."., ; ,
tier concur , \ *
the reqic,\ ;' ..-
tion fane .--.--
deviatio-i r: ,
setting re'-t-i'.
concentraii.".--
more diffK^ji;
with a I lion -..
o
.
(B
(0
a
Figure 2-2. ;:->.:.-:,
26
-------
The two-tiered evaluation technique is
characterized by a change in the slope of the
activity line at the knot. The coordinates of
the Y-axis expressed as an absolute standard
deviation and the X-axis expressed as a mean
value accurately show the percent BSD pic-
torially above the knot. This direct relationship
is a unique characteristic of the plots, which
include data from both mineral and organic
soil samples. Due to wide ranges in con-
centration between the two types of samples,
some plots were modified using scale breaks
in order to incorporate both mineral and or-
ganic soil data. The scale breaks resulted in
a corresponding change in the slope of the
activity line.
The precision MQOs for the MASS were
established for analytical laboratory within-
batch precision of mineral soils and organic
soils. Precision objectives were further defined
for three types of measurement quality samp-
les in accordance with their expected cumula-
tive uncertainty. Field, preparation, and analy-
tical measurement quality samples were used
for this purpose. The precision objectives for
the analytical samples are presented in Table
2-3. The precision objectives for the prepara-
tion samples were derived by multiplying the
lower tier values in the table by 1 and the
upper tier values by 1.5. The precision objec-
tives for the field samples were derived by
multiplying the lower tier values by 2 and the
upper tier values by 3. These factors origina-
ted through interaction between the EMSL-LV
and ERL-C staffs because it was known that
subjecting a soil sample to increasing sources
of measurement error usually results in higher
uncertainty.
Two-tiered precision objectives were not
required for the SAND, SILT, and CLAY para-
meters in the physical parameter group and for
the PHJH2O, PH_002M, and PH_01M parame-
ters in the pH parameter group. Instead, a
single-tier objective was established as a
standard deviation lor the total concentration
range of each of these parameters.
The within-batch precision component
measures the reproducibility of measurement
quality sample data fa*" a given batch of soil
samples analyzed by onp laboratory. Data
from the laboratory auuit samples were used
to assess the analytical within-batch precision
MQOs because the samples were not exposed
Table 2-3. Wlthln-Batch Precision Objectives for the
Analytical Measurement Quality Samples
Parameter
MOIST
SAND*
SILT*
CLAY*
PH H2O
PH 002M
PH_01M
CA CL
MG CL
K CL
NA CL
Al/CL
CA OAC
MG OAC
K OAC
NA_OAC
CEC CL
CEC OAC
AC_BACL
CA CL2
MG CL2
K CL2
NA CL2
FE CL2
AL_CL2
FE PYP
AL PYP
FE AO
AL AO
SI AO
FE CD
AL_CD
SO4 H20
S04 P04
S04_0-32i>
C TOT
N TOT
S TOT
Reporting
units
wt %
II
II
11
pH units
n
H
meq/100g
n
H
H
U
meq/100g
"
11
11
meq/100g
11
11
meq/100g
II
n
"
»
n
wt %
II
"
n
11
11
"
mg S/kg
11
mg S/L
wt %
ii
"
Mineral*
SD %RSD Knot
0.3
3.0
3.0
2.0
0.10
0.10
0.10
0.02
0.02
0.02
0.02
0.20
0.02
0.02
0.02
0.02
0.25
0.25
1.0
0.05
0.005
0.005
0.005
0.01
0.05
0.03
0.03
0.03
0.03
0.03
0.03
0.03
1.5
1.5
0.10
0.05
0.015
0.002
2.5
...
-~
10
10
10
10
15
10
10
10
10
10
10
15
5
10
10
10
15
15
10
10
10
10
10
10
10
10
10
5
10
10
10
12.0
_
0.20
0.20
0.20
0.20
1.3
0.20
0.20
0.20
0.20
2.5
2.5
6.7
1.0
0.05
0.05
0.05
0.07
0.33
0.30
0.30
0.30
0.30
0.30
0.30
0.30
15.0
15.0
2.0
0.50
0.15
0.02
Organic*
SD %RSD Knot
0.3 2.5
._
_
_
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.50
0.10
0.10
0.10
0.10
0.25
0.25
1.0
0.20
0.05
0.025
0.025
0.05
0.20
0.03
0.03
0.03
0.03
0.03
0.03
0.03
1.5
1.5
0.05
0.015
0.002
_
...
10
10
10
10
15
10
10
10
10
10
10
15
5
10
10
10
15
15
10
10
10
10
10
10
10
10
10
10
10
10
12.0
_
_
-~
1.0
1.0
1.0
1.0
3.3
1.0
1.0
1.0
1.0
2.5
2.5
6.7
4.0
0.50
0.25
0.25
0.33
1.3
0.30
0.30
0.30
0.30
0.30
0.30
0.30
15.0
15.0
...
050
0.15
0.02
* Within-batch precision by standard deviation below the
knot and by percent relative standard deviation above
the knot for mineral and organic samples, respectively.
b Particle size and sulfate isotherm parameters were not
analyzed for organic samples.
NOTE: For the preparation samples, lower tier objective
is same as value in table, upper tier objective is 1.5x
value; for the field samples, lower tier objective is 2x
value, upper tier objective is 3x value (Papp et al., 1989).
to sampling error and were assumed to have
negligible preparation error. The analytical
between-batch precision component measures
27
-------
the reproducibility of laboratory audit sample
data for different batches of soil samples
analyzed by one laboratory on different days
or at different times on any one day. The
within-batch variability is usually expected to
be somewhat smaller than the between-batch
variability, although there were no specific
MQOs defined for between-batch precision.
Previous DDRP surveys have shown that
the error variance for many parameters is a
function of analyte concentration (Van Remor-
tel et al., 1988; Byers et al., 1989). It was
discovered that no simple functional relation-
ship could be established except that the error
variance changes with concentration. For this
reason, it was necessary to express the error
variance by a step function where each step
value represents an error variance for a speci-
fic interval of concentration (see Appendix J).
The entire concentration range is divided into
several intervals in such a way that the error
variance within each interval is approximately
independent of the analyte concentration, i.e.,
the error variance within each interval does not
markedly change with analyte concentration
(Miah et al., 1988). For example, if F, . . Fk
are multiple intervals on which the step func-
tion takes values a,2 . . ak2, then the error
variance can be expressed as:
o2 = I,l=w X, of
where: X, is an indicator variable for the ith
interval that allows variance es-
timation across intervals, indepen-
dent of the individual intervals; X;
takes a value of 1 if the analyte
concentration falls in the ith inter-
val, and 0 otherwise.
The overall analytical within-batch preci-
sion, i.e., averaged across batches, was es-
timated using only mineral soil laboratory audit
samples because there were insufficient deg-
rees of freedom to make reliable estimates of
precision for the organic soil batches. The
step function was used for each parameter to
calculate a pooled variance from the B and Bw
horizon laboratory audit pairs of samples. For
comparison purposes, it was advantageous to
express the variance in terms of a standard
deviation in parameter reporting units, hence,
the square root of the pooled variance was
computed to generate a pooled standard
deviation, sp , across the lower range of con-
centration below the knot. The pooled stan-
dard deviation was calculated using the de-
grees of freedom for weighting purposes
according to the formula:
sp = [ I{i=1 } (df, s2) + 3^ df, ] 1/2
where: df, is the degrees of freedom for the
ith interval, and
s,2 is the error variance for the ith
interval.
The pooled RSD value was used to assess
data in the upper range of concentration above
the knot and was calculated in the same
manner except that the pooled standard devia-
tion was divided by the weighted mean and
the quotient was multiplied by 100.
The between-batch analytical precision
was derived by pooling the appropriate labora-
tory audit sample mean values within each
interval of concentration identified by the step
function. The intervals were primarily defined
by the distribution of data for the field dupli-
cates, with additional refinement from the
laboratory audit samples and preparation
duplicates (see Appendix J). The resulting
variability around the pooled mean of each
interval was weighted by the degrees of free-
dom and was subsequently pooled across all
intervals to estimate the average between-
batch precision observed during the survey.
These estimates were compared with the
within-batch precision estimates to determine
the relative magnitude of the batch error as a
treatment effect.
Accuracy
Accuracy is the level of agreement bet-
ween an observed value and the "true" value of
a characteristic or parameter. In this report,
data from the laboratory audit samples are
used to estimate analytical accuracy and data
from the field audit samples are used to
assess accuracy from a system-wide mea-
surement perspective. Natural soils were used
as audit samples because a procedure for
preparing synthetic soils was not established.
Although the true chemical composition and
physical characteristics of the audit samples
were not defined, acceptance criteria for
accuracy were established for the MASS based
on extensive prior analysis of the samples by
multiple laboratories in the two previous DDRP
surveys. The Bw and C horizon audit samples
were common to all three soil surveys. Data
for the B horizon audit sample, used only in
the MASS, were established during a prelimi-
nary evaluation.
28
-------
During the MASS, the audit samples were
not assumed to have a single "correct" refer-
ence value for each parameter. Instead, each
audit sample type was given a range of accep-
table values for each parameter in the form of
an accuracy window which was developed
using the previous DDRP analytical data. Each
accuracy window was initially defined by a
confidence interval placed around a biweight
estimation (Kadafar, 1982) of the mean value
across the two surveys. Additional refine-
ments to the intervals were made by the data
users and QA staff using ANOVA techniques
and exploratory data analysis. From a con-
tractual compliance standpoint, any observa-
tion that occurred within an accuracy window,
i.e., between the lower limit and upper limit,
inclusive, was considered to be unbiased and
acceptable. Any observation below the lower
limit or above the upper limit of the window
was considered inaccurate for compliance
purposes, as shown in Figure 2-3.
c
o>
o
0.
0)
U
R
Figure 2-3. Example of an accuracy window with
upper, lower, and reference values.
Accuracy was assessed by comparing
the MASS analytical data to the accuracy
window acceptance criteria defined for the
different sample types. The aspects of ac-
curacy investigated were bias, laboratory
differences, and laboratory trends. Analytical
bias was considered to be the quantitative
measure of inaccuracy used in the evaluation
of measurement uncertainty. Laboratory
differences and trends served as quantitative
and qualitative evaluations of analytical labora-
tory performance during the MASS.
During the MASS sample analysis, any
audit sample values that fell within the limits
of the accuracy windows were assumed to be
unbiased with respect to the batch acceptance
criteria. For data evaluation purposes, howe-
ver, bias was estimated from the mean of the
accuracy window, i.e., the reference value, for
each parameter. In this framework, analytical
bias for a specific audit sample was defined
as follows:
Bj = I(Yjk-Fy H- n,
where: E, is the bias for the jth audit sample,
Yjk is the average observed value for
the jth audit sample and k observa-
tions,
R, is the reference value [(U, + L) +
2] for the jth audit sample, and
nj is the number of samples of the jth
audit sample type used in the
assessment,
Bias as a function of analyte concentra-
tion was also investigated and it was deter-
mined that the magnitude of the bias in rela-
tion to imprecision was very small. As a
result, no empirical relationship between bias
and imprecision was observed. It was dee-
med unnecessary from a practical viewpoint to
investigate whether concentration dependency
was a demonstrated characteristic of bias.
Therefore, it was assumed that bias was
independent of concentration for the purposes
of this survey.
A calculated bias value for each para-
meter was compared to the corresponding
SDL and, if the bias was less than the SDL,
analytical bias was considered to be negligible.
The weighted average bias for the laboratory
audit samples was used as the bias estimate
for each parameter, such that:
B. = Z,
(n, B) + I n,
where: Ba is the average weighted analytical
bias for the ath parameter.
The relative effect of observations out-
side an accuracy window was evaluated with
respect to the bias estimate. The relative
magnitude of "outlier error" is assessed using
the ratio of outside-window bias to the total
bias estimate. The percentage of samples
that were outside the window also provided a
basis for identifying inordinate outlier effects.
The laboratory difference, d, is defined as
the difference between the mean of a repeated
measurement and the reference value, Ra, for
a given parameter. The individual differences
29
-------
were pooled across audit samples to establish
the performance of each analytical laboratory
in relation to the reference value. The O hori-
zon audit sample was omitted from the pool-
ing because all organic samples were analyzed
only by Laboratory 4. Hence, data for the B,
Bw, and C horizon audit samples were pooled
and weighted by the number of audit samples
of each type for each laboratory. This quantity
was divided by the reference value that had
also been pooled and weighted across audit
samples, and the quotient was multiplied by
100 to yield the average percent difference
from the pooled reference value.
The analysis of variance (ANOVA) was
used to determine significant differences
among the audit sample data reported by the
analytical laboratories. A two-way ANOVA
model (Steel and Torrie, 1960) was used to
test the significance of differences between
laboratory means pooled across audit samp-
les, based on a similar approach detailed in
Schmoyer et al. (1989). A pair-wise compari-
son at the .05 level of significance was per-
formed on the laboratory means using Schef-
fe's multiple comparisons test (Arnold, 1981).
The results of this test were used to identify
laboratory differences of significance and to
define the ranking order for those differences.
The analytical laboratories were subse-
quently evaluated by determining the signifi-
cance of their average deviation from the
reference value. The prior distribution, /v, of
the reference value for each parameter was
assumed to be uniform within each accuracy
window, and the variance, V, of the prior
distribution was calculated as:
V,(M) = (u, - g2 + 12
such that the value "12' is the lowest common
denominator for the expected squared error of
the prior reference distribution and the square
of its mean. The squared deviation, SL2, from
the corresponding reference value was es-
timated for each analytical laboratory as:
SM2 = I(Yjk - R)2 + n,
where: S4 is the squared deviation for the
Lth laboratory and the jth audit
sample.
To allow the presentation of squared deviation
estimates in parameter reporting units, the
square root of each SLj2 term was calculated
and defined as the average deviation from the
reference value.
The mean squared error, E^e2), serves
as the justification for comparison of the
squared deviation and prior distribution vari-
ance terms and was estimated for each analy-
tical laboratory as:
Using data pooled across audit samples, the
squared deviation was assessed with respect
to the variance of the prior distribution and
expressed as a ratio (Box et al., 1978). The
data from those laboratories showing a ratio
of 3:1 or greater were identified as different in
relation to the data from the other laboratories
for a given parameter.
The laboratory audit sample data were
also examined for trends which might be
caused by instrumental drift or other problems
that might have occurred at a particular labor-
atory over the course of the survey. For this
purpose, three-point moving averages (Massart
et al., 1988) of the audit sample data, grouped
by laboratory, were plotted in a temporal
sequence.
Representativeness
The evaluation of representativeness
included: (1) determining whether the routine
samples collected were representative of the
sampling class characteristics, (2) assessing
the sample homogenization and subsampling
procedures by measuring the ability of the
preparation laboratory to prepare representa-
tive subsamples from the bulk soil samples
collected by the sampling crews, and (3)
assessing the ability of the field duplicate
samples to adequately represent the range
and frequency distribution of analyte concen-
trations in the routine samples. This latter
consideration is important when assessing the
effects of concentration dependency on the
calculation of measurement uncertainty.
Soil types identified in the Mid-
Appalachian region were combined into sam-
pling classes that were expected to have
similar physical and chemical characteristics.
Each of the sampling classes could be samp-
led across a number of watersheds in which
they occurred. In this approach, a given soil
sample did not necessarily represent the
specific watershed from which it was col-
lected; rather, it contributed to a set of samp-
les which collectively represented a specific
sampling class for all watersheds within the
30
-------
sampling region. The sampling crew leader
selected a sampling site representing the
designated sampling class and vegetation
class within the designated watershed accord-
ing to the sampling protocols (Appendix A in
Kern and Lee, in press). Both the qualitative
and quantitative data collected were intended
to be representative of so'ls in the region.
During the survey, homogenization of the
samples at the preparation laboratory in
conjunction with subsequent homogenization
at the analytical laboratory were specified to
ensure that a uniform and representative
source of sample was available from which to
select aliquots for analysis. Precision data
from the preparation duplicates were used to
assess the representativeness of samples
created by the preparation laboratory homo-
genization and subsampling procedures, while
data from the laboratory audit samples and
analytical duplicate samples were used to
assess representativeness of the aliquots
used for analysis.
The Kolmogorov-Smirnov two-sample
test, i.e., the KS-statistic, was used to es-
timate the maximum distance between the
cumulative distributions of two data sets
(Conover, 1980). Data sets of the routine
samples (RS) and the field duplicates (FD)
were tested to determine whether the distribu-
tions were similar. The significant KS-statis-
tics, i.e., significant at the .05 level, were
defined according to the critical value, Vc , for
the FD/RS comparison. The critical value is
based on a sample size n, and n2 for the data
sets being compared, where:
Vc = 1.36 [(n, + nj + (n, n2)] 1/2
The algorithm yielded a critical value for FD/RS
of 0.185. If the KS-statistic exceeded the
critical value for a particular pairing, the FD
data set was considered to be possibly non-
representative of the distribution of routine
samples. It was assumed that representative-
ness of the preparation duplicates (PD) data
set in relation to the RS data set would mimic
the FD/RS comparison because the preparation
duplicates were subsampled directly from the
field duplicates and their associated routine
samples. Therefore, the KS-statistic was not
applied to the PD data set.
Completeness
Soil sampling protocols in the MASS
specified the sampling of 100 percent of the
designated pedons and their component hori-
zons to a depth of two meters, with the excep-
tion of certain undecomposed organic hori-
zons, intermittent horizons, horizons less than
three centimeters thick, and horizons contain-
ing dense fragipan material or a prohibitive
amount of stones and boulders (Kern et a!.,
1988). The soil preparation protocols specified
that each batch of samples sent to an analyti-
cal laboratory should include the proper array
of measurement quality samples. The prepar-
ation laboratory was required to complete the
specified analyses and processing tasks for
100 percent of the samples that satisfied the
sample receipt verification criteria. Each batch
of samples sent to a contracted analytical
laboratory was required to contain a specified
configuration of routine samples and measure-
ment quality samples.
The analytical completeness objective
specified the analysis of 90 percent or more of
the parameters for all samples analyzed (Papp
et al., 1989). It was possible to attain 100
percent completeness if a sufficient quantity of
each sample was available to complete all
analyses and subsequent requested reana-
iyses. Analytical completeness was evaluated
using routine data from the MASS verified and
validated data bases.
Comparability
Comparability of data from the three
DDRP surveys was determined to be a com-
plex issue with several levels of detail which
should be considered, such that:
Level 1 data comparability is established
on the basis of statistical evaluation
methods, measurement quality samples,
and the field sampling, preparation, and
analytical methods used.
Level 2 data comparability is established
by the acceptability and useability of the
verified data bases as defined by the data
users.
Level 3 data comparability allows the
direct quantitative comparison of data for
each parameter of interest.
Comparability assessment begins, there-
fore, with the Level 1 data comparison. If the
31
-------
statistical techniques and uncertainty terms
are comparable across surveys, the different
types of measurement quality samples used in
the surveys are compared. If identical types
of samples were used, the field sampling,
sample preparation, and sample analysis
methods are compared. If the methods used
in the surveys were different, the data are not
directly comparable. If identical methods were
used, Level 2 criteria are applied to identify the
least restrictive precision/accuracy quality
.objectives and estimates associated with the
different data bases. It is assumed that the
least restrictive criteria defines the accepta-
bility of ali data used in the comparison. In
addition, a confirmation of the useability of
data generated under the least restrictive
criteria defines those data that are recom-
mended for use by the community of data
users. Therefore, data that satisfy the Levels
1 and 2 criteria can be quantitatively compared
at Level 3. Data not satisfying the Level 2
criteria are highlighted for the data user by
outlining the comparability restrictions of the
data. Ultimately, the data user decides which
comparisons can be made and which user
criteria are appropriate for a given data analy-
sis. An experimental Level 3 assessment,
using the data without regard to Levels 1 and
2 criteria, can be performed to confirm or
reject the hypothesis that highlighted data at
the first two levels are not directly comparable.
The verified analytical data bases for the
NE, SBRP, and MASS surveys were used for
the comparability assessments described in
Section 3. While it is recognized that accuracy
is the preferred measure of data comparability,
precision is sometimes the only attribute that
can be compared. Accuracy, or its best es-
timate, is used in comparability assessment
wherever possible.
As an independent assessment of com-
parability, an interlaboratory comparability
study was conducted to determine whether the
analytical data for the natural soil audit samp-
les analyzed by the contract analytical labora-
tories for the DDRP soil surveys are compara-
ble to data for the same samples analyzed by
other analytical laboratories in the United
States and Canada. A summary of the find-
ings of this study is provided in Section 3, and
a detailed accounting of the study is contained
in a separate report (Fenstermaker et a!., in
preparation).
Uncertainty Estimation
The term "uncertainty describes the sum
of all quantifiable errors associated with a
particular portion of a measurement system or
population of interest. Measurement uncertain-
ty in the MASS data was controlled through
clearly-defined protocols and technical systems
audits, and by the introduction of measure-
ment quality samples during each sample
measurement phase. Each quantifiable com-
ponent of the precision and accuracy attri-
butes was identified individually for the three
different measurement phases of the program
(sampling, preparation, and analysis) and
collectively for overall measurement uncertainty
on a parameter-by-parameter basis. In addi-
tion, an estimate was made, using the routine
sample data set, of the overall data uncertain-
ty in the population of soil samples collected
from the Mid-Appalachian region during this
survey.
Statistical Model
Standard operating procedures and
specified protocols were followed in each
measurement phase of the MASS. Depending
on its limitations or assumptions, each proce-
dure induced a random error for each physical
or chemical characteristic of a soil sample.
The sum of these errors can be defined as the
data collection error, which is treated as a
random variable. This variable can be used in
an additive model (Cochran, 1977) and serves
as an estimate of uncertainty in a data set,
such that:
y = \i +
where: y is an observed sample characteris-
tic,
p is the "true" sample characteristic,
and
e is the data collection error, which is
assumed to be the sum of the
errors generated by the indepen-
dent measurement phases.
The identification of the error distribution
throughout the measurement phases requires
a large number of replicate measurements
which, from a budgetary and logistical stand-
point, can be restrictive. However, a relatively
small number of observations can be used to
estimate the first two moments, i.e., the mean
and variance, of the data collection error. The
moments are sufficient to estimate inaccuracy
32
-------
and imprecision of the routine data in an addi-
tive model. For this report, the standard devi-
ation (SD) in reporting units and the relative
standard deviation (BSD) in percent were used
to assess the precision components. The
analytical bias was considered to be the qua-
ntitative measure used to assess the accuracy
component.
The sum in quadrature of the imprecision
and inaccuracy error terms represents the
measurement uncertainty in the different mea-
surement quality sample data sets, and in-
cludes both measurement and population un-
certainty in the routine sample data set. In
order to make effective use of the MASS data,
the evaluation of measurement uncertainty in
relation to the overall data uncertainty is es-
sential.
The expected squared error was used as
an estimate of uncertainty and was calculated
as the sum of the error variance and the squ-
are of the bias term described previously in
this section and in Appendix J. The error
terms can be defined only for a given analyte
concentration range because the error variance
changes with analyte concentration; however,
the entire concentration range can be parti-
tioned into intervals and a step function is
applied to those intervals in such a way that
each step value is the squared error, Ej(e2), for
a specific interval of the entire concentration
range (Miah et al., 1988).
For most data interpretation uses, it is
preferable to convert the squared uncertainty
values into the standard reporting units of the
parameters of interest. This conversion is
made simply by calculating the square root of
the squared error. To evaluate the data un-
certainties associated with the routine data,
the VEjfe2) terms from each interval were poo-
led and weighted by the proportion of routine
sample concentrations within the correspond-
ing intervals. The resulting uncertainty esti-
mate, 8 or delta, is defined as:
5 = I PiVE^e2) + I p,
where: 8 is the pooled uncertainty for all
intervals over the concentration
range,
P, is the proportion of routine sam-
ples in the ith interval, and
E,(e2) is the mean squared error for the
ith interval, and represents the
sum of the errorvariance and the
square of the bias for the ith inter-
val.
Occasionally, a dearth of measurement
quality sample precision data within one or
more intervals made it impossible to calculate
a variance for that portion of the concentra-
tion range. In those cases, 8 is a conditional
estimate of uncertainty because the sum pro-
portion of samples across all intervals repre-
sented is less than 1.0 and the resulting con-
ditional uncertainty value usually is overesti-
mated. In other situations, certain portions of
the precision data were incomplete, e.g., for
between-batch precision. In these cases, 8 is
considered to be a partial estimate because
specific additive components are missing, and
the resulting numerator variance term is un-
derestimated. As a result of both possibil-
ities, particular intervals were occasionally
excluded from the uncertainty calculations for
some parameters.
Sampling Class/Horizon Groups
The soil mapping units identified in the
watersheds of the Mid-Appalachian region
were too numerous to allow statistically-
based sampling of all soil components. As an
alternative, soils were grouped into sampling
classes, shown in Table 2-4, that were be-
lieved to have similar physical and chemical
properties in relation to soil responses to
acidic deposition. Field sampling sites were
randomly selected from the areas of occur-
rence of each sampling class because the
objective of the sampling was to characterize
sampling classes rather than individual water-
sheds or soil types. Prior to the MASS, the
degree of heterogeneity of physical and chem-
ical properties for individual sampling classes
and soil horizon types was not known. The
QA data analyses of the two previous DDRP
surveys, however, have shown that measure-
ment uncertainty in the routine samples is
confounded with the sampling class/horizon
heterogeneity in the calculated imprecision
estimates. Because these two characteristics
vary widely with respect to each other, the
ratio of measurement uncertainty to overall
sampling class/horizon variability is not con-
stant but can be ascertained through a series
of data set comparisons.
In estimating the relative importance of
measurement uncertainty, mean values and
33
-------
Table 2-4. Sampling Classes Used to Group Different
Soil Types
Table 2-5. Primary Horizon Type* for Sampling
Class/Horizon Groups
Sampling
class
Soil types
CMS Entisols: mine spoils
CFP Entisols: somewhat poorly drained and
wetter
FLW Entisols: moderately well drained and
drier
Inceptisols, Spodosols: no fragipan, flood-
ed
HST Histosols: all
BNS Inceptisols, Spodosols: no fragipan, not
flooded, shallow
BMK Inceptisols, Spodosols: no fragipan, not
flooded, moderately deep, skeletal
BMD Inceptisols, Spodosols: no fragipan, not
flooded, moderately deep, non-skeletal
BND Inceptisols, Spodosols: no fragipan, not
flooded, deep
BXW Inceptisols: Ochrepts
BXP Inceptisols: Aquepts
TWM Ultisols, Alfisols: no fragipan, moderately
well drained and drier, moderately deep
TWO Ultisols, Alfisols: no frapipan, moderately
well drained and drier, deep
TPD Ultisols, Alfisols: no fragipan, somewhat
poorly drained and wetter
TXP Ultisols, Alfisols: fragipan, somewhat
poorly drained and wetter
TXW Ultisols, Alfisols: no fragipan, moderately
well drained and drier
Horizon
type
A
Ag
Ap
AE
Bhs
Bt
Btg
Btx
Bw
Bwg
Bx
Bxg
BA
BC
C
Cg
Cr
Cx
E
EB
Oa
Oe
Routine samples Sampling classes
n3 % represented6
65
3
30
9
6
106
11
27
157
11
47
10
38
42
162
25
9
9
29
7
19
22
7.7
0.4
3.6
1.1
0.7
12.6
1.3
3.2
18.6
1.3
5.6
1.2
4.5
5.0
19.2
3.0
1.1
1.1
3.4
0.8
2.3
2.6
14
2
11
6
4
5
3
3
11
3
4
3
11
9
15
5
4
4
10
5
7
9
" N = 799 (844 total routine samples minus 45 samples
classified as "special interest" samples in the verified
data base.
b There are 148 sampling class/horizon groups.
variances for the sampling class/horizon grou-
ps were determined for each soil parameter
and quantitatively compared to the uncertainty
estimate generated by the measurement
system. There are as many as 22 primary
horizon types associated with each of the 15
sampling classes, forming a total of 148 dis-
tinct sampling class/horizon groups. Table
2-5 presents the number and percentage of
primary horizon types and the number of
sampling classes associated with each hori-
zon.
For this report, four delta values have
been calculated: 8, , 82 , 83 , and S4 values.
The uncertainty estimates are generally ex-
pected to increase with increasing sources of
confounded error. The 84 values are calculated
from the sampling class/horizon groups of
routine samples, while the other three 8 values
are calculated using data from the measure-
ment quality samples described below.
It has been proposed that a measured
value can be considered as essentially error-
less for most uses if the uncertainty in that
value is one-third or less of the permissible
tolerance for its use (Taylor, 1987). The QA
staff examined the relation of measureme.it
uncertainty to overall data uncertainty in the
routine samples, where the overall measure-
ment uncertainty was based on the 83 values
calculated from the field duplicate samples
(see Appendix C). The magnitude of this
uncertainty in relation to the associated 84
data uncertainty values generated from the
sampling class/horizon groups provides the
data user with a basis for assessing the
uncertainty contribution of the measurement
system. For the purposes of this assessment,
the data uncertainty confounded in the sam-
pling class/horizon groups is considered to be
a surrogate for user-defined error tolerance
values. Where 83 is one-third or less of 84,
measurement uncertainty is a negligible con-
tribution to the overall data uncertainty.
34
-------
logram
sample
.-'* com-
,-iproxi-
t:oora-
. of dif-
:(:-i tO
t'-'US
- .";>>y and
'.;-', farnpies
'; .;* msas-
H- -;s;tro! mea-
'.;:.. -:'.'.al,hes were
r'rc ';;") contract
:v-ai8".ir>. Also, the
!i ;-.^?np!o? was bal-
r:--n\ :!/ c; crews and
; :;,u iiv.' f\ nearly
' .;;^;c«8d by one
,! ,- e-,di iabora-
. '".'jrnic batches
-?i'.s9 :he Sabora-
n
PB PE sample
double-blind to the sampling crews, prepara-
tion laboratory, and analytical laboratories. By
definition, a blind sample has a concentration
range that is unknown to the analyst, whereas
a double-blind sample cannot be distinguished
from a routine sample and has a concentration
range that is unknown to the analyst (Taylor,
1987). These samples provided an indepen-
dent check on the QC process and were nsed
to evaluate measurement quality for any given
batch or across all batches. Important chara-
cteristics of the QE natural soiS audit samples
include their similarity to routine samples in
matrix type and concentration level, homogen-
eity and stability, and defined accuracy win-
dows. The QE di'plicate samples were sub-
sampied from specific routine samples. Every
QE sample had a specific purpose in the data
assessment scheme, as described below and
in Table 2-7.
The number and percentage of QE and
routine samples shipped to the analytical
laboratories are shown in Table 2-8. A de-
scription of each QE sarnpte is provided in the
following subsection,
Field Duplicate (FD) SampSe
For every third pedon sampled, a FD
sample was collected from a horizon selected
at random by the sampling crew leader in such
a way that the range of possible soil horizon
types encountered was sampled over the
course of the survey. The sampling crew
placed alternate trowels-fu!! of soil into separ-
ate bags to produce a pair of samples (one
routine, one duplicate) from one specific hori-
zon within a specific pedon. The FDs were
expected to contain the largest amount of
confounded uncertainty of all measurement
quality samples, as noted previously in Table
2-3. Data from the FDs were not specifically
intended to control field sampling imprecision
but were used to control imprecision at the
preparation laboratory and analytical labora-
tories. Individual pairs were used primariy to
estimate system within-batch imprecision, and
were pooied to provide the within-batch com-
ponent of the 83 overall measurement uncer-
tainty, i.e., the confounded uncertainty as-
sociated with sampling, preparation, and
analysis.
.'' the?.'-: >.^surement
,'" -:;!- ' blind or
35
-------
Table 2-6. Distribution of Field Measurement Quality Samples and Routine Samples Among the Sampling Crews
and Analytical Laboratories
Sampling
crew
Field measurement quality samples3
L1
L2
L4
FD C B Bw O FD C B Bw 0 FD C B Bw O
Percent
by crew
Routine sample5 ,
L1
L2
1<
Percent
by crow
WV01
WV02
PA01
PA02
PA03
Percent by Lab
11200
22220
33240
22220
11200
31.0
22220
22220
22220
22220
11020
31.0
22220
22220
32223
32223
11020
38.0
17.2
20.7
27.6
24.1
10.4
100.0
29
46
85
65
35
30.8
57
64
59
52
35
31.6
51
62
78
87
39
10.2
20.4
123
jO.O
" The laboratory measurement quality samples were present in the batches in the same quantity a« tlr-i Moid
samples, except that the PDs are considered in place of the FDs (PD = 2 x FD). (Total = 116 (FD-29. C -P7, B-i
Bw=28, O=6)
b Number and percentage of routine soil samples, respectively. (Total = 844).
Table 2-7. Status and Assessment Purpose of the Measurement Quality Soil Samples
Status of sample*
Sample
type
FD
FAP/FAO
FAL
PD
LAP/LAO
LAL
AD
OCAS
QA
staff
B
K
K
B
K
K
K
Sampling
crew
B
B
B
Preparation
laboratory
B
B
B
B
Analytical
laboratory
DB
DB
DB
DB
DB
DB
B
K
System
PW
PW.A
D,A
Assessm
Sampling
PW.R
PW.A
D,A
entjDurpose6
Preparation
PW
PW.A
D,A
PW.R
Analysis
PW
PW,A
A
PW
PW.PB.A
D,PB,A
PW
A
" K = known concentration, B = blind (unknown concentration), DB = double blind (unknown identity and concentration).
6 0 = detectability/contamination, PW = within-batch precision, PB = between-batch precision, A = accuracy,
R = representativeness.
Table 2-8. Number and Percentage of Quality
Evaluation and Routine Samples
Sample type
FD
FAP/FAO
FAL
PD
LAP/LAO
LAL
Routine6
Mineral
n
27
54
27
54
54
27
795
samples
percent3
2.4
4.9
2.4
4.9
4.9
2.4
72.1
Organic
n
2
6
0
4
6
0
49
samples
percent3
0.2
0.5
0.0
0.4
0.5
0.0
4.4
Percentage based on N = 1,105 (total quality
evaluation and routine samples).
Does not include quality control, comparability, or
special study samples.
Field Audit (FAR or FAO) Sample
The FAR samples are median-range B or
Bw horizon field audit samples that the QA
staff sent in pairs to the sampling crews. The
FAO samples are median-range O horizon field
audit samples that the QA staff sent in tripli-
cate to the sampling crews. The FAP and FAO
samples were processed as if they had just
been obtained from the excavated pit, i.e., the
samples were sieved and bagged in the same
manner as the routine samples at the same
sampling sites where FD samples were col-
lected. The field audit samples subsequently
underwent the same soil preparation and
analysis steps as the routine samples. The
FAP and FAO samples were used in the analy-
tical phase to estimate system precision and
accuracy for mineral and organic batches,
36
-------
MINERAL SOII. BATCHES
1
1
SAMPLING
PREPARATION
ANALYSIS
Routine
sample
PD
PD
Routine
Routine
r~
FD
nples
1
FAL
FD (C)
i r
n
FAL
PD (C)
_r
FD
"i
FAL
PD (C)
2FD/Routine 3 4
PD/Routine FD/PD FAL
i i i
|( B/Bwjl
pair |
1 F;
(B/
i 'Ft
(B/
PE
| no preparation (
ipl , . 1 1 - . ILAP! - . 1 1 |
'Bw) LAL (B/Bw) QCAS
lir (C) pair (A)
Bw) LAL (B/Bw) QCAS
lir (C) pair AD (A)
PAP* LAL' LAP' AD' QCAS"
FD = field duplicate; FAL = low-range field audit; FAP = field audit pair; PD = preparation duplicate; LAL = low-
range laboratory audit; LAP = laboratory audit pair; QCAS - quality control audit sample; AD =* analytical duplicate.
ORGANIC SOIL BATCHES
i
1
SAMPLING
PREPARATION
ANALYSIS
Routine
sample
PD
fr]
FD th
1 1 &
j Routine FD PD | thj
PD
Routine FD PD thi
PD/Routine
ioT-,
3)
ree
no preparation
«>1 | ILAO' | i 1 ' |
-) (0) QCAS
ree three (Oa)
>) (0) QCAS
-ee three AD (Oa)
FD/Routine 3 5 7*89
FD/PD FAO LAO AD QCAS
FD = field duplicate; FAO = field audit triplicate (organic); PD = preparation duplicate; LAO » laboratory audit
triplicate (organic); QCAS = quality control audit sample; AD = analytical duplicate.
System within-batch precision.
Within-batch precision.
Within-batch precision.
System detection limit.
System accuracy and within-/between-batch precision.
Analytical accuracy.
Analytical accuracy and within-/between-batch precision.
Within-batch precision.
Analytical accuracy.
Figure 2-3. Quality evaluation and quality control soil samples for mineral and organic batches.
37
-------
respectively. The B and Bw horizon FAP
sample pairs were alternated between batches
(see Table 2-6).
Low-Range Field Audit (FAL) Sample
The FAL sample is a low concentration C
horizon field audit sample that the QA staff
sent with each pair of FAP samples to the
sampling crews. The FAL sample was handled
in the same general manner as the FAP samp-
les, and was used to identify contamination in
the measurement system and served as an
additional check on system-wide accuracy.
Pooled FAL data from all mineral soil batches
provided an estimate of the system detection
limit (SDL), which was calculated as three
times the pooled standard deviation of the FAL
samples.
Preparation Duplicate (PD) Sample
A pair of PD samples, one split from the
FD sample and one split from its associated
routine sample, was created at the preparation
laboratory immediately following sample homo-
genization and was placed in each batch. The
PDs were used to identify the soil preparation
uncertainty component of measurement uncer-
tainty and as an independent check on within-
batch precision of the analytical laboratories.
These samples characterized the measurement
uncertainty introduced during and after homog-
enization/subsampling at the preparation
laboratory, but did not include uncertainty
related to sample drying or sieving, or any
contamination occurring prior to sample homo-
genization. The uncertainty attributed to
sample preparation was derived after identify-
ing and isolating the uncertainty from sample
analysis. The preparation laboratory activities
were monitored through the evaluation of PD
precision in the initial sample batches. Preci-
sion data from the PD samples were later
used by the QA staff to estimate the 82 con-
founded uncertainty associated with homogeni-
zation/subsampling at the preparation labora-
tory and with sample analysis at the analytical
laboratories.
Laboratory Audit (LAP or LAO) Sample
The LAP samples are median-range B or
Bw horizon laboratory audit samples, identical
to the FAP samples, that the QA staff sent in
pairs directly to the preparation laboratory for
inclusion in each mineral soil batch sent to an
analytical laboratory. The LAO samples are O
horizon laboratory audit samples, identical to
the FAO samples, that were sent in triplicate
by the QA staff directly to the preparation
laboratory for inclusion in each organic soil
batch. The LAP and LAO samples were used
to assess analytical within-batch precision and
accuracy and were used in combination with
the FAP and FAO samples to identify sources
of contamination or method error. Comparison
of the laboratory audit sample data across
batches allowed the QA staff to identify inter-
laboratory differences and to assess the be-
tween-batch component of analytical precision.
Pooled LAP and LAL data were used to es-
timate the 8, uncertainty associated with
sample analysis.
Low-Range Laboratory Audit (LAL) Sample
The LAL sample is a low concentration C
horizon laboratory audit sample, identical to
the FAL, that the QA staff sent directly to the
preparation laboratory for inclusion in each
mineral batch sent to an analytical laboratory.
The LAL samples were used primarily to ass-
ess analytical accuracy at low analyte con-
centrations. In addition, the LAL samples
were evaluated in conjunction with the FAL
samples to identify the source and magnitude
of any suspected sample contamination.
Quality Control Samples
In order to produce data of consistently
high quality, the contract laboratories were
required to analyze certain types of QC sam-
ples that were known to the laboratory staff
and that could be used by the analysts to
identify and control both random and sys-
tematic measurement errors. The QC samples
were non-blind samples procured under con-
tract to assist the laboratories in satisfying
the laboratory MQOs and included soil samp-
les, e.g., analytical duplicates, and non-soil
samples, e.g., reagent blanks. Some of the
QC samples were created by the laboratories
to evaluate instrument calibration and standar-
dization and to identify problems such as
contamination or analytical interference.
Each QC sample had certain specifica-
tions, as listed in Table 2-9, that were required
to be satisfied before data for the parameter
or batch were accepted. The QC samples
38
-------
Table 2-9. Quality Control Requirements for the Analytical Sample Batches
Characteristic
Parameter
Requirement
QC Audit Sample (QCAS)
Analytical Duplicate (AD)
Calibration Blank
number analyzed
measured
Reagent Blank (mean)
QC Check Sample (QCCS)
number analyzed
measured
Detection Limit QCCS
(DL-QCCS)
ALL
ALL
ALL*"
PH_H20, PH_002M, PH 01M
SO4_2 - 16
SO4_32
ALL OTHERS"
ALL
SO4_0 - 32
ALL OTHERS"
ALL"
Within accuracy window
Satisfies precision limits
((samples in batch/10) + 1)
<.CRDL
4.5 <. mean <. 7.5
± 3% of theoretical
± 2% of theoretical
< CRDL
((samples in batch/10) + 1)
± 5% of theoretical
i 10% of theoretical
i 20% of theoretical
Matrix Spike Recovery
Absorption Spike Solution
Ion Chromatography Resolution
Instrument Detection Limit (IDL)
Sample Weight
Sample Volume
S04_0 - 32
ALL OTHERS6
S04_0
S04_2 - 16
SO4_32
CA_CL2
S04JH2O, SO4_P04, SO4_0-32
ALL"-C
ALL
ALL
95% <. x <. 105%
90% <. x <. 110%
< CRDL
± 3% theoretical value
t 2% theoretical value
± 5% theoretical value
Resolution >. 60%
.< CRDL
Measured = ± 5% method
Measured = method
Except moisture, particle size, and pH parameters.
" Except SO4_2-32.
c Except AC BACL.
d Except FS1, CLAY, and AC_BACL.
* Except moisture, particle size, AC_BACL, and SO4_0,4,16 parameters.
were analyzed by each laboratory and allowed
the laboratory manager and the QA staff to
assess whether the physical and chemical
analyses were in control. The QC samples
used in the MASS are detailed in the following
subsections.
Quality Control Audit Sample (QCAS)
The QCAS is a median-range A or Oa
horizon laboratory audit sample that the QA
staff sent directly to the preparation laboratory
for inclusion as the 15th sample in each miner-
al or organic batch, respectively. This sample
had predetermined accuracy windows that
were established from reference values and
confidence intervals computed from previous
DDRP analyses. The windows for each para-
meter were provided to the analytical labora-
tories and were programmed in the computer
data entry system. The QCAS was used to
control bias and to reduce between-laboratory
and between-batch measurement uncertainty.
Data for the QCAS were evaluated by a bet-
ween-batch control chart to ensure that the
results were within acceptable inaccuracy
limits. If the QCAS did not satisfy the ac-
curacy criteria, the affected batch of samples
was re-extracted and reanalyzed for the para-
meter in question.
Analytical Duplicate (AD) Sample
A duplicate subsample of the 25th
sample in each batch (triplicate for pH para-
meters) was selected as the AD sample at the
39
-------
analytical laboratory and was used by the
laboratory manager to ensure that the analyti-
cal within-batch precision MQOs were satis-
fied. Acceptance criteria for the replicate
analyses were based on concentrations, below
the knot, being less than or equal to an ab-
solute standard deviation, or less than or
equal to the percent RSD above the knot. If
data for an AD sample exceeded these criteria,
the batch analysis was halted and the problem
was investigated. Subsequently, additional
samples were analyzed in duplicate or tripli-
cate until the analytical system was brought
under control. At this time, the entire batch of
samples was reanalyzed for the parameter in
question.
Calibration Blank Sample
The calibration blank was generally
defined as a zero mg/L standard containing
only the matrix of the calibration standards
with no analyte present. The two exceptions
to this definition were: (1) the calibration blank
for CA_CL2 which was prepared in deionized
distilled water instead of calcium chloride
solution, and (2) the calibration blank for
AC_BACL which had a concentration requiring
26.0 to 36.0 ml, i.e., 1.3 to 1.8 meq, of titer to
neutralize the buffered base.
One calibration blank per batch was ana-
lyzed immediately after the initial instrument
calibration to check for significant baseline
drift. During batch-wide analysis, calibration
blanks were run before the first sample, after
every 10 samples, and after the last sample of
the batch. The observed concentration of the
blank was required to be less than or equal to
the CRDL, except for AC_BACL If the con-
centration was greater than the CRDL, the
instrument was rezeroed and the calibration
curve was rechecked. Any samples bracketing
the "bad" calibration blank were reanalyzed.
The within-batch data from the calibration
blanks were used to calculate ongoing IDLs
across batches.
Reagent Blank Sample
When an analytical method required
sample preparation, e.g., for extraction, a
reagent blank for each group of samples was
prepared and analyzed. A reagent blank is a
sample composed of all reagents in the same
quantities used in preparing an actual sample
for analysis. Reagent blanks underwent the
same digestion or extraction procedures as an
actual sample, in the absence of a soil matrix
The concentration of the reagent blank was
required to be less than or equal to the CRDL
for all except the particle size, pH, exchange-
able acidity, extractable calcium, and sulfate
isotherm (excluding zero mg SfL) parameters.
No check was performed on the reagent blank
for particle size analysis. All three pH parame-
ters had reagent blanks with values between
4.5 and 7.5, inclusive. The ACJJACL and
CA_CL2 reagent blanks were requiTed to have
concentrations between 1.3 and 1.8 meq and
76 and 84 mg/L, respectively. The sulfate
adsorption isotherm analysis used reagent
blanks that were within ± 3% of the theoretical
concentration for the SO4_2-16 parameters and
within ± 2% of the theoretical concentration for
the S04J32 parameter. If a reagent blank
concentration exceeded the allowable limits,
the source of error was identified and elimina-
ted. A new reagent blank was prepared and
the affected batch of samples was re-extract-
ed and reanalyzed for that parameter.
Quality Control Check Sample (QCCS)
Immediately after an instrument calibra-
tion, a QCCS containing the analyte of interest
at a known concentration in the mid-calibration
range was analyzed. The QCCS was analyzed
to verify the calibration curve prior to any
sample analysis, after every 10 samples, and
after the last sample of a batch. The QCCS
solution was required to be selected from a
stock solution different than that used in the
preparation of standards for the instrument
calibration. The QCCS data were plotted at
each laboratory on a QC chart which showed
if the QCCS analysis exceeded the established
limits, which were defined as a ± percentage
interval for each batch and as 95- and 99-
percent confidence intervals (warning and
control limits, respectively) for the between-
batch results. If a QCCS value exceeded a ±
10 percent interval around the theoretical
concentration (± 5 percent for the sulfate
isotherms) or was outside the control limits on
the QC chart, the QCCS was considered to be
unacceptable. The instrument was recalibrat-
ed and all samples subsequent to the last
acceptable QCCS were reanalyzed.
Detection Limit Quality Control Check
Sample (DL-QCCS)
The DL-QCCS is a low-range QC sample
that contained the analyte of interest at a
specified concentration, usually two to three
times the CRDL The purpose of the DL-QCCS
was to eliminate the necessity of formally
40
-------
determining the detection limit on a daily
basis. The sample was run once per batch
and the measured value was required to be
within 20 percent of the theoretical concentra-
tion. If this criterion was not satisfied, the
source of error was identified and corrected,
and all samples in the batch were reanalyzed
before routine sample analysis was resumed.
Matrix Spike Sample
A matrix spike is a QC sample used to
determine possible chemical interferences that
might affect an analytical result. For liquid
samples, e.g., cation extractions, one matrix
spike sample was prepared for each batch
and each parameter by spiking an aliquot of a
solution with a known quantity of analyte prior
to analysis. The spike concentration was
approximately equal to the endogenous level or
10 times the detection limit, whichever was
larger. The volume of the added spike was
negligible, i.e., less than or equal to one per-
cent of the sample aliquot volume.
For solid samples, e.g., aliquots for
CJTOT, one matrix spike sample was prepared
for each batch and each parameter by adding
a known weight of material containing the
analyte of interest to a sample of known
weight. The spike concentration was twice the
endogenous level or 10 times the detection
limit, whichever was larger. The concentration
of the matrix spike was not allowed to exceed
the linear range of the instrument. Although it
was not negligible, the weight of the spike
material was considered negligible for calcula-
tion purposes.
Matrix spike recoveries were required to
be between 95 and 105 percent, inclusive, for
the sulfate isotherms and between 90 and 110
percent, inclusive, for all other spiked parame-
ters. If the spike recovery was not accep-
table, two additional, different samples in the
batch were spiked with the analyte of interest.
If the recoveries of the two additional spiked
samples failed to satisfy the criteria, the entire
batch of samples was required to be reana-
lyzed for that parameter by the method of
standard additions. All matrix spike data for
the MASS satisfied the criteria and the method
of standard additions was never implemented.
Adsorption Spike Solution
The adsorption spike solutions were: (1)
the extraction solution used in determining
CA_CL2, or (2) the extraction solution stan-
dards used for generation of the SO4_0-32
parameters. For each batch of samples, the
adsorption spike solutions were analyzed in
triplicate. Measured concentrations for the
solution were 76 to 84 mg/L Ca2+ for the
0.002M calcium chloride extraction. Spike
solutions for the sulfate isotherms had values
less than or equal to the CRDL for SO4_0, ±
3% of the theoretical concentration for SO4_2-
16, and ± 2% of the theoretical concentration
for SO4_32. If any adsorption spike solution
did not satisfy these criteria, a new batch of
extraction solution was prepared and any
samples that were equilibrated using the "bad"
solution were re-extracted and reanalyzed.
Ion Chromatography Resolution Sample
An ion chromatography resolution test
was performed once per analytical run by
analyzing a standard that contained concentra-
tions of approximately 1 mg/L each of sulfate,
phosphate, and nitrate. The analysis was run
to ensure "clean" separation of eluant peaks
for the individual ions measured, which is
necessary for proper quantification of sulfate.
The resolution between the sulfate peak and
either the phosphate or nitrate peaks was
required to be greater than 60 percent as
calculated by the following formula:
2(a + b) x 100
where: a is the distance between peaks, in
centimeters, and
b is the width of the peak base, in
centimeters.
If the resolution did not exceed 60 percent, the
column was replaced and the resolution test
was repeated until satisfactory resolution was
obtained.
Data Verification
Different types of errors in the MASS
analytical data base were identified through
the data verification activities and were subse-
quently confirmed or corrected. These includ-
ed:
data entry errors, i.e., values from the
analytical laboratory data packages that
were entered incorrectly;
transcription errors, i.e., data that were
transposed or transcribed incorrectly at
the analytical laboratories;
batch errors, i.e., systematic or sporadic
errors that were discovered when most or
all of the data in specific batches were
outlying; and
41
-------
laboratory errors, i.e., systematic or spora-
dic errors that were discovered when
some or all of the data in batches from a
specific laboratory were outlying.
Details of the data verification procedures are
provided in the following subsections.
Design and Implementation of LEVIS
A computerized Laboratory Entry and
Verification Information System (LEVIS) was
developed for implementation on personal
( .;;np:.iiarf, ,?'iJ we;-; .stilized during the MASS
analytical data collection activities. The LEVIS
program is a two-phase menu-driven system,
in which phase one is an analytical laboratory
data entry and control system and phase two
is an external quality assessment system.
The conceptual design and detailed
specifications for LEVIS were prepared at
EMSL-LV in collaboration with MASS partici-
pants from ERL-C and ORNL. The design
incorporated MASS data management con-
cerns that could affect the LEVIS operation,
ranging from project management issues to
the selection of computer hardware and soft-
ware. The DDRP management team selected
the support staff at ORNL to perform the
computer programming of phase one and
EMSL-LV to develop phase two. Personnel
from both groups worked together to produce
the final design and to maintain a common
understanding of the design characteristics.
Periodic reviews of the phase one programm-
ing implementation were performed to ensure
the correct adaptation of the conceptual de-
sign and project operations. The data base
manager at EMSL-LV served as the primary
point of contact for LEVIS problems discov-
ered by the laboratories.
Compatibility with IBM-AT computer
hardware was essential to the selection of
appropriate personal computer systems.
Specifications included a 40-megabyte hard
drive, 1.2-megabyte floppy disk drive, keyboard,
color monitor, printer, and 2400-baud modem.
A computer system containing all of these
components was installed at each analytical
laboratory. A similar system with a larger
capacity hard disk served as the central com-
puter at EMSL-LV. The central computer
received and maintained data from all three
laboratories.
Phase one included a data entry com-
ponent and two laboratory components of the
verification system: the QC summary reports
and the soil chemistry relationships (SCRs). A
single entry of data at the laboratories, with
100 percent visual verification of the entries,
was required. The LEVIS program generated
hard copy printouts of the input data files
which were checked against the laboratory
raw data sheets by the data entry personnel.
The LEVIS program performed the calculation
of final values, i.e., instrument readings ad-
justed to the soil sample, corrected for mois-
ture content and converted to parameter
reporting units.
Phase one was programmed in dBase III
Plus and then compiled. Communications
software was installed that would facilitate
the transfer of data files from the laboratories
to the central computer at EMSL-LV. This
utility also allowed the central computer to
operate as a remote terminal to the laboratory
computer when the two were connected via
modem, and commands typed on the central
computer were executed on the laboratory
computer. This allowed the QA staff to review
laboratory data files, run programs, select data
files for transfer, and remotely transfer data to
and from the central computer. Archive utility
software was also used to copy and com-
press the entire data base for a batch into one
file suitable for transfer. Through this utility,
a data file size reduction of 80 percent was
common. As the time required for a file trans-
fer is directly related to the size of file, the use
of an archive utility reduced the time required
for data transfer from one hour to less than
fifteen minutes per batch. The backup of the
laboratory data bases to floppy disks also
used the software to reduce the required
backup storage capacity.
Data management personnel from EMSL-
LV and ORNL attended a meeting following the
completion of phase one programming to
review the system. Minor changes were
identified and completed before the system
was installed in a personal computer supplied
by the EPA to each of the three analytical
laboratories. Members of both staffs installed
the equipment and trained the laboratory
personnel. Installation included the hardware
and software set-up as well as additional
system testing. Training included a project
overview and the use of LEVIS hardware and
software, analytical methods, QE/QC sample
design and evaluation, data collection forms,
and a thorough review of all LEVIS compon-
ents and operations. Analytical laboratory
personnel who participated in the training
42
-------
usually included the laboratory manager,
analysts, and data entry operators. To facili-
tate the training, a LEVIS user's guide was
produced (SAIC, 1988) and a "test batch" was
installed on each system.
The programming of phase two did not
begin until phase one was completed and
installed. The data base design and structure
created in phase one was different from the
data base structure required for phase two or
for the final MASS data base structure. The
phase one programming documentation was
used in creating conversion programs to
accomodate data from both phases.
Phase two incorporated two additional
verification programs to evaluate QE samples
for precision and accuracy. Both programs
were implemented on the central EMSL-LV
personal computer. This procedure converted
the phase one data base files to the phase
two data base structure before evaluation
programs were run. Phase two was pro-
grammed in PC-SAS Version 6.03, which had
been selected for structuring all of the com-
pleted DDRP data bases. Although PC-SAS
is not a compiled programming language, the
use of the applications facility module allowed
the entire system to be menu driven so that
the operators never interacted directly with the
programming code.
Analytical laboratories were occasionally
required to reanalyze a batch of samples
because of some identified or suspected QE or
QC discrepancy. The LEVIS program was
designed to accommodate the entry of up to
ten reanalyses of each parameter and main-
tain all versions in the active data base. The
analytical laboratory was required to identify
the analytical run selected for submission to
the QA staff. This was accomplished with a
submission selection option and a submission
tracking file.
Verification Checks
The analytical laboratories evaluated
batch data as part of the laboratory QC proce-
dures, and the QA staff evaluated preliminary
batch data transmitted via modem from the
analytical laboratories. The preliminary data
were usually incomplete because of later
amendments or supplements by the analytical
laboratories, although evaluation of these data
was extremely useful in the early diagnosis of
potential problems which might otherwise have
been justification for reanalysis requests.
Early identification of problems allowed the
laboratories to begin corrective steps prior to
completing analysis of all parameters in a
given batch. Data evaluation of the formal
submission package ensured that the data
were of acceptable quality and that the hard-
copy data package submission was identical
to the corresponding data submitted on floppy
disk. The formal submission data were used
in constructing the MASS final verified data
base.
Verification checks were implemented in
both phases and included QC summary re-
ports, soil chemistry relationships (SCRs), and
evaluations of precision and accuracy. The
data evaluation process in phase one incor-
porated the summary reports and SCRs, which
were available to both the analytical laboratory
and QA staff. The phase two programming,
which incorporated the precision and accuracy
evaluations, was not accessible by the analyti-
cal laboratories and was processed only by
the QA staff.
Quality Control Summary Report
The QC summary report consisted of a
10-page report that listed various data quality
checks for the analytical parameters and a
summary page that allowed a quick overview
of the QC requirements for any potential
deficiencies. The ALSOW mandated that these
requirements be satisfied before a batch of
data could be accepted for payment. The QC
summary report checked the following batch
data characteristics:
number of samples analyzed in each run
and a flag indicating the run number
selected for submission;
mean values and ranges of variability for
each parameter;
replicate sample evaluation;
percent matrix spike recoveries;
calibration blank, reagent blank, QCCS,
and DL-QCCS results;
QCAS accuracy results;
weights and volumes used;
spike solution concentrations; and
instrument detection limits.
The reported data were checked by
LEVIS to ensure that the laboratory had ful-
filled its contractual requirements. Data failing
to satisfy the requirements were automatically
appended with verification flags (see Appendix
A). Confirmation or reanalysis of any flagged
43
-------
parameters was requested if a flag appeared
on any pages of the summary report.
Soil Chemistry Relationships (SCRs)
Sixteen SCRs were checked by LEVIS
during the data verification. The SCRs were
viewed as established empirical relationships
for soil systems and were expected to fail in
fewer than 10 percent of the samples. Ap-
propriate delimiter constraints were built into
each of the SCRs to account for either instru-
ment imprecision or an exceptions to the
theories upon which the relationships were
based, as presented in Table 2-10 below.
Relationships involving particle size
analysis and sulfate isotherm data were not
applied to organic soils because these para-
meters were not required analyses. From a
contractual standpoint, the failure of six or
more relationships in a batch for any SCR
constituted a major flag requiring reanalysis of
the batch, and one to five failures constituted
a minor flag that allowed for the possibility of
reanalysis if other acceptance criteria were not
satisfied.
The SCRs were especially useful in
identifying data entry errors and providing
information about possible anomalies within a
given pedon. Data transcription errors were
more easily identified due to the random failure
of one or more samples to fulfill a particular
relationship. In some cases for which a
parameter was used in multiple relationships,
such as SO4_H2O, it was possible to identify
which parameter data file contained the trans-
posed or questionable values. The results for
a given sample might find one relationship
failing and the other passing, thereby indicat-
ing a problem with the other parameter in the
relationship.
A potential discrepancy was also indi-
cated if the same sample numbers failed
several relationships. In one batch, for ex-
ample, seven samples were flagged for failing
both the pH and CEC relationships, and five of
the samples failed both relationships. Inves-
tigation of the field data forms for these seven
samples revealed that the five samples were
collected from the same pedon, suggesting
that the failures were not related to the labor-
atory analysis but to some anomaly occurring
within that pedon.
Precision Evaluation
With the exception of the phase one
precision estimates for the AD samples and
any internal analyses conducted by the labora-
tories, the analysis of laboratory precision was
performed only by the QA staff who evaluated
data for the FAP/FAO, LAP/LAO, FD, and PD
measurement quality samples. The samples
were evaluated for precision using either an
absolute standard deviation or a percent RSD
depending upon the mean value in relation to
the knot, as described previously. Different
levels of imprecision were established to
account for the varying degrees or types of
error that were expected in the measurement
quality samples. Less precision was expected,
for example, between the FAP samples than
between the LAP samples due to the pos-
sibility of the FAP samples containing field
sampling or preparation error. The LEVIS
program flagged the data for a given parame-
ter if the precision of the replicate samples did
not satisfy the established precision criteria.
Data reported for each pair or triplicate of
samples included the mean, standard devia-
tion, percent RSD, and an identifier code
revealing which of the criteria was used in
evaluating the data for each parameter.
Accuracy Evaluation
With the exception of the phase one
accuracy estimates for the QCAS samples, the
analysis of laboratory accuracy was performed
only by the QA staff using the accuracy win-
dows defined for the MASS laboratory audit
samples. The windows were used to check
the accuracy of the individual LAL samples and
of the mean of the LAP samples for each
parameter. Data values were flagged if the
results were not within the accuracy window.
Data were also evaluated for the FAP, FAL,
and individual LAP samples, although these
results were not used as a basis for request-
ing reanalysis from the analytical laboratories.
Each accuracy printout listed the sample data
value, the accuracy window range, and the
absolute plus or minus difference of the samp-
le value from the acceptable limits of the
window.
Summary pages of the QE and QC
accuracy data were generated for all audit
samples in a batch. Summary pages included
the number of samples outside the accuracy
44
-------
Table 2-10. Soil Chemistry Relationships and Delimiters
Soil chemistry
relationship
Delimiter
PH H20 > PH_002M > PH_01M
CA~CL + MG_CL + K_CL + NA CL < CEC_CL
AL_CL < 0.01 meq/100g
CA OAC + MG OAC + K OAC + NA OAC < CEC_OAC
CEC_OAC > CEC_CL
CEC_OAC + CLAY < 50
CEC_CL + CLAY < 50
MG CL2 < MG CL * 1.10
K_CL2 < K_CL * 1.10
NA_CL2 < NA_CL * 1.10
4 < SO4JH2O + S04 0 < 20
SO4 P04 > SO4 H20*
SO4~32 > SO4 16 > SO4_8 > S04 4 > S04_2 > SO4_0
S04J32N > SO4J6N > S04_8N >"S04_4N > S04_2N > SO4 ON*
7 < C TOT + N TOT < 50
40 < C TOT + S~ TOT < 400
if pH differences were > 0.05 units
if CEC_CL > 1 meq/100g and pH H20 < 7
if PHJH20 > 6.0
if CEC_CL > 1 meq/100g and pH H2O < 7
if CEC_CL > 1 meq/100g and PH~H2O < 7
if CLAY > 1.0 and CEC_CL > 1 meq/100g
if CLAY > 1.0 and CEC_CL > 1 meq/100g
if MG CL2 > 0.008 meq/100g
if K CL2 > 0.003 meq/100g
if NA_CL2 > 0.03 meq/100g
if S04_H2O ;» 2 mg S/kg and SO4_0 * 0.1 mg S/kg
if SO4_PO4 > 1 mg S/kg
if SO4 32N * 7.5 mg S/kg
if SO4~32N * 7.5 mg S/kg
if C TO"T ;» 0.1% and N_TOT * 0.03%
if C TOT ;» 0.1% and S TOT a 0.005%
" SO4_PO4 was multiplied by 1.10 for organic soils
" N = mean spike concentration of initial solution (instrument reading x dilution factor)
windows, the absolute plus or minus differ-
ence from the limits of the windows, and the
mean, maximum, and minimum values of the
absolute differences for the batch. The sum-
mary pages were used to determine whether
an outlier was sample-specific or was consis-
tent throughout all audit sample horizon types
and were helpful in showing where analytical
difficulties might have been occurring at an
analytical laboratory.
Confirmation and Reanalysis
Requests
Following the completion of the initial
data evaluation process for a batch, the QA
staff prepared a summary document of all
flagged parameters, indicating whether a flag
originated from an unacceptable value for the
QC, SCR, precision, or accuracy criteria. The
number and severity of the flags for each
parameter were checked using a QA reanalysis
template (QART) to determine if reanalysis
was required. The templates were constructed
in the manner of a spreadsheet, where rows
specify the evaluation criteria and columns give
the criteria which determine the severity of
measurement uncertainty that had occurred for
each parameter (see Appendix B).
Two levels of magnitude were defined on
each QART, namely, major or minor flags. A
major flag occurred if any two of the following
four evaluations of the laboratory audit samp-
les and preparation duplicates were unaccep-
table: (1) precision of the PD/routine pair, (2)
precision of the PD/FD pair, 3) precision of
the LAP samples or accuracy of their mean
value, or 4) accuracy of the LAL sample. A
major flag was also applied if six or more of
the same SCRs failed in the batch. A major
flag resulted in batch confirmation and/or
reanalysis for that parameter. Minor flags
were applied if only one of the four evaluations
of the laboratory audit samples or preparation
duplicates failed, if the FD or FAP precision
was unacceptable, or if a SCR failed in one to
five instances. Three or more minor flags
were comparable to a major flag for a given
parameter. Reanalysis was not requested if
fewer than three minor flags occurred for any
parameter in a batch. If a major flag or three
minor flags were not identified by the QART
for any parameter, then a sample batch was
deemed to be of acceptable quality. After
confirmation, the data were formally accepted
and transferred into the preliminary verified
data base, and the analytical laboratory was
authorized to receive payment for the analyses.
Data confirmation was performed in-
dependently of the LEVIS program. Where a
data entry error was identified, the analytical
laboratory was requested to re-enter the
correct data in the computer and submit to the
QA staff a new floppy disk and any appropri-
ate LEVIS data forms in which the suspect
value or data correction for the parameter was
checked. This phase of data confirmation did
45
-------
not influence the formal acceptance of the
batch data.
The QA staff compiled verification reports
for each batch data submission. A letter
describing potential discrepancies within the
reported data was sent to each laboratory
after evaluation of the data package. Oc-
casionally, the letter and an accompanying
DDRP Form 500 (see Appendix K) suggested
where errors may have occurred, e.g., trans-
posed numbers or erroneous dilution factors.
The laboratories were required to respond
promptly with confirmation or reanalysis of the
parameters in question. If reanalysis was
necessary, the analysis was performed on the
entire batch of samples for the parameters of
interest rather than on individual samples in
the batch.
Internal Consistency Checks
After completion and receipt of all final
batch data from the analytical laboratories,
two different internal consistency checks were
performed to check for possible data outliers.
Most of the verification checks and evaluations
described previously were performed on data
from the QE and QC samples, and it was
believed that additional evaluations beyond the
SCRs would be useful in identifying suspect
values in the routine soil sample data. The
purpose of these evaluations was to identify,
for each analytical parameter, outlying values
that were not consistent with the majority of
values observed. Confirmation requests of any
suspected outliers identified by these checks
were made to the analytical laboratories, and
the raw data used to generate these values
were subsequently checked for errors in tran-
scription, data entry, calculation, or editing.
Fewer than 10 percent of the suspected out-
liers were changed as a result of this confir-
mation process.
If no correction was made as result of
the checks, each outlying value was flagged in
the analytical data base (see Appendix A). It
was not possible for the QA staff to identify or
verify the cause of every individual outlier with
the laboratories, although confirmation of each
value was requested. A provision was made
in the QART which allowed a major reanalysis
flag to be placed on a batch of data if more
than 10 outliers were identified. However, it
never became necessary to apply this provi-
sion to an analytical laboratory during the
MASS.
Statistical Outlier Check
For the first internal consistency check,
a computer program was used to develop a
correlation matrix for all analytical parameters
measured in the MASS. The strongest correla-
tions, based on the coefficient of determina-
tion (r2), were investigated in more detail (see
Table 2-11). In many cases, the values for one
parameter did not correlate well with values
for any other parameter. Correlations were
not performed on parameters within the same
extract or from the same measurement, e.g.,
CA_OAC values were not correlated with
MGfOAC values even though the resulting r2
value might have shown the strongest correla-
tion. The reason for this decision was that
certain errors, e.g., incomplete extraction,
would not be identified by examining correla-
tions within the same extracting solution.
Table 2-11. Internal Consistency Checks Performed
for Parameter Correlations
Parameter Correlations3
value
SAND
SAND
VCOS
VFS
SILT
SILT
CLAY
PH H2O
PH 01M
CA CL
MG CL
K CL
NA CL
CEC CL
AC BACL
AC BACL
MG CL2
K CL2
AL CL2
FE PYP
AL PYP
FE CD
AL CD
S04 PO4
S04 0
SO4 2
S04 4
S04 8
S04 16
S04 32
C TOT
S TOT
MS
FSI
COS
FS
COSI
FSI
SAND
PH 01M
PH 002M
CA~OAC
MG OAC
K OAC
NA OAC
CEC OAC
AL CL
CEC OAC
MG CL
K OAC
SO4 PO4
FE AO
AL AO
CLAY
AL AO
S04 H20
S04 H20
S04 H2O
S04 H2O
SO4 H2O
SO4 H20
S04 H2O
N TOT
S04 4
0.58
0.80
0.46
0.24
0.43
0.84
0.59
0.83
0.94
0.99
0.99
0.97
0.80
0.76
0.61
0.74
0.70
0.82
035
0.60
0.73
0.59
0.72
0.75
0.99
0.91
0.91
0.93
0.84
0.72
0.68
0.88
a A variable may appear in more than one correlation,
and not all variables were used in the correlations.
46
-------
Application of a Cook's D statistic
(Belsley et al., 1980) was useful in determining
high influence points. This statistic is a ver-
sion of the difference of fit statistic applied in
the internal consistency checks of previous
DDRP surveys, and serves to measure the
change resulting from the deletion of each
point from the routine data set (Van Remortel
et al., 1988). Essentially, the Cook's D statistic
is used to scale and square the difference of
fit statistic to allow extreme values to stand
out more clearly (SAS, 1986).
A list of outlying values was developed
through a multi-stage regression process for
those parameter correlations exhibiting the
highest r2 values. For each parameter pair, a
regression line was fit to the data and a set
of outlying values and leverage points was
generated which included those points having
a Cook's D residual of 1.0 or higher. These
points were temporarily removed from the data
set and the regression was re-run. At this
stage, each point was individually placed into
the second regression; if the outlier had a new
residual value of 3.0 or higher, then it was
positively identified as an outlier. Finally, the
axes of the parameters in the regression were
reversed and the procedure was repeated,
yielding another set of outliers. The two sets,
not mutually exclusive, constituted the final
outlier points that were checked by the QA
staff end underwent confirmation at the analy-
tical laboratories.
A list of outlying values was sent on a
DDRP Form 600A (see Appendix K) to each
analytical laboratory with instructions that
these values be confirmed or corrected by the
laboratory manager. Any outliers that were
not changed to "acceptable" values as defined
by the regression were appended with an "X4"
flag (see Appendix A) in the analytical data
base. Approximately 3 percent of the data
underwent confirmation as a result of this
consistency check.
Pedon/Horizon Outlier Check
For the second internal consistency
check, the analytical data were listed in a data
set using the original soil profile sequence, i.e.,
pedon and horizon, of the samples. Data for
each pedon were visually scanned by soil
scientists for consistency in the magnitude of
the parameter values from one horizon to the
next. This check allowed for subjective identi-
fication of data which appeared to be distinct-
ly aberrant in relation to its horizon type or to
the adjacent horizons within the pedon.
Where individual values appeared to be
very inconsistent with the surrounding horizon
data in a pedon, a list of outlying values were
sent on a DDRP Form 6008 (see Appendix K)
to each analytical laboratory for confirmation
or correction by the laboratory manager. Any
outliers that were not changed to "acceptable"
values as defined by the horizon configuration
were appended with an "X5" flag (see Appendix
A) in the analytical data base. Approximately
2 percent of the data underwent confirmation
as a result of this consistency check.
Overview of the Data Bases
The field sampling and sample prepara-
tion data were entered into separate SAS-
AF raw data files on personal computers at
EMSL-LV. The analytical data were entered
into LEVIS on personal computers located at
the laboratories. All raw data forms were
archived by the laboratories for possible future
reference use. The field data were verified
according to the criteria specified in Appendix
G of the MASS QA Plan (Papp et al., 1989).
The preparation data were verified according
to the criteria specified in Appendix H of the
QA Plan. The analytical data were verified
according to the criteria specified in the AL-
SOW (USEPA, 1988). Raw data verification
was accomplished by a systematic evaluation
of completeness, precision, internal consisten-
cy, and coding accuracy. Apparent discrepan-
cies were appended with data qualifiers, or
flags, unless they could be corrected. After
verification was completed on July 14, 1989,
the data bases were "frozen" and did not
undergo further editing by the QA staff.
Verified data from the analyses con-
ducted during sample preparation were later
appended to the verified analytical data base.
The "frozen" verified field and analytical data
bases were then sent to ORNL to undergo
validation in cooperation with ERL-C and
EMSL-LV personnel. The validation procedures
included a specific assessment of outlying
data points for inclusion or omission in validat-
ed data sets based on assigned levels of
confidence. These outliers warrant special
attention or caution by the data user during
analysis of the survey results. After all data
were evaluated and the suspect values were
confirmed or flagged, the validated data bases
were frozen (Turner et al., in preparation).
47
-------
Section 3
Results and Discuss ion
The data quality estimates presented in
this section are based on the analysis of data
contained in the final MASS verified data base.
The parameter measurements are described
with respect to their detectability, precision,
accuracy, representativeness, completeness,
and comparability. A summary discussion of
measurement uncertainty and its relation to
overall data uncertainty is also provided.
Assessment of Detectability
Data relating to detection limits for
contract requirements, instrument readings,
and system-wide measurement in the MASS
are presented in Table 3-1. The actual report-
ed IDLs, calculated from the calibration blanks,
were less than the corresponding CRDLs for
every parameter in all batches. Therefore, the
MQOs for detectability in the MASS were
completely satisfied. It is apparent that
differences in the theoretical concentrations of
DL-QCCS stock solutions among laboratories
together with temporal variability of the solu-
tions among batches and variation in the
strength of acid titer solution were responsible
for the high IIDLs in certain parameters,
especially AC_BACL. In order to obtain lower
detectability from the IIDL data, it might be
necessary to contractually require the labora-
tories to use stricter criteria for stabilizing the
strength of their DL-QCCS stock solutions.
This approach could, however, preclude efforts
by a particular laboratory to improve the
quality of data by adopting a lower detection
limit target, as was the case with Laboratory
4 in the MASS.
An issue was raised during the DDRP
surveys concerning the type of QC sample that
should be used in assessing instrument detec-
tability. The calibration blank sample has been
considered by some analytical chemists to be
unsuitable for measurement of the IDL be-
cause the limit is based on standard devia-
tions for a zero mg/L blank and cannot be
reliably distinguished from background noise in
the analytical measurement system. As an
alternative, the DL-QCCS could be used to
measure the IDL because the sample contains
the analyte of interest at a concentration high
enough to differentiate from background noise,
yet low enough to be at a meaningful con-
centration for this assessment. Data for the
DL-QCCSs from all DDRP analyses could be
used to determine a more reliable MQO for
instrument detectability in future surveys.
Differences between the IDLs and IIDLs
for most of the other parameters were slight
except for CEC_OAC, AL_CL2, SO4_PO4,
SO4JD, and S_TOT. Low concentrations of
aluminum in the calcium chloride extract con-
tributed to inflation of the DL-QCCS pooled
standard deviations in the zone of con-
centration around the CRDL. The IIDL for the
SO4_PO4 parameter probably was higher
because of inconsistency at the laboratories in
preparing the solution standards for this
parameter. Reasons for differences in the
other three parameters were not clear.
The SDLs calculated from the FAL sam-
ples were much higher than either the IDLs or
IIDLs, as was expected. This assessment of
system-wide detectability differs from the
approach taken in the two previous DDRP
surveys in that low-range field audit samples
were used instead of the 10 percent of field
duplicate samples having the lowest concen-
tration. Using a criterion of 80 percent or
more of the routine sample concentrations
exceeding the SDL as a basis for assessment,
most of the parameters are suitable for all
48
-------
Table 3-1. Defection Limits for Evaluation of Contractual Compliance and for Independent Assessment of
Analytical and System-Wide Measurement
Reporting
Parameter units
CA CL meq/100g
MG CL
K CL
NA CL
AL_CL
CA OAC meq/100g
MG OAC
K OAC
NA_OAC
CEC CLrf meq/100g
CEC~CL"
CEC OACrf
CEC OAC*
AC BACL'
ACJ3ACU7
CA CL2 meq/100g
MG CL2
K CL2
NA CL2
FE CL2
AL_CL2
FE PYP wt %
AL PYP
FE AO
AL AO
SI~AO
FF CD
AL_CD
504 H2O mg S/kg
S04 P04
S04_0 mg S/L
C TOT wt %
N TOT
S TOT
Contractual compliance"
CRDL IDL
0.003
0.005
0.002
0.003
0.015
0.003
0.005
0.002
0.003
0.15
0.075
0.15
0.075
0.25
1.25
0.0050
0.0008
0.0003
0.0004
0.0005
0.0011
0.005
0.005
0.005
0.005
0.005
0.002
0.002
0.50
0.50
0.025
0.010
0.005
0.001
0.0026
0.0015
0.0009
0.0012
0.0111
0.0022
0.0016
0.0009
0.0010
0.0530
0.0175
0.0629
0.0224
0.1738
0.6240
0.0014
0.0004
0.0001
0.0002
0.0004
0.0009
0.0022
0.0029
0.0023
0.0036
0.0026
0.0009
0.0011
0.3339
0.3923
0.0167
0.0056
0.0015
0.0007
Independent assessment*7
IIDL SDL
0.0032
0.0018
0.0012
0.0027
0.0098
0.0035
0.0024
0.0016
0.0026
0.1043
0.0116
0.0785
0.0055
0.5942
2.2500
0.0003
0.0001
0.0004
0.0009
0.0031
0.0030
0.0025
0.0020
0.0026
0.0023
0.0007
0.0008
0.4001
0.7552
0.0350
0.0059
0.0016
0.0021
0.1129
0.0251
0.0180
0.0130
0.3161
0.0548
0.0239
0.0291
0.0105
0.1985
0.2736
0.4513
0.2060
4.3067
0.0861
0.0140
0.0044
0.0034
0.0026
0.0126
0.0416
0.0502
0.1340
0.0637
0.0395
0.0990
0.0272
2.0640
3.7879
0.2901
0.0656
0.0022
0.0077
%RS>SDLC
64.9*
82.6
99.2
40.5*
93.1
76.3*
83.8
96.6
59.4*
100.0
100.0
100.0
100.0
88.3
...
100.0
87.0
98.5
95.9
24.8*
80.5
95.1
94.0
78.9*
94.1
19.2*
99.2
99.4
95.7
97.6
88.9
96.8
99.5
67.1*
" Contract-required detection limit and actual reported instrument detection limit, respectively, in reporting units; based
on instrument analysis of calibration blanks.
" Independent instrument detection limit and system detection limit, respectively, in reporting units; based on analysis
of DL-QCCS and FAL samples, respectively, calculated independently from the contractual requirements.
0 Percent of routine samples with values exceeding the system detection limit (asterisk denotes fewer than 80 percent).
d Detection limits for the titration method (Laboratories 1 and 2 only).
" Detection limits for the flow injection analysis method (Laboratory 4 only).
' Reported for mineral samples.
3 Reported for organic samples.
NOTE: Detection limits not applicable for physical parameters, pH, and the remaining sulfate isotherm parameters.
data uses throughout the range of concentra-
tion. The only exceptions are CA_CL and
CAJDAC, NA_CL and NA_OAC, FE_CL2 and
FE_AO, SI_AO, and S_TOT. Data users should
use caution when assessing these parameters,
as a significant portion of the routine sample
concentrations are less than the corresponding
SDLs and may be difficult to distinguish in
regard to overall system detection and quan-
titation considerations. The routine sample
data for NA_CL, FE_CL2, and SI_AO are par-
ticularly sensitive because less lhan 50 per-
cent of the data for these parameters have
49
-------
concentrations greater than the corresponding
SDL.
Assessment of Precision
The following tables, figures, and text
provide a summary of imprecision estimates
for the routine and QE sample data. The
laboratory audit samples and duplicate sam-
ples were used to establish means and varia-
bilities for pairs of theoretically identical sam-
ples. The routine samples were grouped into
individual sampling class/horizon configura-
tions in order to provide an estimate of soil
heterogeneity of the regional population con-
founded with system-wide measurement
uncertainty. The assessment of precision re-
lates directly to the within-batch MQOs es-
tablished in the MASS QA Plan (Papp et al.,
1989) and to the between-batch precision that
is a characteristic of the batch uncertainty
introduced during sample analysis. Except for
the particle size and pH parameters, the MQOs
have knot values which represent the separa-
tion point for uncertainty expressed as a
standard deviation (SD) in parameter reporting
units for low concentrations and as a relative
standard deviation (BSD) in percent for high
concentrations.
The within-batch precision data are
presented in sequential order of the parame-
ters listed previously in Table 1-1 of Section 1.
For each of the nine parameter groups, a table
of statistics presents the QE and routine
sample data below and above the knot value.
Organic samples were not included in the
analysis because there were insufficient de-
grees of freedom to make a meaningful statis-
tical assessment of these samples. The
tables show the results of the QE sample data
in relation to the achievement of the precision
MQOs for mineral soil samples.
Two figures are presented for each
parameter within each parameter group. The
first figure is a plot of the mean values and
standard deviations of the LAP and LAO sam-
ples in relation to the analytical within-batch
precision MQOs. From a batch acceptance
standpoint, mean values below the MQO line
are considered to have acceptable precision,
while those values above the line had unaccep-
table precision. The second figure is a plot of
the mean and standard deviation for each
sampling class/horizon group of routine sam-
ples. The much higher variability usually
observed in the routine sample data is princi-
pally a result of spatial heterogeneity among
the population of soils within each sampling
class/horizon group. Specifically, horizons of
similarly classified soils in different sectors of
the Mid-Appalachian region, e.g., Pennsylvania
and Virginia, can exhibit dissimilar soil charac-
teristics. The dissimilarities are due to subtle
differences resulting from the local soil-forming
factors of climate, topography, vegetation,
parent material, and time.
Also included in the second figure are
sets of four horizontal lines representing the
within-batch standard deviations for the field
duplicates and preparation duplicates and the
within- and between-batch standard deviations
for the laboratory audit samples. Each set of
lines represents the imprecision estimates
within the partitioned intervals established by
the step function across the entire range of
concentration (see Appendix C). It is assumed
that the intervals within each concentration
range are independent of the type of measure-
ment quality sample represented, even though
there are subtle differences in the distributions
of the samples. The primary purpose of this
figure is to show the contribution of mea-
surement imprecision to the overall variability
of the routine sample data.
A data table corresponding to the step
function statistical procedure applied to each
parameter is given in Appendix C as supple-
mental information for the derivation of the
precision data provided in the plots. Appendix
D presents data, sorted by data set, that
identify QE or routine samples having inor-
dinately high or low values that exert a dis-
proportionate influence on the overall data
quality. These data may be of interest to data
users during the evaluation of specific data
sets represented in the plots or of individual
batches of samples from a given analytical
laboratory.
50
-------
Moisture and Particle Size Analyses
The analytical, preparation, and field
within-batch precision MQOs were satisfied for
the MOIST, SAND, SILT, and CLAY parameters
(see Table 3-2). The individual sand and silt
separates, i.e., fractions, were not subjected to
the batch acceptance criteria used with the
MQOs; hence, the precision of the separates
were not used as a basis for satisfying the
particle size precision objectives (see Appendix
E). Particle size analyses were not performed
on organic soil samples because the inorganic
component for which these analyses were
designed was low or absent in many of these
soils. With few exceptions, a general pattern
of increasing standard deviation with increased
sources of confounded error was observed,
i.e., the standard deviation from the field
duplicate samples exceeded that of the prepar-
ation duplicate and laboratory audit samples.
The upward adjustment of the particle
size analytical MQOs from 1.0 weight percent
in the two previous DDRP surveys to 3.0
weight percent for SAND and SILT and 2.0
weight percent for CLAY in the MASS appeared
to compensate for error in horizons dominated
by one particle size fraction. Therefore, the
less stringent MQOs are considered to be
reasonable expectations for laboratory and
system-wide performance for particle size
analysis.
Imprecision was much lower for COSI
and FSI in the duplicate samples compared to
the laboratory audit samples despite the
increasing sources of error. Also, the audit
sample imprecision for these parameters was
much higher than the corresponding SILT
values. The source of this uncertainty was
investigated and found to be the result of
apparent difficulties with particle size analysis
at Laboratory 2, which had not been using a
270-mesh sieve at the bottom of the nested
sieves during the early stages of batch analy-
sis. Inordinate COSI and FSI values for a LAP
sample in Batch 30106 were responsible for
the higher imprecision, although the precision
for SILT (the silt parameter of primary interest
to the data users) was acceptable. The initial
run and two reanalysis runs of particle size on
this batch provided mixed results for different
particle size fractions, and it was determined
that the second run had the most consistent
analytical results even though the COSI and
Table 3-2. Achievement of Measurement Quality
Objectives for Wlthln-Batch Precision of
Moisture and Particle Size Analysis
Data
set'
LAP
PD
FD
S/H
Parameter
MOIST
SAND
VCOS
COS
MS
FS
VFS
SILT
COSI
FSI
CLAY
MOIST
SAND
VCOS
COS
MS
FS
VFS
SILT
COSI
FSI
CLAY
MOIST
SAND
VCOS
COS
MS
FS
VFS
SILT
COSI
FSI
CLAY
MOIST
SAND
VCOS
COS
MS
FS
VFS
SILT
COSI
FSI
CLAY
df*
27
27
27
27
27
27
27
27
27
27
27
54
54
54
54
54
54
54
54
54
54
54
27
27
27
27
27
27
27
27
27
27
27
618
618
618
618
618
618
618
618
618
618
618
SDC
wt
0.2157
1.1596
0.9116
0.6642
0.7921
0.5946
0.6654
1.8783
4.0549
4.3368
1.1304
0.0707
1.2519
1.5450
0.8262
0.5127
0.7940
0.8538
1.1904
0.9284
1.0083
0.6661
0.1160
2.6239
2.0730
1.1546
0.8578
0.9184
1.1540
2.0266
1.1717
1.3155
1.1231
0.6505
14.2599
4.0964
4.3134
6.3679
7.9281
4.8732
10.2112
5.2974
7.7111
7.3806
MQOC Pairs>MQO
% n %
0.3 5 18.5
3.0 0 0.0
3.0 2 7.4
2.0 1 3.7
0.3 0 0.0
3.0 2 3.7
3.0 2 3.7
2.0 1 1.9
0.3 0 0.0
6.0 1 3.7
6.0 1 3.7
4.0 0 0.0
" LAP = Laboratory audit pairs; PD = Preparation
duplicates; FD = Field duplicates; S/H = Sampling
class/horizon routine samples.
b Degrees of freedom.
c Standard deviation and measurement quality
objective, respectively, expressed in weight percent
for mineral soil samples; a knot point of 12.0 weight
percent was specified for MOIST, although no data
were observed in the upper tier.
51
-------
FSI fractions were erroneous for the LAP
samples. As a secondary check on the quality
of the batch routine sample data, the outputs
from each run were compared. With the
exception of the LAP samples, the silt values
for all samples in the batch were similar
across runs and did not warrant concern for
batch-wide data quality. The QA staff con-
cluded that the overall precision for COSI and
FSI is actually much better than portrayed due
to the inordinate effect of the LAP samples in
Batch 30106.
Figures 3-1 through 3-4 are plots of the
laboratory audit sample data in relation to the
within-batch precision MQOs and of the routine
sample data in relation to selected QE sam-
ples. The plots are provided only for those
physical parameters for which MQOs were
defined, i.e., for MOIST, SAND, SILT, and
CLAY. Appendix E contains the routine data
plots for the remaining particle size parame-
ters. Supplemental information regarding
uncertainty estimates is presented in Appendix
C, and the identification of inordinate data
values is presented in Appendix D.
52
-------
(a)
0.64
0.5-
0.4-
0.3
Q
-20.1-]
~
0.0-
(b)
2.0-1
K
~5
c
o
Q
-o
o
_o
CO
1.5-
.0-
).5-
0.0-
MOIST
Air-Dry Moisture
Laboratory Audit Samples
4567
Mean (wt *)
a Q n B is. A A Bw 0
Routine Samples
MQO
8
0123
Mean (wt «)
+ + + Mineral Routine S/H Groups Field Oups. Lab Audits (Within-Batch)
Organic Routine S/H Groups Preporatfon Dups. Lab Audits (Between-Batch)
Note: One mineral, eight organic sampling class/horizon groups exceed plot boundaries.
Figure 3-1 Range and frequency distribution of MOIST for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
53
-------
44
(a)
SAND
Total Sand
Laboratory Audit Samples
N
-4'
5
C
o
o
'>
MQO
1.
o
in
o.
D a
oo
20
30 40
Mean (wt B)
a a a B A A A Bw
50
60
(b)
254
c
o
? 1CH
\_
o
D
O '
(ft
0-
Routine Samples
H
20
Mineral Routine S/H Groups
40
Mean (wt x)
Field Dups.
Preparation Dups.
60
80
Lab Audits (Within-Batch)
Lab Audits (Between-Batch)
Note: Two mineral sampling class/horizon groups exceed plot boundaries.
Figure 3-2. Range and frequency distribution of SAND for. (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
54
-------
(a)
SILT
Total Silt
Laboratory Audit Samples
-6-
c
o
MQO
o
^
o
CO
0
40
50 60
Mean (wt %)
a D a B ^ A A Bw
70
80
(b)
20 {
15-
C
O
10-
Q
T3
O
? 5
o
CO
0
1
10
20
Routine Samples
30 40
Mean (wt ^)
+ + + Mineral Routine S/H Groups Field Dups.
50 60
Lab Audits (Within-Batch)
Preparation Dups. Lab Audits (Between-Botch)
Note: Two mineral sampling class/horizon groups exceed plot boundaries.
Figure 3-3. Range and frequency distribution of SILT for: (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
55
-------
(a)
CLAY
Total Clay
Laboratory Audit Samples
4-
c
o
-83J
MOO
o
In
oo
0-
o A. a
0
2
6 8
Mean (wt %)
a a a B A A A Bw
10
12
14
(b)
C
O
0)
o
C
o
*~t
(f)
0-
5.
0
Routine Samples
---*.*_
10
20
Mean (wt s)
30
40
-** + Mineral Routine S/H Groups Field Dups. Lab Audits (Within-Batcn)
Preparation Dups. Lab Audits (Between-Batch)
Note: Three mineral sampling class/horizon groups exceed plot boundaries.
Figure 3-4. Range and frequency distribution of CLAY for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
56
-------
Soil pH
The analytical, preparation, and field
within-batch precision MQOs were satisfied for
each of the three soil pH parameters (see
Table 3-3). A general pattern of increasing
standard deviation with increased sources of
confounded error was maintained for each
parameter. The standard deviation showed a
slight change over the measured pH range and
it was necessary to apply the step function to
the data.
Figures 3-5 through 3-7 are plots of the
laboratory audit sample data in relation to the
within-batch precision MQOs and of the rout-
ine sample data in relation to selected QE
samples. Supplemental information regarding
uncertainty estimates is presented in Appendix
C and the identification of inordinate data
values is presented in Appendix D.
Table 3-3. Achievement of Measurement Quality
Objective* for Wlthln-Batch Precision of
Soil pH
Data
set"
LAP
PD
FD
S/H
Parameter
PH H2O
PH 002M
PH_01M
PH H2O
PH~002M
PH_01M
PH H20
PH~002M
PH~01M
PH H2O
PH~002M
PH~01M
df*
27
27
27
54
54
54
27
27
27
618
618
618
SDC
PH
0.0525
0.0518
0.0319
0.0563
0.0599
0.0345
0.0939
0.1017
0.0667
0.4756
0.4562
0.4352
MQOC
units
0.10
0.10
0.10
0.10
0.10
0.10
0.20
0.20
0.20
Pairs>MQO
n
3
1
0
4
1
0
1
2
0
%
11.1
3.7
0.0
7.4
1.9
0.0
3.7
7.4
0.0
* LAP = Laboratory audit pairs; PD = Preparation
duplicates; FD = Field duplicates; S/H = Sampling
class/horizon routine samples.
b Degrees of freedom.
c Standard deviations expressed in pH units for mineral
soil samples.
57
-------
(a)
.0.15-1
c
'0.10
Q)
Q
"O
V.
O
C
o
0.05-
0.00-
4.0
PH_H2O
pH in Water
Laboratory Audit Samples
MOO
Q O
O
CD a
a a
aj
4.5 5.0
Mean (pH units)
a a a B A A A Bw 0
5.5
(b)
1.5-1
-------
(a)
,0.25-f
en
°C
IT
CL
0.20-
o. 10
>
0)
Q w.
-oO.OSJ
o
-t-t
in
0.00-
3.5
PH_002M
pH in 0.002M Calcium Chloride
Laboratory Audit Samples
Q
00
a
4.0 4.5
Mean (pH units)
a a
aB
A A
Bw
MOO
5.0
1.5-
(b)
c
D
X
Q.
C
O
1.0-
So,
o
T3
c
o
-»
C/)
5-
0.0-
Routine Samples
3.0
4.0
5.0
Mean (pH units)
6.0
Lab Audits (Within-Batch)
* + Mineral Routine S/H Groups Reid Dups.
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Botch)
Note: One mineral sampling class/horizon groups exceed plot boundaries.
Figure 3-6. Range and frequency distribution of PH 002M for (a) laboratory audit samples and their relation
to the analytical within-batch precision"measurement quality objective (MOO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
59
-------
(a)
S0.15f
c
3
X
Q.
C
O
'0.10
0)
Q
-o
D
T3
C
O
CO
0.05-
0.00-
3.0
PH_01M
pH in 0.01 M Calcium Chloride
Laboratory Audit Samples
a
a
a aa
a a aa
a
a
MQO
A A A
A A
3.5 4.0 4.5
Mean (pH units)
a a a B A A A Bw 0
5.0
(b)
1.5H
X
Q.
C
'1.0-
a>
Q
T3
O
c
o
<7)
0.5-
0.0-
2.5
Routine Samples
3.0
3.5 4.0
Mean (pH units)
4.5
5.0
Lab Audits (Within-Batch)
+ ++ Mineral Routine S/H Groups Field Dups. - ,
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Botch)
Note: One mineral sampling class/horizon groups exceed plot boundaries.
Figure 3-7. Range and frequency distribution of PH_01M for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimate*
60
-------
Exchangeable Cations in Ammonium
Chloride
The analytical within-batch precision
MQOs were satisfied for all of the cations in
ammonium chloride except for AL_CL above
the knot (see Table 3-4). The MQOs were less
stringent for this cation because it was a new
parameter that was not measured in the
previous DDRP surveys. However, the effect of
data from two of the 27 audit sample pairs in
the upper tier caused an inordinate increase in
the uncertainty to 15.8 percent which slightly
exceeded the MQO. The imprecision estimates
for the preparation and field duplicates were
not affected by inordinate AL_CL data and
easily satisfied their respective MQOs. There-
fore, it is likely that an upper tier precision
objective of 10 percent for the laboratory audit
samples can be satisfied in most cases when
using protocols and acceptance criteria identi
cal to those used to measure AL CL in the
MASS. Likewise, a lower tier objective of 0.10
meq/100g can probably be satisfied for the
same parameter.
The field within-batch imprecision for
CA_CL, while achieving the MQOs for both
tiers, was high in relation to the other cations
of the ammonium chloride extract and in
relation to the imprecision observed in the
laboratory audit samples and preparation
duplicates. The pattern was also noted for
CA_OAC in the ammonium acetate extraction,
as described later in this section. Because of
this apparent incongruity in the precision data,
the possibility of selective sample contamina-
tion during the field sampling or preparation
laboratory operations was investigated. There
was no evidence to suggest a contamination
problem; therefore, the original SCS-SOI-232
field data forms were checked to determine
whether there was a pedon-specific reason for
the variability of particular inordinate field
duplicates. The field duplicate/routine pair for
Table 3-4. Achievement of Measurement Quality Objectives for Within-Batch Precision of the Exchangeable
Cations In Ammonium Chloride
Below the knot^
Data
set"
LAP
PD
FD
S/H
Parameter
CA CL
MG CL
K CL
NA CL
AL_CL
CA CL
MG~CL
K CL
NA CL
AL_CL
CA CL
MG~CL
K CL
NA CL
AL_CL
CA CL
MG~CL
K CL
NA CL
AL CL
df*
27
27
27
32
46
54
50
3
17
23
27
25
1
108
302
542
618
3
SD MQO
~ meq/100g -
0.0116
0.0040
0.0056
0.0074
0.0128
0.0051
0.0035
0.0345
0.0430
0.0136
0.0055
0.0045
0.0233
0.1487
0.2219
0.0661
0.0131
0.6176
0.02
0.02
0.02
0.02
0.20
0.02
0.02
0.02
0.02
0.20
0.04
0.04
0.04
0.04
0.40
Pairs>MQO
n %
1
0
0
0
2
0
0
0
2
0
0
0
0
3.7
0.0
0.0
0.0
4.3
0.0
0.0
0.0
11.8
0.0
0.0
0.0
0.0
df
27
27
22
8
51
10
4
26
510
316
76
615
Above the knotc
RSD
6.8
m
.
15.8
6.7
5.2
.
6.1
23.9
4.8
6/7
142.2
117.6
116.3
59.3
MQO
10
10
10
10
15
15
15
15
15
22
30
30
30
30
45
Pairs>MQO
n %
2 7.4
.
.
2 7.4
1 4.5
0 0.0
0 0.0
1 10.0
0 0.0
0 0.0
" LAP = Laboratory audit pairs; PD = Preparation duplicates; FD = Field duplicates; S/H = Sampling class/horizon
routine samples.
b Degrees of freedom.
0 Standard deviations and relative standard deviations expressed in meq/100g and percent, respectively, for mineral soil
samples below and above the knot point of 0.20 meq/100g (1.3 meq/100g for AL CL); a dot signifies a lack of data
occupying that range.
61
-------
Batch 30113 was collected from an abandoned
cropland site that had been previously limed,
hence, it is possible that lime was irregularly
distributed in the horizon and contributed to
the high observed variability in calcium. It was
also suspected that the field duplicate/routine
pair for Batch 30128 was in some way af-
fected by a sample switch for the cation
analyses even though the QART did not justify
a batch-wide reanalysis to address this dis-
crepancy. If the effect of these two inordinate
pairs was removed from the precision assess-
ment, the field duplicate estimates would be
comparable with the other measurement
quality samples.
There was a general trend of increasing
standard deviation with increased sources of
confounded error. The imprecision estimates
for the preparation duplicates were about the
same as for the laboratory audit samples and
were somewhat less than the field duplicates.
Figures 3-8 through 3-12 are plots of the
laboratory audit sample data in relation to the
within-batch MQOs and of the routine sample
data in relation to selected QE samples.
Supplemental information regarding uncertainty
estimates is presented in Appendix C and the
identification of inordinate data values is
presented in Appendix D.
62
-------
CA_CL
Exchangeable Calcium in Ammonium Chloride
Laboratory Audit Samples
0.5
11.0
Mean (meq/100g)
11.5
12.0
ODD
B
A A
Bw
0
(b)
Routine Samples
S3-
c
o
o 2-
Q
O
~o
c
o
0-
3 4
Mean (meq/100g)
+ + Mineral Routine S/H Groups
Organic Routine S/H Groups
Field Dups.
Lab Audits (Within-Batch)
Preparation Dups. Lab Audits (Between-Batch)
Note: Two mineral, six organic sampling class/horizon groups exceed plot boundaries.
Figure 3-8. Range and frequency distribution of CA_CL for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
63
-------
MG_CL
Exchangeable Magnesium in Ammonium Chloride
(a
0.20^
'5
o
o
\0.15-
cr
0>
§0.10.
-i-J
o
QQ.05.
T3
O
TD
c
£0.00-
00
1
Q
rr
0
0
#$KQ Qffl
i /MOO
* /
/ ,
/
/
/
/
/
/
/
/
/\ /\ /\
T i V V V 1 1 1
0.0 0.1 0.2 3.0 3.1 3.
Mean (meq/100g)
a a a B A A A Bw 0
(b)
Routine Samples
o
o
CT
I21
c
o
-^^
Q
Q
O
ID
C
o
-*<
00
o.
0
Mean (meq/100g)
Lab Audits (Within-Botch)
^ + + Mineral Routine S/H Groups Reid Dups.
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: One mineral, one organic sampling class/horizon groups exceed plot boundaries.
Figure 3-9. Range and frequency distribution of MG_CL for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
64
-------
K_CL
Exchangeable Potassium in Ammonium Chloride
Laboratory Audit Samples
(a)
0.10-
o
2 0.08-
\
o
o
O)
E
0)
Q
(b)
).4-|
.3-
).2-
0.1-
o
T3
I o-o
0.0
Routine Samples
0.2
0.4
0.6
Mean (meq/100g)
Lab Audits (Within-Batch)
" * + Mineral Routine S/H Groups Reid Dups. ~
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: Two mineral, seven organic sampling class/horizon groups exceed plot boundaries.
Figure 3-10. Range and frequency distribution of K_CL for (a) laboratory audit samples and their relation
to the analytical within-batcn precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
65
-------
NA_CL
Exchangeable Sodium in Ammonium Chloride
Laboratory Audit Samples
U.UJ-
cr>
O
0
^002
I
c
g
o
a;
Q
o
0
Jo. oo-
00
_,
Q
a:
o
4
a
i a
A a a a
A 4 aa
a & A a a a
A aa /\
^
i /MOO
* /
/
/
/\ /\
I" I 1 V V V 1 1
0.00
0.05
0.10
Mean (meq/100g)
0.40
0.45
a a a
B
Bw
0
o.03^
en
O
O
cr
c
c
O
a
D
o
c
o
00
>.02
J.01-
0.00-
Routine Samples
* -*.
* + frl**
0.00
0.01
0.02 0.03
Mean (meq/100g)
0.04
0.05
+ * Mineral Routine S/H Groups Held Dups. Lab Audits fWithin-Batch)
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: Three mineral, six organic sampling class/horizon groups exceed plot boundaries.
Figure 3-11. Range and frequency distribution of NA_CL for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
66
-------
(a)
2.0-r
AL_CL
Exchangeable Aluminum in Ammonium Chloride
Laboratory Audit Samples
o
o
CT
0)
1.5-
o
L-
O
Jo.o-l
00
MQO
0
234
Mean (meq/100g)
a a n B A A A Bw 0
o
o
g
>
a>
4-
c
_0
00
0-
(b)
0
Routine Samples
A
t
+ + + *
6 8
Mean (meq/100g)
10
12
14
+ + + Mineral Routine S/H Groups Field Oups. Lab Audits (Within-Batch)
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: Three organic sampling class/horizon groups exceed plot boundaries.
Figure 3-12. Range and frequency distribution of AL_CL for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
67
-------
Exchangeable Cations in Ammonium
Acetate
The analytical and preparation within-
batch precision MQOs were satisfied for the
exchangeable cations in ammonium acetate
(see Table 3-5). A general pattern of increas-
ing standard deviation with increased sources
of confounded error was maintained.
The field within-batch imprecision for
CAJDAC barely satisfied the MQO for the
lower tier and slightly exceeded the MQO for
the upper tier, and was high in relation to the
other cations of the ammonium acetate extract
and in relation to the imprecision observed in
the laboratory audit samples and preparation
duplicates. The pattern was also noted for
CA_CL in the ammonium chloride extraction, as
described previously in this section. Because
of this apparent incongruity in the precision
data, the possibility of selective sample con-
tamination during the field sampling or prep-
aration laboratory operations was investigated.
There was no evidence to suggest a conta-
mination problem, therefore, the original
SCS-SOI-232 field data forms were checked to
determine whether there was a pedon-specific
reason for the variability of particular
inordinate field duplicates. The field
duplicate/routine pair for Batch 30113 was
collected from an abandoned cropland site
that had been previously limed, hence, it is
possible that lime was irregularly distributed in
the horizon and contributed to the high ob-
served variability in calcium. It was also
suspected that the field duplicate/routine pair
for Batch 30128 was in some way affected by
a sample switch for the cation analyses even
though the QART did not justify a batch-wide
reanalysis to address this discrepancy. Remo-
ving the effect of these two inordinate pairs
from the precision assessment makes the field
duplicate estimates comparable with the other
measurement quality samples.
Figures 3-13 through 3-16 are plots of
the laboratory audit sample data in relation to
the within-batch MQOs and of the routine
sample data in relation to selected QE sam-
ples. Supplemental information regarding
uncertainty estimates is presented in Appendix
C and the identification of inordinate data
values is presented in Appendix D.
Table 3-5. Achievement of Measurement Quality Objectives for Wlthln-Batch Precision of the Exchangeable
Cations In Ammonium Acetate
Below the knotc
Data
set"
LAP
PD
FD
S/H
Parameter
CA OAC
MG OAC
K OAC
NAJDAC
CA OAC
MG OAC
K OAC
NAJDAC
CA OAC
MG OAC
K OAC
NAJDAC
CA OAC
MG OAC
K OAC
NA OAC
df6
27
27
54
33
46
51
59
18
23
25
29
113
301
545
754
SD MQO
- meq/100g -
(X0142
0.0053
0.0050
0.0070
0.0095
0.0081
0.0032
0.0394
0.0143
0.0067
0.0043
0.1457
0.2162
0.0640
0.0147
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.04
0.04
0.04
0.04
Pairs>MQO
n %
3
0
0
2
3
2
0
3
0
0
0
11.1
0.0
0.0
6.1
6.5
3.9
0.0
16.7
0.0
0.0
0.0
df
27
21
8
3
9
4
2
505
317
73
Above the knotc
RSD MQO
%_
9.1
9.5
3.3
7.1
33.3
5.6
8.0
143.1
116.2
111.2
10
10
10
10
15
15
15
15
30
30
30
30
Pairs>MQO
n %
4
2
0
0
2
0
0
14.8
9.5
0.0
0.0
22.2
0.0
0.0
" LAP = Laboratory audit pairs; PD = Preparation duplicates; FD = Field duplicates; S/H = Sampling class/horizon
routine samples.
* Degrees of freedom.
c Standard deviations and relative standard deviations expressed in meq/100g and percent, respectively, for mineral soil
samples below and above the knot point of 0.20 meq/100g; a dot signifies a lack of data occupying that range.
68
-------
CA_OAC
Exchangeable Calcium in Ammonium Acetate
Laboratory Audit Samples
0.2
0.4 8.1
Mean (meq/100g)
8.3
8.5
a a a
B
Bw
0
(b)
Routine Samples
c
O
'52
>
a>
K°<
0
23
Mean (meq/100g)
Lab Audits (Within-Batch)
t-» Mineral Routine S/H Groups Held Dups. -
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: Two mineral, three organic sampling class/horizon groups exceed plot boundaries.
Figure 3-13. Range and frequency distribution of CA OAC for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
69
-------
MGJDAC
Exchangeable Magnesium in Ammonium Acetate
Laboratory Audit Samples
(a)
.15-
'cn
O
O
CT
a>0.10-
c
o
-t '
0
'>0.05-
1 _
Q
-O
s_
O
C
-So-
+ -»-
«^«.
j_ +
O
2
^
4-
+ t
t «
t *
* + A
+ » * *
*
*±*+-k. **+ + *
t. +
* *+ \ * + . * .
H- t + J' *
^Bt(-fj^*"l>y *" ^ t" ., ._ -. -_. .- _ -^^--^^.J^^---^^ -
(/) S ! ' 1 ' 1 ' 1 ' r
0.0 0.5 1.0 1.5 2.
Mean (meq/100g)
Lab Audits (Within-Batch)
-*»» Mineral Routine S/H Groups Field Dups. - .......
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: Three mineral, one organic sampling class/horizon groups exceed plot boundaries.
Figure 3-14. Range and frequency distribution of MG OACfor: (a) laboratory audit samples and their relation
to the analytical within-batch preciskxTmeasuroment quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
70
-------
K_OAC
Exchangeable Potassium in Ammonium Acetate
(a)
^0.05--
'CTI
o
2 0.04-
cr
OJ
d Deviation (
D 0 c
D 0 c
ro c
i , 1_ _ ,
^_
a
Jo.OO-
(S)
0.(
Q
O
A aa
a
A S en a, a
^ a a o
" 0
I
i /MQO .
/
/
'
/\ /\ /\
I 1 \x V V I i
30 0.05 0.10 1.10 1.1
5
Mean (meq/100g)
a a a
B
A A
Bw
0
(b)
en
O
O
0)
o.
2-
0)
Q
O
T5
C
O
0.1-
0.0-
0.0
Routine Samples
0.1
0.2 0.3 0.4
Mean (meq/1 OOg)
0.5
0.6
- *' Mineral Routine S/H Groups Field Dups. Lab Audits (Within-Satch)
Organic Routine S/H Groups Preparation Dups.' Lab Audits (Between-Botc
Note: Two mineral, seven organic sampling class/horizon groups exceed plot boundaries.
Figure 3-15. Range and frequency distribution of KJDAC for: (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
71
-------
NA_OAC
Exchangeable Sodium in Ammonium Acetate
^
o
o
Q-
0)
c
o
_"g
(U
Q
TJ
o
0
-t*
CO
.03-
0 02
0.01-
0.00-
QL
O
A
A
n
a
a DO
A a
A a
A a a
AA A O
A A a
a a /\
v^
FQO
/\ /\ .
T - I 1 1 1 V \/ V 1 I
0.00 0.02 0.04 0.06 0.08 0.44 0.46
Mean (meq/100g)
a o
aB
Bw
0
(b
.04
en
o
o
\0.03-
Q)
§0.02
0
0)
S°-01
o
"O
20.00-
4- +
*
,. + +
^ +
+ *
+ i* + * * *
* +*+ ++ * *
- - i- ti 1* T + +" r ' "^ r^~l ' " "
*- + + + -4-4-
0.00
0.01
0.02 0.03
Mean (meq/100g)
0.04
0.05
Lab Audits Within-Batch)
+ * + Mineral Routine S/H Groups Field Dups. ~
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: Eight organic sampling class/horizon groups exceed plot boundaries.
Figure 3-16. Range and frequency distribution of NA_OAC for: (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MOO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
72
-------
Cation Exchange Capacity and
Exchangeable Acidity
The analytical, preparation, and field
within-batch precision MQOs were satisfied for
the CEC and exchangeable acidity parameters
(see Table 3-6). There was a general trend of
increasing standard deviation with increased
sources of confounded error for all parame-
ters.
Figures 3-17 through 3-19 are plots of
the laboratory audit sample data in relation to
the within-batch MQOs and of the routine
sample data in relation to selected QE sam-
ples. Supplemental information regarding
uncertainty estimates is presented in Appendix
C and the identification of inordinate data
values is presented in Appendix D.
Table 3-6. Achievement of Measurement Quality Objectives for Wlthln-Batch Precision of Cation Exchange
Capacity and Exchangeable Acidity
Below the knotc
Data
set"
LAP
PD
FD
S/H
Parameter
CEC CL
CEC OAC
AC_BACL
CEC CL
CEC OAC
AC_BACL
CEC CL
CEC OAC
AC_BACL
CEC CL
CEC OAC
AC BACL
df"
3
19
1
11
4
131
SD MQO
- meq/100g -
0.1389
0.6044
0.1648
0.9879
0.2681
3.3224
0.25
0.25
1.00
0.25
0.25
1.00
0.50
0.50
2.00
Pairs > MQO
n %
0 0.0
1 5.3
0 0.0
t f
0 0.0
df
27
27
27
51
54
35
26
27
16
614
618
487
Above the knotc
RSO MQO
5.2
6.5
6.2
9.0
11.1
8.9
5.5
11.3
9.4
39.0
42.5
46.6
10
10
15
15
15
22
30
30
45
Pairs>MQO
n %
1
2
1
2
5
0
0
1
0
3.7
7.4
3.7
3.9
9.3
0.0
0.0
3.7
0.0
" LAP = Laboratory audit pairs; PD = Preparation duplicates; FD = Field duplicates; S/H = Sampling class/horizon
routine samples.
b Degrees of freedom.
c Standard deviations and relative standard deviations expressed in meq/100g and percent, respectively, for mineral soil
samples below and above the knot point of 2.5 meq/100g (6.7 meq/100g for AC_BACL); a dot signifies a lack of data
occupying that range.
73
-------
CEC_CL
Cation Exchange Capacity by Ammonium Chloride
Laboratory Audit Samples
Mean (meq/100g)
DOE
B
A A A
Bw
O
O
o
T5
L.
O
c
o
in
(b)
Routine Samples
0-
0
* ** *V + **; +
' + * ++ + +
10 15
Mean (meq/100g)
20
25
* Mineral Routine S/H Groups Field Dups. Lab Audits (Within-Batch)
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: One mineral, seven organic sampling class/horizon groups exceed plot boundaries.
Figure 3-17. Range and frequency distribution of CEC CL for: (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned Into concentration intervals of uniform variability
and their relation to pooled precision estimates.
74
-------
CECJDAC
Cation Exchange Capacity by Ammonium Acetate
Laboratory Audit Samples
10
20 30
Mean (meq/100g)
a a a B A A A Bw 0
80
90
(b)
o
o
«MO
c
o
'o
'> c
Q
-O
O
c
o
-t-»
CO
0
Routine Samples
% *
o
10 15 20
Mean (meq/100g)
25
+ + + Mineral Routine S/H Groups Field Dups. Lab Audits (Within-Botch)
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: One mineral, thirteen organic sampling class/horizon groups exceed plot boundaries.
Figure 3-18. Range and frequency distribution of CEC OAC for (a) laboratory audit samples and their
relation to the analytical within-batch precision measurement quality objective (MQO) and (b)
sampling class/horizon routine sample data partitioned into concentration intervals of uniform
variability and their relation to pooled precision estimates.
75
-------
(a)
4-1
AC_BACL
Exchangeable Acidity in Barium Chloride Triethanolamine
Laboratory Audit Samples
O
O
cr
-------
Extractable Cations in Calcium
Chloride
The analytical within-batch precision
MQOs were satisfied for all six extractable
cations in calcium chloride except for MG_CL2
data in the lower tier (see Table 3-7) which
only slightly exceeded the objective. The effect
of one inordinate audit sample pair out of 15
pairs within the tier was responsible for the
failure to satisfy this MQO. The preparation
and field within-batch MQOs were satisfied for
all parameters in this group. The addition of
a lower tier objective for the MASS extractable
cation precision MQOs was helpful in evaluat-
ing the data compared to the earlier DDRP
surveys. There was a general trend of in-
creasing standard deviation with increased
sources of confounded error.
Figures 3-20 through 3-25 are plots of
laboratory audit sample data in relation to the
within-batch MQOs and of the routine sample
data in relation to selected QE samples.
Supplemental information regarding uncertainty
estimates is presented in Appendix C and the
identification of inordinate data values is
presented in Appendix D.
Table 3-7. Achievement of Measurement Quality Objectives for Wlthln-Batch Precision of the Extractable Cations
In Calcium Chloride
Below the knotc
Data
set"
LAP
PD
FD
S/H
Parameter
CA CL2
MG~CL2
K CL2
NA CL2
FE CL2
AL_CL2
CA CL2
MG CL2
K CL2
NA CL2
FE~CL2
ALICL2
CA CL2
MG CL2
K CL2
NA CL2
FE CL2
AL~CL2
CA CL2
MG CL2
K CL2
NA CL2
FE CL2
AL~CL2
df*
27
15
27
25
17
27
54
37
53
54
25
52
27
19
27
27
14
26
618
209
587
618
583
618
SD MQO
- meq/100g -
0.0181
0.0058
0.0013
0.0047
0.0007
0.0047
0.0146
0.0020
0.0014
0.0009
0.0015
0.0047
0.0154
0.0048
0.0028
0.0011
0.0012
0.0093
0.0949
0.0635
0.0197
0.0097
0.0098
0.1042
0.05
0.005
0.005
0.005
0.01
0.05
0.05
0.005
0.005
0.005
0.01
0.05
0.10
0.01
0.01
0.01
0.02
0.10
Pairs>MQO
n %
0
1
0
2
0
0
1
1
2
0
0
0
0
1
0
0
0
0
0.0
6.7
0.0
8.0
0.0
0.0
1.9
2.7
3.8
0.0
0.0
0.0
0.0
5.3
0.0
0.0
0.0
0.0
Above the knot*7
RSD
Hf V
12 3.6
.
2 3.6
.
17 4.5
1 4.0
. .
r .
8 7^8
.
. .
, .
409 78.7
31 125.4
.
22 228.0
MQO
5
10
10
10
15
15
8
15
15
15
22
22
15
30
30
30
45
45
Pairs>MQO
n %
0 0.0
o o!o
1 5.9
0 0.0
0 0.0
.
. .
.
" LAP = Laboratory audit pairs; PD - Preparation duplicates; FD = Field duplicates; S/H = Sampling class/horizon
routine samples.
b Degrees of freedom.
c Standard deviations and relative standard deviations expressed in meq/100g and percent, respectively, for mineral soil
samples below and above the knot point of 1.0 meq/100g for CA, 0.05 meq/100g for MG, K, and NA, 0.07 meq/100g
for FE, and 0.33 for AL; a dot signifies a lack of data occupying that range.
77
-------
(a)
CA_CL2
Extractable Calcium in Calcium Chloride
Laboratory Audit Samples
U.Ub-
"cr
O
o
S"0.04-
£,
c
g
"o
">0.02-
a
o
"O
Jo. oo-
a
QL
O
a
S
a4
A
&
a a *»
tn 4
a 4 4 /\
r" QO
/\ /\ ,.-
0.0 0.2 0.4 0.6 0.8
Mean (meq/100g)
1.4
1.6
D a B
Bw
0
(b)
0.3^
CT>
O
O
CT
0.2-
c
_o
-t-V
o
o
X)
c.
o
.1 -
0.0-
0.2
Routine Samples
0.4 0.6
Mean (meq/100g)
0.8
Lab Audits (Within-Batch)
+ + * Mineral Routine S/H Groups Reid Dups.
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: One mineral, thirteen organic sampling class/horizon groups exceed plot boundaries.
Figure 3-20. Range and frequency distribution of CA CL2 for. (a) laboratory audit samples and their relation
to the analytical within-batch precision~measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
78
-------
(a)
0.03-T
O
O
c
o
g
>
Cl>
Q
~D
L_
O
J o.oo-I
in
0.00
MG_CL2
Extractable Magnesium in Calcium Chloride
Laboratory Audit Samples
A &
A
606
AA
0.02 0.04 0.06 0.08
Mean (meq/100g)
0.68
0.70
a a a
B
Bw
0
(b
0.25-
,- s
en
O
2 0.20.
cr
-------
(a)
K-CL2
Extractable Potassium in Calcium Chloride
Laboratory Audit Samples
o
o
£"0.010-
c
g
| 0.005-
a
T3
0
"D
oO.OOO-
CO
§
o
o
AA a en
a AJJ 4 DD D
i/MQO
I
/v /\ /\
i 1 1 V V V 1 I i
0.00
(b)
0.10-T
0.02 0.04 0.58
Mean (meq/100g)
a a a B A A A Bw 0
Routine Samples
0.60
0.62
C"
|o.08j
cr
a>
J-0.06-I
c
o
"oQ.04-
0.00
0.02
0.04
0.06
0.08
0.10
Mean (meq/100g)
Lab Audits (Within-Batch)
+ * Mineral Routine S/H Groups Field Dups. "
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: One mineral, twelve organic sampling class/horizon groups exceed plot boundaries.
Figure 3-22. Range and frequency distribution of K CL2 for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
80
-------
(a)
0.02-r
o
o
cr
0)
0.01-
Q
O
"O
c
00
0.00
NA_CL2
Extractabie Sodium in Calcium Chloride
Laboratory Audit Samples
/MOO
-T 'VW r
0.02 0.04 0.06
Mean (meq/100g)
a a a B AAA Bw 0
0.30
0.32
0.(
01
o
o
cr
(b)
0.02-
c
O
^0
Q
"D
O
01-
Routine Samples
0.00-
0.00
0.01 0.02
Mean (meq/100g)
0.03
0.04
+ + + Mineral Routine S/H Groups --- Field Dups.
Organic Routine S/H Groups ---- Preparation Dups. -- Lab Audits (Between-Batch)
Lab Audits (Within-Batch)
Note: Five organic sampling class/horizon groups exceed plot boundaries.
Figure 3-23. Range and frequency distribution of NA_CL2 for (a) laboratory audit samples and their relation
to the analytical within-batch precision'measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
81
-------
(a)
en
§0.010
S"o.008-
§0.006^
>0.004j
Q
"20.002J
o
T>
0.000 -I
i ooao
an ODD
FE_CL2
Extractable Iron in Calcium Chloride
Laboratory Audit Samples
'MQO
GO
0.00
0.02 0.04 0.06
Mean (meq/100g)
0.08
a a a
B
Bw
0
(b
.04-
"o>
0
o
>0.03.
QJ
§0.02-
o
cu
^ O.OL
o
c
-0.00^
*
+
+
..
*
+ + + + .
4^- _
i
o
0.00
0.02 0.04
Mean (meq/100g)
0.06
0.08
+ »» Mineral Routine S/H Groups Field Dups. - ,
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Lab Audits (Within-Batch)
Note: Four mineral, five organic sampling class/horizon groups exceed plot boundaries.
Figure 3-24. Range and frequency distribution of FE_CL2 for. (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
82
-------
AL_CL2
Extractable Aluminum in Calcium Chloride
(a)
0.064
en
O
O
S~0.04.
c
O
'I 0.02J
Q
-o
o
c
o
GO
0.00-
0.0
Laboratory Audit Samples
0.1 0.2
Mean (meq/100g)
0.3
a a a
B
Bw
0.4
o.:
(b)
o
o
cr
15
Q)
Q
0.10-
0.05-
^0.00-
0.0
Routine Samples
0.1 0.2
Mean (meq/100g)
0.3
0.4
+ ** Mineral Routine S/H Groups Field Dups. Lab Audits (Within-Batch)
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: Three mineral, seven organic sampling class/horizon groups exceed plot boundaries.
Figure 3-25. Range and frequency distribution of AL CL2 for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MOO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
83
-------
Extractable Iron, Aluminum, and
Silicon
The analytical within-batch precision
MQOs for the extractable iron, aluminum, and
silicon parameters were satisfied except for
SI_AO above the knot (see Table 3-8). Extrac-
table silicon was a new parameter that had
not been measured in the previous DDRP
surveys, and the laboratories apparently had
difficulty in the analysis of SI_AO for upper tier
precision. Unfortunately, there were no upper
tier data for the preparation duplicates or field
duplicates to confirm this hypothesis. There-
fore, the QA staff has concluded that the
upper tier analytical within-batch MQO of 10
percent RSD for this parameter might have
been too restrictive for the laboratories to
satisfy on a sustained basis. An MQO of 15
percent RSD or less may be a better expecta-
tion for the analytical within-batch precision,
since it is thought that the possibility of in-
complete extraction of silicon introduces more
variability into the measurement of this cation
than the other extractable cations.
The preparation and field within-batch
MQOs were satisfied for all parameters in this
group, except for the PD lower tier of FE AO
which only slightly exceeded the MQO." A
general pattern of increasing standard devia-
tion with increased sources of confounded
error was maintained both below and above
the knot.
Table 3-8. Achievement of Measurement Quality Objectives for Wlthln-Batch Precision of Extractable Iron,
Aluminum, and Silicon
Data
set3 Parameter
Below the knotc
SD MQO
_M yy^ O£
Pairs>MQO
n %
Above the knote
df
RSD MQO Pairs>MQO
_ at n w
fQ MM || JQ
LAP
PD
FD
FE_PYP
AL_PYP
FE AO
AL~AO
SI_AO
FE CD
AlTCD
FE_PYP
AL PYP
FE>O
AL_AO
SI_AO
FE_CD
AL_CD
FE PYP
AL~PYP
FE~AO
AL~AO
SI_AO
FE CD
AL~CD
18
32
45
35
41
46
37'
16
22
18
20
23
18
0.0154
0.0133
0.0151
0.0361
0.0189
0.0111
0^0212
0.0213
0.0199
0.0553
0.0273
0.0104
0!0175
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.06
0.06
0.06
0.06
0.06
0.06
0.06
1
3
2
1
1
5.6
3.1
6.7
5.7
2.4
2.2
2.7
0.0
0.0
5.6
5.0
0.0
o!o
27
27
27
27
9
27
27
22
9
19
13
54
17
11
5
9
7
27
9
3.4
5.5
4.2
4.9
12.0
3.1
6.7
5.3
5.0
5.8
5.7
^5.3
3.4
6.1
9.4
7.5
11.8
6.4
3.7
10
10
10
10
10
10
10
15
15
15
15
15
15
15
30
30
30
30
30
30
30
0
1
0
1
3
0
2
0
0
0
1
3
0
0
0
0
0
o'
0
0.0
3.7
0.0
3.7
33.3
0.0
7.4
0.0
0.0
0.0
7.7
5.6
0.0
0.0
0.0
0.0
0.0
0.0
0.0
S/H
FE PYP
AL PYP
FE AO
AL~AO
SI~AO
FE~CD
AL CD
294
539
286
551
614
4
437
0.1869
0.1384
0.2032
0.1152
0.0285
0.1347
0.1070
324
79
332
67
614
181
81.1
111.1
71.9
61.4
44^9
46.6
* LAP = Laboratory audit pairs; PD = Preparation duplicates; FD = Field duplicates; S/H = Sampling class/horizon
routine samples.
* Degrees of freedom.
c Standard deviations and relative standard deviations expressed in weight percent and percent, respectively, for mineral
soil samples below and above the knot point of 0.3 weight percent; a dot signifies a lack of data occupying that range.
84
-------
Figures 3-26 through 3-32 are plots of
the laboratory audit sample data in relation to
the within-batch MQOs and of the routine
sample data in relation to selected QE sam-
ples. Supplemental information regarding
uncertainty estimates is presented in Appendix
C and the identification of inordinate data
values is presented in Appendix D.
85
-------
FE_PYP
Extractable Iron in Sodium Pyrophosphate
Laboratory Audit Samples
KO
-g
C
O
-4-*
O
'>
CD
Q
0
T3
C
O
GO
(a
0,"\ f*
.06-
0.04-
0.02-
0.00-
0
1
, j
Q
ce
o
«
0 0.2
;
X^ 00
"c.
a
a 4
% a
A&
0 a
Q a ^
o
a 4
a
p D A
A
0.4 0.6 0.8 1.
Mean (wt z)
a a o B A A A Bw 0
0.8-1
(b)
-0.6-
c
o
350.
Q
~D
O
O
CO
4.
.2-
0.0-
Routine Samples
:< -
0.0
0.2
0.4 0.6
Mean (wt %)
0.8 1.0
Lab Audits (Within-Batch)
+ + + Mineral Routine S/H Groups Field Oups. -
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Botch)
Note: Three mineral, one organic sampling class/horizon groups exceed plot boundaries.
Figure 3-26. Range and frequency distribution of FE_PYP for: (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
86
-------
(a)
o.isr
o 0.10-1
o>
Q
0.05-
c
o
CO
0.00-1^
0.0
AL_PYP
Extractable Aluminum in Sodium Pyrophosphate
Laboratory Audit Samples
0.2 0.4 0.6
Mean (wt 7.)
a a a B A A A Bw 0
MOO
0.8
1.0
0.6^
(b)
§0.4
0)
Q
0.2
C
O
-t-J
C/)
0.0
0.0
Routine Samples
0.2
0.4
0.6 0.8
1.0
Mean (wt
» Mineral Routine S/H Groups --- Field Dups.
nera oune roups --- e ups. - Lab Audits (Within-Batch)
Organic Routine S/H Groups ---- Preparation Dups. -- Lab Audits (Between-Batch)
Note: Two mineral, three organic sampling class/horizon groups exceed plot boundaries.
Figure 3-27. Range and frequency distribution of AL PYP for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
87
-------
(a)
0.1 CM
c
g
g
Q
o
\_
O
0.06-
0.04-
0.02^
0.00-
0.0
FE_AO
Extractable Iron in Acid Oxalate
Laboratory Audit Samples
'MOO
A/W
0.2 0.8
Mean (wt n)
a a a B A A & Bw 0
1.0
1.2
(b)
0.6-r
o.
0)
Q
4-
0.2-
c
o
-*
00
0.0-
Routine Samples
** *
*
0.0
0.2
0.4 0.6
Mean (wt %)
0.8
1.0
Note:
Reid Dups. Lab Audits (Within-Batch)
Preparation Dups. Lab Audits (Between- Batch)
Four mineral, one organic sampling class/horizon groups exceed plot boundaries.
Mineral Routine S/H Groups
Organic Routine S/H Groups
Figure 3-28. Range and frequency distribution of FE AO for (a) laboratory audit samples and their relation
to the analytical within-batch precision'measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
88
-------
(a)
0.20-f
-0.15.
c
g
-4'
g
'£0.10.
Q
O
O
(7)
0.0
AL_AO
Extractable Aluminum in Acid Oxalate
Laboratory Audit Samples
woo
0.5 1.0
Mean (wt z)
a D D B & & A Bw 0
1.5
0,
0
c
o
'§0
>
a>
Jo-
en
(b)
.5-
.4-
.3-
.2-
.1.
0.0-
0.0
Routine Samples
+ -t
0.1
0.2
0.3
Mean (wt
0.4
0.5
0.6
Note:
Mineral Routine S/H Groups --- Reid Dups.
Lab Audits (Within-Batch)
.
Organic Routine S/H Groups ---- Preparation Dups. -- Lab Audits (Between -Batch)
One mineral, four organic sampling class/horizon groups exceed plot boundaries.
Figure 3-29. Range and frequency distribution of AL AO for (a) laboratory audit samples and their relation
to the analytical within-batch precieion~measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
89
-------
SI_AO
Extractable Silicon in Acid Oxalate
Laboratory Audit Samples
(a]
0.10-
^
^
jgO.08-
c
o
= 0.06-
'>
O
^0.04-
I.
o
§0.02-
CA
o.oo.
0
(b
0/-x /-»
.06-
§0.04
'o
>
0)
Q
T)
,50.02-
"O
c
o
-4 '
(/)
0.00-
0.
_j 1
Q O
cc z
o ^
A
D
MQO
^____ -"
^^^-^~~^^
4 q '
a
4 4 a A
A
sa
A 4A A S a a a
4 A4 °
k a
i ) i * i ' i
.0 0.1 0.2 0.3 0.4 0
Mean (wt *)
a a a B A A A Bw 0
Routine Samples
+ +
*
4.
4-
* *
* + +
+ +
+
* * + *
+ * + +* .
* * <.+*+*
*
* +*.*+»
i. **. + *
T * * + ^
+ *** ++ *
*++ * + + + » * *
*;+*+++ *
- - - - - i. i - ^ J^fc1 TI. » -.A < -.. ^ . ^ii ..!.. .1 . n - - ^ - - - ii ' . -^^-~
* * t +*
* + *+
. ' - * +
1 ' 1 ' 1 ' 1 ' 1 1
00 0.01 0.02 0.03 0.04 O.C
5
5
Mean (wt 7.]
+ * Mineral Routine S/H Groups Field Dups. Lab Audits fWithin-Batch)
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Bated)
Note: Four mineral sampling class/horizon groups exceed plot boundaries.
Figure 3-30. Range and frequency distribution of SI AO for (a) laboratory audit samples and their relation
to the analytical within-batch precision~rneasurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
90
-------
(a)
0.15^
oO.IOj
0)
Q
TJ
O
C
O
-*-*
(T>
0.00^
0.0
FE_CD
Extractable Iron in Citrate Dithionite
Laboratory Audit Samples
a o a B
(b)
2.5-f
2.0-
c
O
0)
Q
.5-
Jo.5-1
CO
0.0-
Routine Samples
-T-*--
o
r~
2
Mean (wt Js)
* -i- Mineral Routine S/H Groups Field Dups. Lab Audits (Within-Batch)
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: Three mineral sampling class/horizon groups exceed plot boundaries.
Figure 3-31. Range and frequency distribution of FE CD for (a) laboratory audit samples and their relation
to the analytical within-batch precision "measurement quality objective (MOO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
91
-------
(a)
0.25^
50.20
c
o
0.15-
Q
T3
O
0.10-
CO
0.0
ALCD
Extractable Aluminum in Citrate Dithionite
Laboratory Audit Samples
0.5
a a a
B
Mean (wt
A A A BW
1.0
0
MQO
1.5
0.3-1
).2-
-------
Extractable Sulfate and Sulfate
Adsorption Isotherms
The analytical within-batch precision
MQOs for the sulfate parameters were satis-
fied except for SO4JH2O, in which the lower
tier MQO was slightly exceeded because of the
inordinate effect of one laboratory audit sam-
ple pair (see Table 3-9). The analytical MQO
for the SO4JD parameter was satisfied, de-
spite the fact that 46 percent of the individual
audit sample pairs exceeded the MQO. This
seems to have been a random error, however,
because there was no evidence of high
between-batch imprecision by a particular
laboratory. Also, there was no pattern of this
type seen in the preparation and field duplicate
samples. The preparation and field within-
batch MQOs were satisfied for all of the
sulfate parameters.
There was a general trend of increasing
standard deviation with increased sources of
confounded error for the sulfate parameters.
Table 3-9. Achievement of Measurement Quality Objectives for Within-Batch Precision of Extractable Sulfate and
Sulfate Adsorption Isotherms
Data
set"
LAP
PD
FD
S/H
Below the knot*7
Parameter
SO4 H2O
S04~PO4
S04 0
S04 2
SO4~4
SO4 8
S04 16
SO4~32
S04 H20
SO4~P04
S04 0
SO4 2
S04 4
SO4 8
SO4 16
S04_32
S04 H2O
S04 PO4
SO4 0
S04 2
SO4 4
SO4 8
SO4 16
SO4~32
SO4 H20
SO4 PO4
SO4 0
S04~2
S04~4
SO4 8
SO4 16
S04~32
df*
11
.
14
1
.
34
2
41
12
19
1
21
7
370
49
546
78
13
1
f
so
1.6719
o!o539
0.0608
1.3116
1.4634
0.0426
0.0404
t
t
1.6254
2.5597
0.1098
0.0734
.
5.1195
4.6368
0.5822
0.7585
0.9980
0.6322
MOO
iVI\riW
1.50
1.50
0.10
0.10
0.10
0.10
0.10
0.10
1.50
1.50
0.10
0.10
0.10
0.10
0.10
0.10
3.00
3.00
0.20
0.20
0.20
0.20
0.20
0.20
Pairs>MQO
n V.
1 1 70
1 9.1
o o!o
0 0.0
2 5.9
1 50.0
0 0.0
0 0.0
.
1 5.3
0 0.0
2 9.5
0 0.0
Hf
Ul
16
27
13
26
27
27
27
27
20
52
13
42
54
54
54
54
8
26
6
20
27
27
27
27
248
569
72
540
605
617
618
618
Above the knotc
RSD MQO
7.3
9.2
5.1
2.0
2.9
2.3
2.6
3.7
6.3
6.6
1.7
2.6
2.5
2.0
1.8
4.7
7.8
10.1
2.6
3.4
4.3
3.9
2.7
6.1
278.6
152.7
287.5
236.5
175.9
115.4
64.8
34.6
10
10
5
5
5
5
5
5
15
15
8
8
8
8
8
8
30
30
15
15
15
15
15
15
Pa Irs > MQO
no/
2
1
6
0
3
1
1
2
1
2
0
0
1
0
1
2
0
2
0
0
0
0
0
1
A>
12.5
3.7
46.2
0.0
11.1
3.7
3.7
7.4
5.0
3.8
0.0
0.0
1.9
0.0
1.9
3.7
0.0
7.7
0.0
0.0
0.0
0.0
0.0
3.7
* LAP = Laboratory audit pairs; PD = Preparation duplicates; FD = Field duplicates; S/H = Sampling class/horizon
routine samples.
b Degrees of freedom.
0Standard deviations and relative standard deviations expressed in reporting units and percent, respectively for mineral
soil samples below and above the knot point of 15.0 mg S/kg for S04JH20 and S04_P04 (2.0 mg S/L for S04 0-32V
a dot signifies a lack of data occupying that range.
93
-------
Figures 3-33 through 3-40 are plots of
the laboratory audit sample data in relation to
the within-batch MQOs and of the routine
sample data in relation to selected QE sam-
ples. Supplemental information regarding
uncertainty estimates is presented in Appendix
C and the identification of inordinate data
values is presented in Appendix D.
94
-------
-6-f
(a)
SO4.H20
Extractable Sulfate in Water
Laboratory Audit Samples
cr>
CO
'4-
c
o
a
2-
o
"O
C
O
-*'
in
0
°a °a
MOO
0
i
10
20 30
Mean (mg S/kg)
ODD
B
Bw
0
40
50
(b)
a>
_*:
\
CO
en
20.
O
-4*
10-
O
T3
C
o
-t '
in
0-
Routine Samples
10
20
30
Mean (mg S/kg)
40
50
+ + + Mineral Routine S/H Groups Field Dups. Lob Audits (Within-Batch)
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: One mineral, six organic sampling class/horizon groups exceed plot boundaries.
Figure 3-33. Range and frequency distribution of S04_H20 for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
95
-------
(a)
CJi
00
01
£
40-
§30-1
"o
c
o
-*'
GO
0-
S04_PO4
Extractable Sulfate in Sodium Phosphate
Laboratory Audit Samples
a B
0 20 40 60 80 100
Mean (mg S/kg)
a a a B A A A Bw 0
MQO
I 20
140
(b)
100-f
oo
CP
80-
o 6°-i
-4'
O
Q
T3
\~_
O
C
o
-t'
00
40-
20-
0-
0
Routine Samples
20
40 60 80 100
Mean (mg S/kg)
120
I 40
+ + + Mineral Routine S/H Groups Reid Oups. Lab Audits (Within-Batch)
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: Two mineral sampling class/horizon groups exceed plot boundaries.
Figure 3-34. Range and frequency distribution of S04 PO4 for. (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
96
-------
SO4_0
Sulfate isotherm 0 mg S/L
Laboratory Audit Samples
^U.ZU-
_J
CO
CJ1
£0.15-
c
o
"o o 10-
Q
aO.05-
~o
c.
o
CO
0.00-
0
.
o
on
0
a
o
Q
a D
3 0.5 1.0 1.5 2
5
4 A
4 MQO
A
a
A
0 2.
a a
Mean (mg S/L)
A A A Bw
(b)
_o n
en ^-. v-/
JL
1C
O I .O -
"o
ILO^
o
?0.5^
o
0.0-
0.0
Routine Samples
0.5
1.0 1.5
Mean (mg S/L)
2.0
2.5
Mineral Routine S/H Groups Field Dups.
Lob Audits (Within-Batch)
Preparation Dups. Lab Audits (Between-Batch)
Note: Four mineral sampling class/horizon groups exceed plot boundaries.
Figure 3-35. Range and frequency distribution of SO4 0 for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and tneir relation to pooled precision estimates.
97
-------
0
SO4_2
Sulfate Isotherm 2 mg S/L
Laboratory Audit Samples
(a)
OH f
.15-
1_T
^\
to
CD
E
~ '0 10
o
o
Q)
?0.05-
o
"O
c
o
GO
0.00-
1
Q
CC
O
a
----- i- - -
i /MQO
^- */^
s^
,s
/°
/^
&
a
4
4 4
a
J
a a
a
° 0
0 gO 4
a 44
Mean (mg S/L)
a a a
B
A A A
Bw
_2.5H
lj"
oo
a>2.0j
1C
O I .D
"g
I 1-0-1
"g 0-5-1
0.0
a»
Routine Samples
. .
1 *
J:v *
1.*+*?+ + \ +
"** +* -f +
4 r^r= =rI-= :^jgjgu.-»fay-^i
0
Mean (mg S/L)
* * + Mineral Routine S/H Groups Field Dups.
Lab Audits (Within-Botch)
Preparation Dups. Lab Audits (Between-Batch)
Note: Four minerol sampling class/horizon groups exceed plot boundaries.
Figure 3-36. Range and frequency distribution of SO4 2 for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MOO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
98
-------
(a)
0.25-1
0*0.20-1
.§0.15-1
2
I 0.10
0.05-
O
-t '
co
0.00-
1
S04_4
Suifate Isotherm 4 mg S/L
Laboratory Audit Samples
D
D o
<* °
O °
r^
3 4
Mean (mg S/L)
a a a B A A A Bw
(b)
Routine Samples
in
en
^E
c
o
-*-*
o
0)
Q
-o 1
o
T)
C
o
o.
v *
+**
234
Mean (mg S/L)
Mineral Routine S/H Groups Field Dups. " uuu nuuiw ittn u-.^../
Preparation Oups. Lab Audits (Setween-Botch)
Note: Four mineral sampling class/horizon groups exceed plot boundaries.
Lob Audits fWithin-Botch)
Figure 3-37. Range and frequency distribution of S04 4 for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MOO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
99
-------
(a)
0.3-r
-Q.2\
o
1E>
0.0-
S04JB
Sulfate Isotherm 8 mg S/L
Laboratory Audit Samples
'MQO
I
4
6 7
Mean (mg S/L)
a n a B
Bw
r~
8
r
10
(b)
Routine Samples
2-
_
-------
(a)
^1.2-r
_i
\
w 1 Q.
en
JE,
c
o
.20.6-1
Q
O
10
0.0
SO4J6
Sulfate Isotherm 1 6 mg S/L
Laboratory Audit Samples
a ft . *
10 14
Mean (mg S/L)
a a
aB
Bw
MQO
r
18
(b)
Routine Samples
CO
|3
"o
>
Q 2
T>
O
"c "I
O
-4'
00
0-
10 14
Mean (mg S/L)
Mineral Routine S/H Groups Field Dups.
Lab Audits (Within-Batch)
Preparation Dups. Lab Audits (Between-Batch)
Note: Three mineral sampling class/horizon groups exceed plot boundaries.
Figure 3-39. Range and frequency distribution of SO4 16 for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
101
-------
SO4_32
Sulfate Isotherm 32 mg S/L
Laboratory Audit Samples
25
Mean (mg S/L)
30
35
o a a
B
Bw
(b)
Routine Samples
00
>§
c
o
6-
.2 4-
a
c
o
00
0-
20 25
Mean (mg S/L)
Mineral Routine S/H Groups Field Dups.
30 35
Lab Audits (Within-Batch)
Preparation Dups. Lab Audits (Between-Batch)
Note: Four mineral sampling class/horizon groups exceed plot boundaries.
Figure 3-40. Range and frequency distribution of SO* 32 for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
102
-------
Total Carbon, Nitrogen, and Sulfur
The analytical, preparation, and field
within-batch precision MQOs were satisfied for
total carbon, nitrogen, and sulfur (see Table
3-10). A general pattern of increasing standard
deviation with increased sources of confoun-
ded error was maintained.
Figures 3-41 through 3-43 are plots of
audit sample data in relation to the DQOs and
routine sample data in relation to the QA
samples. Supplemental information regarding
uncertainty estimates is presented in Appendix
C and the identification of inordinate values is
presented in Appendix D.
Table 3-10. Achievement of Measurement Quality Objectives for Wlthln-Batch Precision of Total Carbon, Nitrogen,
and Sulfur
Below the knotc
Data
set3
LAP
PD
FD
S/H
Parameter
C TOT
N TOT
S_TOT
C TOT
N TOT
S_TOT
C TOT
N TOT
S_TOT
C TOT
N TOT
S TOT
df*
14
5
28
50
46
13
25
23
279
526
527
SD
0.0053
0.0011
0.0119
0.0039
0.0008
0.0502
0.0053
0.0011
0.3334
0.0342
0.0070
MQO
0.05
0.015
0.002
0.05
0.015
0.002
0.10
0.03
0.004
Pairs > MQO
n %
1
0
0
0
2
1
0
0
7.1
0.0
0.0
0.0
4.3
7.7
0.0
0.0
df
27
13
22
26
4
8
14
2
4
339
92
91
Above the knotc
RSD
IV
7.0
3.8
7.8
3.0
1.6
2.2
17.0
7.7
8.4
127.5
56.9
259.8
MQO
10
10
10
15
15
15
30
30
30
Pairs>MQO
n %
2
0
1
1
0
0
1
0
0
7.4
0.0
4.5
3.8
0.0
0.0
7.1
0.0
0.0
" LAP = Laboratory audit pairs; PD = Preparation duplicates; FD = Field duplicates; S/H = Sampling class/horizon
routine samples.
b Degrees of freedom.
0 Standard deviations and relative standard deviations expressed in weight percent and percent, respectively, for mineral
soil samples below and above the knot point of 0.5 weight percent for CJOT, 0.15 weight percent for NJTOT, and 0.02
weight percent for S_TOT; a dot signifies a lack of data occupying that range.
103
-------
0
CJOT
Total Carbon
Laboratory Audit Samples
4 6 38
Mean (wt %)
a a a B A A A Bw 0
42
(b)
4-T
-3-
C
o
52-1
Q
O
"c 1 -
o
£0
0.
Routine Samples
0 2 4 6 8 10
Mean (wt s)
+ + + Mineral Routine S/H Groups Field Dups. Lab Audits (Within-Botch)
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: Two mineral, thirteen organic sampling class/horizon groups exceed plot boundaries.
Figure 3-41. Range and frequency distribution of C TOT for. (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
104
-------
(a)
0.044
^0.03-
c
o
"o
'£0.02-
Q
| 0.01-
o
CO
0.00-
I
0.00
N_TOT
Total Nitrogen
Laboratory Audit Samples
-r-^WV
0.10 0.20 0.30
Mean (wt x)
a a a B A A A Bw 0
1.60
1.70
(b)
o.3r~
O)
Q
TD
C
O
-t-j
CO
).2-
).1 -
0.0-
0.0
Routine Samples
h-
0.2
0.4
Mean (wt
0.6
0.8
Lab Audits fWithin-Batch)
+ + + Mineral Routine S/H Groups --- Field Duos.
Organic Routine S/H Groups ---- Preparation Dups. -- Lab Audits (Between-Batch)
Note: Eleven organic sampling class/horizon groups exceed plot boundaries.
Figure 3-42. Range and frequency distribution of N TOT for (a) laboratory audit samples and their relation
to the analytical withirvbatch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
105
-------
S_TOT
Total Sulfur
Laboratory Audit Samples
0.02
ODD
B
0.04
Mean (wt
Gw
0.18
0.20
A A A
0
(b)
0.08-1
K
1^
C
O
0.06-
o
'$0.04-
Q
cO.02^
o
00
0.00-
0.00
Routine Samples
0.02
0.04 0.06
Mean (wt %)
0.08
0.10
+ + * Mineral Routine S/H Groups Field Oups. Lab Audits (Within-Batch)
Organic Routine S/H Groups Preparation Dups. Lab Audits (Between-Batch)
Note: One mineral, ten organic sampling class/horizon groups exceed plot boundaries.
Figure 3-43. Range and frequency distribution of S_TOT for (a) laboratory audit samples and their relation
to the analytical within-batch precision measurement quality objective (MQO) and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
106
-------
Assessment of Accuracy
The accuracy considerations of impor-
tance to the MASS were investigated using
data primarily derived from the laboratory audit
samples, as described in the following subsec-
tions.
Bias
The B and Bw horizon LAP samples and
the C horizon LAL samples were used to
estimate analytical bias, and the corresponding
FAP and FAL samples were used to estimate
system-wide bias. It was found that the field
and preparation components of system-wide
bias were negligible and that bias was prin-
cipally an analytical laboratory component.
Table 3-11 shows the analytical bias values
expressed in reporting units for the mineral soil
audit samples. Bias was not determined for
the organic audit samples because the low
number of organic batches analyzed resulted
in insufficient degrees of freedom for meaning-
ful statistical analysis.
Analytical bias was negligible when
compared with the SDLs of all 37 parameters
for which detection limits were established.
This comparison could not be made for the 13
physical and pH parameters. For CEC_OAC,
bias exceeded the SDL for Laboratories 1 and
2 which titrated the samples, but bias was
negligible for Laboratory 4 which used flow
injection analysis. Bias for the two newly-
established parameters for the MASS, i.e.,
AL_CL and SI_AO, were both much less than
their respective SDLs. The majority of the
analytical bias estimates for the parameters
were on the positive, or high, side of the
reference value.
As was previously stated, none of the
audit samples were assigned a single refer-
ence value for contractual compliance pur-
poses. Instead, each audit sample was
assigned a range of acceptable reference
values in accuracy windows derived from
confidence intervals constructed around mean
values established from previous DDRP labora-
tory analyses. This approach was taken for
the purpose of monitoring the laboratories to
ensure acceptable data on a batch-by-batch
basis for the duration of the survey. In this
framework, bias within the windows was
acceptable and allowable for each parameter
Table 3-11. Analytical Bias Estimate*
Parameter
SAND
VCOS
COS
MS
FS
VFS
SILT
COSI
FSI
CLAY
PH H20
PH~002M
PHJ31M
CA CL
MG~CL
K CL
NA CL
ALJDL
CA OAC
MG~OAC
K OAC
NA_OAC
CEC CLC
CEC CLrf
CEC OACC
CEC'DAC*
AC_BACL
CA CL2
MG CL2
KCL2
NA CL2
FE CL2
AL~CL2
FE PYP
AL PYP
FE AO
AL AO
SI~AO
FE'CD
AL_CD
S04 H2O
S04 PO4
SO4 0
S04~2
SO4 4
SO4~8
SO4 16
SO4_32
C TOT*
N~TOT*
S~tOT*
Reporting
units
wt %
H
H
"
H
H
«
II
H
pH units
H
H
meq/100g
H
II
II
II
meq/100g
n
H
n
meq/100g
meq/100g
H
«
H
meq/100g
H
11
"
"
"
Wt %
11
H
II
II
II
II
mg S/kg
H
mg S/L
N
H
H
M
II
wt%
11
Analytical
bias
0.3211
0.3948
0.7606
0.5632
-0.5363
-0.8188
0.9283
0.2344
0.2929
-0.6168
-0.0319
0.0277
-0.0216
-0.0028
0.0011
0.0010
-0.0013
-0.0615*
0.0005
0.0007
0.0059
0.0044
-0.0261
0.0217
-0.6132*
0.0523
-0.1611
0.0248
0.0024
0.0005
0.0012
-0.0004
0.0023
-0.0183
-0.0043
0.0099
0.0009
-0.0034s
0.0136
0.0030
0.1055
2.0341
0.0358
0.0446
0.0526
0.0546
0.1053
0.1026
-0.0352
0.0016
-0.0007
' Based on data from B horizon FAP samples only.
b Analyzed by Laboratory 2 only.
c Analysis by titration (Laboratories 1 and 2).
d Analysis by FIA (Laboratory 4 only).
* Exceeds the system detection limit (SDL).
107
-------
from a contractual standpoint. The estimation
of bias at the completion of the survey, how-
ever, included this within-window uncertainty in
the calculations of bias.
For assessment purposes, it was impor-
tant to evaluate the occurrences which were
outside the upper and lower limits of the
windows, and to identify their contribution to
the bias estimates for each parameter across
audit samples. Table 3-12 shows the per-
centage of observations that were outside the
respective accuracy windows and the mag-
nitude of their contribution to the analytical
bias estimates given previously in Table 3-11.
The results show a very wide range in the
ratios of bias for values outside the window
compared to the total analytical bias. Except
for S_TOT, every parameter had fewer than 20
percent of its values outside the accuracy
window. A high second value in the ratio
implies a lower contribution to the bias.
Parameters that had a high percentage of
observations outside the window and ratios
from 1:1 to 1:3 generally had values at varying
distances from the window, while those para-
meters that had a low percentage of observa-
tions outside the window and the same low
ratios generally had only extreme values out-
side the window.
Generally, the contribution of outside-
window values to the analytical bias is low.
Only 11 of the 50 parameters had values
outside the window that contributed more than
one-third to the bias: COSI, FSI, MG_CL,
AL CL, MG OAC, CEC CL, AL PYP, FE_AO,
SI~AO, AL_CD, and SO4_32. "Although the
percentage of samples outside the window for
S_TOT is high, the contribution to overall bias
was relatively low.
Laboratory Differences
The laboratory differences, identified
through a series of assessments of laboratory
means across audit samples in relation to the
reference values, are described in the following
subsections.
Table 3-13 shows the percent differences,
averaged across the B, Bw, and C horizon
laboratory audit samples, of the laboratory
mean values from the pooled reference value
for each parameter. Differences were not
determined for the organic samples because
Table 3-12. Contribution of Outlying Audit Samples
to Overall Analytical Bias
Laboratory audit samples
exceed ina window boundaries
Parameter
SAND
VCOS
COS
MS
FS
VFS
SILT
COSI
FSI
CLAY
PH H20
PH~002M
PH_01M
CA CL
MG CL
K CL
NA CL
AL_CL
CA OAC
MG OAC
K OAC
NAJ3AC
CEC CL
CEC OAC
AC_BACL
CA CL2
MG CL2
K CL2
NA CL2
FE~CL2
ALICL2
FE PYP
AL PYP
FE AO
AL AO
SI~AO
FE CD
AL_CD
S04 H2O
S04 P04
S04 0
S04 2
S04~4
S04~8
S04~16
804^32
C TOT
N~TOT
S~TOT
Percent"
3.7
18.5
6.2
17.3
8.6
2.5
12.3
6.2
12.3
13.6
6.2
7.4
2.5
6.2
6.2
1.2
2.5
3.0
7.4
3.7
14.8
3.7
6.2
1.2
11.1
1.2
17.3
3.7
13.5
8.6
6.2
8.6
16.0
17.3
0.0
4.0
9.9
4.9
16.1
11.1
11.1
3.7
11.1
3.7
4.9
9.9
11.1
0.0
38.3
Ratio"
1 17
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
4
1
1
1
*
1
1
1
1
1
4
1
1
J
1
1
1
1
J
18
12
6
6
47
5
1
1
68
46
4
54
3
1
10
4
1
3
1
10
22
1
44
4
100
24
5
6
4
8
100
1
2
2
4
2
5
9
15
45
12
137
177
2
4
7
a Percentage of samples outside window boundaries.
b Ratio of outside-window to overall bias.
108
-------
only one laboratory analyzed these samples
and no accuracy windows were available.
Also, elemental analysis of C_TOT, N_TOT, and
S_TOT did not exhibit interlaboratory differen-
ces because the analyses were performed only
at Laboratory 2.
For the physical parameters, VFS and
CLAY had the highest laboratory differences
from the reference values. For pH, all labora-
tory differences were two percent or less, and
represented the lowest differences overall for
any parameter group. For the cations in
ammonium chloride, MG_CL and K_CL had the
highest and lowest differences, respectively.
For the cations in ammonium acetate, NA_OAC
and CAJDAC had the highest and lowest
differences, respectively. For the CEC parame-
ters, differences for CEC_OAC were somewhat
higher than for CEC_CL For the cations in
calcium chloride, FE_CL2 and CA_CL2 had the
highest and lowest differences, respectively.
For extractable iron and aluminum, FE_PYP
and AL_PYP had the highest differences. The
differences were less than ten percent for the
extractable sulfate parameters and less than
five percent for the sulfate isotherm parame-
ters.
Generally, Laboratory 1 showed the
smallest differences and Laboratory 4 showed
the largest differences. Across all parameters,
the laboratories showed the lowest differences
for the pH parameters and the highest dif-
ferences for CLAY and FE_CL2.
As a summary assessment for laboratory
differences, an average laboratory difference
from the pooled reference value was calcu-
lated for each parameter group. The labora-
tory differences were all less than six percent
of their respective pooled reference values ex-
cept for the data from Laboratory 1 for the
cations in ammonium acetate and the data
from Laboratory 4 for the cations in calcium
chloride; both average differences were ap-
proximately 10 percent.
Table 3-13 also shows the laboratories
that had significant differences identified by
Scheffe's pairwise multiple comparison test of
laboratory means. Results from the paired
comparisons are presented for laboratories
which were significantly lower, defined at the
.05 level of significance, than one or both of
the other laboratories. About 54 percent of
the parameters (25 of 46 parameters) showed
significant laboratory differences. Four of the
parameters (SILT, PH_002M, CEC_OAC, and
SI_AO) showed all three laboratories to be
significantly different from each other. There
were three parameters where Laboratory 1
was significantly different from the other two
laboratories, five parameters for Laboratory 4
and six parameters for Laboratory 2.
There was no consistent pattern of
differences for any specific laboratory in most
of the parameter groups. However, for the
cations in calcium chloride, Laboratory 2 was
significantly different from the other laborato-
ries in all cases shown. For the sulfate iso-
therm parameters, Laboratory 4 was signif-
icantly lower than the other laboratories in all
four of the cases that showed differences.
These latter two results indicate that, despite
all laboratories adhering to specified protocols
in a strictly controlled QA program, a labora-
tory can show a consistent significant dif-
ference from other laboratories in its sub-
mitted data and still satisfy the batch accep-
tance criteria.
Table 3-13 also presents the average
deviation from the reference value, expressed
in the parameter reporting units, for each
parameter and laboratory. These values
indicate the expectation of the absolute devia-
tion, without regard to sign, exhibited by each
laboratory as a function of bias and impre-
cision in relation to the reference value for
each parameter. Only five parameters were
significantly large, i.e., the ratio of the squared
deviation to the variance of the prior reference
distribution was greater than 3:1. These
parameters included AL_PYP and SO4_PO4 for
Laboratory 1, COSI and FSI for Laboratory 2,
and PH_002M for Laboratory 4.
Laboratory Trends
Plots of the laboratory moving averages
for the laboratory audit samples are presented
in Appendix I. The plots occasionally depict
situations when a particular laboratory showed
an upward, downward, or irregular trend over
time for a given parameter. The Y-axes of the
plots are expressed in divisions of reporting
units but are not enumerated because the
109
-------
Table 3-13. Analytical Laboratory Differences Pooled Across the Laboratory Audit Samples
Parameter
SAND
VCOS
COS
MS
FS
VFS
SILT
COSI
FSI
CLAY
PH H20
PH 002M
PH_01M
CA CL
MG CL
K CL
NA CL
AL_CL6
CA OAC
MG OAC
K OAC
NA_OAC
CEC CL
CEC OAC
AC_BACL
CA CL2
MG CL2
K CL2
NA CL2
FE CL2
AL_CL2
FE PYP
AL PYP
FE AO
AL AO
SI AO6
FE CD
AL~CD
S04 H2O
S04~PO4
S04 0
SO4 2
SO4 4
S04 8
S04 16
S04 32
Reporting
units
wt %
H
II
II
H
H
11
H
II
II
pH units
11
H
meq/100g
"
11
H
ii
meq/100g
"
H
ii
meq/100g
H
11
meq/100g
n
11
n
M
"
Wt %
II
II
II
II
II
H
mg S/kg
II
mg S/L
11
11
n
H
n
Percent
Li
-1
11
12
3
-7
-7
4
6
-1
-18
-1
1
-1
3
12
-1
-7
-3
6
7
12
16
-1
3
-6
5
1
-9
2
-11
16
-1
4
1
1
-7
-1
-4
2
7
4
4
3
3
4
2
difference
Li
5
11
4
4
3
6
-5
-10
-2
6
1
2
-1
-3
-10
5
-5
2
-2
-2
11
12
-1
-13
-4
1
7
15
6
-36
-10
-2
-2
2
-1
7
3
4
1
4
3
1
3
1
1
3
from R
L4
-3
8
11
5
-6
-19
9
8
7
-36
-2
-1
-1
-3
4
1
-1
-1
-3
-2
11
17
-1
-1
8
6
12
3
8
2
27
-8
-5
1
1
-1
1
2
1
-1
1
1
-2
-1
-3
-4
Pairwise lab
difference"
4 < 1,2
4 < 2
2 < 1 < 4
2 < 1.4
4,1 < 2
4 < 1 < 2
4 < 1
4,2 < 1
2 < 4,1
4,2 < 1
2 < 1
2 < 4 < 1
2,1 < 4
2< 1
1,4 < 2
2 < 4,1
2 < 4,1
4 < 1,2
2,4 < 1
2 < 1
1 < 4 < 2
4 < 1,2
4 < 1
4 < 2 < 1
4 < 1,2
Average
LI
1.41
1.17
1.82
1.00
2.10
1.49
2.81
2.47
1.08
0.95
0.11
0.13
0.07
0.04
0.02
0.01
0.01
0.54
0.03
0.01
0.01
0.01
0.19
1.54
1.68
0.07
0.01
0.00
0.01
0.00
0.01
0.04
0.07*
0.08
0.06
0.04
0.13
0.07
2.11
12.96*
0.13
0.14
0.19
0.27
0.73
1,12
deviation
L2
3.37
1.05
1.31
1.53
1.20
2.54
3.81
5.71*
5.27*
1.97
0.07
0.13
0.05
0.02
0.01
0.01
0.01
0.17
0.03
0.02
0.01
0.01
0.36
1.96
1.84
0.06
0.01
0.00
0.00
0.00
0.01
0.04
0.03
0.09
0.04
0.05
0.09
0.08
1.10
7.57
0.11
0.12
0.19
0.18
0.40
1.35
from R
LI
3.52
2.02
3.43
1.78
3.18
2.83
5.39
3.31
2.55
1.95
0.15
0.21*
0.07
0.02
0.01
0.00
0.00
0.18
0.02
0.01
0.01
0.01
0.52
1.80
2.30
0.05
0.01
0.00
0.00
0.00
0.01
0.06
0.05
0.06
0.07
0.01
0.04
0.06
1.91
4.69
0.11
0.13
0.18
0.40
0.69
1.84
R value
- units -
55.65
3.95
8.61
14.26
16.51
12.30
39.81
20.94
19.27
3.91
5.14
4.71
4.56
0.23
0.05
0.06
0.03
3.53
0.20
0.05
0.05
0.03
4.11
11.83
15.40
0.61
0.04
0.01
0.02
0.01
0.02
0.50
0.48
0.67
0.78
0.32
1.10
0.59
12.11
62.17
1.26
2.53
3.87
6.61
12.84
26.07
a Significant at the .05 level of significance by Scheffe's multiple comparisons test.
b Based on differences for B horizon FAP samples only.
* Significantly high (> 3:1) ratio of squared deviation to variance of the prior reference value (R) distribution.
110
-------
window boundaries have been kept confiden-
tial to ensure future use of the soil audit
samples in other programs.
Generally, the trend data for the audit
samples were well within the accuracy win-
dows for most parameters and, in most cas-
es, approached the reference values. There
were a few exceptions, e.g., MG_CL2 for the C
horizon audit sample and K_OAC for the Bw
horizon audit sample, where the laboratories
were grouped near the upper boundary of the
accuracy window. This indicates that the
window boundaries used for the MASS might
not have entirely represented the audit sample
concentrations or that the changes in analyti-
cal procedures for the MASS possibly affected
the comparability of analytical results between
the MASS and the previous DDRP surveys from
which the windows were calculated.
There were also parameters, e.g.,
AC_BACL for the B horizon audit sample,
where one laboratory showed a tendency for
its data to be substantially farther from the
central tendency of the window than the other
laboratories, possibly indicating that the labor-
atory was experiencing difficulty with the
analysis. In many cases, e.g., SILT for the B
horizon audit sample, a general converging
central tendency was observed over time. This
could indicate that the laboratories were
gaining analytical experience at the possible
expense of the survey or that the QC program
was effectively keeping the laboratories in
control of the analysis. There were also a few
parameters, e.g., CEC_OAC for the B horizon
audit sample and CA_OAC for the Bw horizon
audit sample, where a laboratory would initially
exceed the window boundaries and subse-
quently reverse the trend to fall within the
window. This pattern suggests that the
QA program was successful in controlling
between-batch measurement quality.
Finally, there were several parameters,
e.g., SO4_8 and FE_CL2 for the C horizon audit
sample, where the data converged near the
reference value although the boundaries of the
window were relatively wide. This pattern was
most evident for the parameters which were
analyzed by a single laboratory, e.g., N_TOT for
the Bw horizon audit sample.
The trend plot for S_TOT shows an
extremely wide scatter of data outside of the
accuracy window for the Bw horizon audit
sample (see Figure 1-50 in Appendix I). This
characteristic did not initially cause concern
because there was no particular trend as-
sociated with the scatter, and the overall QART
acceptance criteria were satisfied. The subse-
quent assessment of bias, however, indicated
an inordinately high percentage of observa-
tions outside of the window, as shown previ-
ously in Table 3-12. The QA staff discovered
that the Bw horizon window boundaries for
this parameter had been calculated incorrectly
and were very narrow in relation to the normal
scatter of the data and in relation to the other
laboratory audit samples. If real-time assess-
ment of control charts had been a priority
procedure, such an irregular outside-window
scatter would have been evaluated and moni-
tored. Fortunately, the uniform application of
the QART criteria enabled the QA staff to
ensure control of the laboratories' sulfur analy-
sis and the resulting S_TOT data had accep-
table accuracy. Recalculation of the Bw hori-
zon accuracy window indicated that all of the
S_TOT data would have been within the boun-
daries of the corrected window. This is a
good example of the usefulness of control
charts for real-time assessment and for guid-
ing the adjustment of the QA program and
acceptance criteria as a survey progresses.
Assessment of
Representativeness
All pedons sampled were within the
range of morphological characteristics outlined
in their respective sampling classes (Kern and
Lee, in press). Hence, the MQO for represen-
tativeness of the field sampling was satisfied.
The homogenization and subsampling
procedures at the preparation laboratory
produced representative analytical soil sam-
ples of known and accepted quality (Papp and
Van Remortel, in press). More information on
this characteristic of the data can be found in
the precision discussions of this report, where
assessments of the preparation duplicates are
made.
Histograms of the range and frequency
distribution of the routine samples, field
duplicates, preparation duplicates, and nat-
ural audit samples for each of the parameters
are presented in Appendix G. The analyte
111
-------
concentrations in the field duplicates generally
were representative of the routine sample
concentrations. This was important in defining
the confidence a user will have of the data
uncertainty estimates calculated from these
two data sets. The exceptions were mainly
observed in the cations and the extractable
sulfate parameters, and are presented in Table
3-14. A set of POS, PS,, and Pw percentiles are
provided in the table as benchmark values for
the distributions. Of the 15 parameters iden-
tified, most are representative of the range of
concentration but not of the distribution within
the range. The analyte concentrations in the
preparation duplicates generally were represen-
tative of the field duplicate concentrations
because the preparation duplicates were
subsampled directly from the field duplicates.
The audit samples usually encompassed the
overall range of data from the routine samples.
Assessment of Completeness
Soil sampling protocols specified the
sampling of all designated pedons. All of the
150 pedons initially selected were sampled,
resulting in 100 percent completeness (Kern
and Lee, in press). As specified in the prepar-
ation laboratory protocols, each batch of
samples sent to an analytical laboratory con-
tained the appropriate measurement quality
samples, including a pair of preparation dupli-
cate samples that were split from a field
duplicate and its respective routine sample.
The required soil analyses and sample proces-
sing tasks at the preparation laboratory were
performed on 100 percent of the field bulk soil
samples that satisfied the sample receipt
verification criteria (Papp and Van Remortel, in
press).
The number of AO flags and J, M, or X
tags assigned to the 844 routine samples was
used to assess analytical completeness of the
MASS verified data base. These flags and
tags denoted missing data and lack of suffi-
cient sample for analysis, respectively (Papp et
al., 1989). There were no such flags or tags
assigned to the verified data base, indicating
completeness of 100 percent for analysis of all
parameters as shown in Table 3-15. This
satisfied the MQO of 90 percent for analytical
completeness.
Five levels of confidence, ranging from 0
to 4, were assigned during data validation
activities to segregate and classify data in the
validated data base. Data exhibiting levels of
confidence of 0, 1, or 2, i.e., data appended
with fewer than two major flags or fewer than
one major and two minor flags, were used to
assess completeness in the validated data
base (Turner et al., in preparation). Complete-
ness of 98.6 percent or higher for all parame-
ters was achieved for the validated data, as
documented in Table 3-15.
Assessment of Comparability
The comparability issues of primary
importance to the quality of the DDRP data
bases are the: (1) statistical methods, (2)
types of measurement quality samples used to
evaluate the quality attributes, (3) sampling,
preparation, and analytical methods, and (4)
the user-defined data acceptability criteria. If
these characteristics are comparable among
the three DDRP surveys, a quantitative com-
parison of differences among the surveys
allows the user to discern which data may not
be comparable for a given purpose.
Comparison of Statistical Methods
During the data analysis phase of the
NE and SBRP surveys, it was observed that
the error variance for most parameters was a
function of analyte concentration (Van Remor-
tel et al., 1988; Byers et al., 1989). The initial
statistical analysis of the NE region data was
performed using a logarithmic transformation
of the analytical data for each parameter.
This approach tended to inflate the error
estimate for low concentration samples in an
inordinate manner. In the SBRP and MASS
analysis, a step function was used to estimate
error variance such that each step value repre-
sented an error variance for a specific con-
centration interval. The NE data were subse-
quently re-evaluated using this approach so
that the statistical analysis would be consis-
tent among all three surveys.
Comparison of Measurement Quality
Samples
Five audit samples (Oa, A, Bs, Bw, and
C horizons) were used for QE purposes in the
NE and SBRP surveys, and four audit samples
(O, B, Bw, and C horizons) were used in the
MASS. It was discovered at the end of the
112
-------
Table 3-14. Significant Differences In the Distribution of Field Duplicate* In Relation to the Routine Samples
Parameter
Data Set"
Mean
os
P
K
P "
95
a RS = routine samples; FD = field duplicates.
" pos- P50- and P9s are tne 5tn- 50th- and 95th percentiles by data set.
c Kolmogorov-Smirnov statistic, significant at the .05 level; critical value for FD/RS
KS-Statc
MG_CL
K_CL
NA_CL
MG_OAC
K_OAC
NA_OAC
CEC_CL
CA_CL2
MG_CL2
K_CL2
FE_CL2
AL_CL2
FE_PYP
SO4_H20
S04_PO4
RS
FD
RS
FD
RS
FD
RS
FD
RS
FD
RS
FD
RS
FD
RS
FD
RS
FD
RS
FD
RS
FD
RS
FD
RS
FD
RS
FD
RS
FD
844
58
844
58
844
58
844
58
844
58
844
58
844
58
844
58
844
58
844
58
844
58
844
58
844
58
844
58
844
58
0.48
0.30
0.17
0.10
0.02
0.02
0.46
0.27
0.17
0.10
0.02
0.02
8.28
7.63
0.65
0.73
0.12
0.09
0.05
0.03
0.01
0.01
0.11
0.11
0.35
0.33
18.33
19.28
50.28
53.07
0.01
0.01
0.03
0.03
0.00
0.00
0.01
0.01
0.03
0.03
0.00
0.00
2.73
2.42
0.33
0.46
0.01
0.01
0.01
0.00
0.00
0.00
0.00
0.00
0.04
0.11
2.22
4.93
5.06
16.73
0.13
0.04
0.10
0.07
0.01
0.01
0.13
0.05
0.10
0.07
0.01
0.01
7.05
5.21
0.54
0.57
0.06
0.03
0.02
0.02
0.00
0.00
0.09
0.11
0.24
0.26
12.18
14.21
30.23
45.65
1.99
1.84
0.50
0.30
0.06
0.11
1.94
1.63
0.50
0.29
0.06
0.13
19.79
24.41
2.07
2.69
0.41
0.57
0.16
0.18
0.04
0.04
0.27
0.23
1.05
0.99
31.71
43.60
136.92
139.02
0.279
.
0.264
.
0.190
.
0.280
.
0.263
.
0.193
.
0.230
.
0.263
.
0.266
.
0.201
.
0.467
.
0.189
.
0.208
.
0.222
.
0.278
0.185.
SBRP survey that there were insufficient quan-
tities of Oa, A, and Bs audit samples remain-
ing for use as QE samples in the MASS, so a
recently collected B horizon audit sample was
substituted. The Oa and A horizon audit
samples were used for QC purposes in the
MASS, while the Bs horizon audit samples
were used as QE samples at the preparation
laboratory (Papp and Van Remortel, in press).
In the MASS, field audit samples were
included in each sample batch. Mineral soil
batches contained a FAL sample and two FAR
samples, and organic soil batches contained
three FAO samples. The FAR and FAO sam-
ples were used to determine within-batch
system precision and accuracy. A single LAL
sample was included in each mineral soil
batch and was used with the FAL sample to
assess contamination and low-range inac-
curacy, and to estimate the system detection
limits. The use of QCAS samples in the MASS
helped to control error at the laboratories. The
additional audit samples did not directly affect
the comparability of the data from the three
surveys, but aided in the QE and QC activities.
The within-batch precision estimate was
refined by increasing the number of prepara-
tion duplicates from one per batch in the NE
to two per batch in the SBRP and MASS. For
the MASS, one preparation duplicate was split
from the field duplicate and the other from its
associated routine sample.
Comparison of Sampling, Preparation,
and Analytical Methods
Field sampling for the MASS included the
use of standard National Cooperative Soil
Survey (NCSS) characterization and sampling
procedures, specified protocols, and standard
113
-------
Table 3-15. Completeness of Routine Data from the
Verified and Validated Data Bases
Verified data base'
Parameter n percent
MOIST
SAND
VCOS
COS
MS
FS
VFS
SILT
COSI
FSI
CLAY
PH H20
PH 002M
PH_01M
CA CL
MG~CL
KCl
NA CL
AL_CL
CA OAC
MG'OAC
K OAC
NA_OAC
CEC CL
CEC'OAC
AC_BACL
CA CL2
MG CL2
K CL2
NA CL2
FE~CL2
AL_CL2
FE PYP
AL PYP
FE AO
AL AO
SI AO
FE~CD
AL~CD
S04 H2O
SO4 PO4
SO4~0
S04 2
SO4 4
SO4 8
S04~16
S04_32
C TOT
N TOT
S TOT
844
795
795
795
795
795
795
795
795
795
795
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
844
795
795
795
795
795
795
844
844
844
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
Validated data base"
n percent
844
791
795
795
793
794
794
795
794
793
789
844
843
842
837
839
833
836
837
844
843
839
839
840
840
842
839
844
839
837
832
841
844
844
844
844
844
844
844
839
835
793
794
792
794
794
794
842
843
844
100.0
99.5
100.0
100.0
99.8
99.9
99.9
100.0
99.9
99.8
99.3
100.0
99.9
99.8
99.2
99.4
98.7
99.0
99.2
100.0
99.9
99.4
99.4
99.5
99.5
99.8
99.4
100.0
99.4
99.2
98.6
99.6
100.0
100.0
100.0
100.0
100.0
100.0
100.0
99.4
98.9
99.8
99.9
99.6
99.9
99.9
99.9
99.8
99.9
100.0
" The total routine samples (N) is 844; for particle size
and sulfate isotherm analyses, N is 795. For verified
data base, n is samples that underwent analysis;
for validated data base, n is cumulative samples
valid at level of confidence 0, 1, or 2.
reporting forms which ensured that the result-
ing data would be comparable with data
generated from the NE and SBRP sampling
activities. Because of an unclear protocol in
the NE survey, some sampling crews collected
a field duplicate using the alternate trowel-full
method that was intended, while other crews
collected a field duplicate by splitting a com-
posite bulk soil sample. The protocol was
clarified for the other two surveys. As a
result, the field sampling within-batch variabili-
ty is expected to be smaller in the NE survey
than in the SBRP or MASS surveys because
much of the sampling error is masked in the
NE field duplicates (Van Remortel et al., 1988).
Since the routine samples for all three surveys
were collected using similar protocols, the
resulting field data can be compared.
Nearly identical soil preparation methods
were used throughout the three surveys,
although the NE protocols were revised for
clarity in the SBRP survey and again in the
MASS (Appendix A in Haren and Van Remortel,
1987; Appendix A in Papp and Van Remortel, in
press). There were a few differences in the
preparation methods, e.g., the number of
passes through a riffle splitter during homo-
genization was seven in the NE and SBRP
surveys and five in the MASS. There were also
differences inherent in the use of more than
one preparation laboratory, e.g., different types
of sample drying facilities with varying airflow
and humidity conditions. The preparation
laboratory operations were consolidated from
four laboratories in the NE survey and two
laboratories in the SBRP survey to a single
MASS laboratory.
In the MASS, seven physical and chemi-
cal analyses were performed at the prepara-
tion laboratory: field-moist pH, air-dry moisture
content, organic matter by loss-on-ignition,
percent fine gravel, percent medium gravel, and
bulk density determined by clod and known
volume methods. The field-moist pH, organic
matter, and known volume bulk density para-
meters were not measured in the NE and
SBRP surveys.
Five of the 53 analytical parameters
measured over the course of the DDRP were
measured in only one or two of the three
surveys, therefore, the data from these five
parameters are not directly comparable across
all surveys. In each case, however, the data
114
-------
generally satisfied the acceptance criteria for
precision, as there were no accuracy criteria
defined for these parameters. These parame-
ters are the specific surface area (SP_SUR)
and the exchangeable acidity and aluminum in
potassium chloride (AC_KCL and AL_KCL)
parameters measured only in the NE and^BRP
surveys, and the extractable silicon in acid
oxalate (SI_AO) and exchangeable aluminum in
ammonium chloride (AL_CL) parameters meas-
ured only in the MASS.
There was one complete method change
and several modifications made in the DDRP
analytical methods prior to the MASS sample
analysis. One of the extractants for aluminum
was changed from 1M potassium chloride
(AL_KCL parameter) to 1M ammonium chloride
(AL_CL parameter). When extracting calcium
with calcium chloride, samples were shaken
overnight instead of being extracted on a
mechanical extractor (all _CL2 parameters).
The procedure for exchangeable cations and
cation exchange capacity included a two-fold
increase in the amount of soil relative to the
solution volume (all _CL and _OAC parame-
ters). The volume specificity in the exchange-
able acidity procedure was more precise
(AC_BACL parameter). The initial concentra-
tion range requirements for the sulfur solution
standards used in the isotherm measurements
were more stringent (S04_0 - 32 parameters).
Instrumentation requirements were more
specific for the majority of the analyses.
Comparison of User-Defined
Acceptability Criteria
Assessing the comparability of data
collected during the three DDRP surveys was
complicated by differences in the protocols
and MQOs across the surveys. User-defined
data acceptability is reflected in the precision
and accuracy objectives and the overall utility
of data that have satisfied the defined objec-
tives. The precision and accuracy objectives
and acceptance criteria were modified, i.e.,
usually tightened, for the MASS because of
experience gained during the NE and SBRP
surveys. The use of accuracy windows in the
MASS also improved the data evaluation
procedures.
There were several differences in the
MQOs between the NE and SBRP surveys and
the MASS, and overall measurement quality
generally is better in the MASS data bases.
Therefore, it is possible that some of the NE
and SBRP data may not be suitable for the
same data analysis procedures performed on
the MASS data without some form of caveat.
Possible examples include the specific surface
data and the organic soil sulf ate isotherm data
from the NE and SBRP surveys which are not
being utilized by some data users.
Quantitative Comparison of the
DDRP Surveys
Data from the Bw audit sample were
used for general quantitative survey compari-
sons because the Bw was the only median-
range audit sample used for QE purposes in
all three surveys. For each parameter, pooled
mean values for the Bw horizon audit samples
were initially compared across surveys using
an F test as part of ANOVA. Because a large
number of parameters showed a significant
difference among surveys at the .05 level, the
differences were compared to the existing
accuracy windows to determine whether any
of the survey mean values exceeded the limits
of the windows. It was determined that most
of the mean values were safely within the
accuracy windows. As a way to alert data
users to possible comparability constraints,
the QA staff used Scheffe's multiple com-
parison test to identify specific differences in
the survey mean values highlighted at the .01
level of significance, as presented in Table
3-16. Only 10 parameters showed differences
among surveys at this level: AL_CL vs. AL_KCL,
SO4JH2O, SO4_0, SO4_2, SO4_4, CA_CL2,
COSI, FE_CD, and C_TOT. Although 40 of the
50 parameters (80 percent) common to all
three surveys were not significantly different at
this level of significance across surveys, data
users should exercise caution when using data
from these 10 parameters.
An important difference observed was
between the AL_CL and AL_KCL parameters.
The NE and SBRP mean values for AL_KCL
were approximately the same but the MASS
mean value for AL_CL was highly significantly
different. This finding possibly reflects the
change in extractant for exchangeable alumi-
num from 1M potassium chloride used in the
NE and SBRP surveys to 1M ammonium chlo-
ride used in the MASS. The sulfate parameter
differences were largely attributed to the
greater uncertainty in the standard solution
115
-------
Table 3-16. Multiple Comparison of Interlaboratory
Mean Values Using the Bw Audit Sample
Parameter
MOIST
SAND
VCOS
COS
MS
FS
VFS
SILT
COSI**
FSI
CLAY
PH H20
PH~002M
PHJ)1M
CACL
MG~CL
K CL
NA CL
AL_CL-KCL**
CA OAC
MG OAC
K OAC
NA_OAC
CEC CL
CEC OAC
AC_BACL
CA CL2**
MG~CL2
K CL2
NA CL2
FE CL2
AL~CL2
FE PYP
AL PYP
FE~AO
AL AO
FE~CD**
ALlCD
SO4 H20**
SO4~PO4
SO4~0**
S04~2**
S04~4**
S04 8
S04 16
SO4_32
C TOT**
N~TOT
S TOT
Reporting
units
wt%
pH units
H
H
meq/100g
II
II
II
n
meq/100g
H
11
n
meq/100g
H
II
meq/100g
n
n
n
«
n
wt %
II
"
II
"
II
mg S/kg
n
mg S/L
»
"
11
n
11
wt %
n
H
Interlaboratory
mean for Bw sample
NE
2.12
28.10
3.02
4.07
3.62
4.13
13.26
64.86
31.09A
33.75
7.05
5.08
4.71
4.54
0.27
0.05
0.06
0.04
1.53A
0.25
0.05
0.06
0.03
5.18
12.54
17.33
0.51A
0.03
0.01
0.02
0.001
0.004
0.81
0.57
1.03
0.91
1.56*
0.62
20.46*
105.77
2.00*
2.93*
3.86*
6.18
11.32
23.78
1.56*
0.121
0.019
SBRP MASS
2.33 2.34
27.06 25.40
2.62 2.49
4.04 3.77
3.51 3.55
3.91 4.23
13.01 11.40
65.45 68.65
33.71*-8 36.32"
31.73 32.34
7.48 5.95
5.13 5.07
4.72 4.74
4.62 4.57
0.29 0.27
0.05 0.06
0.06 0.06
0.02 0.03
1.42* 1.84B
0.24 0.25
0.06 0.06
0.06 0.06
0.02 0.03
5.79 4.38
13.12 10.99
17.22 16.39
0.60*-" 0.62B
0.04 0.04
0.01 0.01
0.02 0.02
0.001 0.001
0.004 0.010
0.85 0.75
0.58 0.54
1.03 0.99
0.88 0.86
1.88" IT^8
0.67 0.63
22.89B 20.31*
108.97 109.44
2.27" 2.05*
3.22B 3.04*-B
4.30B 4.09*-"
6.54 6.34
11.91 11.87
23.70 24.13
1.51*-8 1.45B
0.110 0.113
0.019 0.020
* B Pairwise differences among surveys highlighted for
significant differences at the .01 level.
** Significant at the .01 level based on Scheffe's
multiple comparison test.
concentrations for the NE and SBRP surveys.
The modified protocol in the MASS resulted in
more accurate initial sulfur extraction solu-
tions. The procedural change from mechanical
extractor to overnight shaking for the deter-
mination of CA_CL2 in the MASS did not
appear to have a significant effect, as the
mean values from the SBRP and MASS were
not very different.
Differences in the data due to treatment
effects of different preparation and analytical
laboratories were present in all comparisons
of the data. For example, it was noted in the
MASS that the fine silt fraction was not com-
pletely separated from the coarse silt fraction
during the early stages of analysis at Labora-
tory 2. This could be a contributing factor to
the higher COSI mean in the MASS. The
significant differences across surveys ob-
served for FE_CD and C_TOT cannot be com-
pletely explained on the basis of differen-
ces among laboratories. Differences in the
instrumentation used to measure C_TOT in the
surveys may also have had an effect.
Interlaboratory Comparison Study
The objective of the interlaboratory com-
parability study was to determine whether the
analytical data for soil audit samples analyzed
by contract soil characterization laboratories
for the DDRP soil surveys are comparable to
data for the same samples analyzed by other
soil characterization laboratories in the United
States and Canada.
The original four DDRP laboratories
constituted the "DDRP1 laboratory group that
used analytical methods and procedures as
specified in the NE and SBRP contractual
statements of work. Data from these labora-
tories were collected in 1985 and 1986. Two of
the original four laboratories were selected to
represent the DDRP laboratories in the study.
Fifteen randomly selected soil characterization
laboratories, including nine in the United States
and six in Canada, were selected to represent
the non-contract or "external" laboratory group.
These two DDRP laboratories followed the
actual methods and procedures used in the NE
and SBRP surveys, while each of the external
laboratories used similar methods but followed
its own specific procedure. In 1988, each
laboratory analyzed two batches containing six
audit samples per batch, with a two-month
116
-------
interval between batches. The audit samples
included six different soil horizon types (A, B.
Bs, Bw, C, and O) from the eastern United
States which had been previously analyzed for
all 51 analytical parameters in the NE and
SBRP surveys. Each laboratory analyzed two
blind replicates of each sample.
Audit sample data for each parameter
were analyzed for various attributes using
primarily youden-pair plots (Youden, 1985)
appropriate F and / statistics. These tests
examined the comparability between the DDRP
and external laboratory groups. The attributes
investigated were time effect, homogeneity of
data between the two groups, protocol varia-
bility, and between-group variability. Summary
statistics were also derived for each soil
parameter by combining data across audit
sample horizons and for each parameter group
by combining appropriate parameters.
The temporal effects of possible degra-
dation, contamination, or procedural changes
on the audit samples were assessed by com-
paring the DDRP data collected in 1985-86 with
the external data collected in 1988. The two
participating DDRP laboratories in the external
group provided the link needed to make this
assessment. There was no time effect ob-
served in the data generated from this com-
parison. Hence, there is no evidence that the
soil chemical and physical characteristics
changed due to degradation, contamination, or
changes in laboratory analytical techniques.
This implies that there was no soil contamina-
tion or degradation under the 4°C storage
conditions and that, if the soils for other
prospective data bases were prepared and
stored prior to analysis in a similar manner as
the DDRP protocols, then the two sets of data
can be compared.
The homogeneity of laboratory data and
protocol variability between DDRP and the
external laboratories were greater for certain
cations and certain soil types. For example,
the DDRP appeared to achieve greater ac-
curacy at the lower concentrations for such
cations as potassium and sodium for all soil
types tested. The differences were revealed
not because the other laboratories did not
exhibit good results, but because the DDRP
laboratories as a group were very accurate
and precise due to exceptionally tight QE and
QC criteria and consistent analytical proce-
dures among laboratories. The difference
merely reveals that the analytical procedures
used by the external laboratories are some-
what variable, though not significantly so in
most cases. A large percentage of laborator-
ies used analytical procedures that were very
similar to the DDRP procedures for most
parameters.
The analytical data for the DDRP labora-
tories were within the ranges generated by the
independent external laboratories and, there-
fore, are generally comparable among the two
groups. The results for most parameters
indicate that the somewhat diverse analytical
procedures are quite robust within a given
analytical method. The parameters were also
classified into ten parameter groups and the
analytical results at this level are comparable
between the DDRP and external data with no
exceptions. The detailed results of the inter-
laboratory comparison study are contained in
a separate report (Fenstermaker et al., in
preparation).
Assessment of Data Uncertainty
Components
The following subsections describe
various data uncertainty components that are
estimated using data from the measurement
quality samples. The magnitude of measure-
ment uncertainty in relation to overall data
uncertainty is also assessed.
Additive Components of Uncertainty
As described previously in Section 2, the
within-batch precision estimates were pooled
over the concentration range using the propor-
tion of routine data in each interval to derive
the within-batch precision component of the S
values (see Appendix C). It was assumed that
each laboratory analyzed the soil samples on
a batch-by-batch basis under conditions speci-
fic to each batch analysis; hence, the between-
batch variability was calculated for each
parameter using the data from the LAP sam-
ples. The bias was demonstrated to be quan-
tifiable using the LAL samples and the means
of the LAP sample pairs, although it was
small for most parameters. The within- and
between-batch variances were added to the
squared bias term and were converted into
reporting units to produce delta values for
117
-------
each data set and parameter. Table 3-17
shows the bias and imprecision estimates and
the contribution of the between-batch and bias
components in relation to the within-batch
imprecision estimates for each parameter for
the analytical component. Ratios are used to
indicate the parameters which show a con-
siderable contribution of the between-batch or
bias components in relation to the within-batch
component.
The within- and between-batch mean
sums of squares and the squared bias com-
ponent of analytical uncertainty can be ex-
perimentally compared with the three mean
squares of a randomized block design. In this
scenario, both the expected value of the
squared bias and expected value of the
between-batch mean square include the ex-
pected value of the within-batch mean square
(Box et al., 1978). If the between-batch mean
square is two times larger than the within-
batch mean square, then the actual between-
batch imprecision can be inferred to be as
large as the within-batch imprecision. Similar-
ly, if the between-batch mean square is more
than three times larger than the within-batch
mean square, then the between-batch impre-
cision can be inferred to be significantly larger
than the within-batch imprecision. Hence, if
the ratio of the between-batch component with
the within-batch component in reporting units
exceeds \/3, i.e., 1.73, then the between-batch
imprecision can be considered significant in
relation to the within-batch imprecision. The
same criteria can be used to examine the
significance of the bias in relation to the
within-batch imprecision.
Under this framework, there is no evi-
dence of significant between-batch or bias
uncertainty in relation to within-batch uncer-
tainty if the ratio is less than 1. If the ratio is
between 1 and 1.73, the observed uncertainty
is negligible and could be due to random
fluctuation. If the ratio is greater than 1.73,
then the between-batch or bias component
could be as large as 50 percent of the within-
batch component.
The ratio of between-batch to within-
batch was greater than 1.73 for four particle
size parameters (SAND, COS, FS, and VFS), all
three pH parameters (PHJH2O, PH_002M, and
PH_01M), two cations from the calcium chlo-
ride" parameter group (CA_CL2 and K_CL2),
and three sulfate isotherm parameters (SO4_2,
S04_8, and SO4J6). Possible reasons for
these parameters to show substantial bet-
ween-batch error could relate to: (1) the
unique aspects of their methods of analysis,
e.g., the particle size parameters, or variability
in the spiking solutions or equilibrium condi-
tions, e.g., temperature, for the sulfate iso-
therm parameters; (2) very low soil concentra-
tions, e.g., the cations in calcium chloride; (3)
a very precise within-batch imprecision es-
timate which magnified the effect of between-
batch imprecision and resulted in a high ratio;
or (4) a wide range of acceptable values in
the accuracy windows. For the pH parame-
ters, for example, the within-batch component
was so small that the between-batch com-
ponent was substantial. In actuality, both
types of uncertainty are low when compared
to the within-batch MQO. Therefore, although
the contribution of the between-batch uncer-
tainty to total analytical uncertainty might be
substantial, the magnitude of the uncertainty
is relatively small. This conclusion also ap-
plies to the CA_CL2 and K_CL2 parameters.
The bias was a relatively minor com-
ponent of uncertainty, and all ratios of bias to
within-batch imprecision were below the criti-
cal value of 1.73. In many cases, bias was
only a small fraction of the within-batch im-
precision.
Effect of Measurement System on
Data Uncertainty
Table 3-18 provides the 8 values of data
uncertainty for the analytical parameters. As
expected, the estimates increased in a pro-
gressive manner from 8, through 84 which are
based on the LAP, PD, FD, and S/H data sets,
respectively. The 8, values contain analytical
laboratory uncertainty confounded with audit
sample preparation uncertainty. The 82 values
contain the confounded uncertainty of sample
preparation and analysis. The 53 values con-
tain the confounded overall measurement
uncertainty of field sampling, preparation, and
analysis. The 84 values contain uncertainty due
to the spatial heterogeneity of the routine
sample population confounded with the overall
measurement uncertainty of sampling, prepara-
tion, and analysis.
It has been proposed that a measured
value can be considered as essentially
118
-------
Table 3-17. Comparison of Measurement Uncertainty Components
Paramotsr
SAND
VCOS
COS
MS
FS
VFS
SILT
COSI
FSI
CLAY
PH H20
PH 002M
PH_01M
CA CL
MG CL
K CL
NA CL
ALjCL
CA OAC
MG OAC
K 0~AC
NA_OAC
CEC CL
CEC~OAC
AC_BACL
CA CL2
MG CL2
K CL2
NA CL2
FE~CL2
ALICL2
FE PYP
AL PYP
FE AO
AL~AO
SI~AO
FE CD
AL~CD
SO4 H2O
S04~PO4
SO4~0
S04 2
S04 4
S04~8
S04 16
S04_32
C TOT
N~TOT
S TOT
Reporting
units
wt. %
II
pH units
H
H
meq/100g
it
it
n
H
meq/100g
»
n
11
meq/100g
n
"
meq/100g
n
H
II
II
II
Wt. %
II
II
II
II
II
II
mg S/kg
11
mg S/L
n
H
II
11
II
Wt. %
H
II
Within-batch SD
1.160
0.912
0.664
0.792
0.595
0.665
1.878
4.055
4.337
1.130
0.052
0.052
0.032
0.021
0.012
0.004
0.006
0.411
0.024
0.014
0.005
0.005
0.306
1.081
1.399
0.018
0.005
0.001
0.005
0.001
0.005
0.024
0.038
0.041
0.056
0.027
0.050
0.058
1.542
8.594
0.085
0.054
0.107
0.136
0.295
0.875
0.209
0.007
0.002
Between-batch SD
2.455
0.993
1.528
1.335
1.779
1.666
2.559
2.446
1.808
1.153
0.112
0.108
0.056
0.026
0.009
0.004
0.004
0.145
0.025
0.009
0.005
0.004
0.264
1.106
1.500
0.038
0.006
0.002
0.003
0.001
0.006
0.032
0.039
0.060
0.037
0.031
0.072
0.055
0.982
6.081
0.099
0.124
0.147
0.260
0.532
1.078
0.098
0.004
0.001
Bias
. units -
0.327
0.402
0.764
0.567
-0.541
-0.810
0.890
0.197
0.313
-0.607
-0.319
0.028
-0.021
-0.003
0.001
0.001
-0.001
-0.062
0.000
0.001
0.006
0.004
-0.123
-0.402
-0.142
0.025
0.002
0.001
0.001
-0.000
0.002
-0.018
-0.004
0.011
0.002
-0.003
0.012
0.003
0.126
2.009
0.037
0.045
0.050
0.052
0.107
0.099
-0.034
0.002
-0.001
Ratio"
B-B/W-B
2.1*
1.1
2.3*
1.7
3.0*
2.5*
1.4
0.6
0.4
1.0
2.1*
2.1*
1.8*
1.2
0.8
1.1
0.8
0.4
1.1
0.6
0.9
0.8
0.9
1.0
1.1
2.1*
1.2
2.0*
0.7
1.1
1.2
1.3
1.0
1.4
0.7
1.1
1.4
0.9
0.6
0.7
1.2
2.3*
1.4
1.9*
1.8*
1.2
0.5
0.5
0.5
Ratio*
Bias/W-B
0.3
0.4
1.2
0.7
0.9
1.2
0.5
0.0
0.1
0.5
0.6
0.5
0.7
0.1
0.1
0.2
0.2
0.1
0.0
0.0
1.2
0.9
0.4
0.4
0.1
1.4
0.5
0.4
0.3
0.6
0.5
0.7
0.1
0.3
0.0
0.1
0.2
0.1
0.1
0.2
0.4
0.8
0.5
0.4
0.4
0.1
0.2
0.2
0.3
" Ratio of between-batch to wit bin-batch variability.
" Ratio of bias to within-batch variability.
* Significant between-batch effect.
NOTE: Delta values were not applicable for the MOIST parameter because no windows were available.
119
-------
Table 3-18. Delta Values and Ratios for Assessment of Uncertainty Components
Parameter
SAND
VCOS
COS
MS
FS
VFS
SILT
COSI
FSI
CLAY
PH H2O
PH~002M
PHJ)1M
CA CL
MG~CL
K CL
NA CL
ALjCL
CA OAC
MG~~OAC
K OAC
NA_OAC
CEC CL
CEC'OAC
AC_BACL
CA CL2
MG CL2
K CL2
NA CL2
FE~CL2
AL~CL2
FE PYP
AL'PYP
FE'AO
AL'AO
SI~AO
FE~CD
ALJJD
SO4 H2O
SO4 PO4
S04~0
S04 2
SO4 4
SO4~8
SO4 16
804^32
C TOT
N TOT
S TOT
Reporting
units
wt %
n
"
H
II
II
II
II
II
II
pH units
H
U
meq/100g
u
n
N
11
meq/100g
11
»
11
meq/100g
n
H
meq/100g
II
»
n
n
n
wt %
II
II
ii
n
11
11
mg S/kg
M
mg S/L
11
N
"
II
II
Wt %
n
n
Mean of
routine samples
44.011
4.538
6.172
9.601
13.930
9.762
38.157
13.183
24.969
17.832
4.703
4.219
4.013
1.276
0.484
0.171
0.022
4.542
1.163
0.459
0.172
0.020
8.281
13.963
15.558
0.651
0.123
0.049
0.011
0.014
0.112
0.351
0.203
0.348
0.209
0.021
1.836
0.272
18.327
50.280
2.089
3.271
4.504
7.167
13.192
26.458
3.212
0.158
0.031
8,
3.422°
1.330
1.326
1.257'
2.009'
1.800C
4.003C
8.230'
4.801'
2.624C
0.078
0.084C
0.055
0.030C
0.015C
0.006C
0.004'
0.414C
0.032°
0.017C
0.007C
0.004'
0.514C
1.407
1.660'
0.042C
0.007C
0.003C
0.004
0.001C
0.009C
0.025'
0.025C
0.046C
0.022'
0.010'
0.108C
0.025C
1.626'
11.877C
0.105'
0.118'
0.165C
0.291C
0.604C
1.280C
0.070C
0.003C
0.001C
Delta
8,
3.457
1.394
1.249
0.987
1.959
1.806
2.881
4.812
2.635
1.675
0.087
0.096
0.058
0.064'
0.014
0.008'
0.005
0.310
0.055
0.019
0.011
0.005
0.535
1.661
1.652
0.040
0.005
0.003
0.002'
0.002'
0.006
0.030
0.022
0.033
0.029
0.014
0.289'
0.026
1.623
6.499
0.096
0.125
0.154'
0.292'
0.560
1.384
0.060
0.005
0.001
Values'
8,
4.234
1.392
1.439
1.223
2.042
2.087
3.381
4.881
2.777
1.945
0.114
0.126
0.079
0.157
0.031
0.007'
0.006
0.324
0.296
0.026
0.009
0.006
0.554
1.633
1.832
0.040
0.008
0.004'
0.002'
0.001'
0.009
0.035
0.027
0.044
0.036
0.013
0.136'
0.025
1.913
7.665
0.137
0.142
0.213'
0.377'
0.624
1.341'
0.185
0.008
0.002'
84
14.557
3.824
4.233
6.112
8.071
5.103
10.582
7.154
8.060
7.553
0.473
0.440
0.410
1.114
0.474
0.100
0.012
2.531
1.091
0.446
0.094
0.014
2.858
4.667
5.582
0.103
0.084
0.022
0.009
0.020
0.095
0.305
0.165
0.268
0.131
0.029
0.826
0.125
27.173
49.421
3.392
5.325
6.325
5.288
6.210
7.105
1.213
0.053
0.027
Ratio
83
1
1
1
1
)
1
1
1
1
1
1
1
1
1
1
1
)
1
1
1
^
1
!
j
^
!
j
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
8<
3
3
3
5
4
2
3
1
3
4
4
3
5
7
15
13
2
8
4
17
10
2
5
3
3
3
10
6
5
5
10
9
6
6
4
2
6
5
14
6
25
38
30
14
10
5
7
7
16
" 8, = calculated from laboratory audit samples; 52 = calculated from preparation duplicate samples; S3
from field duplicate samples; S4 = calculated from sampling class/horizon groups of routine samples.
c ' Conditional or partial estimate, respectively.
NOTE: Delta values were not applicable for the MOIST parameter because no windows were available.
calculated
120
-------
errorless for most uses if the uncertainty in
that value is one-third or less of the permis-
sible tolerance for its use (Taylor, 1987). The
QA staff examined the relation of measure-
ment uncertainty to overall data uncertainty in
the routine samples in order to provide the
data users with a basis for assessing the
relative contribution of the measurement sys-
tem. The 83 values calculated from the field
duplicate samples (see Appendix C) were
compared to the associated 84 data uncertain-
ty values generated from the sampling class/
horizon groups of routine samples. For the
purposes of this assessment, the data uncer-
tainty confounded in the sampling class/
horizon groups is considered to be a surrogate
for user-defined error tolerance values. Where
83 is one-third or less of 84, measurement
uncertainty is a negligible contribution to the
overall data uncertainty.
Under this framework, measurement
uncertainty is negligible for 90 percent (44 of
the 49 parameters measured) of the data. The
five exceptions were the VFS, COSI, NA_CL,
NA_OAC, and SI_AO parameters. Two of
these parameters are particle size fractions for
which no strict precision MQOs were estab-
lished. As a result of extremely low concentra-
tions, the other three parameters showed high
relative measurement uncertainty due to the
inordinate effect of measurement error in the
low concentration range near the detection
limit.
121
-------
Section 4
Conclusions and Recommendations
Quality Assurance Optimization
A number of improvements were made in
the DDRP quality assurance program before
and during the Mid-Appalachian Soil Survey in
order to provide additional evidence that the
various measurement phases of the survey
were in statistical control. The quality as-
surance staff has formulated the following
conclusions and recommendations as sum-
mary findings from the quality assurance
activities for sample analysis conducted in the
survey.
Sample Flow
A single preparation laboratory was
established in Las Vegas, Nevada, in close
proximity to the EPA facility. The one labora-
tory served as a central control and distribu-
tion point for field sampling supplies and
resulted in better distribution of samples
among sampling crews and laboratories than
in previous surveys. It also allowed the es-
tablishment and adherence to strictly-defined
sample preparation protocols and the ability to
track and control progress at the laboratory
on a real-time basis. It is recommended that
sample preparation facilities for future surveys
continue to be consolidated at a single labora-
tory to facilitate quality assurance of the
samples from the field sampling through the
sample analysis phase.
Mineral and organic samples were placed
in separate batches because of differences in
analyte concentrations and in the soil to
solution ratios required for analysis. The
analytical laboratories were able to perform
instrument calibrations and sample analyses
within narrow linear dynamic ranges. This
resulted in improved analytical performance
and allowed the quality assurance staff to
better control and assess the quality of data
generated by the analyses. It is recommended
that mineral and organic samples continue to
be placed in separate batches for future
surveys.
Sample Design
A hierarchical sample design was used
to gain the maximum information from a
manageable number of measurement quality
samples. A variety of these samples were
used in the field, at the preparation laboratory,
and at the analytical laboratories. These
samples allowed assessment of preparation
laboratory precision and accuracy, analytical
and system within- and between-batch preci-
sion, and analytical accuracy. // is recom-
mended that measurement quality samples
continue to be distributed among batches and
analytical laboratories in such a way as to
provide a balanced design for assessment
purposes.
About 25 percent of the soil samples
analyzed in the Mid-Appalachian survey were
measurement quality samples, at a rate of
approximately 10 samples per 40-sample
batch. These samples were used to ensure
that all phases of the measurement system
could be effectively controlled and evaluated.
From a logistical and budgetary standpoint,
however, this frequency of measurement
quality samples would be untenable in many
soil analysis projects. A more appropriate
level might be 10 to 15 percent.
One obvious way to decrease the percen-
tage of measurement quality samples would
be to increase the overall quantity of routine
samples in each batch, e.g., from 40 to 60
122
-------
samples. Advantages of such a move include
fewer batches, an immediate reduction in the
percentage of measurement quality samples
from 25 percent to 17 percent, and a corre-
sponding reduction in the quality assurance
costs for data verification. A disadvantage
would be a reduction in the available degrees
of freedom for statistical estimates of uncer-
tainty across batches.
The quality assurance staff evaluated the
utility of the different measurement quality
samples used in each mineral soil batch. It is
recommended that the following samples be
retained in future surveys because they have
demonstrated the greatest benefit in control-
ling and assessing data quality:
one low-range field audit sample for
measurement system detectivity, accur-
acy, and contamination;
one field duplicate sample for measure-
ment system precision;
one preparation duplicate sample for
preparation precision;
one low-range laboratory audit sample for
analytical detectibility, accuracy, and
contamination;
one pair of median-range laboratory audit
samples for analytical precision and accur-
acy; and
one median-range liquid audit sample for
assessing the soil extraction procedure
(not used in batches to date).
Use of this sample design would result
in a rate of 7 measurement quality samples
per 60-sample batch, i.e., approximately 11
percent. In addition to these seven samples,
contractual requirements would specify the
analysis of standard in-house quality control
samples used by the laboratories to maintain
instrument calibration, detectibility, precision,
and accuracy for each batch. The data neces-
sary to provide detailed uncertainty estimates
would be maintained despite the reduction in
the number or percentage of measurement
quality samples. The only such samples used
in the Mid-Appalachian survey that would be
discontinued are the pair of median-range field
audit samples and the second of the two
preparation duplicates. The quality control
audit sample would shift from a "paid" status
to a standard internal quality control status.
Data Verification
A computerized data entry and verifica-
tion system was developed that allowed the
calculation of final data values and produced
a list of flags and data entry errors for each
batch of samples. This system resulted in
more rapid turnaround for submission of batch
data and completion of the data review and
confirmation and reanalysis phases. The
computer modem link between the laboratories
and the quality assurance staff enabled the
rapid transfer of preliminary and final data. //
is recommended that LEVIS or an equivalent
data entry and verification system be used in
any future soil surveys of this magnitude.
The verification program evaluated the
data with regard to the quality control criteria
and other contractual requirements, thereby
inducing the laboratories to assume much of
the responsibility for identifying and correct-
ing errant data. The use of quality control
charts reduced between-batch and between-
laboratory uncertainty. Computer evaluation of
the double-blind measurement quality samples
was accessible only to the quality assurance
staff and was used in conjunction with the
quality control and summary checks to deter-
mine batch acceptance. Sample payment was
withheld until all contractual requirements were
satisfied, resulting in rapid delivery of data. //
is recommended that these practices be con-
tinued.
Two different internal consistency checks
were performed on the routine sample data
during the final weeks of data verification for
the Mid-Appalachian survey. The first check
used pairwise parameter correlations of relat-
ed parameters to generate a computerized list
of possible outlying values called "outliers".
The second check utilized the original pedon/
horizon configuration of the samples to pro-
duce a separate list of possible outliers. An
average of three percent of all data values
were identified in this manner. The raw data
for each outlier were checked by the laboratory
managers for transcription errors or other
possible errors, e.g., sample switch. Approxi-
mately 10 percent of the outliers, i.e., fewer
than one-half of one percent of the total data
values, were found to be in error. Correction
of these data greatly improved the quality of
data.
123
-------
If it can be determined that relationships
between pairs of parameters, i.e., linear correl-
ations, are comparable among regions of
study, then individual batches of data from
new study regions can be incorporated into an
overall data file and reviewed in real-time to
distinguish outliers for the region. However, if
the correlations do not compare among re-
gions, a statistically significant population of
data must be collected from the new region
before suspect data points from individual
batches can be determined to bd outliers. The
few cases of unacceptable precision nearly
always resulted from one or a few inordinate
values. Future quality assurance efforts
should be directed toward eliminating these
outliers. // /'s recommended that a method be
developed that will clearly identify outlying data
points for real-time quality control purposes.
Data Quality Attributes
Quality assurance experience and chan-
ges in analytical methodology allowed the
measurement quality objectives for the Mid-
Appalachian survey to be, in several cases,
more stringent than the objectives for previous
DDRP surveys. The resulting data quality for
the MASS was generally higher than that of
the earlier surveys.
Detectability
The analytical laboratories satisfied the
contract requirements for instrument detecta-
bility which were based on expected variability
in calibration blank samples. After extensive
evaluation, it was suggested that the expected
variability in low concentration quality control
check samples might be a more suitable
measure of instrument detectability in future
surveys. // is recommended that the quality
assurance staff and data users consider the
development of appropriate instrument detec-
tability objectives using the DDRP between-
batch data from low concentration quality
control check samples instead of calibration
blank samples. Such a move would require
more frequent analyses of the check samples
in future batches, e.g., one check sample for
every 10 routine samples.
Overall measurement system uncertainty
was controlled through precision and accuracy
acceptance criteria for field and preparation
duplicate samples, although measurement
quality objectives were not defined for system-
wide detectability. Although the instrument
detection limit is an integral part of the detect-
ability issue, the actual detection limit that
should be applied to the evaluation of detecta-
bility in the routine sample data set is the
system detection limit. Less than 80 percent
of the Mid-Appalachian routine sample data for
exchangeable calcium and sodium in ammon-
ium chloride and ammonium acetate, extracta-
ble iron in calcium chloride and acid oxalate,
silicon in acid oxaiate, and total sulfur parame-
ters is greater than the corresponding system
detection limit. It is recommended that, in
addition to instrument detectability, system
detectability be addressed in the measurement
quality objectives for future surveys. Although
some consideration has been given to the
development of a soil blank sample, // is
recommended that low concentration audit
samples, entered into the system during the
sampling and analytical phases, continue to be
used as substitutes for soil blank samples.
These audit samples serve not only to identify
contamination and allow the calculation of
instrument and system detection limits, but
also are used to evaluate system-wide accur-
acy as well as providing additional quality
management benefits at the preparation labor-
atory. A comparison of low-range field audit
samples and low-range laboratory audit sam-
ples showed no indication of contamination
from any phase of the measurement system.
Considerable effort was expended during
the course of the three surveys to evaluate
and improve the detectability of various para-
meters. In particular, significant improvement
was obtained for the exchangeable cations
and the sulfur parameters. The detection
limits used in the surveys have been tightened
and refined in conjunction with improvement in
analytical methods and instrumentation.
However, improvement in detectability is a
continuum that should be continually ad-
dressed. For similar studies of sensitive
ecosystems, it is recommended that effort be
directed towards improving detect ability for
future surveys. Additional methods research is
essential to this effort. In addition, guidance
is needed from the data users concerning the
lowest analyte concentrations about which
there is concern. This input could lead to a
substutition of methods, and could reduce the
time and effort expended in assuring the
124
-------
quality of measured values at low concentra-
tions that are not of major concern.
Precision
The precision results indicate that the
analytical within-batch precision objectives
were satisfied in most cases. This shows
that the measurement quality objectives were
well-designed and appropriate to satisfy the
goals of the DDRP. Occasionally an objective
was not satisfied for an upper or lower tier of
a parameter, such as magnesium in calcium
chloride or silicon in acid oxalate. The labora-
tories might have had difficulty in satisfying
certain precision objectives, e.g., an objective
might be unreasonably restrictive or the labor-
atories might have had problems with the
analytical method or sample homogenization.
Another possibility was that the inordinate
effect of outlying data points was responsible
for a large portion of the unsatisfactory preci-
sion. Other parameters, however, showed
better than expected precision. // is recom-
mended that the lower tier precision objective
be lowered from 0.20 to 0.10 milliequivalents
per 100 grams and the upper tier precision
objective be lowered from 15 to 10 percent
relative standard deviation for exchangeable
aluminum in ammonium chloride (AL_CL). It is
recommended that the data users consider
increasing the upper tier precision objective
from 10 to 15 percent relative standard devia-
tion for extractable silicon in acid oxalate
(SI_AO).
Increasing sources of confounded data
collection error led to increased standard
deviations in the precision estimates. How-
ever, the precision of the duplicate samples
prepared by the preparation laboratory was
about the same, on average, as the laboratory
audit samples which did not undergo any
additional subsampling. This indicates that
the preparation laboratory performed very well
in subsampling the bulk soil samples. In
many cases, the error estimates for the prepa-
ration duplicates were less than that for the
audit samples. It is possible that the quality
assurance staff had more difficulty in homo-
genizing the 500-kilogram bulk audit samples
compared to the 5-kilogram routine samples
homogenized at the preparation laboratory.
The audit samples were homogenized and
subsampled using a modified cone-and-quar-
tering technique (Schumacher et al., in review)
and the preparation duplicates were homogeni-
zed and subsampled with five passes through
a closed-bin riffle splitter. This difference has
implications for the preparation of audit sam-
ples, i.e., the need for a more standardized
homogenization method. // is recommended
that the future preparation of audit samples
utilize a large closed-bin riffle splitter during
the subsampling procedure.
For the field duplicates, precision objec-
tives were exceeded for only a few parame-
ters, usually because of the inordinate effect
of one or two field duplicate/routine pairs. In
addition, the high precision for the field dupli-
cates for some parameter groups suggests
that the component of error from soil sampling
is not a large portion of the data collection
error. It is recommended that a review be
made of the measurement quality objectives
for field sampling and sample preparation
because of the lower than expected variability
in these two measurement phases. The
sampling class/horizon groups showed the
greatest variability due to the inherent popula-
tion variability in the routine samples.
Pooling the precision estimates for each
parameter into parameter groups allowed the
within-batch components of precision to be
evaluated with respect to the ability of the
laboratories to achieve the precision objectives.
Table 4-1 is a summary of the overall achieve-
ment of the measurement quality objectives for
analytical within-batch precision for 42 of
the 50 parameters arranged into parameter
groups. A precision index was determined for
each group of parameters by pooling and
weighting the parameter standard deviations
for each of the lower and upper tiers, where
appropriate, and dividing by the respective
measurement quality objective for each tier.
These values were then pooled and weighted
across the two tiers for the group. A precision
index less than 1.0 indicates that the precision
estimate for this parameter group satisfied the
"overall objective" when viewed from the per-
spective of the entire concentration range,
where a lower precision index denotes higher
precision. A precision index of 1.0 or higher
would indicate that the parameter group did
not satisfy the overall objective. This approach
helped to identify which parameter groups
might require further quality assurance empha-
sis in order to redefine the objectives for
future surveys or to reassess the analytical
125
-------
Table 4-1. Precision Indices for Parameter Groups
Based on Pooled Wlthln-Batch Estimates
Parameter group
(parameters included) Precision index*
Particle size analysis
(SAND, SILT, CLAY)
Soil pH
(PH_: H2O, 002M, 01M)
Cations in ammonium chloride
(CA, MG. K. NA: _CL)
Cations in ammonium acetate
(CA, MG, K, NA: _OAC)
Cation exchange capacity
(CEC_CL, CEC_OAC)
Exchangable acidity
(AC_BACL)
Cations in calcium chloride
(CA, MG, K. MG. FE, AL: _CL2)
Extractable iron and aluminum
(FE_, AL_: PYP, AO, CD)
Extractable sulfate
(S04_: H20, PO4)
Sulfate isotherms
(S04_: 0-32)
Total elemental analysis
(C_TOT. N_TOT, S_TOT)
0.52
0.45
0.52
0.48
0.58
0.41
0.32
0.46
0.90
0.58
0.62
* Critical index = 1.00; values exceeding this level are
considered unsatisfactory so far as achievement of
measurement quality objectives is concerned.
procedures for the affected parameters. All
precision indices were less than 1.0, which
suggests that overall precision was very gooc
in relation to the survey objectives.
Accuracy
Accuracy in the Mid-Appalachian survey
was evaluated by quantitatively estimating
analytical bias with respect to a reference
value, defined as the central tendency of an
accuracy window for a given parameter.
Laboratory differences and trends were as-
sessed by comparing mean parameter values
for the laboratories, combined across audit
samples, to their respective reference values.
Analytical bias estimates were pooled
from each parameter into appropriate parame-
ter groups to assess laboratory performance
as a function of analytical methodology. Table
4-2 shows the analytical bias in relation to the
system detection limit for the chemical para-
meters arranged into parameter groups. An
accuracy index was determined for each group
of parameters by averaging the bias values for
all parameters within a group and dividing by
the average system detection limit for the
group. Accuracy indices less than 1.0 show
that the bias for the parameter group was
less than the system detection limit for the
parameter group, signifying negligible bias in
the sample analysis. Accuracy indices of 1.0
and higher show the presence of significant
laboratory bias. The system detection limit
was used as the basis for comparison with
bias because it represents the detection limit
for all phases of the measurement system
including field sampling, sample preparation,
and sample analysis. No measurement quality
objectives for bias were established, therefore,
no overall achievement of bias objectives could
be estimated.
Table 4-2. Accuracy Indices for Parameter Groups
Based on Pooled Bias Estimates
Parameter group
(parameters included)
Accuracy index*
Cations in ammonium chloride
(CA, MG, K. NA : _CL)
Cations in ammonium chloride
(CA, MG, K, NA : _OAC)
Cation exchange capacity
;-": , SL. CEC_OAC)
;- 'changeable Acidity
(AC_BACL)
Cations in calcium chloride
(CA, MG, K, NA, FE, AL : _CL2)
Extractable iron and aluminum
FE_, AL_: PYP, AO. CD)
Extractable sulfate
(SO4_H2O, SO4_PO4)
Sulfate isotherm
(SO4_0)
Total elemental analysis
(C_TOT, N_TOT, S_TOT)
0.04
0.10
0.45
0.04
0.26
0.12
0.36
0.12
0.50
" Critical index = 1.00; values exceeding this level are
considered to show significant bias.
126
-------
The accuracy indices assist in identifying
which parameter groups should undergo
further quality assurance emphasis and can be
helpful in defining and establishing measure-
ment quality objectives for bias in future soil
surveys. All accuracy indices were considerab-
ly below 1.0 for the Mid-Appalachian survey,
which suggests that bias was negligible when
viewed from this system-wide perspective.
As a summary assessment of laboratory
differences, the average deviation from the
reference value was estimated, using multi-
variate analysis, for each laboratory (L1, 12,
and L4) across audit samples and parameters
within each parameter group. The ratio of the
squared deviation to the variance of the prior
reference distribution was evaluated across
audit samples and parameters within each
parameter group. Significance of the differ-
ences is expressed as an index for each
parameter group, where an index of 1.0 or
higher denotes a significant laboratory dif-
ference (see Table 4-3). None of the labora-
tories were significantly different from the
variance of the prior reference distribution for
any parameter group. The similarity in the
indices among laboratories for a given para-
meter indicates that there was similarity in
laboratory performance.
Quality is a continuum and the need for
a specific level of data quality dictates the
measurement quality objectives. These objec-
tives may or may not be attainable with the
current technology. The analytical procedures
for the methods should be examined for their
ability to accomplish the analysis at the level
specified. No single laboratory was consis-
tently superior to the others for all parameters
or parameter groups. Each laboratory appears
to have individual strengths in specific analyti-
cal methods, which is probably a reflection of
the combination of experience, instrumentation,
and laboratory management within each labor-
atory. This resulted in a patchwork of dif-
ferences on a parameter group basis. In order
to control interlaboratory differences in future
surveys, it is recommended to continue the
selection of a specific laboratory to perform
analyses on a parameter basis for those
parameters which use specialized instrumenta-
tion, e.g., total elemental analysis (C TOT,
N_ TOT, S_ TOT).
Table 4-3. Laboratory Difference Indices for
Parameter Based on Pooled Deviations
from Reference Values
Parameter group
Laboratory
difference index*
L1 L2 L4
Particle size analysis
(SAND, SILT, CLAY)
Soil pH
(PH_: H2O, 002M, 01M)
0.21
0.39 0.39
0.29 0.25 0.44
Cations in ammonium chloride 0.56 0.20 0.21
(CA, MG, K, NA: _CL)
Cations in ammonium acetate 0.29 0.35 0.23
(CA, MG, K, NA: _OAC)
CEC and exchangeable acidity 0.27 0.32 0.35
(CEC_CL. CEC_OAC. AC_BACL)
Cations in calcium chloride 0.25 0.19 0.18
(CA, MG, K, MG, FE, AL: _CL2)
Extractable Fe, Al, Si 0.36 0.30 0.24
(FE_, AL_: PYP, AO. CD; SI_AO)
Extractable sulfate
(SO4_: H20, PO4)
Isotherms
(S04_: 0 - 32)
0.60
0.25
0.35 0.23
0.27 0.38
Critical index = 1.00; laboratories exceeding this level
are considered to show significant deviation from the
reference value.
One of the major goals of any quality
assurance program is to reduce the occurrence
of both random and systematic errors, hence,
selection of the best possible laboratories is
of primary importance. As shown in previous
DDRP surveys, analytical performance among
laboratories varies widely and, as demonstrat-
ed for analyses in this and previous DDRP
reports, laboratories can differ substantially in
their performance for certain parameters. // is
recommended that a stringent performance
evaluation process be continued to select
qualified contract analytical laboratories.
Successful quality control of a measure-
ment process results in statistical control of
the resulting data. With this in mind, a quality
control soil audit sample was successfully
incorporated into the quality assurance pro-
gram and allowed the laboratories to monitor
the accuracy of the batch analysis. Lab-
oratories were provided with a range of
127
-------
acceptable analyte concentrations for this
sample and, for each sample batch, the labor-
atories were required for all analyses to satis-
fy contractually-defined accuracy criteria on a
batch-by-batch basis. If the reported analyti-
cal results were not within the designated
accuracy windows, the laboratory was required
to reanalyze a batch of samples until the
reported audit sample values satisfied the
acceptance criteria. This procedure ensured
that each laboratory could meet a rigid stan-
dard for each batch of samples analyzed prior
to final submission of the analytical results.
Between-batch uncertainty and laboratory
differences were markedly reduced through the
use of this sample. It is recommended that
the use of a quality control audit sample be
continued in future surveys.
Data derived from soil analytical methods
which determine extractable or exchangeable
soil constituents may contain errors stemming
from soil extraction as well as from the instru-
ment. The former is assumed to be the main
cause of differences among the laboratories
and is a major reason for conducting inter-
laboratory comparison studies. Without the
ability to distinguish between extraction and
instrument error, however, it is not known
whether one or both types of errors are pres-
ent and where to focus efforts to reduce
systematic bias. It is recommended that liquid
audit samples be incorporated into the quality
assurance program and be used to differenti-
ate between error resulting from soil extraction
and error resulting from instrumentation.
As a result of information gained in this
survey, it is recommended that the accuracy
windows be redefined for future surveys using
data from all three DDRP surveys. It was
observed by several DDRP reviewers that
quality control charts were not used as consis-
tently nor effectively as possible during the
batch analysis. Extensive use of control
charts in real time would have improved the
ability of the quality assurance staff to detect
trends or other differences among the batches.
// is recommended that emphasis be placed
on the use of audit sample control charts by
the QA staff to identify abnormal scatter or
trends outside the accuracy windows during
batch analysis.
Representativeness
In evaluating representativeness of the
quality assurance samples, it is evident that
the field duplicates and preparation duplicates
are representative of the range of concentra-
tion in the range of the routine samples for
most parameters. // is recommended that the
field sampling and preparation laboratory
protocols continue to specify statistically valid
methods for selecting the field and preparation
duplicates. The method should be reiterated
to the sampling and laboratory personnel
during the pre-sampling training session. The
sampling crew leaders should ensure that a
sufficient amount of soil is collected for each
bulk sample during sampling to allow a prepar-
ation duplicate to be subsampled.
Completeness
Sampling of the specified pedons had a
completeness level of 100 percent, and proces-
sing was accomplished for 100 percent of the
soil samples satisfying the sample receipt
criteria at the preparation laboratory. Analyti-
cal completeness in the verified data bases
was 100 percent for all parameters. Sufficient
validated data were generated to make conclu-
sions for each parameter in the Mid-
Appalachian survey data bases.
Comparability
An interlaboratory comparability study
has indicated that soils data collected from
DDRP analytical methods were comparable to
data obtained, using similar methods, from
independent laboratories in the United States
and Canada. This finding has positive implica-
tions for comparability of the DDRP data with
other data bases and is especially important
considering the large geographical aspects of
the DDRP surveys. This comparability aspect
is further evidence of the thoroughness of the
quality assurance program and the appro-
priateness of the analytical design to satisfy
the requirements of the data users.
Comparability was observed to have
several levels of suitability that must be satis-
fied before multiple sets of data can be direct-
ly compared. Numerous differences in data
quality objectives existed between the three
DDRP surveys, and overall data quality is
generally best in the Mid-Appalachian survey.
128
-------
Therefore, some of the analytical data from the
Northeastern and Southern Blue Ridge Province
surveys may not be suitable for the same data
analysis comparisons as the Mid-Appalachian
survey data. In most cases, the sampling,
preparation, and analytical methods and proto-
cols for the three surveys were comparable.
General quantitative comparisons of all
parameters common to the three surveys were
made using the Bw horizon audit samples.
Results of an F test revealed several parame-
ters with significant differences across sur-
veys. // is recommended that data users
exercise caution when using DDRP analytical
data from the following parameters: AL CL
vs. AL_KCL, and SO4_H2O, SO4 O, SO4 4,
CA_CL2, COSI, FE_CD, and C_TOT. Data
comparability across the three surveys is
generally good, as 40 of the 49 parameters (82
percent) common to the three surveys did not
show highly significant differences. // is
recommended that a data comparison be
performed between the two soil extraction
solutions used for exchangeable aluminum in
1Mammonium chloride (AL_CL) and 1Mpotas-
sium chloride (AL_KCL).
Uncertainty Estimation
The step function statistical approach
has been shown to be an effective procedure
for evaluating measurement quality issues in
environmental data spanning a wide range of
concentration. It is recommended that addi-
tional research and development be undertaken
to identify an optimal step function procedure
that is fully compatible with the measurement
quality sample design.
It was determined that measurement
uncertainty is negligible for 90 percent (45 of
the 50 parameters) of the data. The five
exceptions to this finding were the VFS, COSI,
NA_CL, NA_OAC, and SI_AO parameters.
Additional methods research would identify
ways to reduce the contribution of measure-
ment uncertainty in these parameters.
129
-------
References
Altshuller, A. P., and R. A Linthurst. 1984. The
Acidic Deposition Phenomenon and Its
Effects: Critical Assessment Review
Papers. EPA/600/8-83/016bf. U.S. En-
vironmental Protection Agency, Washing-
ton, DC.
Arnold, S. F. 1981. The Theory of Linear Mod-
els and Multivariate Analysis. J. Wiley &
Sons, New York. 475 pp.
Barth, D. S., B. J. Mason, T. H. Starks, and K.
W. Brown. 1989. Soil Sampling Quality
Assurance Guide. Second Edition. EPA/
600/8-89/046. U.S. Environmental Protec-
tion Agency, Las Vegas, Nevada. 260 pp.
Belsley, D. A., E. Kuh, and R. E. Welsch. 1980.
Regression Diagnostics. J. Wiley &
Sons, New York. 352 pp.
Bloom, P. R., and D. F. Grigal. 1985. Modeling
Soil Response to Acidic Deposition in
Nonsulfate Adsorbing Soils. J. Environ.
Qual. 14:489-495.
Blume, L J., B. A. Schumacher, P. W. Shaffer,
K A. Cappo, M. L. Papp, R. D. Van Re-
mortel, D. S. Coffey, and D. J. Chaloud.
In press. Handbook of Methods for Acid
Deposition Studies, Laboratory Analyses
for Soil Chemistry. EPA/600/x-xx/xxx.
U.S. Environmental Agency, Washington
D.C.
Box, G. E. P., W. G. Hunter, and J. S. Hunter.
1978. Statistics for Experimenters. John
Wiley & Sons, New York. 224 pp.
Byers, G. E., R. D. Van Remortel, J. E. Teberg,
M. J. Miah, C. J. Palmer, M. L Papp, W.
H. Cole, A D. Tansey, D. L. Cassell, and
P. W. Shaffer. 1989. Direct/Delayed
Response Project: Quality Assurance
Report for Physical and Chemical Ana-
lyses of Soils from the Northeastern
United States. EPA/600/4-89/037. U.S.
Environmental Protection Agency, Las
Vegas, Nevada. 216 pp.
Chen, C. W., S. A. Gherini, J. D. Dean, R. J. M.
Hudson, and R. A Goldstein. 1983.
Modeling of Precipitation Series, Volume
9. Ann Arbor Sciences, Butterworth Publi-
shers, Boston, Massachusetts. 175 pp.
Church, M. R., K. W. Thornton, P. W. Shaffer,
D. L. Stevens, B. P. Rochelle, G. R. Hol-
dren, M. G. Johnson, J. J. Lee, R. S.
Turner, D. L. Cassell, D. A Lammers, W.
G. Campbell, C. I. Liff, C. C. Brandt, L H.
Liegel, G. D. Bishop, D. C. Mortenson, S.
M. Pierson and D. D. Schmoyer. 1989.
Future Effects of Long-Term Sulfur Depo-
sition on Surface Water Chemistry in the
Northeast and Southern Blue Ridge Pro-
vince: Results of the Direct/Delayed
Response Project. EPA/600/3-89/061, U.S.
Environmental Protection Agency, Wash-
ington, D.C., 887 pp.
Church, M. R. 1989. Predicting the Future
Long-Term Effects of Acidic Deposition
on Surface Water Chemistry: The Direct/
Delayed Response Project. Eos, Trans-
actions, American Geophysical Union
70:801-813.
Cochran, W. G. 1977. Sampling Techniques.
3rd Ed. J. Wiley & Sons, New York. 428
PP-
Conover, W. J. 1980. Practical Nonparametric
Statistics, Second Edition. J. Wiley &
Sons, New York. 493 pp.
Cosby, B. J., G. M. Hornberger, E. B. Rastetter,
J. N. Galloway, and R. F. Wright. 1986.
Estimating Catchment Water Quality
Response to Acid Deposition Using
Mathematical Models of Soil Ion Ex-
change Processes. Geoderma 38:77-95.
130
-------
References (continued)
Driscoll, C. T., and R. M. Newton. 1985. Che-
mical Characteristics of Adirondack
Lakes. Environ. Sci. Technol. 19:101810-
24.
Galloway, J. N., S. A Norton, and M. R.
Church. 1983. Freshwater Acidification
from Atmospheric Deposition of Sulfuric
Acid: A Conceptual Model. Environ. Sci.
Technol. 17:541-545.
Gherini, S. A., L. Mok, R. J. Hudson, G. F. Da-
vis, C. W. Chen, and R. A. Goldstein.
1985. The ILWAS Model: Formulation
and Application. Water, Air, Soil Pollut.
26:425-459.
Haren, M. F., and R. D. Van Remortel. 1987.
Direct/Delayed Response Project: Field
Operations and Quality Assurance Report
for Soil Sampling and Preparation in the
Southern Blue Ridge Province of the
United States, Volume II: Preparation.
U.S. Environmental Protection Agency,
Las Vegas, Nevada. EPA/600/4-87/041.
Kadafar, K. 1982. A Biweight Approach to the
One Sample Problem. Amer. Stat. Assoc.
77(378):416-424.
Kaufmann, P. R., A T. Herlihy, J. W. Elwood,
M. E. Mitch, W. S. Overton, M. J. Sale, J.
J. Messer, K. A. Cougan, D. V. Peck, K. H.
Reckhow, A J. Kinney, S. J. Christie, D.
D. Brown, C. A Hagley, and H. I. Jager.
1988. Chemical Characteristics of
Streams in the Mid-Atlantic and South-
eastern United States, Volume I: Popu-
lation Descriptions and Physico-Chemical
Relationships. EPA/600/3-88/021a. U.S.
Environmental Protection Agency, Wash-
ington, D.C.
Keith, L H., W. Crummett, J. Deegan, R. A
Libby, J.K. Taylor, and G. Wentler. 1983.
Principles of Environmental Analysis.
American Chemical Society Committee
on Environmental Improvement. Anal.
Chem. 55(14) :2210-2218.
Kern, J. S., and J. J. Lee. In press. Direct/
Delayed Response Project: Field Opera-
tions and Quality Assurance Report for
Soil Sampling in the Mid-Appalachian
Region of the United States. EPA/600/
x-xx/xxx. U.S. Environmental Protection
Agency, Corvallis, Oregon.
Lee, J. J., D. A Lammers, M. G. Johnson, M. R.
Church, D. L. Stevens, D. S. Coffey, R. S.
Turner, L. J. Blume, L H. Liegel, and G.
R. Holdren. 1989. Watershed Surveys to
Support an Assessment of the Regional
Effect of Acidic Deposition on Surface
Water Chemistry. Environ. Mgt. 13:95-
108.
Linthurst, R. A, D. H. Landers, J. M. Eilers, D.
F. Brakke, W. S. Overton, E. P. Meier, and
R. E. Crowe. 1986. Characteristics of
Lakes in the Eastern United States,
Volume I: Population Descriptions and
Physico-Chemical Relationships. EPA/
600/4-86/007a. U.S. Environmental Pro-
tection Agency, Washington, D.C., 136
PP-
Long, G. L., and J. D. Winefordner. 1980. Limit
of Detection: A Closer Look at the
IUPAC Definition. Anal. Chem. 55(7):712
Massart, D. L., B. G. M. Vandeginste, S. N.
Deming, Y. Michotte, and L. Kaufman.
1988. Chemometrics: A Textbook. El-
sevier Publishers, New York. 488 pp.
Mian, M. J., R. D. Van Remortel, G. E. Byers, J.
E. Teberg. B. A Schumacher, M. L. Papp,
and L. J. Blume. 1988. A Design for
Estimating Data Uncertainties in a Soil
Survey. Agronomy Abstracts. Soil Sci.
Soc. Amer., Madison, Wisconsin.
National Academy of Sciences. 1986. Acid
Deposition: Long-Term Trends. National
Research Council Commission on Physi-
cal Sciences, Mathematics, and Resour-
ces. Environmental Studies Board.
National Academy Press, Washington,
D.C. 506 pp.
Nikolaidis, N. P., H. Rajaram, J. L. Schnoor, and
K. P. Georgakakos. 1988. A Generalized
Soft Water Acidification Model. Water
Resour. Res. 24:1983-1996.
131
-------
References (continued)
Papp, M. L, R. D. Van Remortel, C. J. Palmer,
G. E. Byers, B. A. Schumacher, R. L.
Slagle, J. E. Teberg, and M. J. Mian.
1989. Direct/Delayed Response Project:
Quality Assurance Plan for Preparation
and Analysis of Soils from the Mid-
Appalachian Region of the United States.
EPA/600/4-89/031. U.S. Environmental
Protection Agency, Las Vegas, Nevada.
225pp.
Papp, M. L, and R. D. Van Remortel. In press.
Direct/Delayed Response Project: Labor-
atory Operations and Quality Assurance
Report for Preparation of Soils from the
Mid-Appalachian Region of the United
States. EPA/600/x-xx/xxx. U.S. Environ-
mental Protection Agency, Las Vegas,
Nevada.
Reuss, J. O., and D. W. Johnson. 1986. Acid
Deposition and the Acidification of Soils
and Waters. Ecological Studies Volume
59. Springer-Verlag, Inc., New York.
SAIC. 1988. User's Guide: Laboratory Entry
and Verification Information System.
Science Applications International Cor-
poration. Oak Ridge National Laboratory,
Oak Ridge, Tennessee.
SAS. 1986. SAS System for Linear Models.
SAS Institute, Inc., Gary, North Carolina.
210 pp.
Schmoyer, D. D., R. S. Turner, and D. A. Wolf.
1989. Direct/Delayed Response Project:
Interlaboratory Differences in the Nor-
theastern Soil Survey Data. EPA/600/
xxx/xxx. U.S. Environmental Protection
Agency, Las Vegas, Nevada.
Schnoor, J. L, N. P. Nikolaidis, and G. E.
Glass. 1986. Lake Resources at Risk to
Acidic Deposition in the Upper Midwest.
J. Water Pollut. Control Fed. 58:139-148.
Shaffer, P. W., R. P. Hooper, K. N. Eshleman,
and M. R. Church. 1988. Watershed vs.
In-Lake Alkalinity Generation: A Com-
parison of Rates Using Input-Output
Studies. Water, Air, Soil Pollut. 39:263-
273.
Shaffer, P. W., and M. R. Church. 1989. Ter-
restrial and In-Lake Contributions to the
Alkalinity Budgets of Drainage Lakes: An
Assessment of Regional Differences.
Can. J. Fish. Aquat. Sci. 46:509-515.
Stanley, T. W., and S. S. Verner. 1985. The
U.S. Environmental Protection Agency's
Quality Assurance Program. In: Quality
Assurance for Environmental Measure-
ments. ASTM STP 867. pp. 12-19. Ameri-
can Society for Testing and Materials,
Philadelphia, Pennsylvania.
Steel, R. G. D., and J. H. Torrie. 1960. Prin-
ciples and Procedures of Statistics.
McGraw-Hill Book Company, New York.
481 pp.
Taylor, J. K. 1987. Quality Assurance of Chem-
ical Measurements. Lewis Publishers,
Chelsea, Michigan. 328 pp.
U. S. Environmental Protection Agency. 1986.
Development of Data Quality Objectives:
Description of Stages I and II. Quality
Assurance Management Staff. U.S.
Environmental Protection Agency, Wash-
ington, D.C. 14 pp.
U.S. Environmental Protection Agency. 1988.
Statements of Work for Soil Analysis in
the Direct/Delayed Response Project Mid-
Appalachian Soil Survey. Attachment A
in Invitation for Bid Solicitations
W802498D1 and W802621D1. U.S. En-
vironmental Protection Agency, Las
Vegas, Nevada.
U.S. Environmental Protection Agency. 1989.
An Overview of the Direct/Delayed
Response Project. In: AERP Status, April
1989. EPA/600/M-89/005. U.S. Environ-
mental Protection Agency, Washington,
D.C.
132
-------
References (continued)
Van Remortel, R. D., G. E. Byers, J. E. Teberg,
M. J. Miah, C. J. Palmer, M. L Papp, M.
H. Bartling, A. D. Tansey, D. L Cassell,
and P. W. Shaffer. 1988. Direct/Delayed
Response Project: Quality Assurance
Report for Physical and Chemical Ana-
lyses of Soils from the Southern Blue
Ridge Province of the United States.
EPA/600/8-88/100. U.S. Environmental
Protection Agency, Las Vegas, Nevada.
251 pp.
Youden, W. J. 1959. Graphical Diagnosis of
Interlaboratory Test Results. Industrial
Quality Control 15(11) 133-137.
133
-------
Glossary
accuracy6
accuracy window6
acid neutralizing capacity6
additive model6
aliquot6
anion5
anthropogenic5
atmospheric deposition6
autocorrelation6
average deviation6
base saturation6
between-batch6
the level of agreement between an observed value and the "true"
value of a soil characteristic
confidence interval established about the reference value of a given
parameter for the purpose of controlling laboratory analytical results
on a batch-by-batch basis
ability of a soil to buffer or neutralize acid-generating conditions
statistical model in which the component effects are considered to
be additive
a subsample of soil taken from a container of homogenized analytical
sample
a negatively-charged ion, as one attracted to the anode in electrolysis
contribution of man-made influences and products to the environment;
influenced by man
placement of natural and unnatural substances into the ecosystem
through wet and dry fall processes
correlation among sequential observations
the average distance in reporting units, without regard to sign, from
the reference value
degree or percent of the total cation holding capacity of a soil that
is occupied by base cations, mainly calcium, magnesium, potassium,
and sodium
the effect of soil samples entering the laboratories in separate
batches of samples and analyzed on different days
(continued)
134
-------
(continued)
bias6
blind sample1
cation5
cation pool6
comparability6
completeness6
concentration dependency8
confounded6
contract-required
detection limit6
control chart1
data qualifier6
data quality objective6
delta value8
systematic and/or random error, with respect to a reference value,
inherent in a method and normally caused by some artifact or
idiosyncrasy of the measurement system, e.g., temperature effects,
extraction inefficiencies, mechanical losses, and calibration errors;
bias may have a positive or negative sign, and several kinds of bias
can exist concurrently
a sample submitted for analysis whose composition is known to the
submitter but is unknown to the analyst; a blind sample is one way
to test proficiency of a measurement process
a positively-charged ion, or one attracted to the cathode in electrolysis
differing level of availability of soil cations which are extracted by
varying the ionic strength of the extracting solutions
the similarity of data from different sources included in a given set
of data and the similarity of methodologies from related projects
across the regions of interest
the quantity of data that is successfully collected with respect to the
amount intended in the experimental design
refers to the increasing variability of soil analytical data with
increasing concentration for some soil parameters
intermingled in such a way as to be inseparable
instrument detectability criterion established for each parameter
in contractual agreement with a laboratory
a graphical plot of test results with respect to time or sequence of
measurement together with limits within which they are expected to
lie when the system is in a state of statistical control
annotation, or flag, applied to a datum and denoting to a possible
data quality discrepancy
user-defined criteria established for each parameter to evaluate
useability of the data
an estimate of uncertainty generated from a data set and used to
evaluate the contribution of uncertainty to a specified data value
(continued)
135
-------
(continued)
detectability6
double-blind sample1
endogenous5
extraction6
flag6
fragipan2
horizon2
hydrophilic6
hydrophobic6
instrument detection limit3
invitation for bid6
knot6
laboratory difference6
linear dynamic range6
the determination of the low-range critical value of a soil characteristic
that a method-specific procedure can reliably discern
a sample known by the submitter but submitted to the analyst in such
a way that neither its composition nor its identification as a check
sample are known
originating from within; the natural composition of a soil
removal of soil constituents by agitation (mixing, shaking,
replacement) of soil analytes using a solution of known ionic strength
and composition
(see "data qualifier")
a dense, brittle subsurface horizon that restricts water movement and
root penetration
a layer of soil of varying thickness, approximately parallel to the land
surface, that has distinct characteristics produced by soil-forming
processes
water-attracting
water-repelling
three times the standard deviation of ten nonconsecutive replicate
calibration blank analyses run on separate days; or, if a signal is not
obtained, three times the standard deviation of 10 nonconsecutive
replicate analyses of a standard whose concentration is four times
the lesser of the actual detection limit or the contract required
detection limit
a document provided to a potential contractor for solicitation in order
to estimate procedures, costs, and intention for performing an
analytical service
a point within the concentration range separating the data evaluated
by absolute and relative measures of precision
the variation between the mean of repeated measurements for a given
laboratory and the reference value for a given parameter
the range of analyte concentration for which the calibration curve is
approximately a straight line
(continued)
136
-------
(continued)
logistics5
measurement quality
objective6
measurement quality
samples6
measurement
uncertainty6
mineral soil sample4
multivariate analysis6
organic soil sample4
outlier1
overall data uncertainty6
pedon6
performance evaluation6
precision6
QART6
quadrature6
relates to the procurement, maintenance, and movement of equipment,
supplies, or personnel
critical level which, if exceeded, is considered to append additional,
and possibly unacceptable, measurement uncertainty to the
corresponding data
quality evaluation or quality control samples which are placed in
each batch of soil samples for assessment purposes
based on delta values calculated from audit samples, preparation
duplicate samples, and field duplicate samples which contain various
configurations of sampling, preparation, and analytical uncertainty
a soil sample consisting predominantly of, and having its properties
determined predominantly by, mineral matter; usually contains < 20%
(by weight) organic matter
analysis in which more than one variable is considered simultaneously
a soil sample which contains 20 percent or more (by weight) organic
matter
a value which appears to deviate markedly from that of other
members of the population in which it occurs
confounded population and measurement uncertainty in the sampling
class/horizon groups of routine samples
a three-dimensional body of soil having lateral dimensions large
enough (1 to 10 square meters in area) to permit the identification
and sampling of its soil horizons
assessment of laboratory capability to perform an analytical service
by conducting and evaluating overall results of an analytical
solicitation and preliminary analysis
the level of agreement among multiple measurements of a soil
characteristic
quality assurance reanalysis template used during data verification
to make decisions relating to the batch acceptance criteria
the square root of the pooled variance and squared bias terms used
to evaluate data uncertainties
(continued)
137
-------
(continued)
quality assurance1
quality control sample6
quality evaluation sample6
representativeness6
sesquioxides3
soil chemistry relationship4
step function6
system detection limit6
uncertainty6
validation3
variability6
a system of activities whose purpose is to provide to the producer
or user of a product or a service the assurance that it meets defined
standards of quality; consists of two separate but related activities,
quality control and quality evaluation
any sample utilized by the analyst to check immediate instrument
calibration or response, whose measurement is expected to fall
within specific acceptance criteria or control limits
double-blind sample placed among routine test samples for the
purpose of assessing and, possibly, controlling the quality of
analytical data produced
the degree to which the data collected accurately represent the
population of interest
generally considered to be iron and aluminum oxides; more
specifically, sesquioxides are combinations of these two minerals
in a one-to-one ratio
a quantitative empirical relationship between two or more soil
chemical variables that is assumed to be valid for most soils
a function which takes stationary values within different segments
of its domain; used to partition the routine soil data into intervals to
assess measurement uncertainty issues
for each chemical variable, except pH, a value calculated from low-
range field audit sample data indicating the critical concentration of
analyte measured in a routine sample that is detectable and
distinguishable from background noise or contamination
a measure of imprecision, bias, or other sources of variability in a
given value
the process of determining the legitimacy of data, involving internal
consistency checks for outlier removal and definition of levels of
confidence
imprecision about a specific characteristic
(continued)
138
-------
(continued)
verification3 the process of confirming the integrity of data, involving discrepancy
checks and precision accuracy evaluations
within-batch6 the effect of soil samples entering the laboratories in separate
batches of samples
1 Taylor, 1987
2 Soil Glossary
3 QA Soils Definitions
4 Glossary of Soil Science Terms
5 Random House College Dictionary
6 Text
139
-------
Appendix A
Analytical Data Verification and Validation Flags
This appendix lists the data qualifiers, or flags, that are assigned to the verified and validated
data bases. The verification flags specify the nature of discrepancies on a batch and sample
basis, while the validation flags are used to characterize more general types of discrepancies that
may affect the level of confidence applied to each datum in the validated data base. The
verification flags are applied only to the analytical laboratory data and are unrelated to similarly-
defined flags used for the preparation laboratory data.
140
-------
Table A-1. Analytical Data Verification Flags
Blank Exception Program
B4" Potential negative sample bias based on laboratory blank data.
B5" Mean of reagent blanks is > CRDL
B6* Laboratory calibration blank is > CRDL
B9" Insufficient number of calibration blanks were run.
Replicate Precision Exception Program
D1" Analytical duplicate (AD/routine) precision is > maximum expected SD or %RSD.
05" Preparation duplicate (PDF/FD) precision is > maximum expected SD or %RSD.
D6" Preparation duplicate (PDF/routine) precision is > maximum expected SD or %RSD.
D7" Field audit pair/triplicate (FAP/FAO) precision is > maximum expected SD or %RSD.
D8" Laboratory audit pair/triplicate (LAP/LAO) precision is > maximum expected SD or %RSD.
09" Field duplicate (FD/routine) precision is > maximum expected SD or %RSD.
Detection Limit Exception Program
L1" Instrument detection limit (IDL) is > contract-required detection limit (CRDL).
Audit Accuracy Program
NO" Low-range laboratory audit (LAL) sample value is > upper accuracy limit.
N1" LAL sample is < lower accuracy limit.
N2** Mean of LAP or LAO is > upper accuracy limit.
N3** Mean of LAP or LAO is < lower accuracy limit.
N4" Quality control audit sample (QCAS) is > upper accuracy limit.
N5** QCAS is < lower accuracy limit.
QCCS Exception Program
Q1" Quality control check sample (QCCS) is > upper control limit.
Q2" QCCS is < lower control limit.
Q3" Insufficient quantity of QCCS were measured.
Q4" Detection limit QCCS is not within 20% of the theoretical concentration.
Spike Program
S1" Percent recovery of matrix spike is above contractual criteria (>105% for sulfate isotherm parameters, >110%
for all other sulfate parameters).
S2" Percent recovery of matrix spike is below contractual criteria (<95% for sulfate isotherm parameters, <90% for
all other sulfate parameters).
S3" Spike solution readings are outside of contractual criteria (sulfate isotherm and extractable calcium parameters
only).
(Continued)
141
-------
Table A-1. Continued
Miscellaneous
AO* Value missing
MO* Value was obtained by using a method that is unacceptable according to the contract.
WO* Air-dry sample weight is not within contractual requirement.
W2* Sample extract volume is not within contractual requirement.
XO* Invalid but confirmed data based on review of QE/QC data.
X1* Invalid but confirmed data: potential gross contamination of sample or parameter.
X2* Invalid but confirmed data: potential gross contamination of pedon.
X3* Possible contamination due to either sampling technique, e.g., bucket augering, or soil amendments, e.g.
herbicides, liming, manure, etc.
X4* Outliers from statistical internal consistency check; raw data and reported data checked only for transcription
errors.
X5* Outliers from manual pedon/horizon internal consistency check; raw data and reported data checked only for
transcription errors.
Soil Chemistry Relationships
C1* PH_H20 < PH_002M if pH differences > 0.05 units.
C2* PHJ002M < PH_01M if pH differences > 0.05 units.
C3* PH_H20 < PH_01M if pH differences > 0.05 units.
C4* CEC_OAC < CEC_CL if CEC_CL > 1 meq/100g and PH_H2O < 7.0
C5* CA_OAC + MG_OAC + K_OAC + NA_OAC > CECJDAC if CEC_CL > 1 meq/100g
C6* CA_CL + MG_CL + K_CL + NA_CL > CEC_CL if CEC_CL > 1 meq/100g
C7* CEC_OAC/CLAY > 50 if CLAY > 1.0% and CEC_CL > 1 meq/100g
C8* CEC_CL/CLAY > 50 if CLAY > 1.0% and CEC_CL > 1 meq/100g
C9* AL_CL > 0.01 meq/100g if PH_H2O > 6.0
H1* NA_CL2 > NA_CL * 1.10 if NA_CL2 > 0.03 meq/100g.
H2* K_CL2 > K_CL * 1.10 if K_CL2 > 0.003 meq/100g.
H3* MG_CL2 > MG_CL * 1.10 if MG_CL2 > 0.008 meq/100g.
11* SO4_32 < S04J6 if SO4_32N £ 7.5 mg S/kg soil.
12* SO4J6 < SO4_8 if S04_32N s 7.5 mg S/kg soil.
13* SO4JB < S04_4 if S04_32N £ 7.5 mg S/kg soil.
14* SO4_4 < SO4_2 if SO4_32N s 7.5 mg S/kg soil.
15* S04_2 < SO4_0 if S04_32N n 7.5 mg S/kg soil.
16* SO4_32N < S04J6N if SO4_32N a 7.5 mg S/kg soil.
17* SO4J6N < S04_8N if S04_32N a 7.5 mg S/kg soil.
(Continued)
142
-------
Tabla A-1. Continued
18* S04_8N < S04_4N if S04_32N a 7.5 mg S/kg soil.
19* S04_4N < S04_2N if SO4_32N a 7.5 mg S/kg soil.
10* S04_2N < S04_ON if SO4_32N s> 7.5 mg S/kg soil.
NC* Not checked because basic initial criteria were not met.
R1* S04_H2O > SO4_PO4 if S04_PO4 > 1 mg S/kg soil (mineral samples); SO4_H2O > SO4_PO4 * 1.10 (organic
samples).
R2* SO4_H2O/SO4_0 < 4 or > 20 if S04_H2O fe 2 mg S/kg and SO4_0 * 0.1 mg S/kg.
T1* C_TOT/S_TOT < 7 or > 50 if CJTOT a: 0.1% and S_TOT * 0.005%.
T2* C_TOT/N_TOT < 40 or > 400 if C_TOT a: 0.1% and N_TOT a 0.03%.
* Sample Flag: flags the affected parameter for specified samples only.
** Batch-wide Flag: flags the affected parameter for all samples in the batch (assumes that measurement quality
samples are representative for the batch).
143
-------
Table A-2. Analytical Data Validation Flags
Interquartile Outlier Program
L1 Value is < Q, - 1.5xIQR but >Q, - 3xIQR
L3 Value is < Q, - 3xIQR
U1 Value is > Q, + 1.5xIQR but Q3 + 3xIQR
Outlier Validation Program
V1 Datum has verification flag type K, W, M, or X
V2 QE/QC data outside expected ranges
V3* Horizon aggregation problem
V4 Misclassification in sampling class
V5* Field data notes indicate possible contamination in the pedon
V6 X4 or X5 verification flag present
V7** Double check for use in aggregating
V8** Internally inconsistent chemistry
V9** Indeterminate problem or negative value
DH Impossible value for parameter
Internal Consistency Program
VD PH_01M s 6.5 and 2 5 CEC_OAC < CEC_CL
VF (0.75 * CEC_CL) < SBC_CL if PH 01M s 6.5 and 2 s CEC_CL;
(0.75 * CEC.OAC) < SBC_OAC if~PH_01M « 6.5 and 2 s CEC_CL
VH CA_OAC < MG OAC; CA_OAC < NA_OAC; CA OAC < K OAC;
CA_CL < MG_CL; CA_CL < NA CL; CA CL < K CL;
CA_CL2 < MG_CL2; CA_CL2 < ~NA_CL27 CA_CL2 < K_CL2
VJ PH_01M < 3 and BS_CL > 10%
VK PH_01M < 3 and BS_OAC > 7%
VL PH_01M > 6 and BS_CL < 50%
VM PHJ01M > 6 and BS_OAC < 40%
VP (CEC_OAC / CEC_CL) < 1.5 if CEC_OAC and CEC.CL a 2 meq/100g and PH_01M < 6.0
VQ AL AO < AL CD; FE CD < FE AO; (AL_AO + FE AO) < (AL PYP + FE_PYP);
(AL_CD + FE_CD) <~(AL_PYP + FE_PYP);
VW (S_TOT * 1000) < S04_PO4 if S_TOT a 0.01% (organic soils)
VX (S_TOT * 2000) < S04_PO4 if S_TOT a 0.01% (organic soils)
VY (S_TOT * 10000) < SO4_PO4 if S_TOT a 0.01% (mineral soils)
VZ (S_TOT * 5000) < S04_PO4 if S_TOT a 0.01% (mineral soils)
(Continued)
144
-------
Table A-2. Continued
HB MG_OAC < MG_CL2 if MG_CL2 a 0.05 meq/100 g
HE CEC_CL < AL_CL2 if AL_CL2 s 0.05 meq/100 g
HG (CA_CL + 0.8) < CA_CL2
HI CA_OAC < 0 meq/100 g; MG_OAC < 0 meq/100 g; NA_OAC < 0 meq/100 g; K OAC < 0 meq/100 g;
CA_CL < 0 meq/100 g; MG_CL < 0 meq/100 g; NA_CL < 0 meq/100 g; K CL < 0 meq/100 g;
CA_CL2 < 0 meq/100 g; MG_CL2 < 0 meq/100 g; NA CL2 < 0 meq/100 g~ K CL2 < 0 meq/100 g;
AL_CL2 < 0 meq/100 g; FE_CL2 < 0 meq/100 g
HJ (((CA_CL2 + MG CL2 + NA CL2 + K_CL2 + FE CL2 + AL CL2) * 5000) - (ZSO4 0-32) * 125) - 4000) / 4000
> 2
HK (((CA_CL2 + MG_CL2 + NA CL2 + K CL2 + FE CL2 + AL CL2) * 5000) - (ISO4 0-32) * 125) - 4000) / 4000
< -0.5 - _ _
* Sample Flag: flags the affected parameter for specified samples only.
** Batch-wide Flag: flags the affected parameter for all samples in the batch (assumes that measurement quality
samples are representative for the batch).
145
-------
Appendix B
Quality Assurance Reanalysis Templates
The following QARTs are used in the evaluation of QE/QC and routine sample data. As a
batch of soils data is received, it is reviewed according to the criteria set forth in the template.
If major uncertainty or a large occurrence of minor uncertainty is identified, confirmation or
reanalysis is requested for the affected parameter and batch.
146
-------
SAND, SILT
Lab Audits
(,
Prep Dups
Field Dupe
t
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur:
1) prep dup/routine pain
SO > 3.0 if value at lower limit
2) prep dup/field dup pair:
SO > 3.0 if value at lower limit
3) mean value lab audits out of accuracy window, or
SD > 3.0 if value at lower limit
4 ) low level lab audit out of accuracy window
NONE
NONE
> 10 outliers
MINOR
If 1 out of 4 of the relationships
listed in the major column occurs
Within batch precision:
SD > 6.0 if value at lower limit
NONE
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SD < 6.0 if valur at lower limit
NONE
< 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Figure B-1. Quality assurance reanalysls template for the determination of SAND and SILT.
-------
CLAY
Lab Audits
&
Prep Dups
Field Dups
t
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of the 4 relationships occuri
1) prep dup/routine pair:
SO > 2.0 if value at lower limit
2) prep dup/ field dup pair:
SD > 2.0 if value at lower limit
3) mean value lab audits out of accuracy window, or
SD > 2.0 if value at lower limit
4 ) low level lab audit out of accuracy window
NONE
If any relationship occurs > 6 times:
CEC OAC/CLAY > 50
if clay > 1.0 and CEC CL > 1 meq/100 g
CEC CL/CLAY > 50
if clay > 1.0 and CEC_CL > 1 meq/100 g
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precision:
SD > 4.0 if value at lower limit
If relationships occur < 6 times
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SD < 4.0 if value at lower limit
If relationships do not occur
< 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Figure B-2. Quality assurance reanalysls template for the determination of CLAY.
-------
PH_H20, PH_OO2M, PH_01M
Lab Audits
t
Prep Dupe
Field Dups
I
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur:
1) prep dup/routine pair:
SD > 0.1 if value at lower limit
2) prep dup/field dup pair:
SD > 0.1 if value at lower limit
3) mean value lab audits out of accuracy window, or
SD > 0.1 if value at lower limit
4) low level lab audit out of accuracy window
NONE
If relationship occurs > 6 times
PH H20 < PHJ302M < PH_01H if difference > .05
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precision:
SO > 0.2 if value at lower limit
If relationship occurs < 6 times
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SO < 0.2 if value at lower limit
If relationship does not occur
< 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Figure B-3. Quality assurance reanalysis template for the determination of PH_H20, PH_002M, and PH_01M.
-------
CA, MG, NA, K_CL, CA, MG, NA, K_OAC
Lab Audits
(,
Prep Dups
Field Dupe
&
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur t
1) prep dup/routine pairt
SD > .02 if value at lower limit
IRSD > 15% if value at upper limit
2) prep dup/field dup pair:
SD > .02 if value at lower limit
IRSD > 15% if value at upper limit
3) mean value lab audits out of accuracy window, or
SD > .02 if value at lower limit
IRSD > 10% if value at upper limit
4) low level lab audit out of accuracy window
none
If relationships occur > 6 times:
CA OAC + MG OAC + NA OAC + K OAC > CEC OAC
if" CEC CL > 1 raeq/100 g
CA CL + MG CL + NA CL + K CL > CEC CL
if" CEC CL > 1 meq/100 g
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precision:
SD > .04 if value at lower limit
IRSD > 30% if value at upper limit
If relationship occurs < 6 times
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precis ion i
SD < .04 if value at lower limit
%RSD < 30% if value at upper limit
If relationship does not occur
< 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Figure B-4. Quality assurance reanalysls template for the determination of CA, MG, K, NA_CL, and CA, MG, K, NA_OAC.
-------
AL CL
Ol
Lab Audits
i
Prep Dupe
Field Dupa
&
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur:
1) prep dup/routine pair:
SO > .20 if value at lower limit
%RSD > 221 if value at upper limit
2) prep dup/field dup pair:
SD > .20 if value at lower limit
%RSD > 22% if value at upper limit
3) mean value lab audits out of accuracy window, or
SD > .20 if value at lower limit
%RSD > 15% if value at upper limit
4) low level lab audit out of accuracy window
NONE
If relationship occurs > 6 times:
AL_CL > 0.1 meq/100 g if pH H2O > 6.0
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precision:
SD > 5.0 if value at lower limit
%RSD > 30% if value at upper limit
If relationship occurs < 6 times
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision!
SD < 5.0 if value at lower limit
%RSD < 30% if value at upper limit
If relationship does not occur
< 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Figure B-5. Quality assurance reanalysis template for the determination of AL_CL
-------
CECJDAC, CEC_CL
Lab Audits
I
Prep Dups
Field Dups
&
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur:
1) prep dup/routine pair:
SD > .25 if value at lower limit
%RSD > 15% if value at upper limit
2) prep dup/field dup pair:
SO > .25 if value at lower limit
%RSD > 15% if value at upper limit
3) mean value lab audits out of accuracy window, or
SD > .25 if value at lower limit
%RSD > 10% if value at upper limit
4 ) low level lab audit out of accuracy window
NONE
If relationship occurs > 6 times
CEC_OAC < CEC_CL
if CEC CL > 1 meq/100 g and PH_H2O < 7.0
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precision:
SD > .50 if value at lower limit
%RSD > 30% if value at upper limit
If relationship occurs < 6 times
5-10 outliers
NOD
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SD < .50 if value at lower limit
%RSD < 30% if value at upper limit
If relationship does not occur
< 5 outliers
U!
10
Reanalyais is requested with occurrence of 1 major flag or 3 major flags
Figure B-6. Quality assurance reanalysls template for the determination of CEC_CL and CEC_OAC.
-------
AC BACL
Lab Audits
t
Prep Dups
Field Dups
i
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur:
1) prep dup/routine pain
SO > 1.0 if value at lower limit
«RSD > 22% if value at upper limit
2) prep dup/field dup pair:
%RSD > 22% if value at upper limit
3) mean value lab audits out of accuracy window, or
SD > 1.0 if value at lower limit
»RSD > 151 if value at upper limit
4) low level lab audit out of accuracy window
NONE
NONE
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precision:
SD > 2.0 if value at lower limit
IRSD > 451 when using upper limit
NONE
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SO < 2.0 when using lower limit
ISOR < 45% when using upper limit
NONE
< 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Figure B-7. Quality assurance reanalysls template for the determination of AC_BACL
-------
CA CL2
Lab Audits
t
Prep Dups
Field Dupo
t
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur i
1) prep dup/routine pair:
SD > .05 if value at lower limit
tRSD > 8% if value at upper limit
2) prep dup/field dup pain
SD > .05 if value at lower limit
tRSD > 81 if value at upper limit
3) mean value lab audits out of accuracy window, or
SD > .05 if value at lower limit
tRSD > 51 if value at upper limit
4) low level lab audit out of accuracy window
NONE
NONE
> 10 outliers
MINOR
If 1 of 4 or the relationships
listed in the major column occurs
Within batch precision:
SD > .10 if value at lower limit
tRSD > 151 if value at upper limit
NONE
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SD < .10 if value at lower limit
tRSD < 15t if value at upper limit
NONE
< 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Figure B-8. Quality assurance roanalysl* template for the determination of CA_CL2.
-------
MG_CL2, NA_CL2, K_CL2
Lab Audits
t
Prep Dupa
Field Dupa
I
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur:
1) prep dup/routine pairt
SO >.005 if value at lower limit
%RSD > 15% if value at upper limit
2) prep dup/field dup pair:
SD >.005 if value at lower limit
%RSD > 15% if value at upper limit
3) mean value lab audits out of accuracy window, or
SD >.005 if value at lower limit
tRSD > 101 if value at upper limit
4) low level lab audit out of accuracy window
NONE
If relationship occurs > 6 times
KG CL2 > MG CL * 1.10 if MG CL2 > 10 CRDL
NA CL2 > NA CL * 1.10 if NA~CL2 > 10 * CRDL
K_CL2 > K_CL * 1.10 if K_CL2 > 10 * CRDL
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precision:
SD > .010 if value at lower limit
IRSD > 30% if value at upper limit
If relationships occur < 6 times
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SD < .010 if value at lower limit
%RSD < 30% if value at upper limit
If relationships do not occur
< 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Figure B-9. Quality assurance reanalysla template for the determination of MG_CL2, K_CL2, and NA_CL2.
-------
FE CL2
Lab Audits
I
Prep Dupa
Field Dupa
I
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur:
1) prep dup/routine pain
SD > .01 if value at lower limit
»RSD > 221 if value at upper limit
2) prep dup/field dup pair:
SD > .01 if value at lower limit
%RSD > 22% if value at upper limit
3) mean value lab audits out of accuracy window, or
SD > .01 if value at lower limit
%RSD > 15% if value at upper limit
4 ) low level lab audit out of accuracy window
NONE
NONE
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precision:
SD > .02 if value at lower limit
%RSD > 45% if value at upper limit
NONE
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SD < .02 if value at lower limit
IRSD < 45% if value at upper limit
NONE
< 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Figure B-10. Quality assurance reanalysls template for the determination of FE_CL2.
-------
AL CL2
Lab Audits
t
Prep Dupe
Field Dups
t
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur:
1) prep dup/routine pain
SO > .05 if value at lower limit
%RSD > 22% if value at upper limit
2) prep dup/field dup pain
SD > .05 if value at lower limit
%RSD > 22% if value at upper limit
3) mean value lab audits out of accuracy window, or
SD > .05 if value at lower limit
IRSD > 15% if value at upper limit
4) low level lab audit out of accuracy window
NONE
NONE
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precision:
SD > .10 if value at lower limit
%RSD > 45% if value at upper limit
NONE
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SD < .10 if value at lower limit
%RSD < 45% if value at upper limit
NONE
> 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Figure B-11. Quality assurance reanalysis template for the determination of AL_CL2.
-------
FE_PYP, AL_PYP, FE_AO, AL_AO, SI_AO, FE_CD, AL_CD
Lab Audits
t
Prep Dupa
Field Dups
t
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur:
1) prep dup/routine pair:
SO > .03 if value at lower limit
IRSD > ISt if value at upper limit
2) prep dup/field dup pair:
SO > .03 if value at lower limit
%RSD > 15% if value at upper limit
3) mean value lab audits out of accuracy window, or
SD > .03 if value at lower limit
IRSD > lot if value at upper limit
4 ) low level lab audit out of accuracy window
NONE
NONE
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precision:
SD > .06 if value at lower limit
IRSD > 30% if value at upper limit
NONE
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SD < .06 if value at lower limit
IRSD < 30% if value at upper limit
NONE
< 5 outliers
Reanalyais is requested with occurrence of 1 major flag or 3 major flags
Figure B-12. Quality assurance reanalysls template for the determination of FE_PYP, AL_PYP, FE_AO, AL_AO, SI_AO, FE_CD, and AL_CD.
-------
SO4_P04, SO4_H20
Lab Audits
d
Prep Dups
Field Dups
t
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur:
1) prep dup/routine pair:
SD > 1.5 if value at lower limit
IRSD > 15% if value at upper limit
2) prep dup/field dup pair:
SD > 1.5 if value at lower limit
IRSD > 15% if value at upper limit
3) mean value lab audits out of accuracy window, or
SD > 1.5 if value at lower limit
IRSD > 5% if value at upper limit
4) low level lab audit out of accuracy window
NONE
If relationship occurs £ 6 times
SO4 PO4 < SO4 H2O
if S04_H20 > 1 mg S/kg soil
SO4 H2O/SO4 0 > 4 or SO4 H2O/SO4 0 > 20
if SO4 H2O > 2 mg S/kg and SO4 0 > . 1 mg S/kg
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precision:
SO > 3.0 if value at lower limit
IRSD > 30% if value at upper limit
If relationship occurs < 6 times
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SD < 3.0 if value at lower limit
IRSD < 30% if value at upper limit
If relationship does not occur
< 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Ul
CO
Figure B-13. Quality assurance reanalysls template for the determination of SO4_PO4 and SO4_H2O.
-------
SO4 0-32
.10 if value at lower limit
%RSD > 81 if value at upper limit
2) prep dup/field dup pair:
SD > .10 if value at lower limit
IRSD > 81 if value at lower limit
3) mean value lab audits out of accuracy window, or
SD > .10 if value at lower limit
%RSD > 51 if value at upper limit
4 ) low level lab audit out of accuracy window
NONE
If relationship occurs > 6 times
504 32 < SO4 16 < SO4 8 < SO4 4 < SO4 2 < SO4 0
If S04_32N >= 7.5 mg S/Kg soil
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occur
Within batch precision:
SD > .20 if value at lower limit
IRSD > 15* if value at upper limit
If relationship occurs < 6 times
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SD < .20 if value at lower limit
IRSD < 151 if value at upper limit
If relationship does not occur
< 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Figure B-14. Quality assurance reanalysis template for the determination of SO4_0-32.
-------
C TOT
Lab Audits
t
Prep Dupe
Field Dups
t
Field Audits
Soil
Chemistry
Relationships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur:
1) prep dup/routine pain
SD > .05 if value at lower limit
IRSD > 15% if value at upper limit
2) prep dup/field dup pair:
SD > .05 if value at lower limit
IRSD > 15% if value at upper limit
3) mean value lab audits out of accuracy window, or
SD > .05 if value at lower limit
IRSD > 10% if value at upper limit
4) low level lab audit out of accuracy window
NONE
If any relationship occurs > 6 times
C TOT/N TOT < 7 or C TOT/N TOT > 50
i7 N TOT > .03 wt% and C TOT > .1 wt%
C TOT/S TOT < 40 or C TOT/S TOT > 400
if s TOT > .005 wt% and C TOT > .1 wt%
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precision:
SD > .10 if value at lower limit
IRSD > 30% if value at upper limit
If relationship occurs < 6 times
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SD < .10 if value at lower limit
IRSD < 30% if value at upper limit
If relationship does not occur
< 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Figure B-15. Quality assurance reanalysls template for the determlaatlon of CJTOT.
-------
N TOT
Lab Audits
t
Prep Dups
Field Dups
t
Field Audits
Soil
Chemistry
Relat ionships
Internal
Relationships
MAJOR
If any 2 of 4 relationships occur:
1) prep dup/ routine pain
SD >.015 if value at lower limit
IRSD > 151 if value at upper limit
2) prep dup/field dup pairt
SD >.015 if value at lower limit
IRSD > 151 if value at upper limit
3) mean value lab audits out of accuracy window, or
SD >.015 if value at lower limit
IRSD > 101 if value at upper limit
4 ) low level lab audit out of accuracy window
NONE
If relationship occurs > 6 times
C TOT/NJTOT < 7 or C_TOT/N_TOT > 50
if N TOT > .03 wtl and C_TOT > .1 wtl
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precis ion i
SD > .03 if value at lower limit
IRSD > 301 if value at upper limit
If relationship occurs < 6 times
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision i
SD < .03 if value at lower limit
IRSD < 301 if value at upper limit
If relationship does not occur
< 5 outliers
Reanalysis is requested with occurrence of 1 major flag or 3 major flags
Figure B-16. Quality assurance reanalyele template for the determination of N_TOT.
-------
S TOT
8
Lab Audits
I
Prep Dups
Field Dups
t
Field Audits
Soil
Chemistry
Relationships
Internal
Relat ionships
MAJOR
If any 2 of 4 relationships occur:
1) prep dup/routine pair:
SD >.002 if value at lower limit
%R5D > 151 if value at upper limit
2) prep dup/field dup pair:
SD >.002 if value at lower limit
IRSD > 15% if value at upper limit
3) mean value lab audits out of accuracy window, or
SD >.002 if value at lower limit
IRSD > 101 if value at upper limit
4 ) low level lab audit out of accuracy window
NONE
If relationship occur > 6 times
C_TOT/SJTOT < 40 or C TOT/S TOT > 400
if S_TOT > .005 wt% and C TOT > .1 wt%
> 10 outliers
MINOR
If 1 of 4 of the relationships
listed in the major column occurs
Within batch precision:
SD > .004 if value at lower limit
IRSD > 301 if value at upper limit
If relationship occurs < 6 times
5-10 outliers
NON
If 0 of 4 of the relationships
listed in the major column occur
Within batch precision:
SD <.004 if value at lower limit
IRSD < 30% if value at upper limit
If relationship does not occur
< 5 outliers
Reanalysia is requested with occurrence of 1 major flag or 3 major flags
Figure B-17. Quality assurance reanalyels template for the determination of SJTOT.
-------
Appendix C
Statistics for Step Function Uncertainty Estimates
This table provides statistical information that supplements the results and discussion for
uncertainty estimation found in Section 3. Provided are data relating to the development of delta
values, including the within-batch and between-batch standard deviations, bias calculations, and
proportions. These data can be used to assess the quality of the routine sample data set on the
basis of the quality of the QE sample data sets. The table is sorted by parameter and subsorted
by data set.
164
-------
Table C-1. Table of Statistics for Step Function Uncertainty Estimates
Data
Parameter set* Delta Interval
MOIST LAP . 0.0-1.0
1.0-12.0
12.0-inf
PD . 0.0-1.0
1.0-12.0
12.0-inf
FD . 0.0-1.0
1.0-12.0
1£0-inf
S/H . 0.0-1.0
1.0-12.0
12.0-inf
SAND LAP 3.4377 0.0-15.0
15.0-35.0
35.0-70.0
70.0-inf
PD 3.4736 0.0-15.0
15.0-35.0
35.0-70.0
70.0-inf
FD 4.2475 0.0-15.0
15.0-35.0
35.0-70.0
70.0-inf
S/H 14.5608 0.0-15.0
15.0-35.0
35.0-70.0
70.0-inf
VCOS LAP 1.3902 0.0-5.0
5.0-20.0
20.0-inf
PD 1.4594 0.0-5.0
5.0-20.0
20.0-inf
FD 1.4582 0.0-5.0
5.0-20.0
20.0-inf
S/H 3.8479 0.0-5.0
5.0-20.0
20.0-inf
COS LAP 1.5563 0.0-5.0
5.0-20.0
20.0-inf
PD 1.4941 0.0-5.0
5.0-20.0
20.0-inf
FD 1.6725 0.0-5.0
5.0-20.0
20.0-inf
S/H 4.3095 0.0-5.0
5.0-20.0
20.0-inf
Mean
3.77
.
0.71
1.69
0^73
1.68
0.79
1.66
2s!o4
46.89
.
10.31
27.09
48.62
77.43
10.42
27.50
48.54
77.38
5.18
27.78
48.53
77.39
3.13
5.65
2.25
8.38
27.55
2.14
7.89
26.45
3.04
7.00
3.79
7.62
2.72
8.09
26.38
2.80
8.07
25.55
3.43
8.04
Withirvbatch
df SD
27
22
32
11*
16
t
130
488
14'
13
4
12
32
6
2
6
16
3
1
161
441
15
21
6
31'
22
1
15
11
1
377
241
14
13
22
30
2
11
15
1
238
380
0.2157
0.0685
0.0722
o!o929
0.1295
0.5066
0.6837
11610
1.1581
0.4650
1.8573
0.9958
1.3317
0.8139
2.2609
2.9892
1.9083
10.2530
13.8975
14.5050
10.5925
0.9093
0.9192
.
0.5337
1.7718
7.1418
0.4705
1.8355
8.6974
2.5102
5.7595
0.4559
0.8321
t
0.3240
0.6559
3.2898
0.2763
1.2918
3.1820
2.5386
5.1207
Pairs>MOO
n (%)
5
.
0
0
.
0
0
.
0
0
0
1
1
0
0
0
1
0
(18.5)
,
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(8.3)
(3.1)
(0.0)
(0.0)
(0.0)
(6.3)
(0.0)
Between-batch
df SD
23 0.0658
25 0.1693
Bias
.
PJJ1
0.216
0.784
13' 3.3233
12 3.2846
25 1.0655
Bias
0.3274
pjil
0.008
0.265
0.698
0.029
30 0.7776
19 1.2623
.
Bias
0.4016
PJQ
0.632
0.365
13 0.5213
36 1.5445
1 5.2326
Bias
0.7639
pjil
0.405
0.595
(continued)
165
-------
Table C-1. Continued
Data
Parameter set*
MS
FS
VFS
SILT
LAP
PO
FD
S/H
LAP
PD
FD
S/H
LAP
PD
FD
S/H
LAP
PD
FD
S/H
Delta
1.3945
1.1487
1.3638
6.1430
2.0833
2.0349
2.1160
8.0891
1.9734
1.9869
2.2467
5.1678
4.1033
3.0487
3.5094
10.6184
Interval
0.0-5.0
5.0-20.0
20.0-inf
0.0-5.0
5.0-20.0
20.0-inf
0.0-5.0
5.0-20.0
20.0-inf
0.0-5.0
5.0-20.0
20.0-inf
0.0-5.0
5.0-30.0
30.0-inf
0.0-5.0
5.0-30.0
30.0-inf
0.0-5.0
5.0-30.0
30.0-inf
0.0-5.0
5.0-30.0
30.0-inf
0.0-5.0
5.0-15.0
15.0-inf
0.0-5.0
5.0-15.0
15.0-inf
0.0-5.0
5.0-15.0
15.0-inf
0.0-5.0
5.0-15.0
15.0-inf
0.0-15.0
15.0-35.0
35.0-60.0
60.0-inf
0.0-15.0
15.0-35.0
35.0-60.0
60.0-inf
0.0-15.0
15.0-35.0
35.0-60.0
60.0-inf
0.0-15.0
15.0-35.0
35.0-60.0
60.0-inf
Mean
3.52
8.09
2.92
12.14
21.75
3.00
1Z05
20.55
3.03
10.60
24.97
4.10
11.67
Z17
14.63
41.72
2.32
14.50
40.00
3.33
13.92
35.97
1188
16.75
Z62
9.35
15.98
2.84
10.27
16.42
3.26
9.19
17.26
50.15
69.78
9.55
28.12
43.57
61.32
7.50
26.62
43.32
60.90
13.17
27.85
43.35
63.25
Within-batch
df SO
14
13
17
36
1
9
17
1
102
500
16
14
13
8
44
2
4
22
1
18
596
4
22
5
11
33
10
6
18
3
7
586
25
14
13
3
15
33
3
1
7
18
1
9
180
427
2
0.2563
1.1101
0.2532
0.5893
0.7778
0.2427
1.0429
0.9192
2.2566
6.7744
9.9827
0.2535
0.8155
f
0.3182
0.6163
2.8745
0.1803
0.8756
2.4042
3.8382
8.0112
9.0569
0.6829
0.5822
0.2567
0.7560
1.4064
0.3096
1.3666
0.7670
3.1279
4.6890
8.3035
2.1340
1.5566
1.8462
0.9707
1.1875
1.3693
0.9899
1.6816
2.2215
1.1314
8.3393
10.4795
10.1497
5.3266
Pairs>MQO Between-batch
n (%) df SD
13 0.3154
12 0.9018
26 1.7526
Bias
0.5670
P(H
0.184
0.782
0.035
13 0.4160
17 1.9016
20 2.1748
Bias
-0.5408
Pfi)
0.040
0.949
0.011
0
45 1.6758
3 1.5091
Bias
-0.8096
P»)
0.028
0.921
0.051
26 1.1879
.
2 (14!3) 12 3.4679
0 (0.0) 12 3.4936
0 (0.0)
0 (0.0)
2 (6.1)
0 (0.0) Bias
0 (0.0) 0.8896
0 (0.0)
1 (5.6)
0 (0.0) PJil
0.017
0.285
0.690
0.008
(continued)
166
-------
Table C-1. Continued
Data
Parameter set*
COSI LAP
PD
FD
S/H
FSI LAP
PD
FD
S/H
CLAY LAP
PD
FD
S/H
PHJH20 LAP
PD
FD
S/H
PH_002M LAP
PD
FD
S/H
Delta
8.2282
4.8094
4.8783
7.1527
4.8079
2.6445
2.7860
8.0627
2.6935
1.8156
2.0510
7.5782
0.0847
0.0933
0.1189
0.4741
0.0883
0.0997
0.1285
0.4404
Interval
0.0-10.0
10.0-25.0
25.0-inf
0.0-10.0
10.0-25.0
25.0-inf
0.0-10.0
10.0-25.0
25.0-inf
0.0-10.0
10.0-25.0
25.0-inf
0.0-10.0
10.0-35.0
35.0-inf
0.0-10.0
10.0-35.0
35.0-inf
0.0-10.0
10.0-35.0
35.0-inf
0.0-10.0
10.0-35.0
35.0-inf
0.0-5.0
5.0-20.0
20.0-inf
0.0-5.0
5.0-20.0
20.0-inf
0.0-5.0
5.0-20.0
20.0-inf
0.0-5.0
5.0-20.0
20.0-inf
0.0-5.0
5.0-inf
0.0-5.0
5.0-inf
0.0-5.0
5.0-inf
0.0-5.0
5.0-inf
0.0-3.5
3.5-4.5
4.5-inf
0.0-3.5
3.5-4.5
4.5-inf
0.0-3.5
3.5-4.5
4.5-inf
0.0-3.5
3.5-4.5
4.5-inf
Mean
2Z55
33.19
6.83
14.21
28.20
6.26
14.02
26.80
7.68
13.91
31.93
28.25
36.47
5.58
23.30
40.79
6.55
23.59
40.97
7.41
24.29
38.31
3.51
6.51
3^88
14.31
31.52
4.20
14.43
32.72
£55
14.63
25.03
4.91
5.11
4.56
5.41
4.57
5.46
4.59
5.26
4!43
4.70
3.36
4.18
5.02
3.35
4.20
5.11
3.30
4.15
4.84
Within-batch
df SD
7
20
14
38
2
6
20
1
69
548
1
24
3
3
41
10
2
20
5
14
546
58
15
12
2
41
11
1
21
5
1
420
197
15
12
45
9
23
4
499
119
10
17
4
45
5
2
23
2
12
507
99
7.5647
1.4724
0.8313
0.8294
2.3162
0.5664
1.3195
0.5657
4.8829
5.3520
0.4950
4.5509
1.8930
1.3398
1.0473
0.6738
0.9220
1.3946
1.1068
3.8103
7.6994
8.4857
0.4865
1.6060
t
0.4610
0.7006
0.5568
0.2828
1.1761
0.9930
0.6364
6.9824
8.1818
0.0381
0.0662
0.0580
0.0468
0.0823
0.1433
0.4285
0.6362
o!o299
0.0611
0.0429
0.0635
0.0318
0.0960
0.1042
0.0738
0.1763
0.3808
0.7434
Pairs>MQO
n (%)
0
1
0
1
0
0
0
0
0
3
4
0
0
1
0
1
0
1
0
0
2
0
(0.0)
(8.3)
,
(0.0)
(2.4)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(25.0)
(8.9)
(0.0)
(0.0)
(25.0)
(0.0)
(5.9)
(0.0)
(2.2)
(0.0)
(0.0)
(8.7)
(0-0)
Between-batch
df SD
26 1.2327
6 5.1822
18 2.3383
Bias
0.1970
PJ1
0.116
0.880
0.004
24 0.4173
22 2.6048
2 1.2984
Bias
0.3131
pm
0.028
0.866
0.107
31 0.5966
10 2.0840
. .
Bias
-0.6070
pm
0.003
0.680
0.317
13 0.0493
37 0.1267
Bias
-0.0319
PJl
0.808
0.192
9 0.0678
40 0.1152
Bias
0.0278
PJl
0.032
0.806
0.162
(continued)
167
-------
Table C-1. Continued
Parameter
PH_01M
CA_CL
MG_CL
K_CL
Data
set'
LAP
PD
FD
S/H
LAP
PD
FD
S/H
LAP
PD
FD
S/H
LAP
PD
FD
S/H
Delta
0.0588
0.0619
0.0818
0.4103
0.0299
0.0642
0.1567
1.1138
0.0145
0.0144
0.0308
0.4737
0.0065
0.0080
0.0074
0.1003
Interval
0.0-4.4
4.4-inf
0.0-4.4
4.4-inf
0.0-4.4
4.4-inf
0.0-4.4
4.4-inf
0.0-0.2
02-1.0
1.0-1.5
1.5-inf
0.0-0.2
0.2-1.0
1.0-1.5
1.5-inf
0.0-02
0.2-1.0
1.0-1.5
1.5-inf
0.0-0.2
0.2-1.0
1.0-1.5
1.5-inf
0.0-0.2
02-1.5
1.5-inf
0.0-0.2
02-1.5
1.5-inf
0.0-02
02-1.5
1.5-inf
0.0-0.2
02-1.5
1.5-inf
0.0-0.04
0.04-0.07
0.07-02
02-inf
0.0-0.04
0.04-0.07
0.07-02
02-inf
0.0-0.04
0.04-0.07
0.07-0.2
02-inf
0.0-0.04
0.04-0.07
0.07-0.2
02-inf
Mean
4.28
4.56
3.97
5.64
3.97
5.65
3.92
4.61
o!si
,
0.06
0.46
3.46
0.07
0.50
1.17
3.70
0.10
0.45
1.24
2.41
0.07
f
0.05
0.43
1.71
0.05
0.44
1.74
0.11
0.56
2.35
o!oe
0.09
.
0.03
0.06
0.13
.
0.03
0.06
0.13
oios
0.06
0.11
0.29
Within-batch
df SD
12
15
52
2
26
1
527
91
27'
32
15
7
17
6
1
3
108
328
43
139
27
t
46
4
4
23
2
2
302
310
6
14
13
11
23
20
.
6
11
10
5
98
439
76
0.0249
0.0365
0.0351
0.0050
0.0679
0.0071
0.3491
0.7618
o!o209
0.0074
0.0253
o!l629
0.0430
0.0642
1.1194
0.1343
0.1487
0.8439
1.2642
2.5039
0.0116
.
0!0128
0.0111
0.0772
0.0136
0.0445
0.0581
0.2219
0.6548
2.5086
0.0043
0.0037
0.0027
0.0045
0.0065
0.0044
0.0066
0.0048
,
0.0102
0.0304
0.0721
0.3378
Pairs>MQO
n (%)
0
0
0
0
0
0
2
m
0
1
.
0
2
0
1
0
1
2
0
0
0
0
0
0
0
0
0
0
0
0
0
.
(0.0)
(0.0)
(0.0)
(0.0)
(0-0)
(0.0)
CM)
m
(0.0)
(6.7)
,
(0.0)
(11.8)
(0.0)
(100.0)
(0.0)
(3.7)
m
(4.3)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
-------
Table C-1. Continued
Data
Parameter set* Delta
NA_CL LAP 0.0043
PD 0.0056
FD 0.0061
S/H 0.0123
AL_CL LAP 0.4125
PD 0.3079
FD 0.3214
S/H 2.5307
CAJDAC LAP 0.0317
PD 0.0551
FD 0.2962
S/H 1.0906
MG_OAC LAP 0.0168
PD 0.0189
FD 0.0258
S/H 0.4462
Interval
0.0-0.01
0.01-0.04
0.04-0.2
0.2-inf
0.0-0.01
0.01-0.04
0.04-0.2
0.2-inf
0.0-0.01
0.01-0.04
0.04-0.2
0.2-inf
0.0-0.01
0.01-0.04
0.04-0.2
0.2-inf
0.0-1.3
1.3-3.0
3.0-6.0
6.0-inf
0.0-1.3
1.3-3.0
3.0-6.0
6.0-inf
0.0-1.3
1.3-3.0
3.0-6.0
6.0-inf
0.0-1.3
1.3-3.0
3.0-6.0
6.0-inf
0.0-0.2
0.2-1.0
1.0-5.0
5.0-inf
0.0-0.2
0.2-1.0
1.0-5.0
5.0-inf
0.0-0.2
0.2-1.0
1.0-5.0
5.0-inf
0.0-0.2
0.2-1.0
1.0-5.0
5.0-inf
0.0-0.2
0.2-0.5
0.5-inf
0.0-0.2
0.2-0.5
0.5-inf
0.0-0.2
0.2-0.5
0.5-inf
0.0-0.2
0.2-0.5
0.5-inf
Mean
0.03
0.07
o!oi
0.02
0.06
.
0.01
0.02
0.06
0.01
0.02
0.05
l!83
3.59
0.42
2.35
3.88
7.86
0.03
2.26
3.86
7.75
0.66
2.06
4.52
7.93
0X7
0.06
0.46
2.26
6.52
0.07
0.53
1.93
6.32
0.10
0.46
1.89
6.64
0.07
0.05
0.37
1.42
0.05
0.33
1.24
0.11
0.34
0.90
Wit bin-batch
df SD
15
12
42
6
2
19'
5
1
397
217
4
15
12
3
18
25
8
1
9
13
4
3
181
340
94
27
33
14
5
2
18
5
3
1
113
326
173
6
27
46
3
5
23
1
3
301
158
159
Olo048
0.0065
0!0027
0.0071
0.0000
0.0034
0.0077
0.0007
0!0074
0.0195
0.0239
o'4926
0.2775
0.0345
0.1757
0.2581
0.3119
0.0233
0.2575
0.2714
0.2503
0.6176
1.7403
2.6185
3.6846
0.0242
o!oo70
0.0296
0.0923
0.4203
0.0394
0.0699
0.8623
0.6548
0.1457
0.8691
1.9978
5.6608
0.0142
o!o095
0.0063
0.0433
0.0143
0.0035
0.0661
0.2162
0.4314
0.9137
Pairs>MQO
n (%)
o'
0
o'
0
0
,
0
0
0
.
1
1
0
0
0
0
0
0
0
0
4
.
2
2
0
0
3
1
1
0
3
3
0
0
0
0
0
(o!o)
(0.0)
(0^0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(6.7)
(8.3)
(o.o)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0-0)
(14*)
.
(6.1)
(14.3)
(0.0)
(0.0)
(16.7)
(20.0)
(33.3)
(0.0)
(11.1)
(6.5)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
Between-batch
df SD
20 0.0027
14 0.0044
11 0.0060
.
Bias
-0.0014
PJll
0.642
0.350
0.008
26 0.0866
13 0.1641
11 0.2159
, .
Bias
-0.0615
Pffl
0.011
0.293
0.523
0.172
26 0.0277
25 0.0229
. .
t ,
Bias
0.0004
Pffl
0.204
0.505
0.281
0.011
51 0.0090
. .
Bias
0.0006
PJil
0.493
0.257
0.250
(continued)
169
-------
Table C-1. Continued
Data
Parameter set* Delta
K_OAC LAP 0.0092
PD 0.0127
FD 0.0109
S/H 0.0948
NA_OAC LAP 0.0062
PD 0.0067
FD 0.0075
S/H 0.0144
CEC_CL LAP 0.5139
PD 0.5345
FD 0.5532
S/H 2.8580
CEC_OAC LAP 1.4645
PD 1.7128
FD 1.6816
S/H 4.6847
Interval
0.0-0.05
0.05-0.1
0.1-0.2
0.2-inf
0.0-0.05
0.05-0.1
0.1-0.2
0.2-inf
0.0-0.05
0.05-0.1
0.1-0.2
0.2-lnf
0.0-0.05
0.05-0.1
0.1-0.2
0.2-inf
0.0-0.02
0.02-0.2
0.2-inf
0.0-0.02
0.02-0.2
0.2-inf
0.0-0.02
0.02-0.2
0.2-inf
0.0-0.02
0.02-0.2
0.2-inf
0.0-2.5
2.5-5.5
5.5-12.0
12.0-inf
0.0-2.5
2.5-5.5
5.5-12.0
12.0-inf
0.0-2.5
2.5-5.5
5.5-12.0
12.0-inf
0.0-2.5
2.5-5.5
5.5-12.0
12.0-inf
0.0-2.5
2.5-18.0
18.0-inf
0.0-2.5
2.5-18.0
18.0-inf
0.0-2.5
2.5-18.0
18.0-inf
0.0-2.5
2.5-18.0
18.0-inf
Mean
0.08
.
O!M
0.07
0.16
0.22
0.04
0.07
0.15
0.21
0.04
0.08
0.15
0.31
o!os
o!oi
0.04
t
0.01
0.04
olot
0.03
4.41
7.38
£28
4.41
7.79
15.12
2.31
4.29
7.62
16.08
2.43
4.68
7.81
14.65
10.99
22.66
8.31
28.13
.
a'ae
26.40
9.38
23.82
Within-batch
df SD
27
.
ie'
28
7
3
7
15
3
2
21
364
160
73
27
49'
5
.
25
2
.
482
136
14
13
3
28
19
4
1
14
10
2
4
161
429
24
14
13
52
2
26
1
582
36
0.0053
t
0.0040
0.0033
0.0199
0.0153
0.0058
0.0073
0.0052
0.0172
0.0197
0.0378
0.1033
0.3414
oiooso
o!o026
0.0051
t
0.0037
0.0070
0.0080
0.0219
0.1794
0.4005
0.1389
0.2426
0.2258
1.9403
0.1648
0.2213
0.4741
0.4530
0.2681
2.0715
2.9760
4.9874
0.4452
1.4873
0.7602
3.5041
10254
0.2185
4.1804
7.9077
Pairs>MQO
n (%)
0
0
0
2
0
0
0
0
0
0
0
0
0
0
0
1
0
1
0
1
0
0
0
0
0
2
4
1
l'
0
(oioj
(o!o)
(0.0)
(28.6)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(o!o>
(oio)
(0.0)
(oioj
(0.0)
.
(oioj
(7.7)
(0.0)
(3.6)
(0.0)
(25.0)
(0.0)
(0.0)
(0.0)
(0.0)
(o'.o)
(15.4)
(7.7)
(50.0)
(3JB)
(0.0)
Between-batch
df SD
26 0.0045
25 0.0049
, .
Bias
0.0059
PJil
0.043
0.563
0.262
0.132
26 0.0028
25 0.0048
Bias
0.0043
PJQ
0.640
0.180
26 0.1239
13 0.2513
12 0.4411
t t
Bias
-0.0123
Pfl)
0.008
0.257
0.680
0.055
26 0.1656
13 1.2431
12 1.8608
Bias
-0.4022
PJil
0.919
0.081
(continued)
170
-------
Table C-1. Continued
Parameter
AC_BACL
CA_CL2
MG_CL2
K_CL2
Data
set*
LAP
PD
FD
S/H
LAP
PD
FD
S/H
LAP
PD
FD
S/H
LAP
PD
FD
S/H
Delta
1.6683
1.6590
1.8373
5.5839
0.0490
0.0473
0.0476
0.1056
0.0075
0.0053
0.0087
0.0841
0.0027
0.0029
0.0039
0.0220
Interval
0.0-6.7
6.7-20.0
20.0-inf
0.0-6.7
6.7-20.0
20.0-inf
0.0-6.7
6.7-20.0
20.0-inf
0.0-6.7
6.7-20.0
20.0-inf
0.0-0.4
0.4-1.0
1.0-inf
0.0-0.4
0.4-1.0
1.0-inf
0.0-0.4
0.4-1.0
1.0-inf
0.0-0.4
0.4-1.0
1.0-inf
0.0-0.05
0.05-0.08
0.08-0.2
0.2-inf
0.0-0.05
0.05-0.08
0.08-0.2
0.2-inf
0.0-0.05
0.05-0.08
0.08-0.2
0.2-inf
0.0-0.05
0.05-0.08
0.08-0.2
0.2-inf
0.0-0.02
0.02-0.05
0.05-inf
0.0-0.02
0.02-0.05
0.05-inf
0.0-0.02
0.02-0.05
0.05-inf
0.0-0.02
0.02-0.05
0.05-inf
Mean
16!30
29.02
5.37
10.50
26.50
5.51
11.28
32.13
5.71
10.84
25.80
o!s7
0.38
0.58
0.36
0.58
0.36
0.53
0.04
0.06
o!o2
0.07
0.15
0.29
0.02
0.07
0.15
0.30
0.04
0.07
0.13
0.28
0.02
0.03
0.01
0.04
0.05
0.01
0.04
o!o2
0.04
0.08
Withitvbatch
df SD
14
13
19
32
3
11
15
1
131
444
43
27
.
2
52
.
1
26
.
8
610
15
12
t
37
9
6
2
19
4
3
1
209
105
277
27
24
3
44
9
1
22
5
.
430
157
31
1.0665
1.6857
0.6044
0.5668
3.1085
0.9879
0.9447
3.0059
3.3224
5.7101
7.7319
oioiai
0.0356
0.0132
.
0.0021
0.0157
.
0.1353
0.0943
0.0058
0.0023
m
0.0020
0.0060
0.0014
0.0099
0.0048
0.0089
0.0074
0.0184
0.0635
0.0743
0.0951
0.1597
0.0013
0.0009
o!ooog
0.0029
0.0021
0.0013
0.0060
o!(X>95
0.0348
0.1006
Pairs>MQO
n (%)
1
0
1
0
0
0
0
0
0
t
1
0
.
0
0
.
1
0
.
t
1
1
0
0
1
0
0
0
0
0
.
0
2
0
0
0
(7.1)
(0.0)
(5.3)
(0.0)
(0.0)
(0-0)
(0.0)
(0.0)
(o!o)
(50.0)
(0.0)
(0.0)
(0.0)
(6.7)
(0.0)
,
(2.7)
(11.1)
(0.0)
(0.0)
(5.3)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
.
(0.0)
(22.2)
(0.0)
(0.0)
(0.0)
Between-batch
df SD
24 0.4623
13 1.4877
12 2.5234
Bias
-0.1415
PJil
0210
0.692
0.097
51 o!()382
, .
Bias
0.0249
PJil
0.027
0.973
39 0.0057
11 0.0048
. .
. .
Bias
0.0024
PJ1
0.330
0.178
0.445
0.047
48 0.0025
2 0.0020
.
Bias
0.0005
PJil
0.668
0.272
0.060
(continued)
171
-------
Table C-1. Continued
Parameter
NA_CL2
FE_CL2
AL_CL2
Data
set* Delta Interval
LAP 0.0046 0.0-0.01
0.01-0.03
0.03-0.05
0.05-inf
PD 0.0023 0.0-0.01
0.01-0.03
0.03-0.05
0.05-inf
FD 0.0024 0.0-0.01
0.01-0.03
0.03-0.05
0.05-inf
S/H 0.0092 0.0-0.01
0.01-0.03
0.03-0.05
0.05-inf
LAP 0.0012 0.0-0.00
0.00-0.02
0.02-0.07
0.07-inf
PD 0.0019 0.0-0.00
0.00-0.02
0.02-0.07
0.07-inf
FD 0.0014 0.0-0.00
0.00-0.02
0.02-0.07
0.07-inf
S/H 0.0195 0.0-0.00
0.00-0.02
0.02-0.07
0.07-inf
LAP 0.0092 0.0-0.03
0.03-0.07
0.07-0.1
0.1-0.3
0.3-inf
PD 0.0066 0.0-0.03
0.03-0.07
0.07-0.1
0.1-0.3
0.3-inf
FD 0.0092 0.0-0.03
0.03-0.07
0.07-0.1
0.1-0.3
0.3-inf
S/H 0.0946 0.0-0.03
0.03-0.07
0.07-0.1
0.1-0.3
0.3-inf
Mean
0.01
0.02
0.05
0.06
0.01
0.03
o!oi
0.03
o!oi
0.02
0.04
0.00
0.01
o!oo
0.00
0.03
0.00
0.00
0.03
0.00
0.01
0.03
0.16
0.01
0.05
0.09
0.02
0.04
0.12
0.21
0.02
0.04
0.12
0.21
0.02
0.05
0.12
0.21
Within-batch
df SD
2
12
11
2
50
4
25'
2
505'
111
2
8
9
22
1
2
12
1
1
469
103
11
22
14
12
1
7
10
29
6
3
6
14
3
59
114
419
26
0.0046
0.0010
0.0067
0.0021
0.0009
0.0012
o!oon
0.0016
0.0074
0.0166
0.0065
0.0007
0.0007
0.0012
0.0035
0.0020
0.0010
0.0014
0.0028
o!o027
0.0163
0.0477
0.3604
0.0039
0.0049
0.0092
0.0021
0.0026
0.0059
0.0025
0.0035
0.0151
0.0072
0.0054
0.0365
0.0579
0.0965
0.3005
Pairs>MQO
n (%)
0
0
2
0
0
0
0
0
0
0
0
0
0
0
0
0
.
0
0
0
0
0
0
0
0
0
0
0
(0.0)
(0.0)
(18.2)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
.
(o!o)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
Between-batch
df SD
27 0.0015
11 0.0020
10 0.0054
1 0.0060
Bias
0.0013
PJil
0.819
0.176
0.005
29 0.0006
8 0.0013
.
Bias
-0.0004
pm
0.715
0.184
0.025
0.035
35 0.0044
11 0.0083
0
.
Bias
0.0023
P(i)
0.104
0.186
0.652
0.056
(continued)
172
-------
Table C-1. Continued
Parameter
FE_PYP
AL_PYP
FE_AO
AL_AO
Data
set* Delta
LAP 0.0321
PD 0.0357
FD 0.0395
S/H 0.3058
LAP 0.0252
PD 0.0222
FD 0.0274
S/H 0.1647
LAP 0.0469
PD 0.0343
FD 0.0448
S/H 0.2684
LAP 0.0224
PD 0.0286
FD 0.0360
S/H 0.1311
Interval
0.0-0.3
0.3-0.7
0.7-inf
0.0-0.3
0.3-0.7
0.7-inf
0.0-0.3
0.3-0.7
0.7-inf
0.0-0.3
0.3-0.7
0.7-inf
0.0-0.1
0.1-0.3
0.3-0.7
0.7-inf
0.0-0.1
0.1-0.3
0.3-0.7
0.7-inf
0.0-0.1
0.1-0.3
0.3-0.7
0.7-inf
0.0-0.1
0.1-0.3
0.3-0.7
0.7-inf
0.0-0.2
0.2-0.3
0.3-0.6
0.6-inf
0.0-0.2
0.2-0.3
0.3-0.6
0.6-inf
0.0-0.2
0.2-0.3
0.3-0.6
0.6-inf
0.0-0.2
0.2-0.3
0.3-0.6
0.6-inf
0.0-0.3
0.3-1.1
1.1-inf
0.0-0.3
0.3-1.1
1.1-inf
0.0-0.3
0.3-1.1
1.1-inf
0.0-0.3
0.3-1.1
1.1-inf
Mean
0.62
0.76
0.18
0.45
1.07
0.18
0.45
1.05
0.19
0.45
0.95
.
0.54
0.84
0.10
0.22
0.36
0.10
0.21
0.36
0.10
0.21
0.37
1.02
_
0.99
0.12
0.23
0.41
0.72
0.13
0.24
0.42
0.71
0.15
0.26
0.41
0.79
o!87
1.41
0.15
0.37
o!i4
0.36
_
o!i7
0.37
Within-batch
df SD
12
15
32
20
2
16
10
1
294
296
28
_
14
13
26
19
9
13
9
5
250
289
79
0
.
.
27
25
10
15
4
14
4
7
2
102
184
292
40
14
13
41
13
20
7
55l'
67
0.0202
0.0266
0.0133
0.0269
0.0244
0.0213
0.0286
0.0474
0.1869
0.3774
0.6662
0.0412
0.0331
0.0116
0.0188
0.0181
.
0!0134
0.0267
0.0335
.
o!o578
0.1812
0.4194
.
o!o413
0.0412
0.0177
0.0246
0.0371
0.0613
0.0250
0.0377
0.0305
0.1043
0.2411
0.3192
0.4292
o!o345
0.0718
0.0189
0.0208
010273
0.0423
o!l152
0.2277
Pairs>MQO
n (%)
0
0
1
0
0
0
0
0
1
0
1
2
0
t
0
0
0
t
t
.
.
0
1
1
0
0
1
0
0
0
o'
1
1
1
1
0
(0.0)
(0.0)
(3.1)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
.
.
(7.1)
(0.0)
(3.8)
(10.5)
(0.0)
(0.0)
(0.0)
(0.0)
.
.
(0.0)
(4.0)
(10.0)
(0.0)
(0.0)
(7.1)
(0.0)
(0.0)
(0.0)
(o!o)
(7.7)
(2.4)
(7.7)
(5.0)
(0.0)
Between-batch
df SD
25 0.0082
11 0.0313
13 0.0532
Bias
-0.0176
pjil
0.478
0.459
0.063
26 0.0140
.
13 0.0379
12 0.0665
Bias
-0.0037
PJl
0.413
0.445
0.141
0.001
25 0.0249
.
.
25 0.0806
Bias
0.0107
PJl
0.182
0.281
0.455
0.081
26 0.0175
13 0.0430
12 0.0555
Bias
0.0018
Ltil
0.874
0.126
*
(continued)
173
-------
Table C-1. Continued
Data
Parameter set* Delta
SI_AO LAP 0.0106
PD 0.0138
FD 0.0137
S/H 0.0292
FE_CD LAP 0.1088
PD 0.2893
FD 0.1367
S/H 0.8260
AL_CD LAP 0.0248
PD 0.0259
FD 0.0250
S/H 0.1247
SO4_H20 LAP 1.6327
PD 1.6279
FD 1.9181
S/H 27.1745
Interval
0.0-0.05
0.05-0.3
0.3-inf
0.0-0.05
0.054.3
0.3-inf
0.0-0.05
0.05-0.3
0.3-inf
0.0-0.05
0.05-0.3
0.3-inf
0.0-0.3
0.3-2.5
^5-3.0
3.0-inf
0.0-0.3
0.3-2.5
2.5-3.0
3.0-inf
0.0-0.3
0.3-2.5
2.5-3.0
3.0-inf
0.0-0.3
0.3-2.5
2.5-3.0
3.0-inf
0.0-0.2
0.2-0.3
0.3-0.8
0.8-inf
0.04.2
0.2-0.3
0.3-0.8
0.8-inf
0.0-0.2
0.2-0.3
0.3-0.8
0.8-inf
0.0-0.2
0.2-0.3
0.3-0.8
0.8-inf
0.0-7.0
7.0-15.0
15.0-18.0
18.0-inf
0.0-7.0
7.0-15.0
15.0-18.0
18.0-inf
0.0-7.0
7.0-15.0
15.0-18.0
18.0-inf
0.0-7.0
7.0-15.0
15.0-18.0
18.0-inf
Mean
0.22
0.35
0.02
0.08
0.02
0.08
0.02
0.08
159
.
t
146
2.77
3.71
146
2.81
3.72
0.22
167
2.63
3.29
0'.63
1.12
0.17
0.26
0.40
0.17
0.26
0.39
0.17
0.27
0.35
12^77
15.90
20.79
3.74
11.85
15.83
23.28
3.54
12.11
16.47
23.05
4.79
11.19
15.94
46.59
Within-batch
df SD
18
9
38
8
19'
4
606
8
27
t
42
5
7
21
2
4
4
505
59
50
14
13
23
14
17
11
7
9
261'
176
181
11
3
13
4
30
10
10
2
17
3
5
74
296
128
120
0.0154
0.0422
0.0087
0.0186
.
0.0085
0.0168
o!o256
0.1124
0.049$
t
0^3193
0.0677
0.1193
o!o935
0.0863
0.2223
0.1347
0.7575
0.9675
13848
010265
0.0792
0.0045
0.0340
0.0136
0.0096
0.0254
0.0146
0.0980
0.1192
0.1635
16719
0.6183
15771
0.4821
1.3852
1.5625
0.7616
0.5884
1.7065
2.5084
0.6322
3.3512
5.4730
6.1761
123.979
Pairs>MQO
n (%)
1
3
0
1
0
0
0
3
0
0
0
0
0
0
2
0
1
0
0
0
0
1'
0
2
0
2
1
0
0
1
0
0
(5^6)
(33.3)
(0.0)
(12.5)
(0.0)
(0.0)
(oio)
(7.'D
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(o'.o)
(15.4)
(0.0)
(7.1)
(0.0)
(0-0)
(0.0)
(0.0)
(9^1)
(0.0)
(15.4)
(0.0)
(6.7)
(10.0)
(0.0)
(0-0)
(5.9)
(0.0)
(0.0)
Betweervbatch
df SD
25 0.0094
16 0.0368
7 0.0572
Bias
-0.0034
P ffl
0.969
0.019
26 0.0312
25 0.0971
.
Bias
0.0123
P_fjl
0.009
0.818
0.095
0.079
26 0.0099
.
13 0.0375
12 0.1044
Bias
0.0032
PJil
0.434
0.269
0.297
25 0.6444
10 14299
1 0.0361
12 11352
Bias
0.1264
fjil
0.145
0.481
0.190
0.184
(continued)
174
-------
Table C-1. Continued
Data
Parameter set* Delta
SO4_PO4 LAP 12.1608
PD 6.8784
FD 7.9622
S/H 49.5081
SO4_0 LAP 0.1111
PD 0.1029
FD 0.1417
S/H 3.3925
SO4_2 LAP 0.1262
PD 0.1337
FD 0.1496
S/H 5.3257
SO4_4 LAP 0.1726
PD 0.1625
FD 0.2191
S/H 6.3261
Interval
0.0-15.0
15.0-50.0
50.0-100.0
100.0-inf
0.0-15.0
15.0-50.0
50.0-100.0
100.0-inf
0.0-15.0
15.0-50.0
50.0-100.0
100.0-inf
0.0-15.0
15.0-50.0
50.0-100.0
100.0-inf
0.0-1.0
1.0-2.0
2.0-inf
0.0-1.0
1.0-2.0
2.0-inf
0.0-1.0
1.0-2.0
2.0-inf
0.0-1.0
1.0-2.0
2.0-inf
0.0-2.0
2.0-2.7
2.7-inf
0.0-2.0
2.0-2.7
2.7-inf
0.0-2.0
2.0-2.7
2.7-inf
0.0-2.0
2.0-2.7
2.7-inf
0.0-2.0
2.0-3.0
3.0-3.7
3.7-inf
0.0-2.0
2.0-3.0
3.0-3.7
3.7-inf
0.0-2.0
2.0-3.0
3.0-3.7
3.7-inf
0.0-2.0
2.0-3.0
3.0-3.7
3.7-inf
Mean
7^25
111.9
4.11
30.82
68.92
121.6
4.62
29.95
68.80
121.8
8.81
29.07
69.10
260.2
l'.33
2.12
0.75
1.38
2.44
0.74
1.41
2.47
0.69
1.44
8.27
1.95
2.16
3.13
1.79
2.29
3.38
1.81
2.35
3.43
1.61
243
5.20
2^96
3.23
4.19
2.58
3.40
4.68
2.58
3.43
4.71
1.65
2.70
3.47
5.93
Wit bin-batch
df SD
15
12
2
29
19
4
1
14
10
2
49
256
292
21
14
13
10
31
13
5
16
6
97
449
72
1
12
14
12
19
23
7
9
11
78
329
211
2
11
14
14
13
27
7
7
13
13
76
233
296
11.0785
3.5698
1.4634
3.8895
Z8575
0.9397
£5597
5.5835
4.7724
5.0722
4.6368
28.2552
54.1880
393.295
0.0539
0.1082
0.0344
0.0449
0.0423
0.0179
0.1254
0.0652
0.4000
0.6145
23.7904
0.0608
0.0437
0.0603
0.0404
0.0704
0.0813
0.0734
0.0991
0.1020
0.7585
0.6049
13.5590
o!l228
0.0895
0.1161
o!l073
0.0741
0.0978
o!l624
0.2384
0.1120
0.9980
1.0777
0.7779
11.6513
Pairs>MQO
n (%)
1
0
1
2
0
0
0
2
0
0
0
6
0
0
0
0
2
0
0
0
0
0
0
0
0
0
0
1'
2
0
1
0
0
0
0
0
(6.7)
(0.0)
(50.0)
(6.9)
(0.0)
(0.0)
(0.0)
(14.3)
(0.0)
(0.0)
(0.0)
(46.2)
(0.0)
(0.0)
(0-0)
(0.0)
(12.5)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
(50.0)
(18.2)
(0.0)
(7.'l)
(0.0)
(0.0)
(0.0)
(0.0)
(0.0)
Between-batch
df SD
26 1.7715
13 9!6793
11 7.0690
Bias
2.0085
PJl
0.108
0.429
0.430
0.033
26 0.1200
12 0.0826
12 0.0498
Bias
0.0365
P(0
0.178
0.700
0.121
37 0.1347
13 0.0858
Bias
0.0445
pjil
0.133
0.505
0.362
1 0.0544
10 0.0926
39 0.1591
Bias
0.0504
PJU
0.023
0.126
0.345
0.506
(continued)
175
-------
Table C-1. Continued
Data
Parameter set* Delta
SO4_8 LAP 0.2954
PD 0.2966
FD 0.3807
S/H 5.2892
SO4J6 LAP 0.6141
PD 0.5706
FD 0.6337
S/H 6.2114
SO4_32 LAP 1.2826
PD 1.3867
FD 1.3435
S/H 7.1054
Interval
0.0-2.0
2.0-5.8
5.8-7.5
7.5-inf
0.0-2.0
2.0-5.8
5.8-7.5
7.5-inf
0.0-2.0
2.0-5.8
5.8-7.5
7.5-inf
0.0-2.0
2.0-5.8
5.8-7.5
7.5-inf
0.0-2.0
2.0-7.5
7.5-1ZO
12.0-14.0
14.0-inf
0.0-2.0
2.0-7.5
7.5-12.0
12.0-14.0
14.0-inf
0.0-2.0
2.0-7.5
7.5-12.0
12.0-14.0
14.0-inf
0.0-2.0
2.0-7.5
7.5-12.0
12.0-14.0
14.0-inf
0.0-2.0
2.0-18.0
18.0-20.0
20.0-30.0
30.0-inf
0.0-2.0
2.0-18.0
18.0-20.0
20.0-30.0
30.0-inf
0.0-2.0
2.0-18.0
18.0-20.0
20.0-30.0
30.0-inf
0.0-2.0
2.0-18.0
18.0-20.0
20.0-30.0
30.0-inf
Mean
523
6.45
t
.
4.82
6.64
8.30
.
4.76
6.55
825
1.55
5.13
6.58
1£31
10^95
1225
6.89
10.03
12.95
15.29
7.09
10.04
13.03
15.41
s!i3
10.78
12.96
19.04
19.70
23.63
17!47
19.89
25.79
31.65
17.45
_
25^68
31.63
_
15J5
19.34
25.11
35.60
Within-batch
df SD
13'
14
_
21
18
15
10
9
8
1
192
323
102
18
9
2
18
20
14
V
9
11
6
1
270
233
114
2
25
5
1
42
6
3
21
3
l'
2
528
87
0.1490
0.1227
0.1159
0.0883
0.1707
0.1999
0.2974
02466
0.6322
1.5537
1.4406
20.4390
02068
0.4184
o!5507
0.1356
0.2247
0.2112
o!l478
0.2643
0.3853
0.3208
13810
2.6221
2.1651
19.4814
2.7652
0.4631
32481
0.2475
0.7429
0.2473
4.3546
0^5748
0.3889
32937
3.5718
3.9387
22.5927
Pairs>MQO
n (%)
1
0
o'
0
0
0
0
0
0
1
1
0
0
0
0
0
0
0
1'
1
1'
0
1
0
1
0
0
(7.7)
(0.0)
(0.0)
(0.0)
(0.0)
(o'o)
(0.0)
(0.0)
(o!oj
(11.1)
(so!o)
(0.0)
(0.0)
(0.0)
(o!o)
(0.0)
(0.0)
(0.0)
(solo)
(4.0)
(2o!o)
(0.0)
(2.4)
(0.0)
(33.3)
(o!o)
(0.0)
Betweervbatch
df SD
12 0.3588
13 0.2314
26 0.2160
Bias
0.0520
PJil
0.003
0298
0.499
0200
16 0.7376
8 0.3194
26 0.4227
Bias
0.1074
Pffl
0.004
0.418
0.358
0.220
0 '.
23 1.2642
26 0.8822
Bias
0.0986
PJil
0.007
0.005
0.827
0.161
(continued)
176
-------
Table C-1. Continued
Data
Parameter set* Delta
C TOT LAP 0.0835
~
PD 0.0715
FD 0.1910
S/H 1.2144
N TOT LAP 0.0036
~
PD 0.0050
FD 0.0084
S/H 0.0534
S_TOT LAP 0.0016
PD 0.0014
FD 0.0018
S/H 0.0267
Interval
0.0-0.5
0.5-2.5
2.5-6.0
6.0-inf
0.0-0.5
0.5^5
2.5-6.0
6.0-inf
0.0-0.5
0.5-2.5
2.5-6.0
6.0-inf
0.0-0.5
0.5-2.5
2.5-6.0
6.0-inf
0.0-0.08
0.08-0.1
0.1-0.3
0.3-inf
0.0-0.08
0.08-0.1
0.1-0.3
0.3-inf
0.0-0.08
0.08-0.1
0.1-0.3
0.3-inf
0.0-0.08
0.08-0.1
0.1-0.3
0.3-inf
0.0-0.00
0.00-0.02
0.02-0.04
0.04-inf
0.0-0.00
0.00-0.02
0.02-0.04
0.04-inf
0.0-0.00
0.00-0.02
0.02-0.04
0.04-inf
0.0-0.00
0.00-0.02
0.02-0.04
0.04-inf
Mean
1.46
4.62
Ol26
0.83
4.80
7.66
0.25
0.79
4.83
7.71
0.31
1.02
3.58
7.55
o!i2
0.22
0.04
0.10
0.26
0.35
0.04
0.10
0.26
0.35
0.04
0.10
0.20
0.40
0.02
0.03
o!oi
0.01
0.03
0.04
0.01
0.01
0.03
t
0.01
0.01
0.03
0.14
Withirvbatch
df SD
14
13
28
22
2
2
13
12
1
1
279
245
50
44
14'
13
45
5
2
2
23
2
1
1
465
61
65
27
5
22
20
26
7
1
11
12
4
172
355
48
43
0.0338
0.2986
oloi19
0.0470
0.0462
0.0809
0.0502
0.1269
0.4992
0.7481
0.3334
0.7593
2.0228
7.7529
o!o053
0.0084
0.0040
0.0025
0.0036
0.0056
0.0055
0.0025
0.0163
0.0290
0.0292
0.0599
0.1297
0.1878
o!ooii
0.0022
.
0.0007
0.0009
0.0007
0.0001
0.0012
0.0011
0.0025
o!o045
0.0080
0.0118
0.2763
Pairs>MQO
n (%)
o'
2
o'
1
0
0
1
1
0
0
1
0
o'
0
0
0
0
0
0
0
0
1
1'
1
0
0
0
0
0
(0.0}
(15.4)
(0.0)
(4.5)
(0.0)
(0.0)
(7.7)
(8-3)
(0.0)
(0.0)
(7!l)
(0.0)
(0.0)
(0.0)
(0.0)
(0-0)
(0-0)
(0.0)
(0.0)
(0.0)
(0.0)
(4.5)
(5.0)
(3.8)
(0.0)
(0-0)
(0.0)
(0.0)
(0.0)
Between-batch
df SD
26 0.0157
13 0.0349
12 0.1975
, .
Bias
-0.0339
p_ia
0.434
0.387
0.107
0.072
26 0.0006
13 0.0043
12 0.0070
.
Bias
0.0017
PJil
0.715
0.109
0.123
0.053
25 0.0004
4 0.0012
20 0.0016
. ,
Bias
-0.0007
p_ia
0.286
0.542
0.100
0.072
* LAP - laboratory audit sample pairs; PD = preparation duplicates; FD
class/horizon groups of routine samples.
inf = open-ended interval, i.e., infinity; df
samples in the ith interval.
field duplicates; S/H - sampling
degrees of freedom; SD = standard deviation; P(i) = proportion of routine
177
-------
Appendix D
Inordinate Data Points Influencing the Uncertainty Estimates
Appendix D provides information on specific data points that have an inordinate effect on
the uncertainty estimates presented in the results and discussion for precision and accuracy in
Section 3. Included is information on each parameter for each data set for the sampling
class/horizon group, the batch/sample number, and the reason for its effect on the estimates. Data
users interested in the quality of data in specific batches will find this table particularly helpful.
The table is sorted by parameter and subsorted by data set.
178
-------
Table D-1. Inordinate Data Polnte Having a High Degree of Influence on the Uncertainty Estimates for the
Data Sets
Data
Parameter set'
MOIST S/H
FD
PD
AS
SAND FD
PD
AS
VCOS FD
PD
AS
COS FD
PD
AS
MS S/H
FD
PD
AS
FS FD
PD
AS
VFS S/H
FD
PD
AS
SILT FD
PD
AS
Sampling
class/horizon
HST /
HST /
HST /
TXW /
TWO/
TXW /
BMK/
FLW /
FLW /
FLW /
TXW /
FLW /
FLW /
FLW /
BXW/
CFP /
TXW /
BMK/
FLW /
FLW /
CFP /
BMK /
CFP /
BNS /
BXW/
BMK /
BMK /
TXW /
TXW /
Oe
Oe
Oe
C
Bw
Bt
C
Bt
C
B
B
C
C
C
C
B
Bw
Bt
C
eg
C
C
B
E
Ag
Bt
Bw
eg
C
C
B
C
Bw
C
C
B
B
Bw
Cr
E
Bw
Bw
Bw
B
B
Bt
Bt
C
C
B
Bw
Batch/sample
30131-14,19
30132-26,34
30127-10
30127-4,23
30113-25,6
30112-37.18
30126-23,5
30118-6
30106-3,34
30117-26,39
30120-34,32
30120-34,38
30120-32,20
30120-23
30106-3,34
30130-10,28
30113-25,6
30120-34,32
30121-1,19
30120-
-------
Table D-1. Continued
Data
Parameter set'
COSI FD
PD
AS
FSI S/H
FD
PD
AS
CLAY S/H
PD
AS
PH_H20 S/H
FD
PD
AS
PH_002M S/H
FD
PD
AS
PH_01M S/H
FD
AS
CA_CL S/H
FD
PD
AS
MG_CL S/H
FD
AS
Sampling
class/horizon
TXW /
BMK /
TXW /
BNS /
TXW /
TXW /
TWO/
BXP /
TXW /
CFP /
FLW /
BNS /
TXW /
TXW /
FLW /
BNS /
BNS /
BMD /
TXW /
BMD/
FLW /
BNS /
TXW /
BMD /
BND /
BNS /
FLW /
TXW /
FLW /
BXW/
TPD /
TPD /
BNS /
TPD /
HST /
Bt
Bw
Bt
C
B
Bw
AE
Bt
Bt
C
C
B
B
Bw
Bw
Bw
Bw
C
C
Bw
BA
Oe
Btx
Btx
Bw
BA
Cr
Oe
Bw
Bt
Bw
Bw
C
BA
Oe
Bt
E
Bw
Bw
C
Oe
Ap
Bt
C
Bw
Bt
Bt
O
Oe
Bw
Oe
B
B
0
Batch/sample
30113-25,6
30117-26,39
30126-23,5
30118-6
30106-4,11
30115-10,28
30113-25,6
30126-18,23
30112-37,18
30118-6
30106-4,11
30106-3,34
30109-23,26
30115-10,28
30115-30,31
30118-6
30109-23,26
30108-39.10
30108-39,6
30107-27.36
30105-39,3
30113-25,6
30105-3.7
30107-12,37
30108-19
30113-25,6
30118-4,17
30123-34,4
30119-20,33
30104-40
30113-25,6
30120-34,32
30128-34,16
30114-8,18
30114-21,1
30131-6,13,31
30131-14,19
30122-6.28
30128-7,24
30131-6,13.31
Reason
large variability
low value
high value
high audit value
negative lab audit value
high field audit value
large variability
large variability
large variability
large variability
high field audit value
high lab audit value
low field audit value
low lab audit value
low field audit value
high value
high value
large variability
high field audit value
high lab audit values
high value
large variability *
large variability
large variability
high field audit value
high value
large variability
large variability *
high value
large variability
high FD value
high lab audit value
low lab audit value
high value
large variability *
large variability
large variability
large variability
high field audit value
low lab audit value
high value
large variability
high value
large variability
large variability
large variability
large variability
high lab audit value
large variability
large variability
large variability
high lab audit value
high field audit values
large lab audit variability
(Continued)
180
-------
Table D-1. Continued
Data
Parameter set'
K CL S/H
FD
PD
AS
NA CL FD
PD
AS
AL CL S/H
FD
AS
CA_OAC S/H
FD
PD
AS
MG_OAC S/H
FD
PD
AS
K_OAC S/H
FD
PD
AS
NA OAC S/H
FD
PD
AS
Sampling
class/horizon
FLW /
TXW /
BMK /
HST /
FLW /
FLW /
TXW /
HST /
BMD /
BMK /
BNS /
FLW /
TXW /
BMD/
FLW /
FLW /
BXW/
HST /
TPD /
HST /
TPD /
BNS /
TPD /
TPD /
TXW /
BMK /
HST /
FLW /
FLW /
HST /
HST /
BMD /
TPD /
TXP /
TPD /
TXP /
HST /
Ap
Bt
C
Oe
O
C
C
Bw
B
B
B
Oe
Oe
Ap
B
B
Oe
Oe
Ap
Bt
Ap
C
Cg
Bw
Oe
Bt
Oe
C
B
B
Bw
Bw
Oe
C
Bt
Bt
C
Oe
Bw
Bw
B
B
Ap
BA
Oe
Oe
C
Bw
A
Bt
Btx
Bt
Btw
Oe
C
Bw
Bw
Batch/sample
30113-25.6
30107-25,34
30131-19,24
30131-6,13,31
30120-34,32
30120-34,38
30107-12,37
30104-19,38
30110-29,39
30116-7,22
30132-26,39
30104-29,14
30122-6,28
30120-10,37
30113-25,6
30104-29,14
30120-34,32
30121-1.19
30128-34,16
30132-26,39
30114-8,18
30131-19,24
30113-5
30128-7.24
30126-4,35
30115-18.36
30114-6.21
30126-18.23
30107-6,1
30131-19,24
30105-32,34
30115-18.36
30122-6,28
30128-7,24
30131-14,19
30131-19,24
30112-38
30125-3,27
30114-8,21
30125-29.1
30114-8.18
30125-1,26
30131-19.24
30123-38
30115-18,36
30129-2,31
Reason
large variability
large variability
large variability
large variability
large lab audit variability
large variability
large variability
high lab audit value
high lab audit value
large field audit variability
low field audit value
large variability
large variability
large variability
low lab audit value
low lab audit value
large variability
high value
large variability
high value
large variability
large variability
large variability
large variability
large variability
high value
large variability
high lab audit value
high field audit value
high field audit value
high lab audit value
large variability
large variability
large variability
large variability
high value
large variability
large variability
high field audit value
high lab audit value
high lab audit value
high field audit value
large variability
large variability
large variability
large variability
high field audit value
high lab audit value
large variability *
high value
large variability
high value
large variability
low value
high field audit value
high lab audit value
high lab audit value
(Continued)
181
-------
Table D-1. Continued
Data
Parameter set'
CEC CL S/H
FD
PD
AS
CEC OAC FD
PD
AS
AC_BACL FD
PD
AS
CA CL2 FD
PD
AS
MG_CL2 PD
AS
K_CL2 S/H
FD
PD
AS
NA_CL2 AS
FE_CL2 S/H
FD
AL_CL2 S/H
FD
PD
AS
Sampling
class/horizon
BNS /
TXW /
7WM/
TWM/
BXP /
BND /
BND /
TXW /
HST /
BMD/
TXW /
FLW /
TXW /
HST /
TXW /
BMK /
TXW /
HST /
FLW /
FLW /
HST /
TWO/
CFP /
BXP /
BXP /
CMS/
TXP /
TXW /
HST /
CMS/
TXW /
TXW /
TXW /
FLW /
TXW /
BXW/
Oe
Bt
C
c
Bx
B
B
B
Bw
Bw
A
Oe
0
Ap
A
Bw
A
C
Bw
Oe
A
C
Bw
Bw
C
A
Oe
B
B
Bw
Ap
BA
Oe
C
C
C
B
B
A
Oe
C
Ap
A
Oe
C
Bx
Oe
Bt
C
A
Bw
Bw
0
Batch/sample
30113-25,6
30127-7,34
30127-5,8
30130-05,4
30108-8,27
30112-8,24
30114-16,26
30106-25,24
30106-24.5
30119-2,13
30131-19,24
30132-14,17,31
30104-29,14
30119-2,8
30124-7.34
30119-2,13
30105-37
30117-5,32
30131-14,19
30119-8,35
30119-19
30119-20,33
30125-24,37
30107-6,1
30119-8,35
30131-14,3
30128-13,29
30128-7,24
30119-20.33
30132-26,39
30112-37,18
30115-39,6
30130-31
30122-,28
30128-13,29
30119-2,8
30132-26,39
30113-25,6
30120-34,32
30119-8,35
30122-19,36
30113-1,23
30131-6,13,31
Reason
large variability
large variability
high value
high value
large variability
low field audit value
large field audit variability
high lab audit value
large variability
large variability
high value
low value
low field audit value
large variability
large variability
large variability
large variability
high field audit value
high field audit value
large variability
large variability
low lab audit value
large field audit variability
high field audit value
large variability
low value
large variability
low lab audit value
high field audit value
large field audit variability
large variability
large variability
large variability
high value
high value
high lab audit value
low lab audit value
low lab audit value
large variability
large variability
large variability
large variability
large variability
large variability
large variability
high value
large variability
large variability
large variability
large variability
large variability
high lab audit value
large lab audit variability
(Continued)
182
-------
Table D-1. Continued
Data
Parameter set'
FE_PYP S/H
AS
AL_PYP S/H
AS
FE_AO FD
PD
AS
AL AO S/H
FD
PD
AS
SI_AO S/H
FD
PD
AS
FE CD S/H
FD
PD
AS
AL_CD S/H
FD
PD
AS
SO4_H2O S/H
FD
PD
AS
Sampling
class/horizon
BXP /
BXP /
BXP /
BXP /
TXW /
7XW /
CFP /
HST /
TXW /
TXW /
TXW /
FLW /
TXW /
BND /
FLW /
BMK /
FLW /
CFP /
TXP /
TWO/
TWO/
BMK /
TXW /
FLW /
BMD /
TWO/
CMS /
HST /
TPD /
BND /
TXP /
HST /
BND /
TXP /
HST /
A
Bw
Bw
Bw
Bx
C
Oe
Bw
Bt
C
Oe
Bt
C
C
Bw
Bw
Oe
Bt
C
Bt
Bw
Bw
B
BC
Bhs
C
C
Bw
Bw
B
C
Bt
Bt
Bw
Bhs
Oe
C
Ap
Bt
B
C
Oe
Bt
Bw
Btx
Oe
Bw
Btx
Oe
C
B
Bw
Bw
O
Batch/sample
30109-29,36
30125-24,37
30125-3,27
30126-18,23
30115-39,30
30132-26,39
30126-23,5
30106-14
30129-36
30105-20,38
30121-3,35
30126-18,23
30120-34,32
30126-23,5
30123-4,11
30125-24,37
30120-10,37
30120-34,32
30115-39,6
30121-3,35
30121-4,21
30124-13,35
30109-14,30
30109-30,38
30109-29,36
30120-<34,32
30104-14,31
30109-30.38
30110-2,38
30123-34,4
30125-29,1
30132-26,39
30123-4,11
30125-29,8
30132-26,34
30106-9
30114-16,26
30119-11,34
30125-24,37
30132-2,6,28
Reason
large variability *
high value
high field audit value
low field audit value
high value
high value
large variability
large lab audit variability
low value
large variability
large variability
low value
high field audit value
high lab audit value
large lab audit variability
large lab audit variability
large variability
low value
high value
low value
large variability
low field audit value
high lab audit value
large variability
large variability
large variability
large variability
high lab audit value
high field audit value
high lab audit value
large variability
large variability
low value
low lab audit value
large variability
large variability
large variability
large variability
low value
large lab audit variability
high value
large variability
large variability
large variability
high value
high value
large variability
high value
large variability
low lab audit value
low lab audit value
high lab audit value
low field audit value
large lab audit variability
(Continued)
183
-------
Table D-1. Continued
Data
Parameter set*
SO4 PO4 S/H
FD
PD
AS
SO4 0 S/H
AS
SO4 2 S/H
AS
SO4 4 S/H
FD
PD
AS
SO4_8 S/H
FD
AS
SO4J6 S/H
PD
AS
SO4_32 S/H
FD
PD
AS
C_TOT S/H
FD
PD
AS
N TOT FD
AS
Sampling
class/horizon
CMS /
TXW /
FLW /
BND /
BMK/
TXP /
CMS/
CMS/
CMS /
TXW /
FLW /
TXW /
FLW /
CMS/
/
TXW /
CMS /
/
BND /
TWM/
CMS/
/
TXW /
TWO/
FLW /
TXW /
CMS/
TXW/
BMD /
BMD/
TXW /
BXW/
HST /
TXW /
C
Bt
C
Bw
C
Btx
B
B
C
C
Bw
C
Bw
C
Bt
Bw
A
Bw
Bw
C
BC
Bt
Bw
C
BC
Bw
C
B
Bw
Bw
C
BC
Bt
Bt
C
Bt
C
Bw
B
B
C
A
Ap
A
Bw
Oe
B
A
Bw
Batch/sample
30113-25,6
30120-34,32
30106-24.5
30107-6.1
30125-1,26
30122-6.28
30124-4,21
30122-8
30119-20,33
30115-10,28
30113-25,6
30124-7,34
30119-8,35
30124-34,32
30105-32.34
30113-25,6
30109-29,36
30111-32.13
30127-7,34
30128-7.24
30111-14.39
30113-1,23
30126-18,23
30109-14,20
30120-34,38
30126-23,5
30122-31
30111-14,29
30110-29,39
30128-7.24
30104-29,14
30118-4.17
30119-2,8
30128-18,34
30131-3,14
30120-10,37
30119-2,8
30121-3,35
Reason
high value
large variability
large variability
large variability
large variability
large variability
low lab audit value
high field audit value
high value
high lab audit value
large field audit variability
high value
low field audit value
high value
large variability
low value
large variability
large variability
high field audit value
high value
large variability
large variability
low field audit value
high value
large variability
large variability
low value
tow field audit value
high lab audit value
high field audit value
high value
large variability
tow value
high value
large variability
low value
low field audit value
low lab audit value
large field audit variability
low field audit value
large variability
large variability *
large variability
large variability
large variability
large variability
large variability
high lab audit value
large variability
tow lab audit value
184
-------
Table D-1. Continued
Data Sampling
Parameter set' class/horizon Batch/sample
S TOT S/H CMS /
HST /
TXW /
AS
C
Oe
A
C 30107-8
B 30114-16,26
Bw 30113-1,23
Reason
large variability
large variability
large variability *
high field audit value
low lab audit value
low lab audit value
" S/H - sampling class/horizon groups of routine samples; FD - field duplicates; PD = preparation duplicates; AS audit
samples.
* Organic soil sample included.
185
-------
Appendix E
Precision Plots for Particle Size Fractions
Following are precision plots for the routine and QA data sets from the VCOS, COS, MS, FS,
VFS, COSI, and FSI parameters. Since these parameters did not have specifically established
DQOs, it was decided to place the routine data plots in the appendix and separate from the audit
sample/routine sample paired plots found Section 3 in the Results and Discussion of the report.
Supplemental information relating to these plots can be found in Section 3 under the parameter
group heading and in Appendices C and D.
186
-------
(a)
2.04
K
-*-1
^
c
o
1.5-
1.0-
).5-
o
-4 '
00
0.0-
0
vcos
Very Coarse Sand
Laboratory Audit Samples
A A
A
2 4
Mean (wt *)
a a a B A A A Bw
i
6
8
0»
Routine Samples
'6-
c
o
£4
Q
T3
L_
0
0-
0246
Mean (wt 7.)
+ *! Mineral Routine S/H Groups Field Dups.
8 10
Lab Audits (Within-Batch)
Preparation Dups. Lab Audits (BetweenBatch)
Note: Three mineral sampling class/horizon groups exceed plot boundaries.
Figure E-1. Range and frequency distribution of VCOS for (a) laboratory audit samples and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
187
-------
(a)
2.0^
1.5-
K
~5
c
O
g
'£ 1.0-
Q
"O
O
cO.Sj
o
to
0.0-
cos
Coarse Sand
Laboratory Audit Samples
Mean (wt %)
a a a B A A A Bw
8
10
(b)
Routine Samples
c
o
Q
T3
i_
O
T5
C
O
-*'
to
10-
8
6-
4.
2
0-
Mean (wt %)
+ Mineral Routine S/H Groups Field Dups.
0 4 8 12 16
Lab Audits (Within-Botch)
Preparation Dups. Lab Audits (Between-Batch)
Note: One mineral sampling class/horizon groups exceed plot boundaries
Figure E-2. Range and frequency distribution of COS for (a) laboratory audit samples and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
188
-------
(a)
MS
Medium Sand
Laboratory Audit Samples
K
-I'
^
C
O
3-
g
52J
o
1
o
-t-J
I/)
a a DD
6 8
Mean (wt ?)
a a a B A A A Bw
10
12
(b)
9-
c
o
a) 6
Q
o
T3
C
O
0-
Routine Samples
0 5 10 15
Mean (wt %)
+ + Mineral Routine S/H Groups Field Dups.
20 25
Lob Audits (Within-Batch)
Preparation Dups Lab Audits (BetweenBatch)
Note: Four mineral sampling class/horizon groups exceed plot boundaries.
Figure E-3. Range and frequency distribution of MS for. (a) laboratory audit samples and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
189
-------
(a)
2.CH
1.5H
c
o
6 1-0
Q
D
"20.5
o
0.0-
FS
Fine Sand
Laboratory Audit Samples
a
a
6810
Mean (wt %)
a a a B A A A Bw
12
r
14
(b)
15-
c
o
"o
'£ 10-1
Q
TJ
O
"c 5-
O
-t'
00
0-
Routine Samples
0 5 10 15
Mean (wt %)
+ + + Mineral Routine S/H Groups Field Dups
20 25
Lab Audits (Withm-Batch)
Preparation Dups. Lab Audits (BetweenBatch)
Note: Four mineral sampling class/horizon groups exceed plot boundaries
Figure E-4. Range and frequency distribution of FS for: (a) laboratory audit samples and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
190
-------
(a)
2.04
c
o
o
'« 1.0-1
Q
0.5-
D
CO
0.0-
VFS
Very Fine Sand
Laboratory Audit Samples
10 12 14
Mean (wt %)
D a D B A A A Bw
16
18
20
(b)
Routine Samples
c
o
0)
a
"O
1_
o
T5
C
_D
Lo
9-
6-
3-
0-
0 5 10 15
Mean (wt %)
+ + + Mineral Routine S/H Groups Field Dups.
20 25
Lab Audits (Withm-Batch)
Preparation Dups. Lab Audits (BetweenBatch)
Note: Two mineral sampling class/horizon groups exceed plot boundaries.
Figure E-5. Range and frequency distribution of VFS for (a) laboratory audit samples and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
-191
-------
(a)
20^
K
-*-i
^1'
c
o
"o
'I 10-
Q
C
O
00
o.
10
COSI
Coarse Silt
Laboratory Audit Samples
20 30
Mean (wt %)
a a a B A A A Bw
40
(b)
K
-4^
^
C
o
9-
o
'^ 6^
Q
T5
i_
O
C
_D
CO
0-
Routine Samples
10 15
Mean (wt %)
20
Mineral Routine S/H Groups Field Dups
Lab Audits (Within-Botch)
Preparation Dups. Lab Audits (BetweenBatch)
Note: Five mineral sampling class/horizon groups exceed plot boundaries.
Figure E-6. Range and frequency distribution of COSI for: (a) laboratory audit samples and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
192
-------
(a)
Fine Silt
Laboratory Audit Samples
c
o
§ 5\
°'r-
20
o a g oo
25 30
Mean (wt %)
a a o B ^ A A Bw
35
40
(b)
§ 10-1
5-
c
_0
(75
0-
Routine Samples
t +
15
25
Mean (wt
Mineral Routine S/H Groups Field Dups.
35 45
Lab Audits (Withm-Batch)
Preparation Dups Lob Audits (Between-Botch)
Note: Three mineral sampling class/horizon groups exceed plot boundaries.
Figure E-7. Range and frequency distribution of FSI for: (a) laboratory audit samples and (b) sampling
class/horizon routine sample data partitioned into concentration intervals of uniform variability
and their relation to pooled precision estimates.
193
-------
Appendix F
General Statistics for the Laboratory Audit Samples
Appendix F is a summary table of the soil chemistry and physical data for the laboratory audit
samples. Included are the data sorted by laboratory and by audit sample type for number of
samples (n), mean concentration, and standard deviation. Supplemental information relating to this
table can be found in the precision discussion and figures in Sections 2 and 3 of the report. Data
are presented for each of the 50 analytical parameters in the order described in Table 1-1 of Section
1 of the report.
194
-------
Tabls F-1. Summary Statistics of Analytical Parameters for Laboratory Audit Samples
Audit Sample*
Laboratory
1
2
4
1
2
4
n
10
8
8
10
8
8
B
Mean
5.01
5.25
5.25
46.73
49.81
44.16
Bw
SD n Mean SD
Air Dry Moisture Percent (MOIST)
0.19 8 2.57 0.12
0.31 10 2.35 0.20
0.22 10 2.53 0.19
Total Sand (SAND) 2.0 - 0.05 mm
1.31 8 23.98 0.87
3.34 10 28.47 1.32
2.80 10 22.48 3.21
C
n Mean
wt. %
9 0.12
9 0.12
9 0.14
wt. %
9 96.21
9 97.30
9 96.91
SD
0.04
0.12
0.05
0.98
0.71
1.45
O
n Mean SD
_t> _ _
6 7.56 0.63
_
Very Coarse Sand (VCOS) 2.0 1.0 mm wt. %
1
2
4
1
2
4
1
2
4
10
8
8
10
8
8
10
8
8
4.95
5.15
4.61
7.87
7.20
7.74
7.80
7.85
8.70
0.96 8 2.54 0.63
0.92 10 3.05 1.04
1.33 10 2.09 1.54
Coarse Sand (COS) 1.0 - 0.5 mm
0.61 8 4.14 0.39
0.98 10 3.92 0.35
0.80 10 3.37 0.74
Medium Sand (MS) 0.5 - 0.25 mm
0.31 8 3.46 0.16
1.09 10 3.65 0.24
1.77 10 3.44 0.53
9 5.62
9 5.06
9 6.30
wt. %
9 16.97
9 15.82
9 17.74
wt. %
9 33.13
9 33.04
9 3Z92
1.29
1.03
2.31
1.45
1.32
4.90
0.78
2.12
221
*»
mm
_ »-
Fine Sand (FS) 0.25 - 0.1 mm wt. %
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
10
8
8
10
8
8
10
8
8
10
8
8
10
8
8
11.57
12.59
10.89
14.52
17.04
12.20
49.87
45.90
52.99
25.53
20.92
27.19
24.34
24.97
25.80
0.69 8 3.69 0.22
0.87 10 4.33 0.28
1.89 10 4.20 0.51
Very Fine Sand (VFS) 0.1 - 0.05 mm
0.69 8 10.19 0.31
1.47 10 13.55 1.24
0.86 10 9.40 1.12
Total Silt (SILT) 0.05 - 0.002 mm
1.19 8 70.15 0.96
3.41 10 63.90 4.25
1.69 10 72.83 2.81
Coarse Silt (COSI) 0.02 0.005 mm
1.18 8 37.48 1.35
9.77 10 32.75 1.55
£10 10 37.60 1.82
Fine Silt (FSI) 0.02 - 0.005 mm
0.84 8 3Z67 1.50
9.11 10 31.15 3.41
2.62 10 3523 2.40
9 30.88
9 34.61
9 31.79
wt. %
9 9.66
9 8.78
9 8.20
wt. %
9 3.70
9 2.30
9 3.01
wt. %
9 3.46
9 1.89
9 2.61
wt. %
9 0.24
9 0.41
9 0.40
1.64
1.72
4.87
1.74
1.63
2.31
0.95
0.77
1.41
0.95
0.70
1.47
0.16
0.21
0.68
~m "
"" "
«*«
.. M -»
_
(Continued)
195
-------
Table F-1. Continued
Laboratory
1
2
4
n
10
a
8
B
Mean
3.40
4.29
2.85
SD
Total
0.47
1.12
1.31
Audit Sample'3
Bw C
n Mean SD n Mean
Clay (CLAY) < 0.002 mm wt. %
8 5.89 0.41 9 0.09
10 7.63 3.22 9 0.40
10 4.69 0.93 9 0.08
SD
0.09
0.30
0.08
0
n Mean SD
mm a
pH in Water (PHJH2O) pH units
1
2
4
10
8
8
4.92
4.91
4.85
0.06
0.03
0.05
8 5.13 0.09 9 5.30
10 5.09 0.07 9 5.47
10 5.07 0.08 9 5.18
0.11
0.07
0.08
6 4.06 0.03
pH in 0.002M Calcium Chloride (PH_002M) pH units
1
2
4
10
8
8
4.49
4.49
4.35
0.04
0.02
0.02
8 4.79 0.11 9 4.99
10 4.76 0.04 9 5.17
10 4.69 0.05 9 4.89
0.18
0.05
0.36
6 3.63 0.01
pH in 0.01M Calcium Chloride (PH_01M) pH units
1
2
4
1
2
4
10
8
8
10
8
8
4.32
4.25
4.28
0.36
0.36
0.35
0.06
0.05
0.05
Calcium in
0.03
0.03
0.03
8 4.55 0.04 9 4.75
10 4.55 0.04 9 4.79
10 4.60 0.04 9 4.72
Ammonium Chloride (CA_CL) meq/100g
8 0.26 0.03 9 0.10
10 0.26 0.01 9 0.06
10 0.27 0.01 9 0.06
0.07
0.05
0.07
0.04
0.01
0.01
6 3.33 0.02
6 11.39 0.51
Magnesium in Ammonium Chloride (MG_CL) meq/100g
1
2
4
1
2
4
1
2
4
10
8
8
10
8
8
10
8
8
0.09
0.08
0.09
0.08
0.09
0.09
0.06
0.06
0.07
0.02
0.01
0.01
Potassium
0.01
0.00
0.00
Sodium in
0.02
0.00
0.01
8 0.07 0.02 9 0.03
10 0.05 0.00 9 0.02
10 0.06 0.00 9 0.03
in Ammonium Chloride (K_CL) meq/100g
8 0.06 0.01 9 0.03
10 0.07 0.00 9 0.03
10 0.06 0.00 9 0.03
Ammonium Chloride (NA_CL) meq/100g
8 0.02 0.01 9 0.01
10 0.03 0.00 9 0.00
10 0.03 0.00 9 0.01
0.01
0.00
0.00
0.01
0.00
0.00
0.01
0.00
0.00
6 3.09 0.13
6 1.13 0.06
6 0.43 0.01
Aluminum in Ammonium Chloride (AL_CL) meq/100g
1
2
4
1
2
4
10
8
8
10
8
8
3.29
3.74
3.42
0.29
0.27
0.30
0.91
0.25
0.34
Calcium in
0.03
0.01
0.03
8 1.62 0.11 9 0.24
10 1.80 0.12 9 0.17
10 1.96 0.09 9 0.21
Ammonium Acetate (CA_OAC) meq/100g
8 0.25 0.01 9 0.10
10 0.26 0.05 9 0.06
10 0.24 0.01 9 0.05
0.14
0.02
0.03
0.03
0.02
0.01
6 5.90 0.13
6 8.30 0.28
(Continued)
196
-------
Table F-1. Continued
Laboratory
1
2
4
1
2
4
1
2
4
n
10
8
8
10
8
8
10
8
8
B
Mean
0.09
0.08
0.09
0.09
0.09
0.09
0.06
0.06
0.07
Audit Samole"
Bw C
SO n Mean SO n Mean SO
Magnesium in Ammonium Acetate (MGJDAC) meq/100g
0.02 8 0.05 0.01 9 0.04 0.01
0.01 10 0.06 0.02 9 0.03 0.01
0.01 10 0.05 0.01 9 0.02 0.00
Potassium in Ammonium Acetate (K_OAC) meq/100g
0.01 8 0.06 0.01 9 0.03 0.01
0.00 10 0.06 0.00 9 0.03 0.00
0.00 10 0.06 0.00 9 0.02 0.00
Sodium in Ammonium Acetate (NA_OAC) meq/100g
0.01 8 0.03 0.00 9 0.01 0.00
0.00 10 0.03 0.01 9 0.01 0.00
0.00 10 0.03 0.01 9 0.01 0.00
0
n Mean SO
6 2.63 0.11
6 1.12 0.03
6 0.45 0.01
Cation Exchange Capacity-Ammonium Chloride (CEC_CL) meq/100g
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
10
8
8
10
8
8
10
8
8
10
8
8
10
8
8
10
8
8
10
8
8
7.16
7.48
7.56
Cation
23.47
20.51
23.81
27.22
28.02
32.26
0.55
0.50
0.52
0.06
0.06
0.06
0.02
0.03
0.02
0.05
0.05
0.05
0.27 8 4.61 0.19 9 0.60 0.06
0.27 10 4.23 0.14 9 0.71 0.13
0.83 10 4.44 0.33 9 0.50 0.07
Exchange Capacity-Ammonium Acetate (CEC_OAC) meq/100g
1.45 8 11.96 1.61 9 1.07 0.09
0.69 10 9.89 0.26 9 0.98 0.23
2.25 10 11.30 0.63 9 0.87 0.08
Barium Chloride-TEA Acidity (AC.BACL) meq/100g
1.57 8 15.36 0.91 9 1.04 0.66
1.90 10 16.02 Z17 9 0.59 0.76
1.59 10 17.33 0.85 9 1.01 0.27
Calcium in Calcium Chloride (CA_CL2) meq/100g
0.03 8 0.59 0.06 9 0.76 0.06
0.02 10 0.59 0.03 9 0.76 0.03
0.01 10 0.63 0.01 9 0.76 0.01
Magnesium in Calcium Chloride (MG_CL2) meq/100g
0.01 8 0.04 0.00 9 0.02 0.01
0.00 10 0.03 0.01 9 0.02 0.01
0.00 10 0.04 0.00 9 0.02 0.00
Potassium in Calcium Chloride (K_CL2) meq/100g
0.00 8 0.01 0.00 9 0.01 0.00
0.00 10 0.01 0.00 9 0.01 0.00
0.00 10 0.01 0.00 9 0.01 0.00
Sodium in Calcium Chloride (NA_CL2) meq/100g
0.01 8 0.02 0.00 9 0.01 0.00
0.00 10 0.02 0.00 9 0.00 0.00
0.00 10 0.02 0.00 9 0.00 0.00
6 26.30 0.87
6 82.72 1.01
6 107.38 2.73
6 1.48 0.03
6 0.69 0.01
6 0.60 0.01
6 0.31 0.01
(Continued)
197
-------
Table F-1. Continued
Laboratory
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
n
10
8
8
10
8
8
10
8
8
10
8
8
10
8
8
10
8
8
10
8
8
10
8
8
10
8
8
10
8
8
B
Mean
0.00
0.00
0.01
0.06
0.05
0.06
0.65
0.62
0.61
0.89
0.81
0.81
0.96
1.03
0.96
1.39
1.37
1.45
0.27
0.40
0.31
1.38
1.46
1.46
1.06
1.14
1.16
13.40
13.00
13.13
Audit Sample*
Bw C
SO n Mean SD n Mean
Iron in Calcium Chloride (FE_CL2) meq/lOOg
0.00 8 0.00 0.00 9 0.00
0.00 10 0.00 0.00 9 0.00
0.00 10 0.00 0.00 9 0.00
Aluminum in Calcium Chloride (AL_CL2) meq/100g
0.02 8 0.01 0.01 9 0.01
0.01 10 0.00 0.00 9 0.00
0.00 10 0.01 0.00 9 0.01
Pyrophosphate-Extractable Iron (FE_PYP) wt. %
0.05 8 0.78 0.05 9 0.03
0.03 10 0.79 0.06 9 0.04
0.02 10 0.71 0.03 9 0.04
Pyrophosphate-Extractable Aluminum (AL_PYP) wt. %
0.07 8 0.53 0.06 9 0.07
0.04 10 0.57 0.04 9 0.05
0.04 10 0.51 0.02 9 0.06
Acid Oxalate-Extractable Ircn (FE_AO) wt. %
0.06 8 1.02 0.13 9 0.06
0.09 10 1.00 0.10 9 0.05
0.03 10 0.99 0.07 9 0.08
Acid Oxalate-Extractable Aluminum (AL_AO) wt. %
0.09 8 0.87 0.05 9 0.08
0.05 10 0.88 0.04 9 0.07
0.06 10 0.87 0.06 9 0.07
Acid Oxalate-Extractable Silicon (SI_AO) wt. %
0.03 8 0.16 0.02 9 0.02
0.05 10 0.26 0.05 9 0.03
0.02 10 0.20 0.02 9 0.03
Citrate Dithionite-Extractable Iron (FE_CD) wt. %
0.15 8 1.76 0.13 9 0.16
0.07 10 1.77 0.09 9 0.15
0.07 10 1.69 0.03 9 0.15
Citrate Dithionite-Extractable Aluminum (AL_CD) wt. %
0.11 8 0.61 0.04 9 0.06
0,13 10 0.67 0.04 9 0.05
0.08 10 0.62 0.01 9 0.06
Water-Extractable Sulfate (SO4_H20) mg S/kg
1.35 8 20.08 3.28 9 3.05
0.81 10 20.80 1.04 9 2.36
3.31 10 20.53 0.96 9 2.67
SD
0.00
0.00
0.00
0.01
0.00
0.00
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.02
0.06
0.02
0.02
0.01
0.01
0.01
0.01
0.03
0.05
0.01
0.01
0.01
0.01
0.51
1.20
0.53
0
n Mean SD
6 0.02 0.00
6 0.27 0.01
6 0.10 0.00
6 0.23 0.00
6 0.14 0.01
6 0.26 0.01
6 0.01 0.00
6 0.23 0.00
6 0.26 0.00
6 44.17 2.63
(Continued)
198
-------
Table F-1. Continued
Audit Samcle"
Laboratory
n
B
Mean
SD
Bw
n Mean SD
n
C
Mean
SD
o
n Mean SD
Phosphate-Extractable Sulfate (SO4_P04) mg S/kg
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
10
8
8
10
8
8
10
8
8
10
8
8
10
8
8
10
8
8
10
8
8
78.80
71.31
78.31
1.30
1.25
1.29
2.24
2.12
2.05
3.28
3.22
3.03
5.43
5.29
4.91
11.38
10.84
9.75
23.81
22.48
21.07
20.12
6.35
2.78
0 mg
0.09
0.08
0.10
2 mg
0.05
0.13
0.08
4 mg
0.11
0.10
0.11
8 mg
0.14
0.25
0.47
16 mg
0.24
0.45
0.55
32 mg
0.93
1.31
1.21
8 115.55 2.69
10 114.39 7.38
10 100.99 3.56
S/L Isotherm Parameter
8 2.06 0.11
10 2.14 0.12
10 2.10 0.07
S/L Isotherm Parameter
8 3.10 0.11
10 3.12 0.06
10 3.16 0.11
S/L Isotherm Parameter
8 4.29 0.12
10 4.24 0.11
10 4.07 0.15
S/L Isotherm Parameter
8 6.55 0.13
10 6.44 0.12
10 6.39 0.37
S/L Isotherm Parameter
8 12.42 0.48
10 11.99 0.19
10 11.70 0.68
S/L Isotherm Parameter
8 24.35 1.07
10 25.29 0.46
10 22.62 2.16
9
9
9
4.96
5.34
4.96
2.04
1.77
1.67
6 47.64 2.45
(SO4_0) mg S/L
9
9
9
(SO4_2)
9
9
9
(S04_4)
9
9
9
(S04_8)
9
9
9
(SO4J6)
9
9
9
0.53
0.46
0.42
mg S/L
2.51
2.37
2.37
mg S/L
4.42
4.40
4.22
mg S/L
8.39
8.21
8.19
mg S/L
16.37
16.20
15.64
0.13
0.09
0.11
0.11
0.15
0.15
0.09
0.20
0.13
0.26
0.14
0.18
0.27
0.37
0.19
: : :
: : :
_
_
" ^ "
(S04_32) mg S/L
9
9
9
31.88
32.56
31.05
0.51
0.91
0.40
Total Carbon (C_TOT) wt. %
1
2
4
10
8
8
4.57
4.60
4.70
0.22
0.20
0.43
8 1.46 0.04
10 1.44 0.04
10 1.49 0.03
9
9
9
0.10
0.11
0.10
0.01
0.02
0.01
6 40.11 0.34
Total Nitrogen (N_TOT) wt. %
1
2
4
1
2
4
10
8
8
10
8
8
0.217
0.220
0.212
0.032
0.032
0.031
0.01
0.01
0.01
0.00
0.00
0.00
8 0.115 0.00
10 0.112 0.01
10 0.116 0.00
Total Sulfur (S_TOT)
8 0.021 0.00
10 0.021 0.00
10 0.021 0.00
9
9
9
wt. %
9
9
9
0.005
0.005
0.006
0.002
0.001
0.001
0.00
0.00
0.00
0.00
0.00
0.00
6 1.69 0.03
6 0.19 0.01
Sample size (n), :v:ean, and standard deviation (SD) for B, Bw, C, and O audit samples.
° - not analyzed by laboratory.
199
-------
Appendix G
Range and Frequency Histograms of the Data Sets
This appendix consists of figures displaying histograms of the range and frequency
distribution of the routine samples (RS), the field duplicates (FD), the preparation duplicates (PD),
and the laboratory audit samples (AS). Supplemental information relating to these plots can be
found under "Representativeness" in Sections 2 and 3 of the report. Histograms are presented for
each of the 50 analytical parameters in the order described in Table 1-1 of Section 1 of the report.
200
-------
RS
1 n
r-r~, p
MOIST
Air Dry Soil Moisture
) R> D C F= r a i- a t i o> 1-1 O i
I i c; 01 t
-------
f r-\
a
SAND
Total Sand
F'D
i- I i c= a "
.-Illll
C.J
W t
Mineral Organic
Figure G-2. Range and frequency distribution for SAND.
B
Bw
C
O
-------
LJtir-io S a m p> I ^ s
vcos
Very Coarse Sand
) F=D CF=>
illi. ----
OO
F=~CZ>
W t
LJ p> I 1 c: a
llll.l. .
CO C>O
*«
Mineral LJ Organic
Figure G~3. Range and frequency distribution for VCOS.
B
Bw
c
o
-------
F5
rv>
Coarse Sand
R> D
a t.
Illla.
MandB
t;:| III
^ 1HUHJH-HL^ m
OCD
C J C 1
CJ C--J
W "t
Wt
AS
no
o
in
-i o -
oo
CT--4
< "t
"t
Mineral Q] Organic
Figure G-4. Range and frequency distribution for COS.
B
Bw
C
O
-------
MS
CO
CL3
F5 S C F? o LJ "t I ri
Medium Sand
F3 O C F3 t-
W
tlom Du^p>lfc;oi"t
-------
FS
Fine Sand
a r-r-> p> I
F3 r-<
00
c--vj c--J
w "t
o>
F~D
/\S
I i
o q
-i a -
~
cJ c"I r-'-o
Mineral [J Organic @ B 0 Bw Q C
O
Figure G-6. Range and frequency distribution for FS.
-------
CZl-
VFS
Very Fine Sand
F= D C
reparation D<_ip>licr;C3"t:
-------
SILT
Total Silt
F?
-t I t i
a r~r~i
CO
O->
a -
E O -
CC3
W
F^O
s o
-o a ~
CO
<-O
W "t ?^
Mineral |_J Organic ^ B
Bw
C
O
Figure G-8. Range and frequency distribution for SILT.
-------
Coarse Silt
^
<> I~T->
t^-> <> uro
^ c--g C--.J
Mineral |_| Organic H B tsl Bw Q C
O
Figure G-9. Range and frequency distribution for COSI.
-------
FSI
Fine Silt
i t: i
a
F= r-<
oo
<3-J
CZ3 = a
O--5
vv i:
ro
it S a m p> I es>
AS
CZ! 2 O
cx-)
00
SO:
"I O -
o
uro
uro
Wt ?>5
Mineral [J Organic H B
Bw C
O
Figure G-10. Range and frequency distribution for FSI.
-------
CLAY
Total Clay
<=> > i "t i r-i
an > p> I
G
p>II
-------
c
ca rm p> I
PHJH2O
pH in Water
R> D
e o
c/->
c^> -^ o
jo
Fo
F~D
fO II i_J n f t:
I i
B o
"-TO
n f t::
Mineral LJ Organic
B
Bw
C
O
Figure G-12. Range and frequency distribution for PH_H20.
-------
PHOO2M
pH in O.OO2M Calcium Chloride
r-r-i f=> I
-------
PH_O1 M
pH in O.O1M Calcium Chloride
F§:
o-o
cu
r~r~i
t F CD n d ii
c^~>
* c=s
t-5
F"D
s o
CT3 -* O -
<^~)
oo
a_5
tT3 -* O -
I "t:
Minera I
Organic ^ B
Bw
C
O
Figure G-14. Range and frequency distribution for PH_01M.
-------
CA CL
Calcium in Ammonium Chloride
a m p> I & s)
s o
-I O
m & c|
ro
at
"I O O
O O
H Mineral |_| Organic
Figure G-15. Range and frequency distribution for CA_CL
B
Bw
C
O
-------
MG_CL
Magnesium in Ammonium Chloride
oo
ro
5)
CZ3
00
r-r-i « cq / 1 O O eg
Minera
Organic
B
Bw
C
OOg
O
Figure G-16. Range and frequency distribution for MG_CL
-------
K_CL
Potassium in Ammonium Chloride
. ir-i-s S a m p> I & sO F3 D CF
D i_i p> I i c^ a t <
ro
F"i l
-^ f^-
B
Bw ^ C
O
-------
_CL
Sodium in Ammonium Chloride
O
a m p> I
L_)pDlic:czit
e o
ro
00
Mineral |_J Organic
B
e <^» OO
BW E2 c
o
Figure G-18. Range and frequency distribution for NA_CL
-------
AL_CL
Aluminum in Ammonium Chloride
CvO
t-«
BB»^__
iii
r~r~i
ro
(0
O
^ 1 O O eg
f't S3 o r-i i p> I «
s o
a o :
1 O -
Mineral | | Organic
B
Bw
C
O
O
Figure G-19. Range and frequency distribution for AL_CL
-------
a
oo
CAJDAC
Calcium in Ammonium Acetate
p> I «s s ) F= D C F3 t-«s |0
-i a o
D
OO
-------
MG_OAC
Magnesium in Ammonium Acetate
RS
-1OO
e o
i-o p> a r a 11 o
P> I I O d t. & :
oo
tZ3
fc-5
ed It S a m p> I
-i o o
CO
O-J
e o -
e o -
oo
1 O O
Mineral | | Organic
I )
B
I 1
Bw
C
O
Figure G-21. Range and frequency distribution for MG_OAC.
-------
K_OAC
Potassium in Ammonium Acetate
L_J C3 I 1 '
6 O
AS
6 O
S O -
-------
NAJDAC
Sodium in Ammonium Acetate
e S a
2
' G>
c~s
S3
oo
R-D C FT
B O
D LJ p> I I <= a -
_^
C=3 3 a r
OO
CT
6-S
e o
= a
+ a
3 Q
2 1=1
-1 O
O
C --- J
c r~>
1 O O
Mineral | | Organic
B
Bw
C
O
Figure G-23. Range and frequency distribution for NAJDAC.
-------
CEC_CL
CEC in Ammonium Chloride
F? S F? o LJ t: f r~i e
i- ca "t f I I c=
r~r~\
C3 = CD :
Mineral |_J Organic
B
Bw
r-n
C
O O
O
Figure G-24. Range and frequency distribution for CEC_CL
-------
CECJDAC
CEC in Ammonium Acetate
F? S
sa
I
r~i
Q ;
C^~J C--J
r-r-i I
s Q
3 O
CI
Mineral LJ Organic ^ B
Bw
C
O
Figure G-25. Range and frequency distribution for CEC_OAC.
-------
AC BACL
oo
Acidity in Barium ChlorideTriethanelamine
m
p> I i c: ca ^
t<
B a
s o
'-^ o -
3 a
s o
-I O
o
J38&L
*~ro
r-r-i«s«q
uro
c~-J
Minera
Organic ^ B
Bw
C
O O eg
O
Figure G-26. Range and frequency distribution for AC_BACL.
-------
F?
s o
F5 co i_i -t r r-i
; en ni pr>
CA_CL2
Calcium in Calcium Chloride
l F=D C F^r-esc
r-o ti o> 11
c^
CD
-» a -
e o
CO S O
i
-------
CL2
Magnesium in Calcium Chloride
FRS
e o
CO B 0
CL5
OO
' C3
fe-5
F5 cs LJ t f i-i
ca m (o
r~n I 1 o a t:
-------
K CL2
Potassium in Calcium Chloride
a
oo
jg so
tCD OO
C-vJ -«=J- tCD
1 O O 01
Mineral J Organic
B
Bw
C
O
Figure G-29. Range and frequency distribution for K_CL2.
-------
Sodium in Calcium Chloride
F§ cz> i_i t f rt
ea i r~i
s o
CO
CO
B O
CO
CL>
oo
' CD
= O
-4 Q -
3 O
2 O
i a
r~r-i ^ cq
r~i i es
-------
FE_CL2
Iron in Calcium Chloride
F?
F> D
p> ci r a 1 1 o> n D LJ p> I i o a t
a m p I
S3
-*» a -
Mineral | | Organic
B
Bw
r-r-i
C
1 O O
O
Figure G-31. Range and frequency distribution for FE_CL2.
-------
F? <=> <
AL_CL2
Alumium in Calcium Chloride
p> I e =O F3 D C F
"t 1 o>
I 1 s a
-» o
3 a -
E O -
-i o -
o
7 O
6 O
-4 O
3 O
2 O -
-III.
O O
CF-f
r~r~i o cq /^ 1 O O
cd 11 ^ a iin pj I
IT
«s a -
s o -
-4 O -
S O -
S3 =50 H
00
4 O
3 O -
20-
1 O -
o
C---J
CZ5 G33
OO-3
1 O O
Mineral | | Organic
B
Bw
c
O
Figure G-32. Range and frequency distribution for AL_CL2.
-------
FE_PYP
Iron in Pyrophosphate
re p>a ra t Fc> r~i
p> I i G ca -t & © ~)
<3J>
\A/ t
F~D
r-r-i prs I
W "t
Mineral | | Organic gsj B M Bw (2 C
O
Figure G-33. Range and frequency distribution for FE_PYP.
-------
AL_PYP
Aluminum in Pyrophosphate
t F
a
B O
CO
OJ
\A/ t
a r~r~i p> I & &
CO
q_>
C3
C3
t~T
B O
3 O
2 O
-I O J
VAX "t
Mineral |_j Organic
B
Bw
C
O
Figure G-34. Range and frequency distribution for AL_PYP.
-------
FE_AO
Iron in Acid Oxalate
F?
CO
q-s
<=>
oo
o>
cs~>
CD
00
O
o
o
W "t 5=6
Mineral | | Organic
B
Bw
C
o
Figure G-35. Range and frequency distribution for FE_AO.
-------
AL_AO
Aluminum in Acid Oxalate
F?
o
1. 1 r-i
a I o s
cj~>
OJ>
vvt
D LJ p> I i c:
~? o
s o
s a
3 o
s o
i a
a
CZ3 CZ3
CTxJ --sj- CO OO
CO
Mineral [J Organic § B
Bw
C
O
Figure G-36. Range and frequency distribution for AL_AO.
-------
SI_AO
Silicon in Acid Oxalate
F5 o i_i t i n
a
1 I i I 1 c ca "t-s
CO
CL>
CxO
C3
3 O -
-4 O '
3 O '
2 O -
i a
OO
| Mineral O Organic
Figure G-37. Range and frequency distribution for SI_AO.
QBw
C
a
-------
FE_CD
Iron in Citrate Dithionite
F5
s a
<=> i_i t i t-i
a r~n p> I
O 1
F'D I <=
CO
CL>
CD
Mineral | | Organic
3
Bw
C
O
Figure G-38. Range and frequency distribution for FE_CD.
-------
AL_CD
Aluminum in Citrate Dithionite
c
a r~r~>
Wt
XX S
I
o -
w t
W "t
Mineral | | Organic
B
Bw
C
O
Figure G-39. Range and frequency distribution for AL CD.
-------
SO4JH2O
WaterExtractable Sulfate
CO
O-J
F? S (T F? o i_j ~t f
-1 a -
BBaM
no
F"i
D >_J (=> I I c= C3 -
p> I
s a
3 O
n
3 O -
<> uro
CT--J r*~>
r-r-1
Mineral |_J Organic
B
Bw
C
O
Figure G-40. Range and frequency distribution for SO4JH2O
-------
SO4_PO4
Phosphate Extractable Sulfate
~t T
a
*-«:
e o
oo
-------
SO4_0
Sulfate Zero Isotherm
a r-n p> I
p>
D(_J p> I i -e
m
Minera
Organic
i \
B
i ^
Bw
C
O
Figure G-42. Range and frequency distribution for SO4_0.
-------
SO4 2
a tii p> | e s
Sulfate Two Isotherm
} F* D CR'i-epacir-a-tfor-i Di_ip>li
p> I f
>
»-«
Mineral | | Organic ^ B KJ Bw Q C
CO
a_>
CP E a
CX-J
i-r-s
r-t-i
>
cj->
cu
c^ sa
SS:
UiT^l < > UT~3 C 3
--
-------
-
SO4 4
F? o i_j-t F r->
Sulfate Four Isotherm
rr-i p I o s ^> F3 D <^ F5 (-& p>
«a-a
Organic ^ B N Bw
rri p> I
CO
CL>
oo
^S
C
O
-------
SO4_8
Sulfate Eight Isotherm
*-£-> <=> i-f-3
oo ocS
Ul
D
-«* O 7
i o n
oo
C=3 2 O
<^O
r-r-i p> I
oo
^s
oc?
r-n
L_
Mineral | | Organic H B H Bw L*A C
O
Figure G-45. Range and frequency distribution for SO4_8.
-------
SO4_16
Sulfate Sixteen Isotherm
F?S
-* a
<=> L_J t F
a>
I i
Mineral [_| Organic
OO
-1 o -
r-n p> I & ss
-* o
rr~i
B
Bw
C
O
Figure G-46. Range and frequency distribution for SO4_16.
-------
SO4_32
Sulfate Thirty Two Isotherm
.
f-i In AolO
F?:
Mineral | | Organic
Sl_
S III oon 1 r>
LJ p> I 1
.-.ill
si
S U f o o n 1 n ?To T
-------
C_TOT
Total Carbon
F?
co s a
F? co i _ i ~t f r~» e S3 a r~r~> p> I
e a
w t
i
W t
a r-r-i ps I
s a -
ao
CO OC3
w "t
Mineral
Organic
B
Bw
C
O
Figure G-48. Range and frequency distribution for CJTOT.
-------
TOT
Total Nitrogen
F?S
a a
=5 O
q_> -4 o -
ca r-Y-i p> I o s
CL> -* O - IHH
s*°-mm
II
j ^ , I I
es o
p> I I o
o :
JBHL^ 1
w t
wt
8
(O
I i c=. a -
s a
5 o
oo
^5 =°
W t
Mineral |_J Organic
B
Bw
Figure G-49. Range and frequency distribution for N_TOT.
-------
S_TOT
Total Sulfur
F?S C F?<=> L_i-t i
s a
t-e
w
a r-r-i
a
S O -;
CO 00
Mineral | | Organic
B
Bw
C
O
Figure G-5Q. Range and frequency distribution for S_TOT.
-------
Appendix H
Technical Systems Audit Reports
This appendix contains copies of the reports written by the QA staff audit teams that
performed technical systems audits at the analytical laboratories. The reports contain information
on laboratory performance, discrepancies identified in the audits, and resolutions to the issues
raised. Laboratories are identified only by number in order to protect their anonymity.
251
-------
AUDIT REPORT FOR LABORATORY 1
Date of Audit; November 17, 1988
Participants: B. Schumacher and R. Slagle - Lockheed-ESC
The laboratory system audit went extremely well due primarily to the large number of analyses
that had been completed by the lab and entered into the LEVIS data entry system and due to a
computer program written by R. Slagle which allowed for the rapid determination of data that were
outsida allowed windows for accuracy and precision. The laboratory personnel were extremely
helpful during the data analysis and laboratory evaluation phases of the system audit. All concerns
expressed during the pre-award audits had been properly addressed and corrected by the
laboratory. Several areas of concerns, problems, and comments were encountered and will be
discussed separately relating first the concern and then the potential solutions to the problem.
The first major problem encountered was due to the change in methods for determining CEC
and exchangeable cations as extracted by NH4OAc and NH4CI. The lab reported and showed that
the method using a polyester pre-plug and the Acrodisc in-line filter resulted in extremely slow
sample processing or a complete blockage of extraction solution through the soil resulting in an
incomplete extraction and thus reanalysis of the sample. The samples in which a solution blockage
occurred were not restricted to the high clay soils as was originally suspected. The lab requested
a resolution of the problem before November 24, 1988 so that the first batch could be competed
by the end of the month.
The second major problem encountered was with the BaCI2-TEA extractable acidity. The
reagent and calibration blanks were well below the expected limit as stated in the SOW. Further
analyses of the routine data resulted in almost consistent values well below the accuracy windows
set for the known audit materials as well as for the lab known QCAS audit sample. Several
possible solutions were discussed and are as follow:
1) the original extraction solution was improperly made,
2) the original extraction solution had been exposed to CO2 resulting
in lower than expected buffering capacities (i.e. Ascarite column
was not working), and
3) theoretical value of 0.808 meq stated in SOW was incorrect.
Reanalysis of the first batch has been requested due to a failure to meet QA/QC requirements. The
laboratory will prepare a new batch of extracting solution to use and will report the quantity of acid
required to bring the solution pH to 8.2 in order to determine if the theoretical value is practical.
The laboratory expressed concern that the allowable ranges for the SO4 isotherm spike
solutions were too tight and that to obtain the required values was a hit or miss process even
using the same standards. Discussions with P. Shaffer (NSI Technology Services Corp., Inc.,
Corvallis, OR) were set to be held to check the possibility of loosening the required window with
the results being told to the laboratory at a later date.
The laboratory also expressed concern about the influence of method changes on the
resultant data for the 0.002M CaCI2 extractions. Their concern was that due to a much greater
contact time between soil and solution that the established accuracy windows would not be met
as was the case for the QCAS sample. The problem was to be discussed with M. Johnson and
P. Shaffer (NSI Technologies, Inc., Corvallis, OR) and answers provided to the lab at a later date.
252
-------
The lab further expressed concern due to a lack of soil received from the preparatory lab for
certain samples. The laboratory manager stated that some samples received only contained about
half a bottle of soil.
Several moisture contents were flagged due to weights out of the allowable limits.
Reanalyses of the flagged samples was requested.
The water extractable SO4 replicate value was out of the allowed window (%RSD > 10%).
The laboratory was informed that another replicate sample was to be chosen and reanalyzed.
A potential problem was identified for the acid-oxalate extractable Fe and pyrophosphate
extractable Al. Three known audit samples were out of the allowable range for both parameters
(Fe to the high side and Al to the low side). Discussions are set to be held with L Blume (EPA -
Las Vegas) and M. Papp (Lockheed-ESC) to determine if reanalysis in warranted.
QC charting was discussed with the laboratory with the final result being that the laboratory
must produce charts for the QCCS, DL-QCCS, and the QCAS samples.
A list of other concerns was provided to us by the lab and is attached. The concerns
primarily relate to changes in methodology among the DDRP soil surveys and whether the changes
will influence the accuracy windows that the laboratory must meet. Also attached is a summary
listing of the number of samples that were out of allowed accuracy ranges, the parameter, whether
the sample value was high or low, and how much the value was out of the accuracy window on
the high or low end.
A list of the scheduling for the first three batches with their associated official receipt dates
(i.e. their start dates for the 60-day processing period) is as follows:
Batch Actual Receipt Date Official Receipt Date
30101 10/18/88 11/7/88
30104 11/1/88 11/17/88
30107 11/17/88 11/21/88
SUMMARY OF AUDIT WINDOW DATA FOR BATCH 30101
Number of samples out Parameter High or Low How much
1 pH H,0 high 0.064
4 pH_01M high 0.99 to 0.15
8 pH_002M high all samples at high end of
window or over the limit
3 CEC_CL low 0.03 to 0.372
3 CA OAC high 0.015 to 0.028
1 CA_CL2 high 0.003
4 MG_CL2 high 0.001 to 0.009
7 AC_BACL low 2.157 to 11.465
3 FE AO high 0.006 to 0.031
3 AL_PYP low 0.027 to 0.603
1 C SAND high 1.023
1 F SAND low 2.07
1 M~SAND low 0.15
1 C_SILT high 1.161
Total number of audit samples checked - 11.
253
-------
AUDIT REPORT FOR LABORATORY 1
Date of Audit: January 25-26, 1989
Participants: M. Papp, G. Byers, B. Schumacher, R. Slagle - Lockheed-ESC,
L Blume - Environmental Protection Agency, Las Vegas, NV
Introduction
The purpose of this report is to document the major issues and concerns raised at the recent
audit of Laboratory 1 which took place on January 25-26, 1989. The goals of this audit were to: (1)
assure that the methods and operations as stated in the contracts were being implemented as
specified, (2) assess the quality of existing data, (3) follow-up on issues raised at past audits, (4)
discuss the process used to evaluate data packages, and (5) answer any questions concerning the
contracts.
Audit Report
An introduction was given by L Blume for the agenda and procedures to be followed during
the audit.
The major issues raised at the pre-award audit were reviewed for clarification and progress
by Laboratory 1. These issues included (1) unclean AA burner heads due to flame problems, (2) the
use of mixed standards for Fe, Al, and Si, (3) the use of a one-point calibration instead of a three-
point calibration for atomic absorption (AA), (4) the non-use of temperature calibration for pH, (5)
open-bin riffle splitter, (6) and use of welder's grade acetylene for AA instead of commercial grade
or better. These issues were discussed and responses indicated that Laboratory 1 had resolved
them. Other issues such as expansion of facilities at the laboratory, chain-of-custody procedures,
and availability of specified equipment for the DDRP analyses were reviewed.
A discussion followed regarding the first batches of soils that were forwarded for analysis.
Analytical issues were discussed after recent printouts of the data were made available by R.
Slagle and B. Schumacher.
Finally, a discussion followed on the Laboratory Entry and Verification Information System
(LEVIS) with R. Slagle. Topics such as cooperation between personnel at Laboratory 1 and EMSL-
LV, transfer of analytical data using LEVIS, the format and requirements of the tables used for the
analytical data, and some calculation and reporting unit concerns were reviewed.
The audit team held a private caucus to evaluate specific issues prior to more discussions
with Laboratory 1 personnel. Included topics were:
(1) QC Charting and QC samples, e.g., missing QCAS and DL-QCCS data, data identification on the
axes, the suggested use of the same stock solution DL-QCCS throughout the survey, documenta-
tion of the DL-QCCS changes, and missing IDL data.
(2) Data analysis, e.g., specific problems in batches 30101, 30104, 30107, and 30110 regarding
confirmation of data, reanalysis of Ca, Mg, SO4 and pH parameters in various extractants.
A laboratory tour was conducted for the audit team. Specific issues were discussed
regarding instrumentation by inductively coupled plasma (ICP) and AA analyses. One specific
problem was resolved. Standards and samples were run twice through the ICP on similar extracts
for which standards were prepared separately. It was recommended that the standards be
combined to save analytical time and reduce errors. Possible contamination by using the parafilm
in the agitation step was discussed. The location of the laboratory gases and other safety issues,
standards, dilutions, integration times, wash sequences, etc., for the various instruments were also
254
-------
reviewed. The facilities for sample preparation and holding areas, sample analysis for chemical and
physical parameters, and sample storage repository were visited.
A debriefing to Laboratory 1 personnel was conducted the next morning on the data review
resulting from the audit team caucus. The issues were thoroughly discussed. More specifically,
problems with the cations in acetate and chloride, the sulfate in phosphate spike solution specified
was too tight, sulfate in water was out of 10% RSD window, the moisture content flags, iron and
aluminum in acid oxalate and pyrophosphate were out of the accuracy windows, the acidity in
barium chloride-TEA results were too low. Also, the pre-plug plus disc restricted the filtering and
modifications would be forthcoming. An insufficient quantity of soil was received for certain
samples and this was to be rectified. The change in the procedure for the calcium chloride resulted
in not able to meet the accuracy windows. The QC charts have to be provided to EMSL-LV.
A presentation was made by the audit team on the reanalysis template by B. Schumacher
and G. Byers and on the soil chemistry relationships, and control charts on blind samples across
all laboratories by M. Papp. This was presented for the benefit and information of Laboratory 1
personnel attending the audit.
Closing remarks and a summary were given by L. Blume.
Agenda for the systems audit for Laboratory 1
1/25/89
2:00-3:00 pm
3:00-4:30 pm
7:00- pm
1/26/89
9:00-10:30 am
Introduction and discussions
Audit team caucus to evaluate the following requested information: control charts,
OCAS, QCCS, DL-QCCS, detection limit raw data, batch 30107 routine data
Laboratory Tour
Debriefing to laboratory personnel on data review findings
- -^» _, . . v vv. .v. ,. ,«£ »** i%«fc*wi MtWI J fj^t <0 WI II IWI Wl I UQL0 I » VI9 W I II IkJII
10:30-12:00 am Presentation to Lab personnel on: reanalysis template, soil chemistry relationships,
control charts of blind samples across all laboratories, true value QCCS for particle
size analysis
12:00-1:15 pm Lunch
1:30-3:30 pm Discussion and closing remarks
255
-------
AUDIT REPORT FOR LABORATORY 1
Date of Audit: May 4, 1989
Participants: B. Schumacher and M. Papp - Lockheed-ESC
L Blume - United States Environmental Protection Agency
Report:
The audit was used mainly to discuss Laboratory 1 billing of miscellaneous expenses and
to back up and retrieve the computer hardware and software provided to the laboratory for
the endurance of the contract.
Billing:
Discussions were held on the equitable billing of certain items by Laboratory 1. The
laboratory did not account for items modified from the original contract that EPA felt made
the analysis of a parameter less difficult, thereby reducing the cost of such analysis. The
parties decided that Laboratory 1 deduct any of their miscellaneous requests from the billing
which had not already been approved and paid.
System Backup:
B. Schumacher received the remaining of official copies of the form 500's. Batch 30125 was
corrected for an invalid submission. The batch was then accepted.
B. Schumacher and the laboratory manager then proceeded to back up the entire database.
Floppies of the system were created and all hardware was boxed for shipping to Lockheed-
ESC.
Since all batches for Laboratory 1 had been accepted, no formal laboratory tour or data
review was necessary.
Sample Shipment:
Two batches, 30104 and 30125 were prepared for shipment to EMSL-LV. These batches will
be sent to Laboratory 3 for reanalysis of certain Ca and Mg parameters. The remaining
batches at Laboratory 1 will be sent to EMSL-LV during the week of 5/8.
256
-------
AUDIT REPORT FOR LABORATORY 2
Date of Audit: November 15, 1988
Participants: L Blume - United States Environmental Protection Agency
B. Schumacher, R. Slagle - Lockheed-ESC
Report:
An on-site audit of Laboratory 2 was conducted on November 15, 1988 by L Blume, B.
Schumacher, and R. Slagle. The purpose of this audit conducted during the first month of analyses
was for the following reasons:
1. Verify that all issues of concern identified during the pre-award were corrected.
2. Discuss, identify, and possibly resolve any methods inconsistencies, data analysis errors,
etc.
3. Resolve any issues pertinent to LEVIS. Test modem transfer of data to EMSL-LV.
4. Discuss issues pertaining to sample scheduling and contract clarification.
1. Pre-Award Issues
a. Stock solutions working standards
They had a separate table for all DDRP standards. Also had new solutions made that
were dated with shelf life and receipt times. New standards will need to be made.
b. Eppendorf automated pipetters needed calibration
The pipettes were not properly calibrated. A recalibration was assured.
c. Instrument logs
Observed they were updated more frequently.
d. Magnesium sulfate reagent
Magnesium sulfate was old (1986) and not desiccated. A new source of MgS04 reagent
was on order and will be used with the first batch.
e. Transport of soils to WAL
WAL will analyze all 0.002 M CaCI2 analytes except K, and the extractable FE and AL
samples. In addition, tracking and custody forms would be used for transport.
f. Particle size analysis (PSA)
PSA will be done in a temperature controlled room and temperature will be monitored in
a blank (DDI water) cylinder.
g. WAL facility
New benchtops will be added, thereby separating working areas and overall cleanliness
was improved.
257
-------
2. Methods Issues and Data Evaluation
Only detection limit data and pH were available due to the fact the lab was just beginning
analyses.
a. CEC filtration problem
Problem - Clogging of samples prevented extraction with NH4C1 in samples 2.13,17,21
(red audit Bt2)
- Further clogging noted during ethanol rinse in sample 9,10,11,12
Notes - Clog occurred in preplug for NH4CI and possibly ethanol
- Removal of Acrodisc in-line filter allowed for easier ethanol rinse although
some breakthrough occurred; no problem-would have been same loss as
Acrodisc
- Two successful runs of Bt2 on samples 4 and 8 using method as stated
in SOW.
Solution- 1. Remove Acrodisc for ethanol rinse
2. Weight reduction to 2.5 g
3.0.3 g of washed cellulose plug with appropriate blank data
The lab is allowed to use option 3 for the 6 red soils. It was suggested to "wet" preplug
with several drops of extracting agent to prevent pressure head created between the polyester and
soils (Cliff Jones - ERC).
Change CEC method page 3-52 step 2: change 45 ml - 35 ml to allow for complete
extract with NH4CI or NH
-------
- A spike method problem was identified through LEVIS. The laboratory was
diluting first, then spiking, instead of the opposite as required. They will correct
this problem.
- A system update was made to the report files and a test of LEVIS communication
to EMSL-LV was successfully made.
- A 10,000 ppm spike solution was used which LEVIS could not accept because only
4 digits were allowed. A note was made in the comment file identifying the true
concentration (i.e., report 1000 but add 10-fold dilution in appropriate column).
4. Sampling Scheduling
Sample shipment was discussed with the laboratory owner stating that EMSL planed on
having all batches shipped to the laboratories by March 1, 1989. In addition discussions were held
on the official start date of the batches the lab had receive, which are as follows:
Actual Official
Receipt Start
Batch Date Date
1 10/18/88 11/11/88
2 11/07/88 11/18/88
3 11/25/88 11/25/88
259
-------
AUDIT REPORT FOR LABORATORY 2
Date of Audit: January 24-25, 1989
Participants: M. Papp, G. Byers, B. Schumacher, R. Slagle - Lockheed-ESC
L. Blume - Environmental Protection Agency
Introduction:
The purpose of this report is to document the issues raised on a recent audit of Laboratory
2 which took place on January 24-25, 1989. The purpose of this audit was to:
- assure methods and operations as stated in each of the contracts is being implemented as
specified;
- to assess the quality of existing data;
- to follow-up on issues raised at past audits;
- to discuss the process used to evaluate data packages;
- to answer any miscellaneous questions concerning these contracts.
The tentative agenda that was followed for this audit was:
8:00-9:00 am Introductory Discussions
9:00-10:30 am Audit team caucus to evaluate the following requested information
-Control Charts
QCAS
QCCS
DL-QCCS
- Detection Limit Raw Data
- Total S,C, and N data for batches 1-3
- Batch 30102, 30106, 30109 routine data
10:30-11:00 am Debriefing to laboratory personnel on data review findings
11:00-12:00 am Presentation to Lab On:
- Reanalysis Template
- Soil-chemistry Relationships
- Control charts of Blind Samples Across all Laboratories
- True value QCCS for particle size analysis
12:00-1:15 pm Lunch
1:30-2:30 pm Visit facility
2:45-5:00 pm Laboratory tour
5:00-7:00 pm Debriefing of audit team findings, and discussion.
The major issues and resolutions were the following:
a. Control Charts
The requested control charts were reviewed. In general, the charts were acceptable but were
lacking data from batch's 30106 and 30109 which at that time were not formally submitted. And
select parameters such as pH and PSA were not charted for the DL-QCCS and the QCAS. We
also discussed and emphasized use of these charts by the lab manager and accessibility to
technicians. This discussion was concluded by requesting formal submission of these charts to
EMSL-LV after completion of the first three batches of data.
260
-------
b. PSA
The silt data was consistently below the accuracy windows and the VFS was usually high. This
occurrence was observed in all data submitted to date for this contract and was also observed in
data from the interlaboratory comparison study. We recommended the following corrective action
steps in priority order:
1. Place a 300 or 325 mesh sieve at the base of the sand fraction sieves. In this manner
remnant clay or silt size material would not contribute to the VFS fraction, and the resulting
silt fractions determined by difference should not be offset low.
2. Reanalysis of affected batches.
3. Run QCAS without filter candle.
4. Cross check quantitation method by using hydrometer for the QCAS sample.
5. Flow chart the individual steps of the method as interpreted by lab and compare this to
other laboratories.
It seems that our first recommendation may have solved the problem based on data
received after completion of the audit.
c. Moisture Content
The precision of the MC determinations observed in the first two batches of data was outside of
the acceptable range. We thoroughly reviewed the handling of the samples for this measurement.
It seems the problem may have been due to a new technician and due to sample handling. We
requested an improved desiccation method and more emphasis on sample handling which the
laboratory had instituted for the third batch of analyses. The problems were not evident after
review of the third batch of data, indicating potential resolution of the problem.
d. SO4-0, SO4-2
The select isotherm points showed accuracy problems primarily biased low for the first two
batches of samples. We discussed the possibility of faulty filters and/or standard preparation
techniques that may account for the problem. When we discussed this problem with the
technicians we observed a grade "A" pipette was not being used for the 2 ppm standard. Also we
were told that a new technician prepared only the SO4-2 and SO4-0 standards while the more
experienced technician prepared all others. We recommended corrective action of the identified
problems followed by reanalysis of the affected batches.
e. CEC Blanks
The mean reagent blank value for CEC_CL was above the required level for the first two batches
analyzed. More extensive washing procedures were applied to the pulp washing of the third batch
of samples which seems to have solved the problem after evaluation of preliminary data.
f. WAL facility
The WAL facility is responsible for all ICP analyses from this contract. Most of the problems
identified from the pre-award audit were rectified. The data generated by WAL has been in very
good shape with no reanalyses requested. The only problem identified was some older stock
standards from 1986 that were stored on a counter adjacent to the ICP. We were advised these
were not being used, although we did recommend getting rid of them.
g. Total S, C, and N
The laboratory had received three batches of samples at the time of our audit visit. We were able
to retrieve preliminary data from the first batch of samples, please note this was before the
261
-------
laboratory manager had reviewed this data. We identified the following problems:
- S_TOT spike recovery slightly above 110% (110.31%)
- reagent blank high (most likely an entry error)
- they ran one less QCCS sample
- relationships
C_TOT>N_TOT: one sample of 25 out
C~TOT/S_TOT: seven samples of 25 out
- a laboratory duplicate sample and matrix spike sample
was not run with the first batch for C and N
In general, the data looked quite good for the first pass. During the laboratory tour, we
emphasized that the matrix spike should be placed at the base of the combustion cell instead of
the top to identify incomplete combustion, if it exists.
Overall, the audit team felt operations at laboratory were going well and as expected. Our primary
concern was based on the isotherm and particle size analysis problems, although upon completion
of the audit, we felt we had made much progress in solving these problems.
262
-------
AUDIT REPORT FOR LABORATORY 2
Date of Audit: May 2-3, 1989
Participants: B. Schumacher and M. Papp - Lockheed-ESC
L Blume - United States Environmental Protection Agency
P. Shaffer - NSI Technology Services Corporation
Report: 5/2/89
received from laboratory formal submission of CMS batches 30119, 30120, and 30121
informed laboratory of acceptance of Batch 30116 for total CNS
discussions were held concerning C:N and C:S ratios
- three potential reasons for failure of ratios:
a) wrong expectations at start of project
b) C_TOT values too low
c) N_TOT and S_TOT values too high
- it was noted that C TOT values are in lower 1/3 of accuracy windows while N_TOT
and S_TOT contents"were around mean of the accuracy window therefore C_TOT
is suspected to be too low
- lab stated it suspected possible contamination of samples from pyrite which was
visually identified in some samples
- not likely from our point of view because:
a) not mine spoil areas
b) horizons are oxidized Bt, Btx, Bx, and C horizons therefore
most likely pyrite would already be weathered out from the
pedon
- lab stated there was a possibility of NJTOT contamination due to atmospheric
N in system although it should have been purged effectively
- discussed QC charts
- N_TOT tight for accuracy and values close to mean
- C_TOT for B and Bw horizons spread out on chart; biased low with mean;
C horizons biased low with mean
- S_TOT tight for accuracy for C horizon; Bw mean at high warning limit;
B horizon had mean biased low with 9 samples out of accuracy window
on the low side
- further discussions to be held tomorrow with L Blume
: laboratory lab tour was given for both contracts
5/3/89
M. Papp gave review of data for routine batches 30124 and 30127
lab stated that batch 30130 would probably be ready for our initial review on Monday
discussions were held concerning Fe and Al problem in _AO, _PYP, and _CD extracts
263
-------
- reviewed QC control charts
- _PYP extracts for LAP were running out of accuracy window high
- ~AO extracts for LAP were out of accuracy window high - even QCAS sample
- _CD extracts for LAP were running out of accuracy window high
- lab brought up issue of temperature increases and how that influences results
- lab noted up to a 10 or more degrees increase in temperature in shaking
room between start and finish of extraction procedure
- also have increase in ambient temperature between start of project and
now
- lab stated it had a change in the method of operations between start and current
analysis process for _CD extraction
- previously lab would make _CD solution, weigh samples, and then extracts
- now lab weighs samples, makes _CD solution, and extracts
lab noted increased Fe contents with increased time of extraction
- lab stated it is running reanalysis of batch 30124 for Fe and Al with shaking room
door open to prevent or at least help eliminate the drastic temperature rise over
time
Problem Resolutions:
CNS
1) do a comparative laboratory study to check reproducibility of methods and results
a) Laboratory 3 will do a batch of samples for total C, N, and S following the
contractual methods used at Laboratory 2
b) P. Shaffer will find a lab to do total C, N, and S analyses
- preferably M. David for SJTOT and T. Strickland for N_TOT and C_TOT
- no particular method was" stated
c) Laboratory 2 will determine total C and N using coulometric and Kjeldahl methods,
respectively
- Laboratory 2 will send 1 aliquot of ground soil samples to Laboratory 4
and 2 aliquots to P. Shaffer for further distribution
- batches selected were batch 30114 and as an alternative 30119
2) batches with more than 6 soil chemistry relationships failing will not be called for
reanalysis at this time as agreed on by all parties present pending the results of the
comparative laboratory study
Fe and Al
1) will wait for results of reruns of batch 30124 and new results of batch 30130 to determine
if 'cooler" temperatures had an influence on results especially of the B horizon LAPs
2) lab will FAX data for batch 30130 on Friday - batch will be 6 samples short and possibly
a QCCS sample short due to misplacement of samples
264
-------
AUDIT REPORT FOR LABORATORY 3
Audit Date: November 9, 1988
Participants: L. Blume - United States Environmental Protection Agency
B. Schumacher, R. Slagle - Lockheed-ESC
C. Palmer - Environmental Research Center-LV
Introduction:
A laboratory audit was conducted on Wednesday November 9,1988 of Laboratory 3. Analytical
services are being provided by this laboratory to the Direct/Delayed Response Project in support
of the Mid-Appalachian Soil Survey. The objectives of the audit were to:
1) Evaluate progress since the pre-award audit;
2) Examine preliminary data;
3) Discuss the implementation of LEVIS (Laboratory Evaluation and Verification Information
System);
4) Discuss the use of quality control charts;
5) Expand the awareness of data user needs to the laboratory personnel.
Report
1) Progress since the pre-award audit
Approximately 50% of the parameters for the first batch of samples had been completed at the
time of this audit. The laboratory was experiencing problems with the heavier clay soils
clogging in the pre-plug during the analysis of cation exchange capacity and exchangeable cations.
A recommendation was provided of reducing the weight of sample for these soils down to 2.5
grams. The QA officer will provide a letter of approval to substitute the sample size from 5.00
grams down to 2.50 grams for individual samples experiencing clogging problems.
2) Examination of preliminary data
A review of the data indicated that several data entry errors were occurring. For example,
sample #3 for PH_002 M of the first batch had a pH reading of 4.43 on the raw data sheet but
was entered as 4.35. Sample #2 for CA_CL2 was recorded as 51.7 on the raw data sheet, but
had been entered as 51.4. It was noted that verification of the data entry process had not taken
place and that this would be addressed.
The QCCS sample for Ca in CA_CL2 was not in mid-calibration range as requested in the
protocols nor did the calibration standards span the range of concentrations in the soil extracts.
The laboratory staff indicated that the quality of the results would not be affected due to the
linearity of response of the ICP analysis for Ca. They noted that any inter-element interferences
are not correctable if high concentration standards are used a clear standard is needed. It was
felt that the instrument could correct for any inter-element problems with the present
standards.
The laboratory is to provide proof of this linearity up to 100 ppm for one batch of samples for
evaluation by the QA staff. All appropriate data will be included. After review of the results, a
decision will be made and the laboratory notified.
265
-------
A review of the QC data indicated that the QCAS was below the acceptable window for
FE_PYP. The IDL's for AL_CD and FE_CD had been determined on 11/2, 11/3, and 11,7 but were
recorded as being completed on 11/4.
A review of the QE sample data (blind samples) indicated an overall very good precision for the
laboratory. A few blind audit samples we found to be outside of previously determined
windows. An overall evaluation of the results will be undertaken once the batch is formally
submitted to see if these results warrant the reanalysis of the batch for any given parameter.
3) Implementation of LEVIS
The completed LEVIS has been completed and was delivered to the laboratory on Monday,
November 7. It was noted that a separate program is being developed for the organic soil batches.
It is structured the same as the LEVIS for mineral soils and will be installed when the first batch
of organic soils is sent.
4) Quality control charts
C. Palmer gave a presentation on control charting along with recommended charting procedures
for the laboratory. The charting of the QCAS was particularly encouraged as a means of reducing
between batch variability. Concern was expressed by the laboratory over the potentially large
amount of effort charting could require. It was noted that computer programs exist to assist with
charting tasks.
5) Meeting data user needs
The laboratory staff was provided with a short explanation of the purpose of the DDRP
Mid-Appalachian soil survey and the role of soil measurements in the final interpretations. Special
attention was given to the importance of high quality data to the overall success of the program.
Summary Comments and Recommendations:
The purpose of this audit was to make an early assessment of the performance of the
laboratory in the analysis of routine samples and identify and correct problems at an early stage.
Evaluation of results was limited to approximately 50% of the parameters from the first batch that
had been sent to the laboratory.
Specific problems that were identified included the need for checks to prevent data entry errors
and an evaluation of the standards in use for the analysis of CA_CL2. Additional consideration is
to be given to modifying the CEC and exchangeable cation procedures as modifications to the filter
pulp method have created clogging problems for certain soils.
It appears that LEVIS will be successfully implemented in this laboratory. The importance of
control charting was introduced to the laboratory as well as suggestions for implementing charting
for certain key quality control samples.
In general terms, the audit team was pleased at the progress of the laboratory with the new
soil samples, their cooperative attitude in working with the DDRP QA staff, and their commitment
to high quality data.
266
-------
AUDIT REPORT FOR LABORATORY 3
Date of Audit: March 12, 1989
Participants: B. Schumacher, G. Byers, R. Slagle, Lockheed-ESC
The audit schedule was to spend the major part of the day with the manager of the
Chemistry Division, and the supervisor of the Environmental Chemistry subdivision, discussing
issues of concern. The remainder of the time would be spent on an audit tour of the laboratory
and closing discussions.
The meeting commenced at 9:00 a.m. with an introduction by B. Schumacher of the
purposes and goals to be accomplished for the audit. G. Byers then gave an overview of the
various procedures used at Lockheed to assess the QA/QC requirements prior to officially accept-
ing the data. His presentation included a detailed discussion using overheads of the templates
used in assessing the data for acceptance and the procedures involved in reanalysis, a detailed
discussion of the sample design for the quality assurance samples used in the DDRP on a batch
basis, and the internal consistency procedures. B. Schumacher then used a batch of soil analytical
results (batch 30114) as an example for the detailed stepwise procedures used in evaluating soil
data on a batch basis.
Considerable discussion followed regarding problems and issues related to Laboratory 3
data and the specific analytical and reporting issues for data having been assigned minor and
major flags. Issues such as reanalysis of the parameter PHJH2O, assembling the windows for the
QCAS sample (Sample #15) for the organic soil batches, instrumentation issues and other details
were discussed. A copy of the handwritten report summarizing the reanalysis requests for batch
30114, as well as copies of the previous reports where flags occurred, were provided to the
laboratory manager. In addition, B. Schumacher received all initial IDL data for the Lockheed-ESC
data files and verification of the 100 ppm standard required by the laboratory for CACL2 (see
previous audit report).
G. Byers was escorted by the environmental chemistry supervisor on a laboratory tour, of
the facilities used for the ODRP soils. Detailed discussion took place regarding the various
procedures and instrumentation used by the laboratory in the soil analyses.
R. Slagle worked with the computer technician on the installation of a version of the
computer program for the organic soil analysis reporting system using the LEVIS data entry
system. Any problems involved with LEVIS were discussed. In addition, R. Slagle archived the
laboratories soils data bases, removed an expanded memory board, and updated LEVIS for the
mineral soils.
After lunch, B. Schumacher, R. Slagle and the laboratory manager joined G. Byers and the
laboratory supervisor and toured the soil sample entry room, soil sample preparation area, and soil
sample storage room in the basement. Discussions continued as the laboratory representatives
completed the tour of the soil analysis section and other facilities of interest. The group then met
for a final summary of the audit.
Overall, the operations and soil analysis system used by laboratory 3 for the DDRP soils
were very satisfactory. The laboratory has had a history from the beginning of the program of
providing prompt and satisfactory data from the quality assurance aspect. The laboratory staff
was encouraged to keep up the very high caliber of data analysis.
267
-------
The laboratory procedures used in the soil analytical methods by the laboratory for the
DDRP soils were being followed. Protocols were being met. No problems were found in the
analytical phase of the soil analysis system.
The storage, handling, identification, and retrieval system for soils samples used by the
laboratory were satisfactory. The organized manner and overall cleanliness of the facilities were
also noted.
Overall, the performance of laboratory computer data processing facility in the use of the
LEVIS reporting system was excellent. No problems were found.
268
-------
Appendix I
Plots of Moving Averages for Laboratory Audit Samples
This appendix contains plots of three-point moving averages showing laboratory trends for
the MASS data collected from the B, Bw, and C horizon laboratory audit samples. Each plot
depicts the temporal distributions of the moving averages in relation to the boundaries of the
accuracy windows and their associated reference values. No conclusions have been made on the
significance of differences seen in the plots, or which of the apparent trends may be of importance
to the data users.
269
-------
o
ii
o>
en
o
o
C7>
C
O
2
MOIST
Air-Dry Moisture
BHorizon
Bw (Horizon
C Horizon
U - Upper Limit L = Lower Limit Ft = Reference Value
Figure 1-1. Moving average plota of laboratory trenda In B, Bw, and C horizon audit samples for MOIST.
270
-------
s
o
<
o>
s
o
5
O
cn
o
en
c
SAND
Total Sand
B Horizon
..L
Bw Horizon
.. L
C Horizon
L
U - Upper Limit I. = Lower Limit R = Reference Value
Figure 1-2. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for SAND.
271
-------
s
o
O
5
s
o
<0
CT
O
01
o
L
-------
s
o
o
i
o
0)
en
o
CT"
c
S
o
COS
Coarse Sand
B Horizon
\
.. L
Bw Horizon
C Horizon
U - Upper Limit L = Lower Limit R = Reference Value
Figure 1-4. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for COS.
273
-------
S
o
II
>
o
en
o
en
c
MS
Medium Sand
B Horizon
Bw Horizon
C Horizon
,-L
U - Upper Limit L = Lower Limit R - Reference Value
Figure 1-5. Moving average plota of laboratory trenda In B, Bw, and C horizon audit aamplea for MS.
274
-------
s
o
en
O
S
0)
en
o
en
c
o
S
o
o
2
FS
Fine Sand
B Horizon
i 1
Bw Horizon
C Horizon
L
U = Upper Limit L = Lower Limit R = Reference Value
Figure 1-6. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for FS.
275
-------
s
o
CN
V
cn
o
s
o
(N
HI
en
o
ID
01
O
VFS
Very Fine Sand
B Horizon
L
Bw Horizon
C Horizon
L
U - Upper Limit L = Lower Limit R = Reference Value
Figure 1-7. Moving average plota of laboratory trenda In B, Bw, and C horizon audit aamplea for VFS.
276
-------
3
O
o
ID
£
SILT
Total Silt
B Horizon
Bw Horizon
C Horizon
L
-2 R
._ L
L
U = Upper Limit L = Lower Limit R = Reference Value
Figure 1-8. Moving average plota of laboratory trenda In B, Bw, and C horizon audit aamplea for SILT.
277
-------
s
o
s
q
oj
0)
en
S
O
COSI
Coarse Silt
BHorizon
Bw Horizon
.. L
C Horizon
U - Upper Limit L = Lower Limit R = Reference Value
Figure 1-9. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for COSI.
278
-------
3
O
s
O
<
o>
o>
c
FSI
Fine Silt
B Horizon
._ L
Bw Horizon
_L
C Horizon
U - Upper Limit L = Lower Limit R = Reference Value
Figure MO. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for FSI.
279
-------
5
O
-------
X
Q.
o
II
>
CT>
O
X
a
en
o
01
c
a.
PH_H20
pH in water
B Horizon
L
Bw Horizon
C Horizon
L
U = Upper Limit L = Lower Limit R = Reference Value
Figure 1-12. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for PH_H20.
281
-------
.2
'c
13
Q.
O
II
>
c
3
Q.
O
II
O
cn
c
'
I
Q.
>
'-6
PH_002M
pt-i in 0.002M Calcium Chloride
B Horizon
Bw Horizon
C Horizon
_L
U - Upper Limit L = Lower Limit R « Reference Value
Figure 1-13. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for PH_002M.
282
-------
c
3
X
Q.
>
T>
<
x
a
o
2
c
3
X
a.
>
o
PH_01M
pH in 0.01M Calcium Chloride
B Horizon
L
Bw Horizon
..L
C Horizon
L
U - Upper Limit L = Lower Limit R - Reference Value
Figure 1-14. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for PH_01M.
283
-------
CA_CL
Exchangeable Calcium in Ammonium Chloride
B Horizon
o
o
cr
II
en
c
'
en
O
O
cr
i)
E
o
o
n
o>
en
o
O
o
cr
01
E
ID
q
cn
en
c
'
Bw Horizon
.. L
C Horizon
U = Upper Umit L = Lower Limit R = Reference Value
Figure 1-15. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for CA_Ct_
284
-------
MG_CL
Exchangeable Magnesium in Ammonium Chloride
B Horizon
o
o
o
o
II
0)
o
<
o
o
cr
-------
K_CL
Exchangeable Potassium in Ammonium Chloride
B Horizon
o
o
tr
0)
E
o
o
I!
o>
o
O
o
CT
4)
q
o
li
<
en
o
o
o
CT
V
E
.- L
Bw Horizon
-L
C Horizon
._ L
U - Upper Limit L = Lower Limit R =» Reference Value
Figure 1-17. Moving average plot* of laboratory trenda In B, Bw, and C horizon audit aamplea for K_Cl_
286
-------
o
o
o
o
o
o
cr
q
d>
o
o
o
o
II
<
en
NA_CL
Exchangeable SodiumTn Ammonium Chloride
B Horizon
Bw Horizon
.. L
C Horizon
U = Upper Limit L = Lower Limit R = Reference Value
Figure 1-18. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for NA_CI_
287
-------
AL.CL
Exchangeable Aluminum in Ammonium Chloride
B Horizon
o
o
cr
E
d
ii
o
\_
-------
CA_OAC
Exchangeable Calcium in Ammonium Acetate
BHorizon
O
o
cr
11
eg
O
en
o
o
5
CT
O
O
cr
v
CM
o
o
II
O
o
cr
-------
MGJ3AC
Exchangeable Magnesium in Ammonium Acetate
B Horizon
O
o
cr
<
.-L
Bw Horizon
..L
C Horizon
L
U - Upper Limit L = Lower Limit R = Reference Value
Figure 1-21. Moving average plot* of laboratory trend* In B, Bw, and C horizon audit sample* for MGjOAC.
290
-------
KJDAC
Exchangeable Potassium in Ammonium Acetate
B Horizon
O
O
cr
<
Bw Horizon
C Horizon
L
U = Upper Limit L = Lower Limit R = Reference Value
FIgura I-2Z Moving average plota of laboratory trenda In B, Bw, and C horizon audit aamplaa for K_OAC.
291
-------
o
o
cr
-------
CEC_CL
Cation Exchange Capacity by Ammonium Chloride
B Horizon
o
o
CT
1)
E
r\i
o
en
c
o
o
cr
c
Bw Horizon
..L
C Horizon
U Upper Limit L = Lower Limit R - Reference Value
Figure 1-24. Moving average plota of laboratory tranda In B, Bw, and C horizon audit aamplaa for CEC_CL
293
-------
CECJ3AC
Cation Exchange Capacity by Ammonium Acetate
o
o
cr
0)
E
o
O
o
cr
-------
AC_BACL
Exchangeable Acidity in Barium Chloride Triethanolamine
o
o
t
o
o
o
£
o
en
o
O
o
X.
>
o
B Horizon
Bw Horizon
C Horizon
L
U = Upper Limit L = Lower Limit R = Reference Value
Figure 1-26. Moving average plota of laboratory trend* In 8, Bw, and C horizon audit aamplea for ACJ3ACL.
295
-------
o
o
HI
01
o
en
C
o>
o
o
\
0>
o
o
o
cr
>
<
01
c
CA_CL2
Extractable Calcium" in Calcium Chloride
B Horizon
Bw Horizon
C Horizon
U - Upper Limit L = Lower Limit R - Reference Value
Figure 1-27. Moving average plota of laboratory trenda In B, Bw, and C horizon audit sample* for CA_CLZ
296
-------
o
o
cr
<
,c
o
a-
o
o
cr
0)
E
o
o
MG_CL2
Extractable Magnesium in Calcium Chloride
B Horizon
Bw Horizon
._ L
C Horizon
L
U = Upper Limit L = Lower Limit R = Reference Value
Figure 1-28. Moving average plota of laboratory trenda In B, Bw, and C horizon audit aamples for MG_CL2L
297
-------
o
o
-------
C7>
O
O
CT
0)
E
fi
o
O
o
<
en
c
>
o
o
o
IT
0)
E
O
O
cn
o
en
o
O
o
o
6
I!
NA_CL2
Extractable Sodium in Calcium Chloride
B Horizon
Bw Horizon
C Horizon
R
L
U - Upper Limit L = Lower Limit R = Reference Value
Figure 1-30. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for NA_CL2.
299
-------
FE_CL2
Extractable Iron in Calcium Chloride
B Horizon
o
o
cr
c
'
o
en
o
o
V
E
(N
O
O
ci
II
o>
o
C7>
C
C Horizon
..L
L
U - Upper Limit L = Lower Limit R = Reference Value
Figure 1-31. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for FE_CL2.
300
-------
O
o
O
o
en
o
o
O
o
o
II
-------
q
ci
II
q
d
o>
o
o
2
FE_PYP
Extractable Iron in Sodium Pyrophosphate
BHorizon
m
q
ci
o
0)
z
o
Bw Horizon
C Horizon
L
U = Upper Limit L = Lower Limit R = Reference Value
Figure 1-33. Moving average plot* of laboratory trends In B, Bw, and C horizon audit aamplea for FE_PYP.
302
-------
AL^PYP
Extractable Aluminum in Sodium Pyrophosphate
B Horizon
m
o
o
II
0>
>
<
en
c
>
O
s
ID
O
O
en
O
cn
c
'
o
o
II
Bw Horizon
C Horizon
.-L
L
U = Upper Limit L = Lower Limit R = Reference Value
Figure 1-34. Moving average plots of laboratory trend* In B, Bw, and C horizon audit samples for AL_PYP.
303
-------
o
ii
O
II
v^
<
C
O
V
0>
en
C
o
FE_AO
Extractable Iron in Acid Oxalate
B Horizon
Bw Horizon
C Horizon
._ I
U - Upper Limit L = Lower Limit R = Reference Value
Figure 1-35. Moving average plota of laboratory trenda In B, Bw, and C horizon audit aamplea for FE_AO.
304
-------
O
O
0)
z
CT
C
q
d
0)
en
O
O
2
K
"$
O
O
en
C
AL_AO
Extractable Aluminum in Acid Oxalate
B Horizon
..L
Bw Horizon
C Horizon
U Upper Limit L = Lower Limit R = Reference Value
Figure 1-36. Moving average plots of laboratory trend* In B, Bw, and C horizon audit samples for AL_AO.
305
-------
O
O
O
0>
>
<
CM
q
6
II
s
-------
s
in
o
o
o
o
II
en
O
>
O
3
m
o
o
en
C
FE_CD
Extractable Iron in Citrate Dithionite
B Horizon
Bw Horizon
C Horizon
L
U - Upper Limit L = Lower Limit R = Reference Value
Figure 1-38. Moving average plots of laboratory trend* In B, Bw, and C horizon audit sample* for FEjCD.
307
-------
5
in
o
o
en
o
en
c
i
m
q
d
(0
en
o
AL_CD
Extractable Aluminum in Citrate Dithionite
B Horizon
K
"S
q
ci
II
o>
C7>
c
Bw Horizon
L
C Horizon
L
U m Upper Limit L = Lower Limit R = Reference Value
Figure 1-39. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for AL_CD.
308
-------
en
f.
o
-5.
E
o
00
01
E
o
cr-
a
o
2
SO4JH2O
Extractable Sulfate in Water
B Horizon
Bw Horizon
L
C Horizon
L
U - Upper Limit L = Lower Limit R = Reference Value
Figure 1-40. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for SO4_H2O.
309
-------
CO
Oi
£
o
LO
II
o
CO
cn
E
q
ui
<
0>
c
CO
cn
E
O
m
0!
o>
o
o
SO4_PO4
Extractable Sulfate in Sodium Phosphate
B Horizon
--L
Bw Horizon
L
C Horizon
U " Upper Limit L = Lower Limit R = Reference Value
Figure 1-41. Moving average plot* of laboratory trends In B, Bw, and C horizon audit samples for SO4_PO4.
310
-------
o
ii
01
c
o
5
01
E
o
II
01
o
o>
c
o
2
o
II
o>
o
SO4_0
Sulfate Isotherm 0 mg S/L
B Horizon
Bw Horizon
R
L
C Horizon
U = Upper Limit L = Lower Limit R = Reference Value
Figure 1-42. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for SO4_0.
311
-------
CO
E
o
II
>
1)
^
CO
en
E
o
II
en
o
en
c
CO
en
£
O
II
SO4_2
Sulfate Isotherm 2 mg S/L
BHorizon
Bw Horizon
C Horizon
U = Upper Limit L - Lower Limit R = Reference Value
Figure 1-43. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for SO4_2.
312
-------
CO
0>
CM
o
E
CNJ
6
o
II
<
CT>
E
CM
d
V
en
O
cn
c
SO4_4
Sulfate Isotherm 4 mg S/L
B Horizon
L
Bw Horizon
.-L
C Horizon
U = Upper Limit L = Lower Limit R = Reference Value
Figure 1-44. Moving average plots of laboratory trends In B, Bw, and C horizon audit sample* for SO4_4.
313
-------
en
E
ID
O
o
o
C Horizon
L
U - Upper Limit L - Lower Limit R - Reference Value
Figure 1-45. Moving average plota of laboratory trenda In B, Bw, and C horizon audit aamplea for SO4_8.
314
-------
E
uT>
O
II
o
in
O
o
V
>
<
CT>
C
E
in
o
01
o
o
SO4J6
Sulfate Isotherm 16 mg S/L
B Horizon
L
Bw Horizon
C Horizon
L
U » Upper Limit L = Lower Limit R - Reference Value
Figure 1-46. Moving average plota of laboratory trenda In B, Bw, and C horizon audit aamplea for SO4_16.
315
-------
00
O!
£
o
00
en
E
o
o
SO4_32
Sulfate Isotherm 32 mg S/L
B Horizon
Bw Horizon
C Horizon
CO
cr>
O
-------
s
o
o
I!
o
2
O
II
en
c
C_TOT
Total Carbon
BHorizon
Bw Horizon
..L
C Horizon
._ L
U * Upper Limit L - Lower Limit R = Reference Value
Figure 1-48. Moving average plota of laboratory trenda In B, Bw, and C horizon audit aamplea for CJTOT.
317
-------
p
d
o>
c
'
o
q
d
o
K
Is
O
o
II
v
CT
O
N_TOT
Total Nitrogen
B Horizon
Bw Horizon
L
C Horizon
2-2-T"2 2 2 2 2 r
U » Upper Limit L = Lower Limit R = Reference Value
Figure 1-49. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for NJTOT.
318
-------
o
o
o
2
(M
o
o
o
IN
o
o
o
O
2
S_TOT
Total Sulfur
B Horizon
Bw Horizon
C Horizon
U = Upper Limit L = Lower Limit R = Reference Value
Figure 1-50. Moving average plots of laboratory trends In B, Bw, and C horizon audit samples for SJTOT.
319
-------
Appendix J
Step Function Approach for Estimating Data Uncertainties
The following internal report describes the step function approach that was used by the
quality assurance staff for estimating the uncertainty from data collection error. A linear model is
used to define measurement imprecision and bias as the principal quantitative characteristics of
measurement uncertainty quantified during the routine data analysis. In addition, estimates are
made of confounded population and measurement uncertainty as a function of specific
configurations of regional soils data. Induced uncertainties at different data collection stages are
estimated using soil chemistry data from replicate samples and audit samples.
320
-------
Internal Report
Step Function Approach for Estimating Data Uncertainties
M. J. Miah, R. D. Van Remortel, G. E. Byers, M. L. Papp, and J. E. Teberg, Lockheed
Engineering & Sciences Company, Las Vegas, Nevada, and L. J. Blume, U.S. Environmental
Protection Agency, Las Vegas, Nevada.
Introduction
The effect of measurement error on the overall quality of environmental data
should be carefully evaluated because of the potential for this error to influence the
results of a study. The measurement error can be considered to be the sum of several
components of error that occur at different, independent stages of the data collection
process. The error can be characterized using data from a large number of replicate
measurements within the sample population. Budgetary and logistical constraints,
however, usually do not allow replication of each sample measurement. The next
best alternative is to select a representative subsample of the routine samples for
replication purposes.
Measurement error can be defined as a random variable or, in some cases, as
a sequence of random variables. The error characteristics can be expressed as various
combinations of the random variables. For this study, the primary characteristics
that are associated with the measurement uncertainty are imprecision and inaccuracy
[1]. Using an additive model, the observed characteristic of a routine sample is
assumed to be the sum of the true sample characteristic and a random value of the
measurement error. The imprecision is defined as the variance of the measurement
error and the inaccuracy is defined as the bias, or expected value of the measurement
error. The overall measurement uncertainty of a data set is a function of its
measurement imprecision and inaccuracy.
In any quality assurance design, the data generation process should be
controlled or managed in such a way as to achieve predetermined measurement
quality objectives with respect to the error characteristics. For certain parameters,
imprecision and inaccuracy may vary across the range of observed analyte
concentration. In this event, it is possible to segment the entire concentration range
into a series of different, mutually exclusive intervals. Imprecision and inaccuracy
can be estimated for each interval of observed analyte concentration and all of the
individual estimates can be combined to provide a pooled estimate of overall
measurement uncertainty.
321
-------
Operational Framework
The quality assurance design described herein was developed for use in a soil
characterization study of various watersheds in the Mid-Appalachian region of the
United States. The study is the third in a series of three regional surveys conducted
as part of the U.S. Environmental Protection Agency's "Direct/Delayed Response
Project". The objective of the project is to make regional predictions of the response
time of various lake and stream watersheds to current or projected rates of acidic
deposition. The two previous studies focused on the Northeastern region and
Southern Blue Ridge region of the United States.
In the Mid-Appalachian survey, soil sampling classes were configured on the
basis of intensive soil mapping conducted in the region. Each sampling class consists
of discrete soil types, having similar geomorphic and developmental characteristics,
which are grouped together to provide a representative sampling base across
watersheds within the region. Soil pits are excavated at randomly selected sites
within the appropriate sampling classes, and soil samples are collected from the
component soil horizons within each exposed pedon. It is expected that the soil
physical and chemical characteristics of samples from different pedons belonging to
a specific sampling class are influenced by the horizon type and by natural spatial
variability. It is also possible that overall measurement uncertainty is dependent on
a specific soil characteristic.
The quality assurance design must be versatile enough so that the measure-
ment uncertainty can be estimated irrespective of its dependence upon soil charac-
teristics or configurations of soil variables. For this study, a modified hierarchical
design [2] is used to estimate overall measurement uncertainty for the soil characteris-
tics of interest. This approach assumes that the errors associated with different
measurement operations are independent of each other. Measurement errors in the
survey are generated during three independent operations: field sampling, sample
preparation, and sample analysis. The generalized model used in the quality
assurance design is as follows:
Y = X + e
where: Y is the observed value of a soil characteristic in a soil sample,
X is the true value of the characteristic in that sample, and
e is the measurement uncertainty confounded in the sample,
and the primary components of the measurement uncertainty, e, are defined as:
e = ( + ep + ea
where: e{ is the field sampling error,
ep is the sample preparation error, and
ea is the sample analysis error.
322
-------
Assessment of Precision
The imprecision components of the measurement error are estimated through
an analysis of variance (ANOVA), where the variance of the measurement error,
V(e), is defined in the following relationship:
V(e) = c/f + a2p + cr2,
where:
-------
imprecision that becomes conditional if data within one or more levels of concentra-
tion are missing.
The ANOVA for replicate sample data provides an estimate of overall
measurement imprecision as well as an estimate of the specific imprecision associated
with the field sampling and laboratory operations. At this juncture, the sample
preparation and sample analysis imprecision are confounded and cannot be separated.
An ANOVA table identifying the variance components is provided in Table 1 below.
Table 1. Analysis of Variance Components
Sources
of variance
Average
Sum of
square
nY2
df
1
Mean
square
Expected value of
the mean square
Field duplicates 2(Y;j - Y,)2 1
Preparation duplicates 22(Yj. - Y )2 2
Total Y;j2 4
mf
m,
2cr2f + a2 + a2
After obtaining an estimate of measurement imprecision for each dupli-
cate/routine sample pair, it is necessary to combine the individual estimates to obtain
a pooled estimate of measurement imprecision for the data set. A fundamental
problem is that this estimate is dependent on the analyte concentration of the
associated routine samples, i.e., the error variance becomes cr2; for the "i"th analyte
concentration. A scatter diagram of the mean of replicate measurements versus the
error variance can provide a useful approximation of the relationship between
analyte concentration and measurement imprecision. It may be that the relationship
is not very distinct except for an indication that measurement imprecision varies with
changes in the observed analyte concentration, such that:
a2 : a -<0,K]
where: fl is the set on which the routine sample observations take values, and
K is the highest variance in the range.
The dominated convergence theorem is used to define a sequence of step
functions (j>n = 2,i=1>n) o-2xFi , such that 4>n converges to a2 irrespective of its func-
tional form [4]. The overall measurement imprecision is also a random variable
belonging to the weighted partition { F;: a2j } i=1>n for a large number of partitions.
To regulate the effects of concentration dependency, the entire concentration range
can be partitioned into n mutually exclusive intervals Fj, F2,... Fn. These intervals
are not necessarily of equal concentration length and are such that within each
interval, F( , the error variance is independent of analyte concentration; therefore,
UFj = fl , and F.OFj = <5 for all i^j
where: 3> is an empty set.
324
-------
From this, the expected variance can be defined in the following manner:
E(a2) = /a2dG
= 2[i=i,n| cr2i E(xFi)
where:
So,
where:
Xm
G
is 1 (if Y e Fs) or 0 (if Y £ F,"), and
is a distribution function.
E(XFi) = P(FO
P(Fj) is the proportion of routine samples within the interval Ff.
For every interval, the variance can be estimated from the replicate sample
data set and the corresponding proportion can be estimated from the routine sample
data set. Cluster analysis is useful for defining the number of intervals and their
endpoints. The expected variance across intervals, as defined above, can be used as
a measure of uncertainty due to imprecision. The rationale for this is as follows:
Scenario 1:
Suppose it is found that imprecision is relatively constant for different
analyte concentrations, i.e., there is no evidence for rejecting the
hypothesis that cr2 = A, Le., that V(e) is a constant function on the routine
data set fl. Then:
E(a2) = / a2 dG
= AS(i=1,nl E(XFi)
= A, because F; and Fj are disjoint for every i ^ j,
and P(UFj) = P(fl) = 1
Scenario 2: Suppose it is found that imprecision is not constant for different analyte
concentrations, i.e., there is no evidence for rejecting the hypothesis that
V(e) is a two-step function, as follows:
A, if Y Fj
a2 =
A2 if Y e F2
Scenario 3:
then: E(a2) =
If cdOO percent of the routine samples is within Fj, the above expression
becomes E(a-t) = aA^ + (l-a)A2 . The convex combination of A! and A2
is a reliable estimate of measurement imprecision over the concentration
range. In this scenario, cdOO percent of the routine data have
imprecision equal to Aj and (l-a)lOO percent of the data have imprecision
equal to A2 .
Suppose it is found that imprecision is continuously changing on the
space ft where the observed routine sample characteristic Y takes values.
325
-------
Since cr2 > 0, there exists a sequence of step functions, 6n , as defined
previously, such that
-------
of inaccurate data may indicate that there is nontrivial bias in the data set. In the
Mid-Appalachian survey, contractual requirements for field and laboratory reference
samples are used to control bias to a negligible level. To accomplish this, the mean
sum of squares of the reference sample results should be small enough so that the
measurement bias can be considered negligible with 95 percent confidence, that is:
I Y - X | < Z* V E((f).
If the expected variance is known or can be estimated a priori by using reference
data, then exception windows for the reference samples can be generated.
In the additive model, the bias component of measurement uncertainty is
estimated through an analysis of the expected error, E(e), as defined in the following
relationship:
E(e) = Bf + Bp + Ba
where: Bf is the bias from field sampling,
Bp is the bias from sample preparation, and
Ba is the bias from sample analysis.
If suitable values for the reference samples are known, then the expected error
is calculated as:
E(e) = M- - E(Y)
where: \L is the reference value, and
E(Y) is the expectation of the observed values.
In this study, a standard soil reference material consisted of natural non-
synthetic soil, making it difficult to establish a single reference value. Instead, a
range of acceptable values defined by confidence intervals from previous charac-
terizations is used in lieu of the reference value. All observed values within that
range are assumed to be equally reliable and, therefore, have zero bias. Accordingly,
the error term can be described as:
Y - U if Y > U
e ={ 0 if L < Y < U
Y - L if Y < L
where: U is the upper reference value, and
L is the lower reference value.
So, E(e) = /Y>u (Y - U) f(Y) dY + /YU(Y-U) + 2Y
-------
to vary with concentration, then the step function technique described in the
imprecision section is used to pool the biases from different concentrations.
In the Mid-Appalachian survey, nearly equal quantities of standardized
reference samples were allocated to each sampling crew (field audit samples) and
analytical laboratory (laboratory audit samples). System-wide bias can be evaluated
using the field audit samples. Throughout the data acquisition process, it is assumed
that the primary sources of bias are derived from procedural or instrumental
differences among the analytical laboratories. Inaccuracy stemming from instrument
calibration, sample contamination, etc., can cause laboratory measurements to be
biased high or low in relation to a reference value. If the laboratory bias is constant
across different analyte concentrations for the population of laboratories, then a
single audit sample concentration is sufficient to control and estimate the analytical
bias.
For this study, it is assumed that bias does not change on the set H from which
the routine samples take values. However, if bias differs significantly between
analytical laboratories, then it can be considered to be a random variable which
takes different values for different laboratories. The expected error in that case is:
E(e) = E{EL(e)}
where: EL(e) is the expected error for laboratory L.
Therefore,
Ej(e) = EL(e) = BL for laboratory L,
and: E[EL(e)] = 2 BLPL.
Accordingly, bias can be estimated using the reference sample data from each
analytical laboratory. The relative frequency of the routine samples analyzed by
each laboratory can be used to estimate probabilities.
Laboratory Differences
It is possible that laboratory inaccuracy may differ significantly from one
laboratory to another. In this event, laboratories can be evaluated for the
significance of these differences. The estimates of conditional error expectations can
be used to test the significance of the analytical bias. In the Mid-Appalachian survey,
the homogeneity of conditional error expectations among laboratories were studied
by using appropriate t tests or F tests. The interlaboratory differences comparison
is a laboratory evaluation aspect and not a data evaluation aspect, hence, no error
component criteria can be derived for the data uncertainty estimates.
Laboratory Trends
It is possible that inaccuracy is caused by contamination effects, instrumental
drift, or some other systematic or irregular effect over time. For each laboratory,
these effects can be determined by examining temporal intralaboratory trends.
Standard regression techniques applied to the laboratory reference sample data were
328
-------
used to identify any temporal patterns in the observed analytical laboratory
uncertainty.
Estimation of Overall Measurement Uncertainty
The overall measurement uncertainty is calculated as a 52 term defined for
the expected squared error using the summation of the appropriate imprecision and
bias terms. The expected variance is a function of within-batch and between-batch
imprecision and the expected error is a function of bias across laboratories. The
expected squared error can be described as follows:
52 = E(e2) = E{Efc*)}
= E [Es{e - E^e)}2 + E2^)]
= E2 + EiEfc)}
+ VB(e)} + E{E2L(e)}
Accordingly, the appropriate operator is placed on the squared error terms
within each interval, which allows the overall measurement uncertainty to be
expressed in the reporting units of the variable of interest, as follows:
5 = E V cii2 + Ei2(e)
= E VV^e) + VB(e) + E2(e)
The estimate of measurement imprecision is the pooled value of the error
mean sum of squares across different concentration ranges. In some surveys, it may
not be possible to estimate inaccuracy due to bias and, consequently, the total error
mean sum of squares. In this event, the error variance can be pooled knowing that
the result provides a measure of uncertainty due to measurement imprecision only.
Results and Discussion
Quality assurance data from the DDRP Mid-Appalachian soil survey were
used to estimate the overall measurement uncertainty associated with the soil
chemistry data. The modified hierarchical design implemented in the survey allowed
the individual components of overall measurement uncertainty for the entire range
of analyte concentration to be estimated. One exception to this is the analytical
imprecision estimate, which was based on data from three types of laboratory audit
samples encompassing the low- and median-ranges of concentration only. This
estimate, however, was deemed more reliable than the estimate calculated from
duplicate samples provided by the laboratories.
The measurement uncertainty was defined as the expected squared error as
previously described and was derived by estimating cr2w , cr2B , and E2(e) across specific
partitioned intervals of analyte concentration. The duplicate/routine sample and
laboratory audit sample data sets were used to estimate imprecision, and the
laboratory audit sample data sets were used to estimate bias.
As previously described, three separate data collection stages were evaluated
for their respective data measurement uncertainties. These are expressed as 8j , 52 ,
329
-------
and 83 values which represent the analytical uncertainty, the confounded sample
preparation and analytical uncertainty, and the confounded field sampling, sample
preparation, and analytical uncertainty, respectively. In order to obtain uncertainty
estimates for the routine sample population as defined by soil sampling class/horizon
configurations, the standard deviations from all possible configurations were pooled
for a given soil characteristic. This pooled uncertainty estimate, 8«, generally shows
a marked increase relative to the confounded three-stage measurement uncertainty
estimated as 83. A relative comparison of 83 and 84 can be used as one possible
approach for estimating the magnitude of measurement error in relation to the
overall variability in the routine sample population.
As expected, the overall uncertainty increased with increasing sources of
confounded uncertainty such that 82 < 83 < 84. For some parameters, however, 8j >
82 . In these situations, a lack of laboratory audit sample data in some of the
intervals made it impossible to construct a reliable estimate of the uncertainty across
the intervals. The Sj estimates are considered to be conditional on those intervals for
which uncertainty estimates are available.
Table 2 provides two different examples of the overall measurement
uncertainty estimates, 83 , for the concentration intervals within which the
measurement uncertainty is considered to be constant. For each interval, estimates
of within-batch imprecision, between-batch imprecision, bias, expected squared error,
and the empirical estimates of the proportion of routine samples in each interval are
provided. It was expedient to pool the error characteristics for the different
Table 2. Selected overall measurement uncertainty estimates for different intervals
Parameter Interval CTW orB E(e) V/E(e2) P(F)
Extr. sulfate
(mg S/kg)
Exch. calcium
(meq/lOOg)
<7.0
7.0-15.0
15.0-18.0
>18.0
<0.2
0.2-1.0
1.0-5.0
>5.0
0.588
1.707
2.508
0.632
0.039
0.070
0.862
0.655
0.644
1.430
0.036
1.135
0.028
0.023
m
0.022
0.022
0.022
0.022
0.002
0.002
0.002
0.002
0.872
2.227
2.508
1.299
0.048
0.074
0.862*
0.655*
0.145
0.481
0.190
0.184
0.204
0.505
0.281
0.011
83 = 1.913
83 = 0.296*
CTW is the within-batch imprecision for interval
CTg is the between-batch imprecision for interval
E(e) is the bias for interval
VE(62) is the measurement uncertainty for interval
P(F) is the empirical proportion of routine samples in interval
83 is the overall measurement uncertainty across intervals, calculated from field duplicate samples
is a partial estimate (underestimate) due to incomplete data in one or more intervals
330
-------
concentration intervals, hence, the corresponding summary 5 value pooled across the
intervals is also given for the two parameters of interest.
It is evident that the data for extractable sulfate generally have high precision
across the range of concentration, in relation to the survey objective of 30 percent
precision for the field duplicate samples. The estimated overall measurement
uncertainty for sulfate is approximately 1.91 milligrams sulfur per kilogram of soil
(mg S/kg), which is small in relation to the mean sulfate concentration of 17.24 mg
S/kg. The estimated overall measurement uncertainty for exchangeable calcium is
approximately 0.30 milliequivalents per 100 grams of soil (meq/lOOg), which is large
in relation to the mean calcium concentration of 0.93 meq/lOOg. This relatively large
uncertainty can be attributed to the generally low concentrations of exchangeable
calcium in the population of soil samples collected.
Table 3 provides a summary of data uncertainty estimates for the various
stages of measurement and the sample population. In nearly all cases, 8< > 83 > 62,
while Si and 82 are irregular in their magnitude. Commonly, the 8, value is relatively
high because it is a conditional estimate based on incomplete sample data. To
compare the data uncertainties of any two data sets, a data user may wish to relate
the data uncertainty measures with the population mean of the routine samples
which is provided for reference in the table. A comparison of 83 and 84 values can
be used as a basic estimate of the relative effect of measurement uncertainty on the
overall uncertainty of the routine data.
Table 3. Uncertainty estimates associated with the Mid-Appalachian survey.
Parameter
Soil pH
Exch. calcium
Exch. magnesium
Exch. potassium
Exch. sodium
Exch. capacity
Exch. acidity
Extr. iron
Extr. aluminum
Extr. sulfate
Total carbon
Reporting
units
pH units
meq/lOOg
meq/lOOg
meq/lOOg
meq/lOOg
meq/lOOg
meq/lOOg
weight %
weight %
mg S/kg
weight %
Data uncertainty estimate"
St 82 83 &4
0.078
0.032**
0.017**
0.007**
0.004**
1.407
1660*
0.025*
0.025**
1.626*
0.070**
0.087
0.055*
0.019*
0.011*
0.005*
1.661
1.652
0.030
0.022**
1.623
0.059**
0.114
0.296*
0.026*
0.009*
0.006*
1.633
L832
0.035
0.027**
L913
0.185*
0.473
L091*
0.446*
0.094*
0.014*
4.667
5582
0305
0.165*
27.175
L213*
Mean"
4.751
0.930
0.418
0.133
0.016
7.242
11197
0347
0.190
17244
L438
8, = imprecision due to analytical error, computed from laboratory audit samples
O2 = imprecision due to preparation and analytical error, computed from preparation duplicates
8, = imprecision due to field sampling, preparation, and analytical error; computed from field duplicates
54 = imprecision due to measurement sources and sample population uncertainty; computed from routine samples grouped
in sampling class/horizon configurations
mean concentration of the soil samples in the survey
partial estimate (underestimate) due to incomplete data for one or more intervals
conditional estimate (overestimate) due to lack of data in one or more intervals
331
-------
Summary and Conclusions
The objective of the modified hierarchical design was to control and estimate
measurement precision and accuracy during the Mid-Appalachian survey. These
estimates are very useful in assessing the quality of environmental data. When the
soil characteristics of a survey region are expected to be wide ranging, it is likely that
the measurement imprecision will vary for different intervals across the range of
concentration for most variables. In this event, measurement uncertainty can be
estimated separately for each of the intervals, and the overall measurement
uncertainty can be estimated by a step function which pools the different intervals.
The relative effect of measurement uncertainty on the overall data uncertainty in
the routine samples can be assessed using the weighted average of the step function
with respect to the routine samples. The data from the Mid-Appalachian survey
suggest that data uncertainty, with respect to imprecision, increases with increasing
confounded sources of uncertainty.
In any survey design, measurement uncertainty should be controlled at an
acceptable level. Unless there is evidence from previous data to reject the hypothesis
that the measurement bias is constant for different intervals of analyte concentration,
one source of reference soil material in a single concentration interval for each
variable is sufficient to control and estimate measurement bias.
(Although the research described in this article has been supported by the United
States Environmental Protection Agency through contract 68-03-3249 to Lockheed
Engineering & Sciences Company, it has not been subjected to Agency review and
therefore does not necessarily reflect the views of the Agency and no official
endorsement should be inferred.)
References
[1] M.J. Miah and J.M. Moore, Parameter design in chemometry, Chemometrics and
Intelligent Laboratory Systems, 3(1988)31-37.
[2] G.E.P. Box, W.G. Hunter, and J.S. Hunter, Statistics for Experimenters, J. Wiley
& Sons, New York, 1978, pp. 572-582.
[3] J.K. Taylor, Quality Assurance of Chemical Measurements, Lewis Publishers,
Chelsea, Michigan, 1987, 328 pp.
[4] W. Rudin, Real and Complex Analysis, McGraw-Hill, New York, 1974, pp. 22.
332
-------
Appendix K
Data Tracking and Verification Forms
This appendix contains some of the data tracking and verification forms used by the QA staff
to document batch sample data checks performed by the analytical laboratories during the MASS
sample analysis. Included are examples of the DDRP Form 102 (shipping form), DDRP Form 500
(batch confirmation/reanalysis form, and DDRP Forms 600A and 600B (internal consistency
confirmation forms).
333
-------
DIRECT/DELAYED RESPONSE PROJECT (DDRP)
SHIPPING FORM 102
DATE RECEIVED
BY DATA MGT
D D M M M
[3 D M M M V V
Preo Lab ID Dale Received _ .._._.
Batch ID Dale Shioped
Analytical Lab ID
SAMPLE NO
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
BULK SAMPLES
SHIPPED
RECEIVED
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
PH < $
RETURNED TO LEMSCO
(CHECK Y IF YES)
Signature nf Preparation 1 ahnratnrv Manaqor _
Comments'
Figure K-1. DDRP Form 102 (shipping form).
334
-------
OAlt SENT
OME RECEIVED
DORP SOIL SURVEY
fORM iOO
Batch ID
Data Confirmation/Reanalysis Rpquest t-orm
Contractor Analytical laboratory Laboratory Supervisor
The following suspect data values require:
Confirmation (See I)
Reanalysis (See II)
PARAMETER
! ! SUSPfCT [RECONFIRMED
DORP , SAMPLE ; ORIGINAL | NEW
FORM NO. i ID j VALUE I VALUE
Explanation
CONTRACT
ANALYTICAL
LABOR A TOR V
lEHSCO
I. Confirmation Request: Old ANY values change: Yes No
If yes, reason (note above in explanation column):
(A) Reporting Error (C) Original reported value did not change
(B) Calculation Error (0) Data Previously Omitted
(E) Other - Explain
Whether values are changed or not, submit supporting RAW DATA.
Additional Cements Regarding Confirmation:
II. Reanalysis Requested Due to:*
OA (External) Data
QC (Internal) Data Indicated Below:
IDL > CROL
Matrix Spike Recovery Outside Criteria
Replicate Precision (X PSD) Outside Criteria; Insufficient Number-of Replicates
Blank > CRDL (Reagent; Calibration)
QCCS Outside Criteria (DL: Low; High)
1C Resolution Value Below feOX
Air Dry Sample Weight Ouside Criteria
Total Sample Volume. Aliquot Volume, or Dilution Volume Outside Criteria
Standard Relationships Out of Range
Additional Comments Regarding Reanalysls:
All appropriate data forms including OC data forms must be submitted in support of reanalysis.
Date form completed' must reflect date of reanalysis.
FOR LEMSCO USE ONLY:
PRE-VERIFICATION
POST-VERIFICATION
NUMBER OF VALUES SUBMITTED
NUMBER OF VALUES CHANGED
0600C
Figure K-2. DDRP Form 500 (data conflrmatlon/reanalysis request form).
335
-------
DDRP FORM 600A
Parameter Name:
Laboratory:.
DDRP Form
BATCH
No:
SAMPLE
NUMBER
SUSPECTED
VALUES
CORRECTED
VALUES
COMMENTS
LABS LESC
LABORATORY MANAGER
Figure K-3. DDRP Form 600A (Internal consistency confirmation form).
336
-------
DDRP FORM 60OB
Parameter Name:
Laboratory:.
DDRP Form
BATCH
No:
SAMPLE
NUMBER
SUSPECTED
VALUES
CORRECTED
VALUES
COMMENTS
LABS LESC
LABORATORY MANAGER
Figure K-4. DORP Form 600B (Internal consistency confirmation form).
337 *U.S. GOVERNMENT PRINTING OFFICE: 1990 -7*8-
-------
|