April 2011
   Environmental Technology
   Verification Report


   INSTRUMENTATION NORTHWEST INC.
   AQUISTAR® TEMPHION™ SMART SENSOR
   AND DATALOGGER NITRATE-SPECIFIC ION-
   SELECTIVE ELECTRODE FOR
   GROUNDWATER REMEDIATION MONITORING
             Prepared by

             Balteile
            //?L? Husincss oj Innovation
          Under a cooperative agreement with

             f/EPA
ET1/ET1/ET1/

-------
                                    April 2011
 Environmental Technology Verification
                Report

   ETV Advanced Monitoring Systems Center

    INSTRUMENTATION NORTHWEST INC.
AQUISTAR® TEMPHION™ SMART SENSOR AND
    DATALOGGER NITRATE-SPECIFIC ION-
 SELECTIVE ELECTRODE FOR GROUNDWATER
        REMEDIATION MONITORING
                   by
     Andrew Barton, Chris Gardner, Dale Rhoda, and Amy Dindal, Battelle
              John McKernan, U.S. EPA

-------
                                         Notice

 The U.S. Environmental Protection Agency, through its Office of Research and Development,
funded and managed, or partially funded and collaborated in, the research described herein.  It
 has been subjected to the Agency's peer and administrative review. Any opinions expressed in
 this report are those of the author(s) and do not necessarily reflect the views of the Agency,
 therefore,  no official  endorsement  should be  inferred.  Any  mention  of  trade  names  or
 commercial products does not constitute endorsement or recommendation for use.
                                           11

-------
                                      Foreword

The EPA is charged by Congress with protecting the nation's air, water, and land resources.
Under a mandate of national environmental laws, the Agency strives to formulate and implement
actions leading to a compatible balance between human activities and the ability of natural
systems to support and nurture life. To meet this mandate, the EPA's Office of Research and
Development provides data and science support that can be used to solve environmental
problems and to build the scientific knowledge base needed to manage our ecological resources
wisely, to understand how pollutants affect our health, and to prevent or reduce environmental
risks.

The Environmental Technology Verification (ETV) Program has been established by the EPA to
verify the performance characteristics of innovative environmental technology across all media
and to report this objective information to permitters, buyers, and users of the technology, thus
substantially accelerating the entrance of new environmental technologies into the marketplace.
Verification organizations oversee and report verification activities based on testing and quality
assurance protocols  developed with input from major stakeholders and customer groups
associated with the technology area. ETV consists of six environmental technology centers.
Information about each of these centers can be found on the Internet at http://www.epa.gov/etv/.

Effective verifications of monitoring technologies are needed to assess environmental quality
and to supply cost and performance data to select the most appropriate technology for that
assessment. Under a cooperative agreement, Battelle has received EPA funding to plan,
coordinate, and conduct such verification tests for "Advanced Monitoring Systems for Air,
Water, and Soil" and report the results to the community at large. Information concerning this
specific environmental technology area can be found on the Internet at
http ://www. epa.gov/etv/centers/centerl .html.
                                           in

-------
                                 Acknowledgments

The authors wish to acknowledge the support of all those who helped plan and conduct the
verification test, analyze the data, and prepare this report.  The U.S. Department of
Agriculture/Agricultural Research Service (USDA/ARS) National Laboratory for Agriculture
and the Environment, and specifically Tom Moorman, Beth Douglas, and Jerry Hatfield,
provided a tremendous amount of support for this verification test, including reference analyses,
field site access, and laboratory support.  Dr. Stuart Nagourney (New Jersey Department of
Environmental Protection [NJDEP]) and Michael Brody (EPA) provided extensive input to the
planning of the verification test and participated in document review. We also acknowledge the
contributions to this report from Kati Lentz and Rose Filoramo, students at the College of New
Jersey, who performed data and statistical analyses under the direction of Dr. Nagourney. Data
storage,  graphing, and mapping capabilities were performed by Mark Kram and Cliff Frescura of
Groundswell Technologies, Inc.  Quality assurance (QA) oversight was provided by Michelle
Henderson, EPA, Amy Bowman, NJDEP, and Zachary Willenberg, Rosanna Buhl, and Kristen
Nichols, Battelle.  The authors also thank Charles Spooner, EPA, Ken Wood, DuPont, and Dr.
Jacob Gibs, U.S. Geological Survey  for their review of the test/QA plan and/or this verification
report.
                                          IV

-------
                                       Contents
                                                                                   Pat
Foreword	iii

Acknowledgments	iv

Figures	vi

Tables	vi

List of Abbreviations	viii

Chapter 1 Background	1

Chapter 2 Technology Description	2

Chapters Test Design and Procedures	4
        3.1    Introduction	4
        3.2    Nitrate Analysis Reference Method	4
        3.3    Test Design	5
       3.3.1   Laboratory Testing Stage	5
       3.3.2   Field Testing Stage	6
        3.4    Test Procedures	9
       3.4.1   Accuracy	9
       3.4.2   Variability	10
       3.4.3   Duplication	10
       3.4.4   Effect of Changes in Water Quality	10
       3.4.5   Operational and Sustainability Factors	10
        3.5    Analysis of Baseline Concentrations	11
        3.6    Verification Schedule	11

Chapter 4 Quality Assurance/Quality Control	12
        4.1    Laboratory Sample Analysis QA/QC	12
       4.1.1   Instrument Calibration Checks	12
       4.1.2   Initial Calibration Checks	12
       4.1.3   Laboratory Reagent Blanks	13
       4.1.4   Laboratory Fortified Blanks	13
       4.1.5   Laboratory Fortified Sample Matrix	13
       4.1.6   Laboratory Duplicate Samples	13
        4.2    Field QA/QC	15
       4.2.1   Field Duplicate Samples	15
       4.2.2   Equipment Blanks	15
        4.3    Audits	15
       4.3.1      Performance Evaluation Audits	15
       4.3.2      Technical Systems Audit	16
       4.3.3      Data Quality Audits	17

-------
Chapters  Statistical Methods	18
        5.1   Accuracy	18
        5.2   Variability	19
        5.3   Duplication	20
        5.4   Effect of Changes in Water Quality	20

Chapter 6  Test Results	21
        6.1   Accuracy	21
        6.2   Variability	25
        6.3   Duplication	25
        6.4   Effect of Changes in Water Quality	27
        6.5   Operational and Sustainability Factors	29

Chapter 7  Performance Summary	31

Chapters  References	33


                                       Figures

Figure 2-1. Aquistar® TempHion™ Smart Sensor	3
Figure 2-2. Sensor Collection and Data Transmission Schematic	3
Figure 3-1. Laboratory Test Cell Configuration	6
Figure 3-2. Schematic Showing Well and Sensor Layout in Bioreactor	7
Figure 3-3. Figure Showing Nitrate Sensors in Bioreactor	8
Figure 3-4. Nitrate Sensor with attached Conventional Sample Tubing	8
Figure 6-1. Time Series Graph of Concentrations in Well 1 (Inlet)	30
Figure 6-2. Plume Map Showing Dissolved Nitrate Concentrations in Test Cell	30
                                       Tables

Table 3-1. Summary of Nitrate Sensor Laboratory Testing	6
Table 3-2. Summary of Nitrate Sensor Verification  Samples	9
Table 3-3. Summary of Background Water Quality	11
Table 3-4. Verification Schedule	11
Table 4-1. Summary of Laboratory and Field QA/QC Samples for Nitrate Results	14
Table 4-2. Summary of PE Audit Results	16
Table 6-1. Summary of Phase 1 t-Test and MAE Results	22
Table 6-2. Summary of Phase 2 t-Test and MAE Results for all Sensors Combined	23
Table 6-3. Summary of Field Testing t-Test and MAE Results for all Wells	23
Table 6-4. Well-Specific MAE Estimates	23
Table 6-5. Summary of RPD for Laboratory Test for All Nitrate Concentrations	26
Table 6-6. Summary of RPD for Field Test	26
Table 6-7. Summary of Variability in Phase 1 Laboratory Test	27
Table 6-8. Summary of Duplication Results in Laboratory Test	28
Table 6-9. Summary of the Effects of Changes in Water Quality 	28
                                          VI

-------
                                   Appendices

Appendix 1: Well-Specific Time Series Nitrate Concentration Plots	34
Appendix 2: Accuracy Data Plots	38
                                        vn

-------
                               List of Abbreviations
AMS        Advanced Monitoring Systems
ANOVA     analysis of variance
ARS         Agricultural Research Service

bgs          below ground surface

CCC         continuing calibration check

DI           deionized

ECC         end calibration check
EPA         U.S. Environmental Protection Agency
ETV         Environmental Technology Verification

1C           ion chromatography
ICAL        initial calibration
ICC         initial calibration check
INW         Instrumentation Northwest, Inc.
ISE          ion-selective electrode

LFB         laboratory fortified blank
LFSM       laboratory fortified sample matrix
LRB         laboratory reagent blank

MAE        mean absolute error
MDL        method detection limit
MS/MSD     matrix spike/matrix spike duplicate
MSE        mean square error

NELAC      National Environmental Laboratory Accreditation Conference
NJDEP      New Jersey Department of Environmental Protection
NRMRL     National Risk Management Research Laboratory
NTU        nephelometric turbidity unit

PVC         polyvinyl chloride

QA          quality assurance
QC          quality control
QCS         quality control sample
QMP        Quality Management Plan

RFIC        reagent-free ion chromatography
RPD         relative percent difference
                                         Vlll

-------
TQAP        Test/QA Plan
TSA         technical systems audit

UCL         upper confidence limit
USD A       U. S. Department of Agriculture

VTC         Verification Test Coordinator
                                         IX

-------
                                      Chapter 1
                                     Background


The U.S. Environmental Protection Agency (EPA) supports the Environmental Technology
Verification (ETV) Program to facilitate the deployment of innovative environmental
technologies through performance verification and dissemination of information. The goal of the
ETV Program is to further environmental protection by accelerating the acceptance and use of
improved and cost-effective technologies. ETV seeks to achieve this goal by providing high-
quality, peer-reviewed data on technology performance to those involved in the design,
distribution, financing, permitting, purchase, and use of environmental technologies.

ETV works in partnership with recognized testing organizations; with stakeholder groups
consisting of buyers, vendor organizations, and permitters; and with the full participation of
individual technology developers. The program evaluates the performance of innovative
technologies by developing test plans that are responsive to the needs of stakeholders,
conducting field or laboratory tests (as appropriate), collecting and analyzing data, and preparing
peer-reviewed reports. All evaluations are conducted in accordance with rigorous quality
assurance (QA) protocols to ensure that data of known and adequate quality are generated and
that the results are defensible.

EPA's National Risk Management Research Laboratory (NRMRL) and its verification
organization partner, Battelle, operate the Advanced Monitoring Systems (AMS) Center under
ETV. The AMS Center recently evaluated the performance of Instrumentation Northwest, Inc.'s
(INW's) nitrate-specific ion-selective electrode (ISE) for measuring nitrate concentrations in
groundwater.  This evaluation was carried out in collaboration with the U.S. Department of
Agriculture / Agricultural Research Service (USDA/ARS) National Laboratory for Agriculture
and the Environment.

-------
                                      Chapter 2
                              Technology Description


The objective of the ETV AMS Center is to verify the performance characteristics of
environmental monitoring technologies for air, water, and soil. This verification report provides
results for the verification testing of INW's Aquistar® TempHion™ Smart Sensor and
Datalogger nitrate-specific ISE. The following description of the sensor is based  on information
provided by the vendor. This section describes INW's Aquistar® TempHion™ Smart Sensor and
Datalogger, which can be outfitted with sensors for temperature, pH, specific ions (chloride,
bromide, or nitrate), or redox elements.  The products used in this verification test were outfitted
only with nitrate ISEs and temperature electrodes, and the test focused only on the nitrate ISE.
The following technology description was provided by the vendor and was not verified in this
test.

The Aquistar® TempHion™ Smart Sensor is a submersible water quality sensor and datalogger
capable of measuring and recording pH, specific ions, redox, level and temperature. Each unit
comes with a thermistor-based temperature element and a pressure/level element, with the option
of adding up to three pH, ISE, or redox  elements. The TempHion™ Smart Sensor logs data,
operates on low power, and comes with its own software. The sensor has two digital output
protocols, Modbus or Sdil2; both  options are license-free digital communication  languages.
Several TempHion™ sensors, or a combination of TempHion™ sensors and other INW  Smart
Sensors, can be  networked together and controlled from one location, either directly from a
computer or through a WaveData® Wireless Data Collection  System.  The sensor used in this
verification test is shown in Figure 2-1.  Data were collected  automatically using  a cellular
modem link, INW's auto-collection program (Aqua4Push), and Groundswell Technology's
visualization software.  A schematic showing the data collection and transmission process is
shown in Figure 2-2.

The TempHion™ Smart Sensor can be powered internally with two AA alkaline batteries, or
with an auxiliary power supply for data intensive applications. The unit is programmed  using a
laptop or desktop Windows®-based computer via its RS485/RS232 or USB port and INW's
Aqua4Plus utility software.  Once programmed, the unit will  measure and collect data internally
on a variety of time intervals. The internal processor in the TempHion™ Smart Sensor allows
for calibration using the calibration utilities in INW's Aqua4Plus software.  Once calibrated, the
calibration data  are stored in non-volatile memory within the Smart Sensor and are applied to the
collected data.

-------
                              .Top-cap/cable
                              harness
                                 2 AA batteries

                                 Smart Sensor recording
                                 circuit board 5-channels
                                 with digital output
                                 Nitrate
                                 module
               Embedded
               Thermistor

                Patented
                reference
                electrode
        Figure 2-1. Aquistar® TempHion™ Smart Sensor
          Data  Flow Overview
Located at EPA ETV Site
Located Off Site
                    Copyright 2010 Groundswell Technologies
  Figure 2-2. Sensor Collection and Data Transmission Schematic

-------
                                      Chapter 3
                            Test Design and Procedures
3.1    Introduction

This verification test was conducted over a nine-week period beginning in April 2010 and ending
in July 2010, according to procedures specified in the Test/QA Plan (TQAP) for Verification of
Nitrate Sensors for Groundwater Remediation Monitoring (1). As indicated in the test/QA plan,
the testing conducted satisfied EPA QA Category III requirements. The test/QA plan and/or this
verification report were reviewed by:
       •  Stu Nagourney, NJDEP
       •  Kenneth Wood, DuPont
       •  Michael Brody, U. S. EPA
       •  Charles Spooner, U.S. EPA (test/QA plan only)
       •  Jacob Gibs, U.S. Geological Survey (report only).

The verification was based on comparing the nitrate concentration results from the Aquistar®
TempHion™ nitrate ISE to those from a laboratory-based reference method. The reference
method for nitrate analysis was ion chromatography (1C), performed by USDA/ARS according
to EPA Method 300.1 "Determination of Inorganic Anions by Ion Chromatography" (2) (see
Section 3.2). The nitrate sensors were calibrated using a one- or two-point calibration method
through INW's proprietary Aqua4Plus software for Microsoft Windows.  The nitrate sensors
were verified in the laboratory by challenging the sensors with solutions of known nitrate
concentrations with and without the addition of selected interference parameters. The sensors
were then deployed in the field for nine weeks. Sensor output was verified against groundwater
samples collected and analyzed using the EPA laboratory method.

3.2    Nitrate Analysis Reference Method

All conventional groundwater samples collected were analyzed following the EPA laboratory
method for determination of nitrate. Samples were collected in 125 mL plastic containers that
were rinsed with deionized (DI) water, preserved by refrigeration to ± 2°C, and analyzed within
48 hours of collection. The collection of the samples was the responsibility of USD A and
Battelle staff. For the reference analysis, a Dionex ICS-2000 Reagent-Free Ion Chromatography
(RFIC) System was operated by USDA staff according to instrument procedures (see Appendix
E of the TQAP) and the manufacturers' instructions, including those for warm-up and
stabilization time before testing. The USDA laboratory was responsible for coordinating the
analysis of the samples with associated QA/quality control (QC). Calibration and maintenance

-------
documentation for the Dionex ICS-2000 and all results of the reference analyses were provided
as part of the data dissemination process.  A laboratory audit addressing 1C data collection was
performed by the NJDEP (see Section 4.2.2) according to guidelines provided by the 2003
National Environmental Laboratory Accreditation Conference (NELAC) Standard.

3.3    Test Design

INW's nitrate sensor was verified based on the following performance parameters:

       •  Accuracy
       •  Variability of readings
       •  Duplication of readings
       •  Effect of nitrite, turbidity, and chloride on nitrate sensor readings
       •  Operational and sustainability factors, including ease of use, downloading of data,
          timely dissemination of data, and environmental impacts of using nitrate sensors for
          real-time remote data collection.

The verification test involved two separate stages: a laboratory testing stage in which sensors
were challenged with known nitrate concentrations and interference parameter concentrations,
and a field application stage in which several sensors were placed in monitoring wells and
streamed data to a remote server. Nitrate sensor concentration data were compared to laboratory
1C data to determine  a number of verification parameters including accuracy and variability.

3.3.1   Laboratory Testing Stage

The laboratory stage  of the verification test was performed in the USDA/ARS laboratory, and
involved challenging the nitrate sensors by measuring solutions of known nitrate concentrations
in two clear polyvinyl chloride (PVC) test cells measuring 4-ft high with a 2-inch diameter
(Figure 3-1). One test cell contained a single nitrate sensor suspended approximately 6 inches
below the base of the test cell, whereas the other test cell contained two duplicate sensors
suspended at the same depth as in the first test cell.

During Phase 1 of the laboratory stage, nitrate solutions of known concentration were added to
the test cells and sensors were programmed to begin collecting data in one-minute intervals for a
predetermined period of time (20 minutes) for each nitrate concentration.  After sensor data
collection, a water sample for 1C analysis was collected from each cell through the attached
stopcock located at the base of the test cell.  Phase 2 of the experiment followed the same
methods, but interference parameters (chloride, nitrite, and turbidity) at varying concentrations
were also added to the nitrate solutions.  The sensors were programmed to collect readings  at one
minute intervals for a period of 10 minutes for each nitrate/interference parameter concentration
combination performed during Phase 2. Further details  of the experimental  design for the
laboratory stage are located in the Test/QA Plan for Verification of Nitrate Sensors for
Groundwater Remediation Monitor ing (1). Table 3-1 provides a summary of the nitrate and
interference parameter concentrations used in the laboratory testing.

-------
                      Figure 3-1.  Laboratory Test Cell Configuration
                 Table 3-1.  Summary of Nitrate Sensor Laboratory Testing
Phase
1
(Nitrate Only)
2
(Nitrate Plus
Interference)
Interference
Parameter
None
Chloride
Nitrite-N
Turbidity
Interference
Parameter
Concentration
Chloride = ND
Nitrate-N = ND
Turbidity = ND
lOOmg/L
500mg/L
2,500 mg/L
Img/L
2 mg/L
4 mg/L
1NTU
5NTU
Nitrate-N Concentration" (mg/L)
1.0 (4.4)
1.0 (4.4)
1.0 (4.4)
1.0 (4.4)
3.0(13)
3.0(13)
3.0(13)
3.0(13)
6.0 (26)
12 (53)
12 (53)
12 (53)
12 (53)
-
-
-
a:  Equivalent nitrate concentrations in parentheses
NTU = nephelometric turbidity unit
ND = non detect
3.3.2  Field Testing Stage
Field testing consisted of nitrate sensor deployment in an end-of-tile bioreactor located at the
Kelly Farm research site in Ames, IA.  The bioreactor is an excavated below-grade cavity filled

-------
with wood chips at the downstream end of a series of subsurface tiles that are used to promote
drainage in the surrounding agricultural area. The tile drainage water is routed through the
bioreactor, where the wood chips naturally support populations of microorganisms that remove
the nitrate through denitrification.  The bioreactor contains inlet and outlet piping, and seven
monitoring wells that are screened from 2 to 6 feet below ground surface (bgs) and used to
monitor water quality (see Figure 3-2).
                                      I
                                     Inlet
   Welll (Inlet)
    Nitrate Sensor ID: W1_ISE(3)
    Sensor Depth: 4 ft bgs
   Well 2
    Nitrate Sensor ID: W2_ISE(5)
    Sensor Depth: 4 ft bgs
   WellS
    Nitrate Sensor IDs: W3_ISE(7), W3_ISE2(14)
    Sensor Depths: 3 ft bgs, 5 ft bgs
   Well 4
    Nitrate Sensor ID: W4_ISE(9)
    Sensor Depth: 4 ft bgs


   WellS
    Nitrate Sensor ID: W5_ISE(11)
    Sensor Depth: 4 ft bgs

   Well6
    Nitrate Sensor ID: W6_ISE(13)
    Sensor Depth: 4 ft bgs

                                             Welll
         Wei I 2
                                             Well 3 (Shallowand Deep Sensor)
         Wei 14, WellS
         Wei I 6
Outlet
                              Wells with Sensors
                              Well without Sensors
Not to scale
                                           North
           Figure 3-2.  Schematic Showing Well and Sensor Layout in Bioreactor
Sensors measured continuous nitrate concentrations in 15-minute intervals from seven locations
within the bioreactor (Figures 3-2 and 3-3) for a period of nine weeks.  Sensors were deployed in
the inlet to the bioreactor, and in four 2-inch PVC monitoring wells (two sensors were deployed
in one of the wells at varying depths to evaluate vertical gradients within the test cell) (see Figure
3-2 for sensor deployment depths).  Data were transmitted wirelessly to the vendor's server, and
then forwarded to a Web site for download.  Conventional groundwater samples were collected
weekly for the nine-week deployment period using a low-flow purging technique, whereby
dedicated tubing was installed in each well and attached to the sensor with a zip tie so that the
tubing inlet was located at the same depth where the sensor reading was collected (see Figure 3-
4). In addition  to the weekly monitoring, two days of intensive conventional sampling events
                                             7

-------
were performed at the beginning and end of the test, during which samples were collected once
per hour for eight hours. It should be noted that although nitrate sensors were deployed in seven
locations, only four of the locations (Wells 1, 2, 3 [shallow and deep], and 6) were used for data
evaluation, as outlined in the TQAP. In addition, water-level sensors were placed in each of the
monitored wells. Data from the additional sensors and from the additional wells were used for a
broader evaluation performed by NJDEP to evaluate the spatial distribution of water quality
within the test cell, and was conducted separately from this ETV test; accordingly, the data
generated from these sensors were not evaluated in this verification report.
                Figure 3-3. Figure Showing Nitrate Sensors in Bioreactor
          Figure 3-4.  Nitrate Sensor with attached Conventional Sample Tubing

-------
3.4    Test Procedures
Comparisons were made between nitrate concentrations measured using the nitrate sensor and
those measured in the laboratory using the EPA laboratory method from analysis of samples
collected using the conventional groundwater sampling technique. It is assumed that the nitrate
concentration value in the sample collected using the conventional sampling technique
represented the actual or target nitrate concentration present in the well against which the nitrate
sensor concentration was being evaluated. Table 3-2 summarizes the types and numbers of
samples that were used to verify the performance of INW's nitrate sensors. The test procedures
used to evaluate the performance of the nitrate sensors are presented in the following
subsections.

                Table 3-2. Summary of Nitrate Sensor Verification Samples
Sample Type
Phase 1 laboratory
water samples
Phase 2 laboratory
water samples
Field groundwater
samples
User observations
Approximate
Number of Samples
or Readings
80
240
200
All
Associated QC
Samples
Equipment rinsates
"Field" duplicates
Laboratory QA/QC
Equipment rinsates
"Field" duplicates
Laboratory QA/QC
Equipment rinsates
Field duplicates
Laboratory QA/QC
Not Applicable
Uses
Accuracy, variability,
duplication, user
agreement, operational
factors
Accuracy, duplication,
effect of changes in
water quality, user
agreement, operational
factors
Accuracy, effect of
changes in water
quality, user agreement,
operational factors
Operational factors
3.4.1   Accuracy

Prior to deployment and testing, each nitrate sensor was calibrated by the vendor. Immediately
after calibration, the sensor was programmed to take a few readings while the sensor was still in
the reference standard to verify the accuracy of the initial calibration. The accuracy of the nitrate
sensor in the field and in the laboratory was determined by comparing nitrate sensor readings to
simultaneous measurements made using conventional (low flow) groundwater sampling
techniques.  The comparison of accuracy was made statistically and graphically by plotting
nitrate sensor readings (concentrations) against the nitrate concentrations measured in
groundwater samples collected using conventional techniques.

In the laboratory testing (see Tables 3-1 and 3-2), nitrate concentrations were generated from a
concentrated stock solution spanning the range of anticipated field concentrations,  and evaluated
with the nitrate sensors and conventional EPA method sample analysis.  Additionally, in Phase 2
of the laboratory testing, concentrations of interference parameters (chloride, nitrite, and
turbidity) were introduced into the test cells to evaluate the ability of the nitrate sensor to
accurately measure nitrate concentrations under simulated field conditions.

-------
3.4.2  Variability

Variability of nitrate sensor concentration readings refers to the consistency in reported nitrate
concentrations. Variability was assessed in Phase 1 of the laboratory evaluation using 20
readings made by each of three sensors deployed in separate test cells at four reference nitrate
concentrations (see Table 3-1). Variability was equated to drift and expressed as percentage
change in concentration as a function of time compared to the reference concentrations.

3.4.3  Duplication

The degree of agreement of nitrate concentrations reported simultaneously using duplicate nitrate
sensors was assessed in the laboratory in the two test cells.  As discussed in Section 3.2.1, one
test cell housed two nitrate sensors to evaluate intra-well duplication within the test cell, and a
second test cell housed one nitrate sensor to evaluate inter-well duplication between the two test
cells. The three nitrate sensors were synchronized and programmed to record nitrate
concentrations at one-minute intervals throughout Phase 1 and Phase  2 of the laboratory test to
directly compare nitrate concentration data.

3.4.4  Effect of Changes in Water Quality

The effect of water quality (i.e., interference parameters) on nitrate sensor response to nitrate
concentrations was evaluated in Phase 2 of the laboratory testing by exposing nitrate sensors to
constant nitrate concentrations under different water quality conditions. The laboratory testing
schedule is described in Section 3.2.1 and summarized in Table 3-1.  The ability of the nitrate
sensors to accurately measure nitrate concentrations was evaluated under each of the 24 different
scenarios. In addition to the laboratory testing, conventional groundwater field and associated
QC samples were analyzed for the presence and level of nitrite as nitrogen (nitrite-N) until
negligible concentrations (<1 mg/L) were verified in successive monitoring events. An initial
sampling event also was  conducted at each well in the test cell prior to sensor deployment to
evaluate background concentrations of nitrate, nitrite, and chloride in groundwater (see Section
3.5).

3.4.5  Operational and Sustainability  Factors

Operational factors associated with use of the nitrate sensors were evaluated based on the
comments and observations of verification test staff (i.e., Battelle and USD A) in laboratory and
field testing.  Such observations addressed the convenience and environmental impact of using
the nitrate sensors, the completeness of nitrate sensor readings (percent data collected), their
reliability under differing conditions, the apparent consistency of nitrate sensor readings, and
acceptability as a groundwater monitoring tool. Observations also included any noted biofouling
at the end of the field testing period. In addition, data dissemination  was evaluated, including
ease of data transmission, timeliness of data dissemination, ease of data downloading, and
usability of downloaded data.  Cost for the nitrate sensor and associated data transmission
equipment also was reported.
                                            10

-------
3.5    Analysis of Baseline Concentrations
Prior to the field verification testing, conventional groundwater samples were collected by field
personnel from Battelle and USDA from each location within the test cell to evaluate
background nitrate concentrations. In addition, the samples were analyzed for nitrite and
chloride to better understand the background water quality. The background laboratory analyses
were performed by USDA.  The results from these analyses are summarized in Table 3-3.  As
discussed in Section 3.4.4, groundwater samples were analyzed for nitrate during the initial
stages of the field test to understand baseline levels before starting the field testing.  Because
nitrite concentrations were consistently well below the 1 mg/L threshold throughout the first day
of intensive sampling, continued monitoring for nitrite was unnecessary.

                     Table 3-3.  Summary of Background Water Quality
Sampling Location
Well 1 (Inlet)
Well 2
Well 3 S
Well 3D
Well 6
Nitrate (mg/L)
8.59
8.54
8.51
7.82
8.61
Nitrite (mg/L)
0.01
0.02
0.02
0.02
0.03
Chloride (mg/L)
14.27
14.47
14.20
14.02
14.17
3.6    Verification Schedule

Table 3-4 summarizes the schedule for verification testing, data analysis, and reporting.

                              Table 3-4. Verification Schedule
                 Date
                 Verification Activity
            April 19-22, 2010     Completed Performance Evaluation Audit
            April 26-29, 2010
Completed Phase 1 and Phase 2 of the laboratory evaluation
Completed Technical Systems Audit
Installed nitrate sensors in test cell
Collected initial groundwater samples from test cells for analysis
  of nitrate and potential interference parameters
Began field test
Completed first day of initial intensive sampling event
April 29, 2010
through
July 12, 2010
May 3, 2010
July 13-14, 2010
September 2, 2010
September 8-22, 2010
Performed field test
Completed two Audits of Data Quality
Completed second day of initial intensive sampling event
Completed final intensive sampling event
Completed final Audit of Data Quality
Peer review of draft report
                                             11

-------
                                      Chapter 4
                        Quality Assurance/Quality Control


QA/QC procedures were performed in accordance with the Quality Management Plan (QMP)for
the ETV Advanced Monitoring Systems Center (3) and the TQAP for this verification test (1).
QA/QC procedures and results are described below.

4.1    Laboratory Sample Analysis QA/QC

Quality of the laboratory reference nitrate measurements were ensured by a calibration of the
Dionex ICS-2000 RFIC before testing began. A pre-testing calibration curve was prepared for
each analytical run; the curve was required to be linear with the coefficient of determination (R2)
greater than or equal to 0.995 before proceeding with analysis.  Calibration was verified
throughout the analytical run by inserting calibration check standards and reagent blanks with
every set of 10 samples.  The calibration and all verifications are incorporated into the run
alongside the samples and visually evaluated by the instrument operator to meet the reference
laboratory's QC criteria. A complete description of the USDA's current Dionex ICS-2000 RFIC
analytical procedures including equipment, standards, reagents, and calibration is included in
Appendix E of the TQAP.  The following subsections summarize the results  of the laboratory
sample analysis QA/QC procedures.

4.1.1   Instrument Calibration Checks

Instrument calibration checks were performed on each batch of samples submitted for laboratory
analysis. A nitrate-N standard concentration of 10 or 20  mg/L was used for instrument
calibration. Three separate calibration checks were performed, one during the initial portion of
the laboratory run (initial calibration check [ICC]), one during the laboratory run (continuing
calibration  check [CCC]), and one near the end of the laboratory run (end calibration check
[ECC]).  If the determined concentrations were not within 90% to 110% of the stated values,
performance of the determinative step of the method was unacceptable and would be repeated.
The results of the instrument calibration checks are summarized in Table 4-1, and indicate that
the instrument calibration check QC criteria were met for all samples.

4.1.2   Initial Calibration Checks

To establish the ability to generate acceptable precision results, the laboratory analyzed 10
replicates of a mid-range standard within the range of anticipated field concentrations as an
initial calibration (ICAL) check. The results of the replicates were used to compute the average
percent recovery and the standard deviation for the analyte. A linear calibration curve with the
                                           12

-------
R2 greater than or equal to 0.995 is required for acceptance. The results of the ICAL checks are
summarized in Table 4-1, and indicate that the ICAL QC criteria were met.

4.1.3   Laboratory Reagent Blanks

A laboratory reagent blank (LRB), consisting of filtered DI water, was included in each
laboratory batch run. Although the acceptance criteria for the LRB were not defined in the
TQAP or in the laboratory protocol,  discussions with USDA ARS personnel indicated that QC
criteria were met if LRB concentrations were below the nitrate-N method detection limit (MDL)
(<0.3 mg/L).  The LRB results are summarized in Table 4-1, and indicate that the LRB QC
criteria were met with the exception of the absence of LRB samples in the third and eighth field
sampling events.

4.1.4   Laboratory Fortified Blanks

A laboratory fortified blank (LFB), consisting of filtered DI water spiked to a nitrate-N
concentration of 20 mg/L, was included in each laboratory batch run. QC criteria were met if the
determined concentrations were within 85% to 115% of the stated value.  The LFB results are
summarized in Table 4-1, and indicate that the LFB QC criteria were met.

4.1.5   Laboratory Fortified Sample Matrix

Laboratory fortified sample matrix (LFSM), or matrix spike/matrix spike duplicate (MS/MSD)
samples, were prepared and analyzed at a rate of 5% of the total number of samples.  For this
analysis, the selected field water sample was divided after filtering and aliquotted into Dionex
PolyVials and stored in the refrigerator. Following analysis to determine the background
concentration, these reserved samples were spiked with a concentrated solution of nitrate-N to
achieve concentrations 2 to 3 times above background with a minimal (<2%) change in sample
volume. QC criteria were met if the determined recovery was within 75% to 125% of the stated
value.  The LFSM (MS/MSD) results are  summarized in Table 4-1, and indicate that the LFSM
QC criteria were  met.

4.1.6   Laboratory Duplicate Samples

Duplicate analyses were performed in 10% of the field samples, with QC criteria being met if the
relative percent difference (RPD) was ±10%.  The laboratory duplicate sample  results are
summarized in Table 4-1, and indicate that the laboratory duplicate sample QC criteria were met
for all samples.   It should be noted that the concentrations of many of the laboratory duplicate
samples collected during the final intensive sampling events were well below the MDL, so the
RPD was not calculated.
                                          13

-------
               Table 4-1. Summary of Laboratory and Field QA/QC Samples for Nitrate Results
Sampling Event
Phase 1 Laboratory Test
Phase 2 Laboratory Test
Initial Intensive Sampling Event
Day 1
Initial Intensive Sampling Event
Day 2
Field Sampling Event 1
Field Sampling Event 2
Field Sampling Event 3
Field Sampling Event 4
Field Sampling Event 5
Field Sampling Event 6
Field Sampling Event 7
Field Sampling Event 8
Field Sampling Event 9
Final Intensive Sampling Event
Day 1
Final Intensive Sampling Event
Day 2
Date
04/27/10
04/28/10
04/29/10
05/03/10
05/10/10
05/17/10
05/24/10
06/01/10
06/07/10
06/17/10
06/22/10
06/28/10
07/06/10
07/13/10
07/14/10
Laboratory QA/QC
Instrument Calibration
Check Result (%)
ICC
99.6
100
98.8
101
99.5
99.6
100
99.8
100
100
101
100
100
100
100
100
100
ccc
99.7
101
99.6
100
101
99.3
101
101
100
101
101
101
102
99.6
99.7
100
100
101
100
100
101
101
101
101
100
99.6
100
101
101
100
ECC
101
101
100
101
99.0
101
99.7
100
101
101
99.9
100
101
101
101
99.9
102
ICAL
(R2 Value)
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
LRB
(mg/L)
0.005
0.007
0.006
0.005
0.006
0.005
0.008
NA
0.004
0.001
0.009
0.005
NA
0.033
0.001
0.001
0.001
LFB
(%)
99.8
99.8
100
100
100
101
99.8
99.7
101
100
99.7
101
101
101
100
99.8
99.5
99.4
99.7
99.9
99.6
99.5
99.9
99.8
99.8
99.5
99.9
100
99.9
100
101
101
101
LFSM
(%)
102
93.8
102
101
102
101
98.5
101
103
103
102
101
101
102
113
102
103
99.2
101
101
101
101
101
101
100
100
99.8
100
Laboratory
Duplicate
(RPD)
0.05
2.8
0.42
0.80
7.1
2.0
0.77
3.8
0.43
3.7
0.03
2.1
0.04
0.58
2.4
0.59
0.21
2.4
4.4
8.2
0.03
1.9
0.02
0.09
5.0
NC
NC
NC
NC
NC
NC
0.19
Field QA/QC
Field
Duplicate
(RPD)
NS
0.03 0.17
2.6 0.49
0.19 4.6
3.3 0.47
0.58 4.6
0.24
1.3
3.8
4.2
0.17
1.1
NS
NS
NS
3.1
NS
0.04
0.62
NS
NS
14
NC
NC
NC
NC
NC
Rinsate
Blank
(mg/L)
0.02
0.01
0.03
0.02
4.1
2.4
0.03
0.01
0.01
0.01
0.02
0.02
0.02
0.04
0.01
<0.01
NA - not analyzed         NS - no sample collected
NC - not calculated; laboratory duplicate concentrations were at or below the method detection limit.
                                                          14

-------
4.2    Field QA/QC

During field and laboratory groundwater sampling activities, QC samples, including field
duplicates and equipment blanks, were collected to ensure the reliability of field data.  The field
QC samples are discussed in the following sections.

4.2.1   Field Duplicate Samples

Duplicate groundwater samples were collected at a frequency of one for every 10 samples (i.e.,
10%) to evaluate the reproducibility of analytical results. If 10 samples were not collected
during a sampling event, one duplicate sample per sampling event was collected. Duplicate
samples were collected simultaneously with the original sample into identical sample containers.
The QC criteria were met if the RPD was ±10%.  The field duplicate sample results are
summarized in Table 4-1, and indicate that the QC criteria for sample collection frequency was
not met on several occasions, and the RPD was slightly exceeded on one occasion.  It should be
noted that the concentrations of many of the field duplicate samples collected during the final
intensive sampling events were well below the MDL, so the RPD was not calculated.

4.2.2   Equipment Blanks

Equipment blanks, also referred to  as rinsate blanks, were collected to evaluate the potential for
sample cross-contamination from the sampling equipment used. Equipment rinsate blanks were
collected daily during sampling to  ensure that nondedicated groundwater sampling equipment
had been decontaminated effectively.  Daily equipment blanks were collected after collection of
at least one field sample and after the equipment was decontaminated. The equipment blank for
groundwater sampling equipment was laboratory-provided DI water that was passed through or
over the sampling equipment used  to collect samples (i.e., Teflon® polyethylene tubing).  The
QC criteria were met if the analytical results from the equipment blank sample were <2 mg/L
nitrate-N. The equipment rinsate sample results are summarized in Table 4-1, and indicate that
the QC criteria were not met on two occasions near the beginning of the field study, and based
on the elevated concentration, may be indicative of improper rinsate blank sample collection.

4.3    Audits

Three types of audits were performed during the verification test: a performance evaluation (PE)
audit of the laboratory analysis method,  a technical systems audit (TSA) of the verification test
performance, and three data quality audits. Audit procedures are described further in the
following subsections.

4.3.1   Performance Evaluation Audits

A PE audit was performed to confirm the accuracy of the laboratory nitrate analysis reference
method.  Prior to the laboratory and field investigations, five blind  samples of varying nitrate
concentrations within the range of  anticipated field concentrations were generated from a stock
solution and shipped to the USDA  ARS  laboratory on 19 April 2010 for analysis on 21 April
2010. Table 4-2 summarizes the results  of the PE audit of laboratory nitrate analysis reference
method, showing the stock solution generated nitrate concentration, the laboratory 1C nitrate
concentration,  and the RPD between the two concentrations. Table 4-2 shows that all of the
RPD values for generated nitrate concentrations were below 4%, and  a graphical plot of the data
                                           15

-------
indicated an R2 value of 1.00  The PE audit results were within the target RPD tolerances of
10% set forth in Appendix E of the TQAP.
                        Table 4-2. Summary of PE Audit Results
Generated Nitrate-N
Concentration
(mg/L as N)
6.0
3.0
12
0
1.0
Laboratory 1C Result
(mg/L Nitrate-N)
6.09
3.03
12.5
0.026
1.03
RPD
1.5
1.0
3.8
Not applicable
o o
3.3
4.3.2   Technical Systems Audit

A Quality Auditor from the NJDEP (Amy Bowman) conducted a ISA at the USDA ARS
laboratory and field test site on 27-29 April 2010 to ensure that the verification test was being
conducted in accordance with the TQAP (1) and the AMS Center QMP (3).  This audit was
designed to achieve the following objectives:

       •  Evaluate all activities related to the installation and verification testing of the nitrate
          sensors

       •  Review laboratory elements of the TQAP for Verification of Nitrate Sensors for
          Groundwater Remediation Monitoring, prepared by the ETV Program

       •  Assess data from Performance Evaluation samples analyzed by 1C
       •  Audit laboratory operations and 1C instrument operations at the USDA ARS research
          facility for method and QA/QC compliance.

During this TSA, the NJDEP Quality Auditor performed  a review of 1C data and operations.
Issues related to run logs and inclusion of the required QC samples were reviewed with the
laboratory analyst, and the run logs were revised to include the required QC samples. In
addition, a review of the laboratory sample receipt procedures was performed, and sensor testing
and data recording were observed in the laboratory. The  initial groundwater sampling event (for
collection of background nitrate and interference parameter concentrations) was observed,  as was
a partial installation of a nitrate sensor array into one of the field monitoring points.  During
sensor installation, the vendor was interviewed about sensor installation and calibration in the
field.

The TSA of both the laboratory and field testing portions resulted in eight findings and nine
observations. The corrective actions taken in response to significant findings of the TSA were as
follows;

       •  Inclusion of a temperature blank to measure sample temperature upon sample receipt
          in the laboratory
                                           16

-------
       •  Revision of the initial intensive sampling schedule from a Thursday/Friday schedule
          to a Thursday/Tuesday schedule to ensure the sample holding time was not exceeded

       •  Revision of the laboratory run logs to include QC samples required by the analytical
          method.

The remaining findings and the observations noted documentation errors, need for improvements
to the manner in which samples were processed, and QC sample frequency deficiencies. The
findings and observations were discussed onsite with the field team and subsequently with the
verification test coordinator (VTC) and project team via a conference call; immediate changes
based on the discussed improvements were implemented.

A TSA report was prepared, and a copy was distributed to the EPA.

4.3.3   Data Quality Audits

Records generated in the verification test received a review from a technical person independent
of the person generating the data before these records were used to calculate, evaluate, or report
verification results.  Data were reviewed by a Battelle technical  staff member involved in the
verification test. The person performing the review added his/her initials and the date to a  hard
copy of the record being reviewed.

All of the verification test data were  reviewed for quality by the VTC, and at least 10% of the
data acquired during the verification test were audited.  The data were traced from the initial
acquisition, through reduction and statistical analysis, to final reporting to ensure the integrity of
the reported results. All calculations performed on the data undergoing the audit were checked.

Three data quality audits were performed. The  first data quality audit, which covered the
laboratory test investigation, resulted in 11 findings, five observations, and two
recommendations.  The first audit results were related to laboratory and field QC procedures,
laboratory reporting and documentation issues,  laboratory instrument calibration procedures, and
data transcription errors. The second data quality audit,  which covered the first intensive
sampling event, resulted in three findings and three observations. The second audit results were
related to laboratory and field QC procedures and documentation errors.  The third data quality
audit, which  covered the weekly field sampling events and the second intensive sampling  event,
resulted in four findings and two observations.  The third audit results related to laboratory and
field QC procedures and documentation errors.  The data quality audit issues were addressed
with procedural changes and references to the TSA audit recommendations that were
implemented to refine the field and laboratory QC procedures.

Three data quality audit reports were prepared, and copies were  distributed to the EPA.
                                           17

-------
                                       Chapter 5
                                  Statistical Methods
The statistical methods used to evaluate the quantitative performance factors listed in Section 3.1
are presented in this chapter. The methods described below are consistent with those outlined in
the approved TQAP; the additional methods not outlined in the TQAP were selected by a
Battelle statistician to provide an additional data evaluation approach, and were based on several
iterations of representative statistical methods.  Qualitative observations also were used to
evaluate verification test data.

5.1    Accuracy

Accuracy was determined by comparing nitrate sensor readings to water samples collected
during laboratory and field testing.  The water samples were analyzed in the laboratory using
EPA-approved analysis methods. The accuracy of the nitrate sensor concentrations with respect
to the laboratory measured concentrations was assessed graphically and by evaluating the
differences between paired concentrations (concentration residuals) from measurements
collected simultaneously at the same location. The nitrate sensor concentration reading (reported
every 15 minutes) collected closest in time to the collection of the reference monitoring sample
was initially used for paired comparison.

Two statistical measurements were used to assess the accuracy: (1) inference about the mean
difference,  and (2) estimation of the mean absolute error (MAE). The inference about the
observed difference included estimation of the mean difference and a statistical hypothesis about
whether the mean difference was equal to or different from zero (using a paired-sample t-test).
The hypothesis tests were conducted on the natural log of the concentration data in order to more
closely approximate the normal or Gaussian distribution and thereby satisfy the assumptions of
the t-test. The log of the laboratory concentration was subtracted from the log of the sensor
concentration and the null hypothesis was that the resulting difference had a mean of zero. If the
p-value of the hypothesis test was below 0.05, the null hypothesis was rejected and there was
strong evidence to suggest that the sensor and laboratory concentrations were not  equal. If the  p-
value was larger than 0.05, there was not strong evidence to reject the null hypothesis.  (It should
be noted that even when there are no differences in two underlying population means,
differences in random  samples drawn from those populations [due only to sampling error] should
be expected.)

When considering the  p-value for a specific hypothesis test, the p-value is the proportion of times
that a difference of the magnitude observed in these data, or larger, would be expected by chance
due only to sampling error if the null hypothesis were true (the null hypothesis often states that
there is no underlying  difference between the two population means). If the p-value is  smaller
                                            18

-------
than 0.05, the difference is large enough to be expected fewer than five out of every 100
experiments (5/100 = 0.05), even if there are no underlying differences between the two groups.
In these cases, the null hypothesis was rejected and it was concluded that there was noteworthy
evidence of an underlying difference.

The MAE was calculated for the concentration differences as follows:
                       n
                    1V
             MAE = -  >  (laboratory concentrationj — nitrate sensor concentrationj |
where n is the number of paired nitrate concentration measurements. The MAE was used to
represent the average absolute difference in the two measurement methods.

The statistical analyses for accuracy, as outlined in the TQAP (1), called for hypothesis tests
where the null hypothesis was strict equality between the mean sensor and lab measurements.
Based on discussions with personnel involved with sensor technology and nitrate field
monitoring, and the desire to further evaluate the overall objective of this nitrate sensor
evaluation, the null hypothesis was modified to evaluate whether the sensor measurements were,
on average, within a select percentage (i.e., 25%) of the lab results.  To perform this additional
evaluation, the RPD for each sensor reading was computed, and an estimate of the upper
confidence limit for that quantity was made. For example, if the upper confidence limit was
below 25%,  a null hypothesis that says that the average sensor error was smaller than 25% would
not be rejected.  It should be noted that these calculations are not intended to be used for strict
acceptance criteria or for apportioning the variance in observations between different
components of variation. For the purpose of evaluation, an RPD value of <20% was considered
to represent general agreement between sampling methods.  The RPD was calculated as follows,
and assumes that the laboratory concentration is the accepted (benchmark) concentration value
for comparison purposes:
         RPD =
100 x
                       (nitrate sensor concentration — laboratory concentration)
                                       laboratory concentration
The accuracy estimates were calculated separately for Phase 1 and Phase 2 of the laboratory
evaluation, for the two intensive hourly sampling events at the beginning and end of the field
evaluation, and for the weekly sampling conducted during the field evaluation. In addition,
comprehensive accuracy estimates were calculated using all of the paired data sets from the field
and laboratory evaluation.  Well-specific MAE values also were calculated.  Time series plots
showing sensor and conventional monitoring data collected during the field investigation also
were used to evaluate the accuracy.

5.2    Variability

Variability was assessed by observing the spread of nitrate sensor readings made at constant
nitrate concentrations (equated to drift) using stock solutions in the laboratory portion of the
investigation.  Variability of the nitrate sensor concentration readings was evaluated using data
from Phase 1  of the laboratory evaluation using the multiple readings (20) made by each of three
sensors deployed in separate test cells at four reference nitrate concentrations (1, 3, 6, and 12
                                           19

-------
mg/L nitrate-N). Variability was expressed as the standard deviation of the sensor concentration
readings calculated two ways: (1) using the reference concentrations as the average (mean)
values for comparison, and (2) using the measured mean concentrations (geometric mean from
log-transformed results) from the samples.  Standard deviation values were calculated for each of
the four reference concentrations.

5.3    Duplication

Nitrate sensor duplication of readings was assessed by comparing nitrate  sensor readings made
by placing duplicate nitrate sensors in a single test cell in the laboratory so they were exposed to
identical concentrations simultaneously during Phase 1 and Phase 2 of the laboratory tests.  The
degree of agreement of nitrate concentrations reported simultaneously using duplicate nitrate
sensors was assessed in this laboratory evaluation.  The first test cell housed two nitrate sensors
to evaluate intra-well duplication within the test cell, whereas the second test cell housed a single
sensor to evaluate inter-well duplication between the two test cells.  The degree of agreement
between each pair of reported nitrate concentrations (inter-well and intra-well) was assessed by
calculating the intra-well mean square error (MSB) and inter-well MSB using a random-effects
analysis-of-variance (ANOVA) model.

5.4    Effect of Changes in Water Quality

The effect of nitrite, turbidity, and chloride on nitrate sensor readings was assessed in the
laboratory by comparing nitrate sensor readings exposed at constant nitrate concentrations with
varying interference parameter (nitrite, turbidity, and chloride) concentrations. Consistency of
nitrate sensor readings over time was assessed in the accuracy of nitrate sensor readings made in
the field and in the laboratory over time, verified against laboratory analyses. The field testing
evaluation (particularly the readings collected near the end of the field period) served to evaluate
the effects of sensor fouling on the accuracy and duplication performance parameters.
The effect of changes in water quality on nitrate sensor performance was assessed using the data
from Phase 2 of the laboratory evaluation by calculating the accuracy, variability, and
duplication of nitrate sensor readings (Sections 5.1 through 5.3) of the test data at each of the
water quality conditions outlined in  Table 3-1. The results were compared to indicate whether
changes in nitrate, turbidity, and chloride concentrations have any apparent effect on the nitrate
sensor performance at constant nitrate concentrations.  Accuracy or variability results that
differed by more than 5% or nitrate sensor duplication results that differed by more than 20%
were taken as evidence of a significant water quality effect.
                                            20

-------
                                       Chapter 6
                                      Test Results
The statistical methods used to evaluate the quantitative performance factors listed in Section 3.2
are presented in this chapter.  Qualitative observations also were used to evaluate verification test
data.

6.1    Accuracy

As discussed in Section 5.1, several statistical measurements were used to assess the accuracy,
including the paired t-test, and calculation of the MAE. Table 6-1 presents the MAE and t-test
results for Phase 1 of the laboratory investigation and a graphical representation of the data are
presented in Appendix 2. The MAE increases with increasing nitrate concentrations, ranging
from 0.13 to 1.7 mg/L for summing the data from the sensors for each target concentration. The
overall MAE for Phase 1 was below 1 mg/L (0.79 mg/L). As noted in Section 5.1, the
hypothesis tests for equality between the sensors and laboratory values was strict, with no margin
for disagreement. For every sensor at every concentration during Phase 1, the hypothesis of
equality was rejected, although the hypothesis was accepted when summing the sensor data for a
nitrate-N concentration of 1 mg/L.

Table 6-2 presents the MAE and t-test results for Phase 2 of the laboratory investigation and a
graphical representation of the data are presented in Appendix 2.  Similar to that observed in
Phase 1, the MAE increased with increasing nitrate concentrations. Excluding the chloride
interference at a concentration of 2,500 mg/L, the MAE ranged from 0.33  to 0.96 mg/L, 0.60 to
2.0 mg/L, and 2.3 to  5.3 mg/L for target nitrate levels of 1, 3, and 12 mg/L, respectively.  The
chloride interference at a concentration of 2,500 mg/L posed a problem for two of the three
sensors regardless of the underlying nitrate  concentration. The MAE for the chloride
concentration of 2,500 mg/L was 6, 7, and 9 mg/L for target nitrate levels  of 1, 3, and 12 mg/L,
respectively.  The Phase 2 data show that varying the nitrite and turbidity levels have little effect
on the sensor performance at the respective nitrate concentrations, whereas increasing chloride
concentrations have a significant effect on sensor performance.  Table 6-2 shows that the
hypothesis of strict equality was rejected for the majority of nitrate and interference parameter
concentration combinations.

Table 6-3 presents the MAE and t-test results for the field investigation and a graphical
representation of the  MAE  data are presented in Appendix 2. For the field evaluation, all wells
were summarized into a single set of descriptive statistics for the  respective test period (initial
and final intensive sampling and weekly sampling) and the overall field test.  The MAE increases
with time, from 3.3 mg/L during the initial intensive sampling to  7.3 mg/L for the final intensive
                                            21

-------
sampling.  The overall field test MAE is 5.9 mg/L.  The hypothesis of strict equality was rejected
for the intensive sampling events and the overall field test, but was not rejected for the weekly
sampling portion of the field test.

                     Table 6-1.  Summary of Phase 1 t-Test and MAE Results
Target Nitrate-N
Concentration (mg/L)
1
3
6
12
1
o
5
6
12
All
Sensor
W3-1
W3-2
W4
W3-1
W3-2
W4
W3-1
W3-2
W4
W3-1
W3-2
W4
All
All
All
All
All
N
20
20
20
20
20
20
20
20
20
20
20
20
60
60
60
60
240
t Statistic
-13.7
9.53
32.3
23.3
23.4
No variation1
No variation1
-62.5
-17.1
-390
-41.3
55.2
1.83
36.5
-12.0
-6.83
-4.66
p-Value
0.0001
0.0001
0.0001
0.0001
0.0001
0.0001
O.0001
O.0001
0.0001
O.0001
O.0001
0.0001
0.073
O.0001
0.0001
0.0001
O.0001
Decision
Reject
Reject
Reject
Reject
Reject
Reject
Reject
Reject
Reject
Reject
Reject
Reject
Fail to reject
Reject
Reject
Reject
Reject
MAE
(mg/L)
0.13
0.11
0.15
0.25
0.26
0.35
1.8
1.2
0.24
3.2
1.4
0.54
0.13
0.29
1.1
1.7
0.79
     1 - Upon peer review of the information presented in this table, it was noted that the issue of 'no variation' could be
     addressed by doing a multi-way analysis of variance rather than a sequence of t-tests.  The multi-way analysis would
     compute a single pooled estimate of variation using all of the data and use the pooled estimate to calculate the effects
     associated with of sensor and nitrate concentration. Although it is agreed that this would be a reasonable approach, it
     would mask the informative result that under two of the conditions presented in the table, a sensor gave 20
     consecutive identical readings. The first figure in Appendix 2 illustrates graphically that the sensors were very
     consistent in Phase 1, if not always accurate.
                                                    22

-------
Table 6-2. Summary of Phase 2 t-Test and MAE Results for all Sensors Combined
Target Nitrate-N
Cone. (mg/L)
1
3
12
Interference
Parameter
Chloride
Nitrite
Turbidity
Chloride
Nitrite
Turbidity
Chloride
Nitrite
Turbidity
Interference
Parameter Cone.
lOOmg/L
500 mg/L
2,500 mg/L
Img/L
2 mg/L
4 mg/L
1NTU
5NTU
100 mg/L
500 mg/L
2,500 mg/L
1 mg/L
2 mg/L
4 mg/L
1NTU
5NTU
100 mg/L
500 mg/L
2,500 mg/L
Img/L
2 mg/L
4 mg/L
1NTU
5NTU
N
33
47
33
33
33
33
33
33
33
33
33
33
33
33
33
33
33
33
33
33
33
33
33
33
t Statistic
0.412
17.9
21.6
-3.40
-5.13
-4.32
-1.20
1.17
-6.50
1.85
14.0
-8.25
-8.47
-8.74
-4.21
-2.71
-4.19
-1.11
5.17
-6.47
-5.74
-9.41
-4.06
-2.53
p-Value
0.683
O.0001
0.0001
0.002
O.0001
0.0001
0.240
0.249
O.0001
0.074
0.0001
O.0001
0.0001
0.0001
O.0001
0.011
0.0001
0.276
0.0001
0.0001
O.0001
O.0001
O.0001
0.017
Decision
Fail to reject
Reject
Reject
Reject
Reject
Reject
Fail to reject
Fail to reject
Reject
Fail to reject
Reject
Reject
Reject
Reject
Reject
Reject
Reject
Fail to reject
Reject
Reject
Reject
Reject
Reject
Reject
MAE
(mg/L)
0.33
0.95
6.4
0.17
0.32
0.35
0.46
0.96
0.60
1.1
6.8
0.90
0.80
1.2
1.1
2.0
2.3
2.8
9.2
4.3
2.7
5.2
4.0
5.3
   Table 6-3.  Summary of Field Testing t-Test and MAE Results for all Wells
Field Data
Period
Initial Intensive
Sampling
Weekly Sampling
Final Intensive
Sampling
All
N
63
55
80
187
t Statistic
-6.10
-0.312
3.80
2.52
p-Value
O.0001
0.7583
0.0003
0.0127
Decision
Reject
Fail to reject
Reject
Reject
MAE
(mg/L)
3.3
7.0
7.3
5.9
                                    23

-------
Table 6-4 presents the well-specific MAE estimates for the field evaluation. The data in Table 6-
4 show that the comparison of nitrate concentrations in Well 2 exhibited the lowest MAE values
for the initial intensive sampling event and the weekly sampling event, data that are supported by
the time series data plots presented in Appendix 1.  Well 3D exhibited the lowest MAE values
for the final intensive sampling event.  Several extreme sensor concentration values that differed
significantly from the paired laboratory concentration value resulted in high MAE values for
Well 1 (weekly sampling) and Well 6 (initial intensive sampling).

                        Table 6-4.  Well-Specific MAE Estimates
Well
1
2
3S
3D
6
MAE (mg/L)
Initial
Intensive
Sampling
6.7
1.1
7.2
8.5
570
Weekly
Sampling
87
2.8
7.8
4.4
13
Final Intensive Sampling
(Day 1, Day 2)
9.4, 9.5
4.6, 11
1.7, 10
0.74, 2.0
11, 14
Table 6-5 presents a summary of the RPD evaluation for the Phase 1 and 2 of the laboratory test,
and lists the mean RPD, a 95% upper confidence limit (UCL) for the RPD, and the minimum,
maximum, and 25th, 50th, and 75* RPD percentiles for each different phases of the laboratory
test.  In Phase 1, when there are no interferences present, the 95% UCL for RPD was 15%; this
indicates that the sensor measurements were within 15% of the laboratory 1C concentrations on
average and they were always within 31% (maximum) of the laboratory-derived concentrations.
The average RPD for Phase 1 met the <20% criteria discussed in Section 5.1 that indicates
reasonable agreement between the INW sensor and the 1C measurements. In Phase 2, the 95%
UCLs ranged from 27% to 77%. The 95%UCL for the highest chloride parameter level was
substantially greater, indicating  a large interference.

The data from the field portion of the experiment are characterized by a number of extreme
sensor concentration values that differ significantly from the paired laboratory concentration
value. The final intensive sampling event of the field portion of the experiment was especially
problematic because many of the laboratory concentrations were near the limit of detection, so
the RPD was significantly high. Table 6-6 presents a summary of the RPD evaluation for the
field test, including parameters similar to those presented in Table 6-5.  The data presented in
Table 6-6 indicates that the 95% UCL is above 100% for each of the three evaluation periods
when all data are included in the analyses. Due to the presence of significant number of extreme
sensor concentration values in the data, additional analyses were performed to selectively
remove the most extreme sensor and laboratory concentration pairings for the first intensive
sampling and the weekly sampling periods.  In the first intensive sampling event, the 95% UCL
was 47% when the nine most extreme values were removed. For the weekly data collection
period, the 95% UCL was 100% when the two most extreme values were removed.  For the
second intensive sampling event, the number and magnitude of extreme sensor concentrations
                                          24

-------
was such that the selective removal of a reasonable percentage did not result in a significant
reduction in the 95% UCL.

It should be noted that the 95% UCL calculations use the simplifying assumption that the
observations in each portion of the experiment are statistically independent.  That is to say that
they ignore the clustering by factors like sensor, well, laboratory measurement, and target nitrate
concentration. The goal of these calculations is to investigate broadly the rough order of
magnitude of a "buffer zone" that could be built into a null hypothesis in this experiment to have
it not be rejected, and these RPD calculations are useful for that purpose. However, it should be
noted that these calculations are not intended to be used for strict acceptance criteria or for
apportioning the variance in observations between different components of variation.

Time series  plots showing nitrate concentrations measured weekly using laboratory 1C methods
and at 15-minute increments using the INW nitrate sensors are included in Appendix 1.  With the
exception of Well 3S, nitrate sensor data were capable of reporting relative changes in nitrate
concentration over nine continuous weeks.

6.2    Variability

Variability of the nitrate sensor concentration readings was evaluated using data from Phase  1 of
the laboratory evaluation using the multiple readings (20)  made by each of three sensors
deployed in separate test cells at four reference nitrate concentrations (1, 3, 6, and 12 mg/L
nitrate-N).  Variability was expressed as the standard  deviation of the sensor concentration
readings calculated two ways: (1) using the reference (laboratory)  concentrations as the average
(mean) values for comparison, and (2) using the measured mean concentrations from the
samples. The variability results are  summarized in Table 6-7, and  show that the variability in
readings increases with increasing nitrate concentrations, ranging from 0.13 to 1.5 mg/L using
the mean values, and from 0.13 to 2.0 mg/L using the target values.

6.3    Duplication

Nitrate sensor duplication of readings was assessed by comparing nitrate sensor readings made
by placing duplicate nitrate sensors in a single test cell in the  laboratory so they were exposed to
identical concentrations simultaneously during Phase  1 and Phase 2 of the laboratory tests. The
degree of agreement between each pair of reported nitrate concentrations (inter-well and intra-
well) was assessed by calculating the intra-well MSB and  inter-well MSB using a random-effects
ANOVA model.

Table 6-8 summarizes the results of the duplication analyses, and indicates that there is strong
evidence for different mean levels (MAE ^ 0) from sensors both within (intra) and between
(inter) wells. The inter- and intra-well MAE values for Phase 1 all were below 1 mg/L, but
increased above  1 mg/L in a significant majority for the Phase 2 testing.  All of the p-values
except one intra-well difference are  significantly lower than 0.05, indicating the inter- and intra-
well results  are not equal.
                                            25

-------
Table 6-5. Summary of RPD for Laboratory Test for All Nitrate Concentrations
Phase
1
2
Interference
Parameter
NA
Chloride
Nitrite
Turbidity
Interference
Parameter Cone.
NA
lOOmg/L
500 mg/L
2,500 mg/L
Img/L
2 mg/L
4 mg/L
1NTU
5NTU
N
240
99
113
99
99
99
99
99
99
Relative Percent Difference (RPD)
Mean
14%
24%
60%
336%
28%
27%
40%
39%
68%
Standard
Error
Mean
1%
2%
4%
34%
2%
2%
2%
3%
5%
95%UCL
15%
27%
67%
392%
32%
30%
44%
43%
77%
Minimum
2%
1%
16%
15%
<1%
8%
0%
2%
2%
25th
Percentile
8%
7%
29%
86%
12%
11%
11%
15%
51%
Median
13%
25%
38%
241%
18%
19%
57%
18%
53%
75th
Percentile
18%
41%
113%
351%
52%
48%
60%
67%
76%
Maximum
31%
46%
147%
966%
72%
60%
63%
80%
205%
                Table 6-6. Summary of RPD for Field Test
Period
Initial
Intensive
Sampling
Weekly
Sampling
Final
Intensive
Sampling
Extreme Data
Points Removed
None
2
9
None
1
2
None
N
72
70
63
45
44
43
80
Relative Percent Difference (RPD)
Mean
618%
151%
40%
437%
237%
85%
> 1,000%
Standard
Error
Mean
370%
40%
4%
249%
153%
9%
>1,000%
95%UCL
>1,000%
217%
47%
847%
489%
100%
> 1,000%
Minimum
<1%
1%
<1%
2%
2%
2%
98%
25th
Percentile
9%
9%
8%
47%
44%
41%
100%
Median
52%
46%
18%
78%
77%
76%
> 1,000%
75th
Percentile
82%
81%
77%
139%
127%
109%
> 1,000%
Maximum
> 1,000%
>1,000%
86%
>1,000%
>1,000%
253%
> 1,000%
                                  26

-------
              Table 6-7. Summary of Variability in Phase 1 Laboratory Test
Target Nitrate
Concentration (mg/L)
1
3
6
12
N
20
20
20
20
StdDev (mg/L)
(Ref=Mean)
0.13
0.05
0.69
1.5
StdDev (mg/L)
(Ref=Target)
0.13
0.06
1.4
2.0
6.4    Effect of Changes in Water Quality

The results of the Phase 2 laboratory data evaluation of accuracy, variability, and duplication
(see Sections 6.1 through 6.3, respectively) were compared to indicate whether changes in
nitrate, turbidity, and chloride concentrations have any apparent effect on the nitrate sensor
performance at constant nitrate concentrations.  Accuracy or variability results that differed by
more than 5% or nitrate sensor duplication results that differed by more than 20% were taken as
evidence of a significant water quality effect.

Table 6-9 summarizes the changes in MAE, standard deviation computed both ways (reference
equal to the mean, and reference equal to the laboratory target concentration), and intra- and
inter-well MSB resulting from changes in water quality (i.e., different interference parameter
concentrations).  The information presented in Table 6-9 indicates that the changes in MAE,
standard deviation (computed both ways), and MSB for both intra-well and inter-well
measurements exceed the threshold differences of 20% and 5%.
                                           27

-------
Table 6-8. Summary of Duplication Results in Laboratory Test
Phase
1
2
Interference
Parameter
NA
Chloride
Nitrite
Turbidity
Interference
Parameter
Concentration
NA
lOOmg/L
500 mg/L
2,500 mg/L
Img/L
2 mg/L
4 mg/L
INTO
5NTU
Target Nitrate
Concentration
(mg/L)
1
3
6
12
1
3
12
1
3
12
1
3
12
1
3
12
1
3
12
1
3
12
1
3
12
1
3
12
Intra-well
MSB
(mg/L)
0.64
0
0.22
0.35
2.4
0.74
1.6
1.4
2.0
2.4
6.8
4.3
3.8
2.2
2.0
0.32
5.6
1.3
2.2
1.8
4.0
4.0
1.2
0.81
0.45
7.8
12
1.3
Inter-well
MSB
(mg/L)
0.40
0.01
0.98
0.85
2.0
0.67
1.2
1.7
1.5
0.85
2.0
2.9
2.6
0.70
0.66
5.8
0.02
0.21
0.75
4.9
0.72
1.2
11
16
11
8.8
3.0
18
Intra-well
P
O.001
0.704
0.001
0.001
O.001
0.001
O.001
O.001
0.001
O.001
0.001
O.001
O.001
0.001
O.001
0.019
0.001
0.001
O.001
O.001
0.001
O.001
0.001
O.001
O.001
O.001
O.001
0.001
Inter-well
P
O.001
0.001
0.001
0.001
O.001
0.001
O.001
O.001
0.001
O.001
0.001
O.001
O.001
0.001
O.001
0.001
0.001
0.001
O.001
O.001
0.001
O.001
0.001
O.001
O.001
O.001
O.001
0.001
                           28

-------
              Table 6-9.  Summary of the Effect of Changes in Water Quality
Interference
Parameter
Chloride
Nitrite
Turbidity
Target Nitrate
Concentration
(mg/L)
1
3
12
1
3
12
1
3
12
Change in
MAE
>1,000%
> 1,000%
294%
106%
51%
94%
109%
72%
33%
Change in
StdDev
(Ref=Mean)
775%
687%
178%
60%
61%
33%
117%
65%
47%
Change in
StdDev
(Ref = Target)
777%
418%
133%
57%
67%
61%
144%
39%
24%
Change in
Intra-well
MSB
396%
479%
147%
215%
201%
> 1,000%
564%
> 1,000%
188%
Change in
Inter-well
MSB
18%
333%
209%
>1,000%
247%
664%
23%
453%
60%
6.5    Operational and Sustainability Factors

The TempHion™ Smart Sensor was calibrated and installed in each well during the laboratory
and field test by a representative from INW without significant problems. The calibration
procedure was simple to perform, and was taught to field personnel from Battelle and the USDA
ARS laboratory prior to deployment for the field test. During the test, operator observations on
sensor performance were recorded on field activity and sampling logs.

Nitrate sensor data from each well were transmitted wirelessly to the vendor's server on  15
minute increments, and subsequently forwarded on to a web site for download for near real-time
viewing and analysis (see Figure 2-2 for data transmission schematic). During the field
investigation, the nitrate sensors each achieved a 100% data collection standard, indicating
completeness in data collection. A power outage at the test site did result in a stoppage in data
transmittal, but the nitrate sensor data were stored internally within the sensors, and subsequently
recovered.  The Web site used for data download was easily accessible and data were provided in
usable format (i.e., comma-delimited worksheets and Microsoft Excel spreadsheets).  The web
site used for data download and storage also provided real-time graphics nitrate sensor
concentrations, including well-specific time series plots (see Figure 6-1) and plume maps (see
Figure 6-2). Upon removal from the test wells, no biofouling was noted on any of the sensors.

Review of the laboratory data collected during the field test showed that the nitrate
concentrations were at the very low end of this range (typically below 15 mg/L). According to
the vendor, the nitrate sensors used in the verification test were capable  of detecting and
reporting nitrate-N concentrations ranging from approximately 3 to 10,000 mg/L.  The vendor
indicated that tailoring the operational range of the sensors to anticipated field conditions (i.e.,
smaller range) and performing a three-point calibration procedure could improve the accuracy of
the sensors.

The cost of a single TempHion™ nitrate sensor at the time of this verification test was $1,495.
                                                  TA/f
Remote groundwater well monitoring using TempHion   nitrate sensors have potential cost and
long-term sustainability advantages with fewer field visits and reduced sampling and laboratory
                                           29

-------
analysis costs. The nitrate sensors used in this investigation were programmed to collect nitrate
measurements at 15 minute intervals, which equates to 129,600 readings during a traditional
quarterly sampling schedule. Yearly monitoring costs for a single well monitored conventionally
on a quarterly schedule can exceed $2,000, including labor, equipment, disposal of purge water,
and frequent transportation to and from the field site.
     Figure 6-1.  Time Series Graph of Sensor Nitrate Concentrations in Well 1 (Inlet)
   Figure 6-2. Plume Map Showing Dissolved Sensor Nitrate Concentrations in Test Cell
                                           30

-------
                                      Chapter 7
                               Performance Summary


The evaluation of the accuracy of the nitrate sensors indicated that the MAE typically increased
with increasing nitrate concentrations. In Phase 1 of the laboratory investigation, the overall
MAE was below 1 mg/L, and in Phase 2, the MAE ranged from 0.33 to 0.96 mg/L, 0.60 to 2.0
mg/L, and 2.3 to 5.3 mg/L for target nitrate levels of 1, 3, and 12 mg/L, respectively (excluding
the chloride interference at a concentration of 2,500 mg/L, which had MAE values above 6
mg/L). For the field evaluation, the MAE increased with time, from 3.3 mg/L during the initial
intensive sampling to 7.3 mg/L for the final intensive sampling, with an overall MAE of 5.9
mg/L.  The hypothesis of strict equality was rejected for every  sensor at every concentration
during Phase 1, for the vast majority of nitrate and interference parameter concentration
combinations in Phase 2, and for the intensive sampling events and the overall field test, but was
not rejected for the weekly sampling portion of the field test.

The RPD evaluation indicated that the sensor measurements were within 15% of the laboratory
1C concentrations. The average RPD for Phase 1 met the <20% criteria discussed in Section 5.1
that indicates general agreement between sampling methods. In Phase 2 RPD evaluation, the
95% UCLs ranged from 27 to 77%, with one exception. The data from the field portion of the
verification indicated extreme variations between paired sensor concentrations and laboratory
analytical results. The accuracy readings indicate strong correlation between methods in Phase
1, but suggest that accuracy is reduced when elevated levels of interference parameters are
introduced, and decreases with time during field deployment.  With the exception of Well 3S,
nitrate sensor data were capable of demonstrating relative changes (trending) in nitrate
concentrations over nine continuous weeks based on time series plots.

-------
Variability in nitrate sensor readings increases with increasing nitrate concentrations, ranging
from 0.13 to 1.5 mg/L using Phase 1 mean values, and from 0.13 to 2.0 mg/L using Phase 1
target values. Nitrate sensor duplication of readings indicates that there is strong evidence for
different mean levels (MAE ^ 0) from sensors both within (intra) and between (inter) wells.
During Phase 1, the inter- and intra-well MAE values all were below 1 mg/L, but increase above
1 mg/L in a significant majority of the Phase 2 test.  All of the p-values except one for one intra-
well difference are significantly lower than 0.05, indicating the inter- and intra-well results are
not equal. When evaluating the effect of changes in water quality, changes in MAE, standard
deviation (computed both ways), and MSE for both intra-well and inter-well measurements
exceed the threshold differences of 20% and 5%.

The TempHion™ Smart Sensor was calibrated and installed in each well during the laboratory
and field test by a representative from INW without problems. The calibration procedure was
simple to perform, and was taught to field personnel from Battelle and the USDA ARS
laboratory prior to deployment for the field test.  Nitrate sensor data from each well were
transmitted wirelessly to the vendor's server on 15-minute increments, and subsequently
forwarded on to a web site for download for near real-time viewing and analysis with 100% data
collection reported. The web site used for data download was easily accessible and data were
provided in usable format with real-time graphics capabilities. Upon removal from the test
wells, no biofouling was noted on  any of the sensors.  The vendor indicated that tailoring
operational range of the sensors to anticipated field conditions could improve the accuracy of the
sensors.

The cost of a single TempHion™ nitrate sensor at the time of this verification test was $1,495.
Remote groundwater well monitoring using TempHion™ nitrate sensors have potential  cost and
long-term sustainability advantages with fewer field visits and reduced sampling and laboratory
analysis costs.

When reviewing the results of the  nitrate sensor performance analysis, consideration should be
given to the objectives of the long-term monitoring effort and the threshold concentration for
nitrate, particularly considering the percentage of error associated with sample collection and
laboratory analysis. These errors,  in addition to the magnitude of the threshold of the site-
specific nitrate concentration and the cost and sustainability of the technique, should be  taken
into account when selecting a sampling approach for long-term monitoring.
                                           32

-------
                                    Chapter 8
                                    References
(1) Battelle. 2010. Test/QA Plan for Verification of Nitrate Sensors for Groundwater
   Remediation Monitoring for the ETV Advanced Monitoring Systems Center. Environmental
   Technology Verification Program, April.

(2) EPA. 1997.  Method 300.1: Determination of Inorganic Anions in Drinking Water by Ion
   Chromatography.

(3) Battelle. 2008.  Environmental Technology Verification Program Advanced Monitoring
   Systems Center Quality Management Plan (QMP) for the ETV Advanced Monitoring
   Systems Center, Version 7.0. December.
                                         33

-------
                   Appendix 1




Well-Specific Time Series Nitrate Concentration Plots
                        34

-------
o


0
41
U

O
U
     5/2/2010
                                Field Analysis: Well 1
 2/22/2010         6/11/2010         7/1/2010

       Date of Simultaneous Sampling
                                                       -1C
                                                       -Sensor
7/21/2010
   0
    5/2/2010
                               Field Analysis:  Well 2
                                                                          -1C
                                                                          •Sens
2/22/2010         6/11/2010        7/1/2010

       Date of Simultaneous Sampling
7/21/2010
                                        35

-------
  25
  20
E
a

315
n
O
                               Field Analysis: Well 3S
                                                                        -1C
                                                                        'Sensoi
    5/2/2010        2/22/2010        6/11/010        7/1/2010

                         Date of Simultaneous Sampling
7/21/2010
                               Field Analysis: Well 3D
   0
    5/2/2010        2/22/2010         6/11/010          7/1/2010

                        Date of Simultaneous Sampling
 7/21/2010
                                        36

-------
Field Analysis: Well 6
35 -
30 -
a
-S 25 -
\l
0
«J
1 15 .
c
o
§ 10 -
u
5 .
0 -
^""^*
r^
I
•
/v
•v^^^^y :
*— (-,
_ , *^
* * * * *

-1C
fr-Sensor
5/2/2010 2/22/2010 6/11/010 7/1/2010 7/21/2010
Date of Simultaneous Sampling
37

-------
    Appendix 2




Accuracy Data Plots
        38

-------
                 Phase 1 Accuracy Evaluation Plot
           Individual Value Plot of Sensor Nitrate (ppm)
14-
12-
Q. 10-
0.
^ Q _
E
£ 6-
o
8 4-
&
2-
0-




XMK
IMK
w
          Sensor
Target Nitrate (ppm)
                               39

-------
       Phase 2 Accuracy Evaluation Plot - Target Nitrate Level of 1 mg/L
   E
   a.
   a.
   o
   M



   I
                Individual Value Plot of Sensor Nitrate (ppm)

                            Target Nitrate Level = 1 ppm
            10H
             6-
             4-
2-
                                     m

Interference Level
                fV
                                                             \
                                                              \
                                     40

-------
        Phase 2 Accuracy Evaluation Plot - Target Nitrate Level of 3 mg/L
E
o.
a.
re

              Individual Value Plot of Sensor Nitrate (ppm)

                          Target Nitrate Level = 3 ppm
14-




12-



10-



 8-



 6-




 4-



 2-



 0-
     Sensor
Interference Level    ^>
                                                    <\
                                     41

-------
        Phase 2 Accuracy Evaluation Plot - Target Nitrate Level of 12 mg/L
E
a
a
^>

a>
*>
re
o
10

I
              Individual Value Plot of Sensor Nitrate (ppm)

                         Target Nitrate Level = 12 ppm
JU-
25-
20-
15-
10-
5-
0-
w

M *^
f
3MMt



^^tf^^v^ «^fe
^99C uftift )Mi{^ «R


MR
J* *
M
^^^
j^K ^^^
                                                                       12
     Sensor
   Interference Level

                                        42

-------
MAE During Field Sampling
     Sensor Minus Lab
10
    15       5    10    15       5
    Consecutive Sample Number
10
15
           43

-------