EPA
i lor
Survey
)R 97232
United States
Environmental Protection
Agency
Environmental Research
Laboratory
Athens GA 30605
EPA 600 7-78-198
(Vtnber 1978
Research and Development
Multiple Regression
Modeling Approach for
Regional Water
Quality Management
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U S Environmental
Protection Agency, have been grouped into nine series These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4 Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8 "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control technologies for energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
This document is available to the public through the National Technical Informa-
tion Service, Springfield. Virginia 22161.
-------
EPA-600/7-78-198
October 1978
MULTIPLE REGRESSION MODELING APPROACH
FOR REGIONAL WATER QUALITY MANAGEMENT
by
David J. Lystrom
Frank A. Rinella
David A. Rickert
Lisa Zimmermann
Geological Survey
U.S. Department of the Interior
Portland, Oregon 97232
Interagency Agreement No. EPA-IAG-D5-0792
Project Officer
Ray R. Lassiter
Environmental Systems Branch
Environmental Research Laboratory
Athens, Georgia 30605
ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
ATHENS, GEORGIA 30605
-------
DISCLAIMER
This report has been reviewed by the Environmental Research Laboratory,
U.S. Environmental Protection Agency, Athens, Ga., and approved for publica-
tion. Approval does not signify that the contents necessarily reflect the
views and policies of the U.S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
11
-------
FOREWORD
Environmental protection efforts are increasingly directed towards pre-
venting adverse health and ecological effects associated with specific com-
pounds of natural or human origin. As part of this Laboratory's research on
the occurrence, movement, transformation, impact, and control of environmen-
tal contaminants, the Environmental Systems Branch studies complexes of en-
vironmental processes that control the transport, transformation, degradation,
and impact of pollutants or other materials in soil and water and assesses
environmental factors that affect water quality.
The effects of changes in land use on water quality are of increasing
concern to environmental planners and managers who are concerned with asses-
sing and controlling nonpoint source pollution. This report describes the
development of a methodology for estimating the background water quality of
rivers in the United States and its application to the Susquehanna River
basin. The technique may prove useful for describing the extent of regional
water pollution in most areas of the United States and for determining whether
more detailed data and models are needed.
David W. Duttweiler
Director
Environmental Research Laboratory
Athens, Georgia
iii
-------
ABSTRACT
A statistical approach is used in this study to assess the spatial vari-
ability of water quality among 80 subbasins of the Susquehanna River basin in
Pennsylvania and New York. Water quality, for purposes of this study, is de-
fined by 17 characteristics of calculated annual yields or mean concentrations
of suspended sediment, dissolved solids, and various chemical species of ni-
trogen and phosphorus. The water quality characteristics are related exper-
imentally to 57 basin characteristics compiled from available sources of
data. The 57 basin characteristics were selected to account for nonpoint
sources of pollution or to describe processes that control the 17 water qual-
ity characteristics. The six general categories of basin characteristics are
climate, topography, geology, soils, streamflow, and land use.
Multiple-linear-regression equations were developed to relate water
quality characteristics (dependent variables) to basin characteristics (inde-
pendent variables). Usable regression equations were developed for 14 of
the 17 water quality characteristics. These equations explain from 56 to 89
percent of the variation of the water quality characteristics with standard
errors of estimate ranging from 17 to 75 percent. The 14 regression equations
can be used to estimate water quality at any stream site in the study region.
These equations are also used to simulate generalized effects of specific
basin characteristics on water quality. For example, simulated ranges of
background water quality characteristics can be generalized by mathematically
removing the land use variables from the regression equations. Comparison of
simulated ranges of background water quality to observed ranges gives a gen-
eral indication of the effects of the land use variables. For example, ob-
served nitrate yields are as much as 20 times greater than simulated back-
ground yields. This increase is believed to be a result of animal wastes,
the application of chemical fertilizers, and of increasing urbanization. Land
use variables affected by human activities and economic development had meas-
urable impacts in all 14 of the usable regression functions.
It is concluded that this is a useful screening technique to assess the
gross effects of land use and other basin variables on water quality in the
Susquehanna River basin. On the basis of these results, it appears that
similar regression analysis techniques might be applicable to other regions.
This report was submitted in fulfillment of interagency Agreement
No. EPA-IAG-D5-0792 by the U.S. Geological Survey under the sponsorship of
the U.S. Environmental Protection Agency. The report covers the period
from June 30, 1975, to December 31, 1977, and work was completed as of
May 1978.
IV
-------
CONTENTS
Foreword iii
Abstract iv
List of Figures vi
List of Tables vii
Introduction 1
Background 1
Purpose and Scope 2
Basin Setting 3
Physiography and Geology 3
Climate and Hydrology 3
Land Use 5
Approach Concepts 5
Regression Models 7
Selection of Independent Variables 7
Water-Quality Characteristics 8
Suspended Sediment 11
Dissolved Solids 15
Nitrogen and Phosphorus I8
Basin Characteristics 21
Climate 22
Topography 22
Geology 23
Soils 24
Streamflow 25
Land Use 26
Multiple-Regression Analysis 28
Sensitivity of Independent Variables 30
Validity of Regression Models 30
Accuracy of Regression Models 34
Independent Testing of Regression Models 35
Applications of Regression Models 37
Generalized Applications 37
Specific Applications 37
Limitations of the Regression Models 37
Discussion and Conclusions 40
Selected References 42
Appendix 1. Water-Quality Characteristics 49
Appendix 2. Basin Characteristics 51
Appendix 3. Average Soil Characteristics of the Principal Soil
Associations in the Susquehanna River Basin 57
Appendix 4. Annual Tonnages, by County, of Commercial Fertilizer and
Animal Wastes Expressed as Nitrogen and Phosphorus 59
-------
LIST OF FIGURES
Figure Page
1. Physiographic Provinces of the Susquehanna River Basin 4
2. Schematic Diagram of Regional Water-Quality Assessment
Illustrating Multiple-Regression Approach 6
3. Location of Stream-Sampling Sites in the Susquehanna
River Basin 9
4. Suspended Sediment Load Versus Stream Discharge,
Crooked Creek at Tioga, Pa 12
5. Comparison of Computed and Published Suspended
Sediment Loads for Streams in the Susquehanna
River Basin 14
6. Dissolved Solids Concentration versus Stream
Discharge, Chemung River at Chemung, N.Y 16
7. Dissolved Solids Load Versus Stream Discharge,
Chemung River at Chemung, N.Y 17
8. Nitrogen Concentration Versus Stream Discharge,
Tioga River at Tioga, Pa 19
9. Phosphorus Concentration Versus Stream Discharge,
Tioga River at Lambs Creek, Pa 19
vi
-------
LIST OF TABLES
Table Page
1. Results of Multiple-Linear-Regression Analysis of
Logarithmic-Transformed Variables 29
2. Ranges of Observed Variables and Regression Weights
and Selected Correlation Coefficients of Independent
Variables 31
3. Testing of Regression Models 36
4. Observed Ranges of Water Quality Yields and Concentrations
and Background Ranges Simulated by Regression Models .... 38
vii
-------
MULTIPLE REGRESSION MODELS
FOR REGIONAL WATER QUALITY ASSESSMENT
INTRODUCTION
The concern over change in our environment that led to recent Federal
legislation has also created an urgent need for practical methods to assess
the relationship of water quality to land use. In response to the need, this
report describes the application of regression techniques to describe the im-
pact of land use on stream water quality in the Susquehanna River basin,
Pennsylvania and New York.
Background
The 2-year study summarized by this report was funded by the U.S.
Environmental Protection Agency (EPA). The project objective was to develop
a methodology for estimating the background water quality of rivers in the
United States. Background water quality is needed as a basis for (1) asses-
sing the level of culturally related nonpoint source pollution, (2) developing
realistic water quality standards, and (3) formulating legislation concerning
pollution abatement.
The project outline was formulated by a joint team from the U.S. Geolo-
gical Survey and EPA. Four water quality properties—suspended sediment, dis-
solved solids, nitrogen, and phosphorus—were selected for study because of
wide concern about their impacts on stream water quality in rural areas under-
going rapid development. Suspended sediment, as an indicator of erosion and
sedimentation, is considered by many to be the nation's most critical nonpoint
source pollutant. Dissolved solids are of concern in heavily irrigated areas.
Nitrogen and phosphorus from urban areas, agricultural fertilizer, animal feed
lots, and irrigation return flow may stimulate eutrophication in streams and
impoundments.
Specific objectives outlined for methodology development were:
1. Develop a methodology that is quickly and easily applicable for one large
region, using existing data.
2. Provide a means to assess the general effects of land use on water quality
and to estimate gross background streamflow quality.
-------
3. Demonstrate the application of the methodology in layman's terms.
After the project outline was established, the authors began a survey of
possible methodologies and study basins. Statistical and digital process-
modeling techniques were quickly highlighted as the most promising methodolo-
gies. The statistical approach was chosen as the preferred method because the
study required short-term results using existing data. The statistical ap-
proach was viewed as a first step, providing (1) initial answers on several
key land-use and water-quality problems and (2) a basis for evaluating the
need for more intensive assessments which might involve digital modeling and
the collection of additional water-quality data.
Selection of the study basin involved consideration of available data on
water quality, land use, and various characteristics of climate and terrain.
Land-use and water-quality data were limited in many areas of the country.
Through a screening process the Susquehanna River basin in Pennsylvania and
New York was selected for the analysis.
Purpose and Scope
The purpose of this report is to (1) document the methods used to compile
water-quality characteristics and the basin characteristics that affect water
quality, and (2) demonstrate the feasibility of using multiple-regression
analysis for regional water-quality assessment. The reported regression mod-
els are used to assess the generalized effects of land use on regional water
quality. This approach may be useful in most areas of the United States for
describing the extent of regional water pollution and for determining whether
more detailed data and models are justified to evaluate the management alter-
natives needed to fulfill water-quality objectives.
Multiple-linear regressions are developed by standard statistical tech-
niques. These regressions relate the spatial variations in water quality
among 80 subbasins of the Susquehanna River basin to selected characteristics
of climate, physiography, and land use. Water quality is represented here by
yields and concentrations of suspended sediment, dissolved solids, and various
species of nitrogen and phosphorus. The criteria for selecting and computing
water-quality and basin characteristics are described in detail. Computed
values of these characteristics are tabulated in the appendixes.
The regressions developed in this study generally represent the processes
that affect regional water quality. The sensitivity of regression models to
land use and natural basin characteristics is analyzed to minimize misuse.
The accuracy, conceptual viability, and limitations of the regressions are
discussed and examples are described to illustrate selected applications to
management problems. In the examples, the culturally induced characteristics
of land use are hypothetically removed from the regressions to provide indi-
rect estimates of background water quality. By this approach, simulated
ranges of background water quality are computed for subbasins throughout the
study region. These results are used to define the relative effects of land-
use variables on water quality and to estimate the expected ranges of water
quality that would occur if land use approximated predevelopment conditions.
-------
Planners and managers can also use the regression models to estimate water-
quality characteristics for any subbasin of the Susquehanna River basin based
on basin characteristics compiled from available data sources and information
provided in the appendixes.
The regression models are tested by comparing observed water-quality
characteristics to corresponding simulated results. The observed character-
istics were computed from limited water-quality data collected during the
1976 and part of 1977 water years. These data were not used in determining
the regression coefficients.
The approach used in this study is empirical and therefore direct appli-
cability of the results is limited to the Susquehanna River basin and hydro-
logically similar adjacent areas. However, the general methodology is poten-
tially applicable to any river basin or study region for which adequate data
are available to define water quality and the appropriate basin characteris-
tics.
BASIN SETTING
The Susquehanna River, which empties into the Chesapeake Bay, drains the
largest basin along the east coast of the United States (area 27,510 nd.2) , of
which 76 percent is in Pennsylvania, 23 percent in New York, and about 1 per-
cent in Maryland (Rudisill, 1976).
Physiography and Geology
The Susquehanna River basin spans four physiographic provinces (see
fig. 1.): (1) the Appalachian Plateaus, (2) the Valley and Ridge, (3) the
Blue Ridge, and (4) the Piedmont (Fenneman, 1928). The rocks of the Appalach-
ian Plateaus province are nearly horizontal and are of Devonian, Mississippi-
an, and Pennsylvanian age. They consist of alternating shale, siltstone,
sandstone, limestone, and bituminous coal. The northeast part of the Appa-
lachian Plateaus consists of flat-topped mountains and deeply incised stream
valleys. The Valley and Ridge Province is underlain by folded and faulted
rocks. The Valley and Ridge Province is characterized by a sequence of alter-
nating shale, sandstone, and limestone of Paleozoic age which forms steep
mountains and ridges separated by valleys. Only a small part of the Blue
Ridge Province, which is underlain by crystalline rocks and contains deep,
well-drained soils, lies within the Susquehanna River basin. The Piedmont
Province consists of both uplands and lowlands, the former underlain by crys-
talline rocks and the latter by limestone, sandstone, and shale. The Piedmont
generally has terrain that is gently rolling to hilly, and it has deep, well-
developed soils.
Climate and Hydrology
The climate in the Susquehanna River basin is moderate. The length of
the growing season ranges from 120 to 200 days and averages about 150 days.
The growing season is shortest in parts of the Appalachian Plateaus and is
longest near the mouth of the Susquehanna (Johnson, 1960; Kauffman, 1960).
-------
APPALACHIAN PLATEAUS
»s
^''£''~^
AND
RIDGE
PIEDMONT
Figure 1,—Physiographic provinces of the Susquehanna River basin (Fenneman, 1928).
-------
.-. -X1--
INDEX MAP SHOWING
THE SUSQUEHANNA RIVER
s^V-vx • *\ \l
'- ••—' BINGHAMTONV. |V
'\^' v^r^
* APPALACHIAW. PLATEAUS
0 10 JO 30 40 MiLtS
h—r-H' i1 i -4-
0 10 20 30- 40 60
f Bay
Figure 1.-Physiographic provinces of the Susquehanna River basin (Fenneman, 1928).
4
-------
REPLACEMENT PAGE
FOR
MULTIPLE REGRESSION
MODELING APPROACH FOR
REGIONAL WATER
QUALITY MANAGEMENT
EPA-600/7-78-198
Please insert replacement page 4 (back of this sheet)
in your copy of Multiple Regression Modeling Approach for
Regional Water Quality Management (EPA-600/7-78--198) .
&EPA
Official Business
Penalty for Private Use
$300
Postage and Fees Paid
Environmental Protection Agency
F.PA335
f*'
uSEuu!
United States
Environmental Protection
Agency
Environmental
Research Laboratory
Athens GA 30605
ERIC
MS J CIRCIELLO
012500022
AGY,
Return this sheet if you do NOT wish to receive this MaterialD,
or if change of address is needed D (indicate change, including zip code).
EPA Form 1510-1
-------
Average annual precipitation ranges from 32 inches in the northwestern part
of the basin to 44 inches in the southern and east-central part, with a basin
average of 40 inches.
About 50 percent of the precipitation over the Susquehanna River basin
appears in the stream as runoff. The month^to«month variation in streamflow
generally is more extreme than the variation in precipitation because of the
large losses to evaporation and transpiration during the hot summer months
and the impermeability of the soil during winter.
Streamflow is composed of water that reaches the stream by direct over-
land flow and by ground-water inflow which sustains base runoff. During base-
flow periods the dissolved-solids concentration of the Susquehanna River is at
a maximum because the chemical quality of the river water is affected by evap-
oration, ground-water inflow, and coal-mine drainage. As streamflow increases,
the dissolved-solids concentration is lowered by dilution from direct runoff
(Anderson, 1963).
Land Use
In tne study region, climate, soils, and topography have influenced the
use of the land for many decades. Where the soils are productive the flat-to-
rolling countryside was commonly cleared for cultivation. Forests cover most
of the land where the soils are poor or the slopes are too steep for cultiva-
tion.
Water quality in the Susquehanna River basin is greatly influenced by
agriculture and the degree and type of urbanization and industrialization.
In addition, streams receiving water from coal-mine fields are low in pH and
high in iron, sulfate, and dissolved solids. Relatively little water is con-
sumed by industry in the basin. About 60 percent of the Susquehanna River
basin is covered by forest, 31 percent is used for agriculture, and 4 percent
is urban (determined from Rudisill, 1976, p. 5, 13, 20, 31, 39, and 45).
APPROACH CONCEPTS
Water quality varies temporally and spatially within stream systems.
These variations are a result of many complex processes which are controlled
in large part by climate, physiography, and land use. Some of these control-
ling processes are well known; however, many are poorly known and some may
still be unidentified.
The approach used in this study focused on establishing empirical rela-
tionships between water-quality characteristics and basin characteristics.
The first step was to establish a conceptual framework for compiling available
data. Water-quality and basin characteristics must be defined for a time
period during which land-use and management techniques have remained relative-
ly stable. Based on discussions with land-management and planning agencies
in the basin, the 10-year interval from 1966 to 1975 was selected as the
study period. Water-quality characteristics are defined by weighted or aver-
age annual concentrations, or average annual yields occurring during this
-------
period. Similarly, basin characteristics represent unique aspects of land-
use, physical, and climatic conditions existing during the period. A sim-
plified schematic diagram of the approach concepts is shown in figure 2.
The multiple-linear-regression approach (illustrated by the example in
figure 2) is commonly used by hydrologists to define regional variations of
streamflow as a function of basin characteristics. This method was applied
extensively in 1969 and 1970 in a nationwide U.S. Geological Survey (USGS)
program to provide a means for estimating streamflow characteristics of un-
gaged basins. (See Thomas and Benson, 1970; and Benson and Carter, 1973.)
Similar studies have related water-quality characteristics to basin charac-
teristics. (See for example Branson and Owen, 1970; Flaxman, 1972; Hindall,
1976; and Steele and Jennings, 1972.)
The multiple-regression approach provides a means of estimating water-
quality characteristics at unsampled stream sites and of estimating the gen-
eral effects of natural and cultural aspects of drainage basins on water qual-
ity. The principal advantage of this approach is that a multiple-regression
model can be developed on the basis of available data and can be applied to a
large region to define the general magnitude and possible causes of selected
water-quality characteristics. From a regional vantage point, the approach
provides information for reaching decisions on how to resolve certain water-
pollution problems, and for determining where there is need for more sophis-
ticated studies and the collection of more detailed data.
COMPUTE
BASIN
CHARACTERISTICS
X'S
c
DATA MATRIX
COMPUTE
WATER-QUALITY
CHARACTERISTICS
Y's
Y=aX1b1X2b2
j
Figure 2.—Schematic diagram of regional water-quality assessment illustrating
multiple-regression approach.
-------
Regression Models
In this study the multiple-linear-regression technique is used to define
spatial variations in water-quality characteristics as a function of the phys-
ical, climatic, and land-use aspects of stream drainages. The general form of
a multiple-linear regression is
y = a + 2yf2 + b2x2 +.., bnxn (i)
where Y is a water-quality characteristic (dependent variable) , the X's are
basin characteristics (independent variables), a is the regression constant,
the Jb's are regression coefficients, and n is the number of basin character-
istics. Nonlinear relationships between hydrologic variables have frequently
been found to be linear if the variables are transformed to logarithms (Ben-
son and Carter, 1973, p. 17). The general form of a log-transform regression
is
log Y = log a + b log X + b2log X2+ . . . b^log Xn (2)
An equivalent form of equation 2 is
Y = a x.blx, b2... Xbn (3)
1 2 n
Because the logarithm of zero is undefined, a constant, such as 1.0, is added
to all independent variables that could feasibly be zero. For example, the
percent of agriculture (LU2) is zero for some basins used in this study. The
method of computing the a and b constants is explained by Riggs (1968, p. 12
18) . A system of statistical computer programs (STATPAC) was used to trans-
form variables, compute regression coefficients, and perform other statisti-
cal tests (Sower and others, 1971).
Selection of Independent Variables
Selection of basin characteristics to be compiled for the analyses was
based primarily on conceptual knowledge of the dominant sources and processes
that affect water quality. Because implementation of the approach depends on
availability of data, it was necessary in some cases to use a surrogate as an
index of a variable that actually controls the particular process. For exam-
ple, the percent of basin urbanized is a surrogate that can be used to define
the effects of domestic sewage effluent on nutrient concentrations. Percent
urbanization, however, is also a descriptor, of overland urban runoff. It is
important to recognize the limitations of surrogates to properly qualify
assumptions about cause-and-effect relationships.
The process of selecting the most significant independent variables for
each regression was complicated by the large number (57) of potential varia-
bles. Consequently, several trial-andrerror regressions had to be computed
for each water-quality parameter to derive the best equations. The final
selection of a set of independent variables to form each regression equation
-------
was based on considerations and statistical criteria as follows;
1, Each independent variable must be statistically significant at the 95-
percent level according to Students t-test of significance (Draper and
Smith, p. 305, 1966).
2. A combination of selected independent variables, as compared to other
possible combinations, should (a) have the lowest standard error of
estimate and (b) explain the greatest percent of variance of the depen-
dent variable.
3. Combinations of cross-correlating independent variables (correlation
coefficients greater than 0.6 or 0.7) should be minimized.
WATER-QUALITY CHARACTERISTICS
Available data were used to define one or more characteristics of sedi-
ment, dissolved solids, nitrogen, or phosphorus for 80 stream sites in the
Susquehanna River basin. The sources of water-quality data used for this
study were (1) the USGS WATSTORE water-quality computer file, (2) the USGS
WATSTORE daily-values (streamflow) computer file, and (3) USGS annual publi-
cations "Water Resources Data" for Pennsylvania and New York, Part 2, 1966 to
1975. Figure 3 shows locations of the 80 stream-sampling sites and indicates
which water-quality characteristics were computed for each site.
All water-quality data were transferred to magnetic tapes to facilitate
computation of characteristics by use of computer programs written for this
study. Although some additional data were available from other sources,
these data were not used because there were differences in sampling proce-
dures and laboratory-analysis techniques that might have caused inconsist-
encies among the data.
The methods of computing and selecting water-quality characteristics
used for this study are based on: (1) the need for methods that are adapt-
able nationwide, (2) adaptability to the multiple-regression-analysis ap-
proach, and (3) availability of data. Several possible water-quality charac-
teristics were excluded because of insufficient data. Two general criteria
for including a water-quality characteristic in this study were: (1) a mini-
mum of 20 sampling stations in the study region and (2) at least 10 samples
collected at each station during 1 or more years within the study period.
The 17 water-quality characteristics considered in this study are as
follows:
1. Suspended-sediment yield (SEDYLD): The average annual load per unit of
contributing drainage area for the period of water years 1966 to 1975
(excluding 1972), in (tons/mi2)/yr. Data for water year 1972 were ex-
cluded because of the extreme effect of tropical storm Agnes on sediment
loads. The rationale for excluding 1972 is discussed under "Computation
of suspended-sediment loads."
-------
INDEX MAP SHOWING
THE SUSQUEHANNA RIVER BASIN J
f^f-1618860
1618500
1518850
BINGHAMTO
1515050
1515000
O SCRANTON
WILLIAMS PORT
EXPLANATION
"•,'
I66760).15682°P
0 10 2O 30 40 50 KILOMETERS
HAVRE DEGRAC
Ammonia
Phosphorus concentration
Phosphorus concentration
and load
4r Orthophosphate concen-
tration
A Sediment concentration
and load
A Dissolved-solids concen-
tration and load
•A Nitrogen concentration
Nitrate concentration
Chesapeake Bay
Figure 3.-Locations of stream-sampling sites in the Susquehanna River basin.
-------
2. Suspended-sediment concentration (SEDCONC): The discharge-weighted aver-
age sediment concentration for the period of water years 1966 to 1975
(excluding 1972, see above), in mg/L.
3. Dissolved-solids yield (DSYLD): The average annual load of dissolved sol-
ids per unit of drainage area for the period of water years 1966 to
1975, in (tons/mi2)/yr.
4. Dissolved-solids concentration (DSCONC): The discharge-weighted average
annual dissolved-solids concentration for the period of water years 1966
to 1975, in mg/L.
5. Coefficient DSEXP of the transport curve relating dissolved-solids load,
L£S, in tons/day, to instantaneous discharge, (?, in ft3/s. This rela-
tionship is defined by the equation log I^s ~ log (DSCOEF) + (DSEXP)
log Q. (See equation 13.)
6. Coefficient DSCOEF of equation 13 described above.
7. Nitrogen concentration (NAVE): The average total nitrogen concentration
for each sampling site for the period of water years 1970 to 1975, in
mg/L as N.
8. Standard deviation (NSD) about the average total nitrogen concentration
(NAVE) for each sampling site for the period of water years 1970 to
1975, in mg/L as N.
9. Nitrate concentration (N03AVE): The average total nitrate concentration
for each sampling site for the period of water years 1970 to 1975, in
mg/L as N.
10. Standard deviation (N03SD) about the average total nitrate concentration
(N03AVE) for each sampling site for the period of water years 1970 to
1975, in mg/L as N.
11. Nitrate yield (N03YLD): The average annual nitrate load per unit of
drainage area for the period of water years 1966 to 1975, in (tons/mi2)/
yr as N.
12. Ammonia concentration (NH4AVE): The average total ammonia concentration
for each sampling site for the period of water years 1970 to 1975, in
mg/L as N.
13. Phosphorus concentration (PAVE): The average total phosphorus concentra-
tion for each sampling site for the period of water years 1970 to 1975,
in mg/L as P.
14. Standard deviation (PSD) about the average total phosphorus concentration
(PAVE) for each sampling site for the period of water years 1970 to
1975, in mg/L as P.
10
-------
15. Phosphorus yield (PYLD): The average annual phosphorus load per unit of
drainage area for the period of water years 1966 to 1975, in (tons/mi2)/
yr as P.
16. Orthophosphate concentration (P04AVE): The average total orthophosphate
concentration for each sampling site for the period of water years 1970
to 1975, in mg/L as P.
17. Standard deviation (P04SD) about the average total orthophosphate concen-
tration (P04AVE) for each sampling site for the period of water years
1970 to 1975, in mg/L as P.
Water-quality characteristics are tabulated in appendix 1 for 80 stream-
sampling sites in the Susquehanna River basin. The following sections de-
scribe in detail the methods for computing each of the water-quality charac-
teristics.
Suspended Sediment
Twenty-eight stream stations in the Susquehanna River basin have adequate
data for computing average annual sediment loads for the study period (1966 to
1975 water years). Only one of these, Juniata River at Newport, Pa., has a
complete record of daily loads. Twelve additional stations have published
daily sediment loads for 1 or more years during the study period. The pre-
dominant source of available data is miscellaneous sediment concentrations in
the U.S. Geological Survey's WATSTORE water-quality computer file.
Computation of Suspended-Sediment Loads
Average annual suspended-sediment loads are computed for the study period
by the sediment-transport curve method. This method was shown by Miller
(1951) to provide a useful method of computing annual sediment loads, and was
also used for a previous stream-sediment appraisal in the Susquehanna River
basin (Williams and Reed, 1972).
Sediment-transport curves are based on the relationship between sediment
loads and discharges for each stream station. Daily sediment-load data were
not available for most of the 28 stations; consequently, instantaneous loads
were calculated for each instantaneous concentration and discharge by the
equation
Ls - 0.0027 CSQ (4)
where: Ls is the instantaneous sediment load in tons/day, Cs is the instan-
taneous sediment concentration in mg/L, Q is the instantaneous discharge in
ft-Vs, and 0.0027 is a units conversion constant. :
A computer program, REGPLOT, was developed for this study to plot instan-
taneous sediment loads (from eq. 4) versus instantaneous discharge as shown in
figure 4. This program includes a least-squares curve-fitting routine for
log-transformed linear and quadratic regression equations
11
-------
log L = log a + b log Q
(linear)
.2
log L = log a + b log Q + c (log Q)~ (quadratic)
S
(5)
(6)
where Ls is instantaneous sediment load in tons/day; Q is instantaneous stream
discharge in ft-Vs; and a, b, and c are regression coefficients. A transport
curve for each stream station was defined by a single log-linear equation
(eq. 5), or by a series of straight-line segments manually fitted to portions
of a quadratic curve (eq. 6). The primary criterion for establishing a
sediment-transport curve was a minimum of 10 data points that are reasonably
well distributed over the range of daily discharges. Transport curves were
10,000
DC
UJ
Q_
p
Z
Q
I-
Ul
5
a
IU
CO
a
LU
Q
Z
Ul
a.
s
V)
tn
O
LU
Z
<
Z
<
to
1,000
100
10
1 -
0.1
1 10 100 1,000 10,000
INSTANTANEOUS DISCHARGE, IN CUBIC FEET PER SECOND
Figure 4.-Suspended-sediment load versus stream discharge for Crooked
Creek at Tioga, Pa. (station 1518500).
12
-------
not used if the range of daily discharges extended more than one-half of a
log cycle higher than the plotted data points.
Once a sediment-transport curve has been defined for each station, the
long-terra mean annual sediment load is generated by computer from the curve
by using records of daily discharge. A computer program, LOAD, is used to
generate daily loads and to summarize monthly, annual, and 10-year average
loads. This program utilizes a magnetic tape of daily discharges extracted
from the U.S. Geological Survey's WATSTORE computer filing system. Definition
of the sediment-transport curve is input on punched cards specifying log-
linear regression coefficients (eq. 5) or a table listing the end points of
each manually fitted straight-line segment of the quadratic curve (eq. 6).
Tropical storm Agnes, which occurred in June, 1972, produced floods hav-
ing recurrence frequencies ranging from 2 to more than 100 years (Bailey
and others, 1975, Table A-l). Because the extreme sediment loads occurring
during this event are atypical of an average 10-year period, 9-year average
annual loads, excluding 1972, were also computed. By comparison, the nine-
year averages were as little as one-tenth the 10-year averages. Both 9- and
10-year average loads were related to several experimental sets of basin char-
acteristics by regression analysis. (Refer to technique described in the
section, "Multiple-regression analysis.") It was found that an acceptable
regression model could be established for the 9-year average sediment load.
However, none of the experimental regression models tested for the 10-year
load was successful,(as indicated by low percentages of explained variation).
Consequently, the 9-year load was selected for the study.
Accuracy of the Generated Sediment Loads
The scatter of data points about most of the sediment-transport curves
was large; sometimes standard errors of estimate were as great as + 100 per-
cent. The accuracy of the generated annual loads and long-term averages is
dependent on the assumptions that (1) transport curves represent the entire
study period, and (2) the technique for fitting transport curves is unbiased.
The first assumption is supported by experience indicating that the transport
curves generally did not change greatly over the 10-year period. Bias in
curve fitting can be tested by comparing annual loads computed by the trans-
port-curve method with published annual loads. Annual suspended-sediment
loads published in the annual USGS data reports are based on a systematic
sampling program in which sediment concentrations are determined daily, and
more frequently during periods of high flow. Figure 5 shows annual sediment
loads, generated from transport curves, plotted against published annual loads
for daily sediment stations. This plot represents 22 annual loads for 13 USGS
stations. The least squares regression line in figure 5 very nearly coincides
with the line of equal values, indicating that there is no appreciable bias in
the curve-fitting technique. The standard error of estimate of the computed
loads as compared with the published loads is about + 31 percent. Broadly
interpreted, this indicates that about two-thirds of the computed loads are
within + 31 percent of the published loads.
13
-------
10,000,000
ID
£ 1,000,000
(/>
o
100,000
5 5
sg
So
LU UJ
LU %
o. 5
z
10,000
1,000 -
100
NOTE: Standard error of estimate is 31 percent
Least squares regression line
I
100 1,000 10.000 100,000 1,000,000
ANNUAL SUSPENDED-SEDIMENT LOADS, IN TONS PER YEAR
OBTAINED FROM PUBLISHED SEDIMENT LOADS
10,000,000
Figure 5.—Comparison of computed and published suspended-sediment loads for
streams in the Susquehanna River basin.
Loads computed for miscellaneous sediment sampling sites have no compa-
rable published data. However, the errors of estimate may be somewhat larger
than those shown in figure 5 because these transport curves are based on
fewer samples which, in some cases, did not define the entire range of stream-
flow.
Computation of Sediment Concentrations
Sediment concentrations vary substantially over time, with high concen-
trations resulting from flood runoff. It is therefore difficult to describe
sediment-concentration variations adequately using a single characteristic.
In this study the discharge-weighted average sediment concentration was selec-
ted as an index value. It is computed by program LOAD according to the
equation
14
-------
SEDCONC = s (7)
0.9860
where SEDCONC is the average annual discharge-weighted sediment concentration
in mg/L; ~LS is the average annual sediment load in tons/yr; Q is the average
daily stream discharge, in ft-Vs; and 0.986 is a units conversion constant.
Because of the method of computation, the accuracy of the average annual
discharge-weighted sediment concentrations is limited to the accuracy of the
computed average annual sediment loads.
Dissolved Solids
Available Data
Dissolved-solids loads and concentrations were computed for 26 stream
stations in the Susquehanna River basin for the study period. Dissolved-
solids concentrations and specific-conductance data were obtained from the
USGS WATSTORE water-quality computer file. Dissolved-solids concentrations
were determined by the residual on evaporation (DSroe) method from unfiltered
water samples. DSroe concentration data were augmented with dissolved-solids
estimates made by use of linear-rregression relationships of DSroe with sum of
dissolved-solids constituents (DS8UIQ) and with specific conductance (COND).
Values for DSsum in the Susquehanna River basin were consistently lower
than those for DSroe, and consequently could not be interchanged. Computer
program REGPLOT was used to plot DSroe concentrations versus DSSum concentra-
tions and to compute a least-squares-regression equation for each station hav-
ing 10 or more paired analyses. The resulting equations were of the general
form
DSroe = a + b(DSsum) <8)
where a and b are regression coefficients determined for each station. These
coefficients were computed and used to augment DSroe data for six stations;
the average standard error of estimate was 8 percent.
The regression coefficients a and b in equation 8 did not vary apprecia-
bly among stations. Therefore, a regional model relating DSroe to DSSUm
was also computed based on 456 available analyses at 25 sampling sites in the
study region. The resulting regional equation
DS - 4.5 .+ 1.06 DS . (9)
roe sum
has a standard error of 10 percent about the mean of DSroe. This regional
equation was used to augment DSroe data for two stations, which had less than
10 dual data points available to define a station equation (eq. 8).
A similar procedure was used to augment DSroe data based on available
specific-conductance data. This method utilized the linear-regression
. 15
-------
equation
DS
roe
a + i(COND)
(10)
where a and b are regression coefficients determined for each sampling station.
These coefficients were computed and used to augment DSroe data for 14 sta-
tions; the average standard error of estimate was 8 percent.
A regional equation was also computed by program REGPLOT for DSroe versus
COND based on 1,441 paired analyses at 27 stations. The regional equation is
DSroe = 1.04+0.62 (COND);
(ID
it has a standard error of estimate of 14 percent. The regional equation
(eq. 11) was used to augment DSroe data for 10 stations.
The procedures and rationale for developing station and regional equa-
tions for DSroe versus COND are described in detail by Lystrora and others
(1978).
The data-augmentation procedures used in this study effectively increased
the number of analyses for DSroe from 719 to 1,547 and increased the number of
usable stations from 19 to 26.
Computation of Dissolved-Solids Loads and Concentrations
Average annual dissolved-solids loads were computed by the same trans-
port-curve method used for suspended-sediment loads. Unlike sediment concen-
trations, dissolved-solids concentrations in Susquehanna streams generally de-
crease with increased streamflow. Figure 6 is a typical example of the rela-
tionship of dissolved-solids concentrations to streamflow.
O 1,000
F
c
z
C/9
Q
85 100
«9 M
O 5
UJ 7
5*
02
Sd
01
Z
• •
1010
NOTE: Standard error of estimate Is 17 percent
Correlation coefficient is 0.87
I
100 1.000 ! 0,000
INSTANTANEOUS DISCHARGE, IN CUBIC FEET PER SECOND
100,000
z Figure 6.-Dissolved-solids concentration versus stream discharge for the Chemung
River at Chemung, N.Y. (station 1531000).
16
-------
For the purpose of plotting dissolved-sollds transport curves, instan-
taneous dissolved-solids loads (L, )t in tons/day, are computed by the
. . cts
equation
ds
= 0.0027
(12)
where cds is an instantaneous dissolved-solids concentration in mg/L; Q is the
instantaneous discharge in ft-Vs; and 0.0027 is a units conversion constant.
Program REGPLOT is used to plot transport curves and to compute log-linear
regression equations of the form
log L, = log (DSCOEF) + (DSEXP) log Q
CLS
(13)
where Lds and 0 are as explained for equation (12), and DSCOEF and DSEXP are
regression coefficients for each station. The log-linear regressions provided
a good fit for all dissolved-solids transport curves. A typical dissolved-
solids transport curve is shown in figure 7. Average annual dissolved-solids
loads and discharge-weighted average dissolved-solids concentrations are com-
puted by program LOAD for the period of water years 1966 to 1975, as described
for computation of sediment characteristics.
100,000
W
I
UI
10,000
1,000
o
(A
O
LU
z
!
z 100
NOTE: Standard error of attlmatt l» 17.8 parcant
Correlation coefficient It 0.99
_L
I
100 1,000 10.000 100,000
INSTANTANEOUS DISCHARGE, IN CUBIC FEET PER SECOND
Figure 7.—Dissolved-solids load versus stream discharge for the Chemung
River at Chemung, N.Y. (station 1531000).
For experimental purposes the regression coefficients in equation (13)
(DSCOEF and DSEXP) were also included in this study as water-quality charac-
teristics. (See appendix 1.) These coefficients define a unique dissolved-
solids transport curve for each streamflow sampling station. Therefore, if
17
-------
each coefficient could be defined regionally as a function of basin character-
istics, an estimated dissolved-solids transport curve (equation 13) could be
used to generate daily dissolved-solids loads for any stream station, provided
daily discharges were available.
Accuracy of Dissolved-Solids Loads and Concentrations
Data on dissolved-solids load, with which the computed annual dissolved-
solids loads could be compared, are not available. The accuracy of generated
annual loads, however, is considered on the basis of the general accuracy of
transport curves (such as the one depicted in figure 7). The average standard
error of estimate of daily loads for the 26 transport curves was 18 percent.
The accuracy of the 10-year-average loads should be better than the standard
error of the transport curves because of the compensating effect of summing
daily loads to obtain annual loads. A similar assumption for annual sediment
loads was verified in "Accuracy of'the generated sediment loads." Because of
the method of computation, the accuracies of the discharge-weighted average
dissolved-solids concentrations are similar to the accuracies of generated
dissolved-solids loads.
Nitrogen and Phosphorus
The same methods were used for compiling nitrogen and phosphorus infor-
mation and, therefore, these two constituents are discussed together. The
characteristics of nitrogen and phosphorus evaluated in this study were based
on unfiltered samples. The characteristics evaluated were total nitrogen (N),
nitrate (N<>} as N), ammonia (NH4 as N), phosphorus (P), and orthophosphate
(P04 as P). Collectively, these constituents are referred to as nutrients.
Available Data
Only available nutrient data from water year 1970 to the end of the study
period (water year 1975) were utilized because of uncertainties over methods
used in the handling and analysis of water samples for nutrients prior to 1970.
Average concentrations based on a minimum of 10 seasonally spaced samples
per station were computed for the five nutrient species. The number of sta-
tions representing each species is as follows:
Number of
Species stations
N 27
N03 58
NH4 46
p .49
POA 20
Average annual loads were computed only for total nitrate and total phos-
phorus. Loads were not computed for the other three nutrient species because
fewer than 20 of these nutrient-concentration stations have daily discharge
data. For the purpose of defining the variability of nutrient concentrations
the standard deviations about the mean concentrations were also included.
18
-------
Computation of Average Nutrient Loads and Concentrations
Nutrient transport curves were found not to be useful for computing loads.
The utility of nutrient-transport curves had been questioned initially when
regression analysis of nutrient concentrations versus discharge resulted in very
low correlation coefficients. Figures 8 and 9 are typical plots of nitrogen
and phosphorus concentrations versus discharge. The mean of the correlation
coefficients for all stations were 0.44 and 0.35 for N and P, respectively.
10
cc
o -1
o o:
z u
g°-
§(/)
5
Z Z
1 0.1
NOTE: Correlation coefficient is 0.51
I
I
10 100 1,000
INSTANTANEOUS DISCHARGE, IN CUBIC FEET PER SECOND
10,000
Figure 8.-Nitrogen concentration versus stream discharge for the Tioga River at
Tioga, Pa. (station 1518000).
o
1
UJ CC
«
X S
8*" cc
C3
P
2?
1
0.1
.01
NOTE: Correlation coefficient Is 0.36
_L
JL
J_
10 100 1,000
INSTANTANEOUS DISCHARGE, IN CUBIC FEET PER SECOND
10,000
Figure 9.-Phosphorus concentration versus stream discharge for the Tioga River
at Lambs Creek, Pa. (station 1516820).
19
-------
To resolve the question of validity of the relationships between concen-
tration and discharge for nutrient species, the analyses of variance (ANOVA)
test (Mendenhall, 1971) was applied using computer program REGPLOT. The ANOVA
test was used to determine whether the variation about the least-squares-
regression curves relating concentration to discharge is significantly differ-
ent from the variation about the mean concentrations. In this test, the
variance about the regression curve, V±, and the variance about the mean con-
centration, V2t are computed for each station. The variances, Vj and ^2, are
explained by
(C0 - Cc)2 , and (-14)
N - 2
N
V, = 1
2 N - 1
(15)
where N is the number of concentration-discharge observations, Co is an obser-
ved concentration, cc is the corresponding concentration computed by the
regression curve, and ~C0 is the mean concentration. The ratios of variances,
V1/V2* are then compared to standard F-distribution values for the 95th per-
centile of significance. The resulting number of significant differences are
as follows:
Nutrient Number of Number of significant
characteristic stations differences^ at 95-percent level
N 27 3
N03 36 2
NH4 27 0
P 34 2
P04 20 0
According to the definition of the F-distribution at the 95th percentile, 5
percent of the variances would be significantly different if discharges and
nutrient concentrations were drawn from random numbers. The results in the
above table indicate that, on the average, there is no significant difference
between the variation about the least-squares-regression curves and the vari-
ation about the mean concentrations. Because of this finding, it was decided
for this study that nutrient parameters should be calculated from average
nutrient concentrations rather than from nutrient transport curves. Conse-
quently, average nutrient loads, Ijj, in tons/year, were computed for each
station by using the equation
Ln = 0.986 CnQ (16)
where Cn is the average nutrient concentration is mg/L, Q is the 10-year
mean daily discharge in ft3/s, and 0.986 is the units conversion constant.
20
-------
Accuracy of Nutrient Characteristics
The accuracy of average nutrient concentrations and loads is difficult
to evaluate. Errors in these characteristics may be related to (1) discrete
time sampling, (2) field sampling techniques, (3) sample storage, and (4) lab-
oratory analysis. The relative effect of the last three error sources is
generally minimal, although certain properties such as NH^ concentrations are
sometimes subject to considerable error. The effect of sampling at discrete
time intervals (error type 1) is quite variable and is dependent on the dis-
tribution of sample coverage during critical periods or extreme events. Al-
though accuracies of the nutrient characteristics cannot be evaluated directly,
some inferences can be made from the standard errors of estimate, derived from
the regional multiple-regression analysis, which are discussed later.
BASIN CHARACTERISTICS
A basin characteristic as used in this report is a numeric value defin-
ing some unique aspect of a drainage basin. The basin characteristics ini-
tially considered were those related to processes known to control sediment,
dissolved solids, nitrogen, and phosphorus in streams. The characteristics
compiled for this study were limited, however, to those for which data were
available.
Two basic procedures are used in computing basin characteristics. First,
a basin characteristic is averaged by area weighting within each drainage area
to account for spatial variations. Area-weighted averages are computed by
overlaying a grid of a known scale on a map depicting a specific characteris-
tic such as basin slope. The values of the characteristic at grid intersec-
tions are summed and averaged. The grid-overlay method is used also to deter-
mine proportional areas for some characteristics, such as land use, by count-
ing the grid intersections falling in each specific land-use category within
a drainage basin. The proportion of each land use is, in this example, com-
puted by dividing the number of grid intersections overlying a land use by the
total number of intersections in the basin.
The second procedure for computing basin characteristics requires that
time-variable characteristics, such as climate or streamflow, must represent
a long-term average, or more specifically for this study the 10-year period
(1966-75). However, if year-to-year changes of a characteristic are known to
be small and the period of data available is short, the characteristic is com-
puted for 1-year during the study period.
A total of 57 basin characteristics was compiled in this study. They are
divided into six categories as follows: (1) climate, (2) topography, (3) ge-
ology, (4) soils, (5) streamflow, and (6) land use. Data sources and methods
of computing each basin characteristic are discussed for each category in the
following sections. Basin characteristics are tabulated in appendix 2 for
80 subbasins of the study region.
21
-------
Climate
Five climatic characteristics were computed from isohyetal and isother-
mal maps using an area-weighting technique. The following are climatic char-
acteristics and data sources used in this study:
1. Mean annual precipitation (PRECIP), in inches, from basin characteristics
published in Page (1970) and Darmer (1970), and from isohyetal maps
based on 1931-1960 precipitation data (Flippo, 1977, plate 2; and
Dethier, 1966).
2. Twenty-four hour rainfall intensity having a 2.33-year recurrence inter-
val (124,2), in inches, measured from an isohyetal map by Reich,
McGinnis, and Kerr (1970, fig. 8) with modifications made by the USGS
office in Harrisburg, Pa. (H. Flippo, personal commun.).
3. Mean annual snow accumulation (SN), in inches, from basin characteristics
published in Page (1970), a map prepared by U.S. Weather Bureau (1964)
for Pennsylvania, and a map for New York by the U.S. National Oceanic
and Atmospheric Administration (NOAA) (1972, p. 18).
4. Mean minimum January temperature (MINJAN), in degrees Fahrenheit, from
Page (1970), Darmer (1970), a map for New York prepared by the NOAA
(1972, p. 21), and a map for Pennsylvania prepared by the USGS office
in Harrisburg based on 1931 to 1952 temperature records (H. Flippo,
personal commun.).
5. Rainfall erosivity factor (R) according to the universal soil-loss equa-
tion (Wischmeier and Smith, 1965).
Topography
The following eight topographic characteristics are extracted from pub-
lished sources or computed from maps as follows:
1. Total drainage area (AREA), in square miles, obtained from the latest USGS
streamflow data reports or measured by counting grid intersections of a
known scale overlain on l:250,000-scale topographic maps.
2. Contributing drainage area (CONTDA), in square miles, is the total drain-
age area minus the area upstream from lakes and reservoirs, measured by
the grid-overlay method using l:24,000-scale topographic maps, or from
Williams and Reed (1972).
3. Main channel slope (SLOPE), in feet per mile, determined from elevations
at the 10- and 85-percentiles of the distance along the channel from
the gaging station to the divide (Benson, 1962). Data sources are
Darmer (1970), Page (1970), and l:250,000-scale topographic maps.
4. Average basin slope (BSLOPE), in feet per thousand feet, based on the av-
erage of 25 or more slopes taken at points on an equal-spaced grid
22
-------
pattern overlain on l:250,000-scale topographic maps.
5. Percent of basin having slopes greater than 20 percent (SLGT20), based on
25 or more points from an equal-spaced grid pattern overlain on
l:250,000-scale topographic maps.
6. Area of lakes and ponds (STOR), in percent of drainage basin, determined
from l:24,000-scale topographic maps, Darmer (1970), and Page (1970).
7. Mean basin elevation (ELEV), in thousands of feet above mean sea level,
was determined from 25 or more equal-spaced grid points on 1:250,000-
scale topographic maps.
8. Drainage-density index (DDI), in miles per square mile, is the ratio of
the total length of channels divided by the drainage area as determined
from 1:24,000-scale topographic maps.
Geology
Nine geologic characteristics used in this study are based on generalized
geologic maps of Pennsylvania (Socolow, 1960) and New York (Hollyday, 1969).
Characteristics representing geologic units, listed as items 1-6 below, have
been selected on the basis of (1) broad groups of formations caused by similar
processes and thus having similar physical properties, and (2) specific rock
types that could have regional effects on the water-quality characteristics
under study. The proportion of a basin underlain by each geologic unit was
determined using the grid-overlay method.
Selected ground-water characteristics, numbered 7-9 below, are included
in addition to the geologic units. The ground-water characteristics are based
on median ground-water values for each rock formation according to Seaber and
Hollyday (1965, 1966), Seaber (1968), and Hollyday (1969). Area-weighted av-
erages of ground-water characteristics were computed by the grid-overlay meth-
od. Geologic and ground-water characteristics used in this study are:
1. Undifferentiated sedimentary geologic units (SED), expressed as percent of
drainage area.
2. Undifferentiated metamorphic and igneous geologic units (METIG), expressed
as percent of drainage area.
3. Limestone and dolomite (LIMDOL) geologic units, expressed as percent of
drainage area.
4. Coal formations (COAL), expressed as percent of drainage area.
5. Triassic sedimentary geologic unit (TRIAC), expressed as percent of drain-
age area. .
6. Area glaciated (GLAC), in percent of drainage area (Fenneman, 1928).
23
-------
7. Median dissolved-solids concentration of ground water (GEOTDS), in mg/L.
8. Median nitrogen concentration of ground water (GEON), in mg/L.
9. Median specific capacity of shallow wells (SPCAP), in (gal/min)/ft of
drawdown.
Soils
Twenty-one area-weighted average soil characteristics were computed from
generalized soil-association maps and associated soils data. The soil charac-
teristics used in this study were tabulated and keypunched on computer cards
for each soil series in the study region. Chemical and mechanical data defin-
ing the first 12 soil characteristics listed below were obtained from U.S.
Soil Conservation Service (SCS) (1974a, 1974b), Cunningham and others (1972,
57 p.), Cunningham and others (1972, 805 p.), Ciolkosz and others (1972,1974),
Peterson and others (1968, 1972), and Ranney and others (1970, 1972). Data
for the remaining nine soil characteristics were obtained from SCS standard
soil-interpretation forms (SCS-soils-5) for each soil series (U.S. SCS, 1971).
A computer program (SOILS) was developed to compute a table of average soil
characteristics for each soil association in the study region based on known
percentages of the major soil series comprising each soil association. The
21 soil characteristics are tabulated in appendix 3 for the principal soil
associations in the basin.
The next step in the procedure was to determine percentages of soil asso-
ciations in each drainage basin. These percentages were measured by the grid-
overlay method using generalized soil-association maps for Pennsylvania (U.S.
SCS, 1972) and for New York (Arnold and others, 1970). Computer program SOILS
was then used to compute area-weighted average soil characteristics for each
basin based on the percentages of soil associations and the table of soil
characteristics (appendix 3).
The extensive selection of soil characteristics is intended for experi-
mental purposes because the characteristics that control the water-quality
processes in the soil profile generally are not well known. The soil charac-
teristics used in this study are as follows:
1. Clay content (CLAYA) of the A horizon, in percent by weight.
2. Silt content (SILTA) of the A horizon, in percent by weight.
3. Soil nitrogen (SOILNA) in the A horizon, in milliequivalents per 100 grams
(meq/100 g).
4. Soil-nitrogen (SOILNG) concentration in the A, B, or C horizon, whichever
is greatest, in meq/100 g.
5. Extractable-acidity (XACIDA) concentration in the A horizon, in meq/100 g.
6. Extractable-acidity (XACIDG) concentration in the A, B, or C horizon,
whichever- is greatest* in meq/100 g.
24
-------
7. Extractable-cations (XCATA) concentration in the A horizon, in meq/100 g.
8. Extractable-cations (XCATG) in the A, B, or C horizon, whichever is great-
est, in meq/100 g.
9. Cation-exchange capacity (CECA) of the A horizon, in meq/100 g.
10. Cation-exchange capacity (CEC6) of the A, B, or C horizon, whichever is
greatest, in meq/100 g.
11. pH (PHA) of the A horizon (in H20).
12. pH (PHL) of the A, B, or C horizon, whichever is lowest (in HO).
13. Soil erodibility (KA) of the A horizon according to the universal soil-
loss equation (Wischmeier and Smith, 1965).
14. Permeability (PERMA) of the A horizon, in in/hr.
15. Permeability (PERML) of the least permeable soil horizon, in in/hr.
16. Hydrologic soil groups (HSG) according to SCS. Soil groups A, B, C, and
D are arbitrarily equated to 1, 2, 3, and 4, respectively.
17. Available water capacity (WATCAP), computed as a depth-weighted average
of the A, B, and C soil horizons, in inches of water per inch of soil.
18. Depth to bedrock (BDRK), in inches.
19. Proportion of soil (LT200A) in the A horizon that passes the No. 200 mesh
sieve, in percent by weight.
20. Gravel content (GRAVA) in the A horizon, in percent by weight.
21. Stones greater than 3 inches (STONEA) in the A horizon, in percent by
weight.
Streamflow
The six streamflow characteristics used in this study are based on the
USGS WATSTORE computer file of mean daily flows and peak flows. Flood-
frequency characteristics were computed by USGS computer program J407 which
is based on Bulletin No. 17 of the Hydrology Committee of the U.S. Water
Resources Council (1976). Streamflow characteristics used in this study are
as follows:
1. Mean annual stream discharge (MAQ10) for the period of water years 1966 to
1975, in ft3/s.
2. Mean annual discharge (MAQ9) for the period of water years 1966 to 1975,
excluding 1972, in ft3/s.
25
-------
3. Largest peak discharge (PK10) for the period of water years 1966 to 1975,
in ft3/s.
4. Peak discharge having a recurrence interval of 2 years (P2), in ft-Vs,
based on the period of record for each station.
5. Peak discharge having a recurrence interval of 25 years (P25), in ft^/s,
based on the period of discharge record for each station.
6. Ratio of the largest peak discharge during the study period to the peak
discharge having a recurrence interval of 10 years based on the period
Of discharge record (PK10/P10).
Land Use
Characteristics of land use in this study are described using designated
level I categories according to Anderson, Hardy, Roach, and Witmer (1976).
Percentages of land uses in Pennsylvania were measured by the grid-overlay
method using l:250,000-scale land-use maps. These maps, based on 1974 aerial
photography, are preliminary copies prepared by the U.S. Geological Survey's
Land Information and Analysis office. Land-use percentages for basins in New
York were computed by Cornell University using the Land Use and Natural Re-
source inventory (LUNR) of New York State (Crowder, 1974). Computations were
made by digital computer from a data-storage system utilizing one-square-
kilometer grid cells. LUNR data are based on 1967 and 1968 aerial photography.
In addition to defining land use by categories, a land-cover index (C-
factor) was also used as a land-use characteristic. The C-factor is a ratio
of soil loss from land cropped under specific conditions to the corresponding
loss from tilled, continuous fallow as used in the universal soil-loss equa-
tion (Wischmeier and Smith, 1965). Area-weighted average C-factors were com-
puted based on generalized values of C for agriculture, urban, forest, and
extractive land uses (John Robb and others, oral and written commun., SCS,
HarriSburg, Pa., 1976).
Level-I land-use categories and the C-factor used in this study are as
follows:
1. Percent of drainage area urbanized (LU1).
2. Percent of drainage area under agriculture (LU2).
3. Percent of drainage area forested (LU4).
4. Percent of drainage area covered by water (LU5).
5. Percent of drainage area in a disturbed condition such as extractive,
strip mines, construction (LU7).
6. Average basin C-factor according to the universal soil-loss equation (C).
26
-------
Two additional characteristics of agricultural management were compiled
to quantify chemical fertilizers and animal wastes applied to each basin.
These characteristics are:
7. Tons of phosphorus applied per basin (AGP); includes estimates of chemi-
cal fertilizer and animal waste, in tons per year, as phosphorus.
8. Tons of nitrogen applied per basin (AGN); includes estimates of chemical
fertilizer and animal waste, in tons per year, as nitrogen.
Characteristics of agricultural fertilizer are intended to be rough indi-
cators of the combined effects of chemical fertilizer and animal wastes on the
nutrient levels in streams. The annual nutrient application in each basin,
expressed in tons of nitrogen (AGN) or phosphorus (AGP), was computed for each
basin by the equation
n
AGP or AGN = Agr I (T.) (P±) (17)
where: Agr is the area of agricultural land in the basin, in square miles;
P± is the fraction of county i in the basin; T+ is a loading density (see
explanation below) for county i in tons per.year of nitrogen or phosphorus per
square mile of agricultural land; and n is the number of counties or parts of
counties in .the basin. The nutrient-loading density factor for each county,
TJ;, is based on the equation
+ Ta , (18)
Acp
where: Tc is the annual tonnage of chemical fertilizer for each county, ex-
pressed as nitrogen or phosphorus (U.S. Dept. of Agriculture, 1973 p. 208-211;
New York State Dept. of Agriculture and Markets, 1969; U.S. Dept. of Commerce,
1972a, table 19); Ta is the annual tonnage of animal ;wastes for each county,
expressed as nitrogen or phosphorus (see explanation below) ; Acp is the area *
of cropland and pasture for each county, -in square. miles (State Conservation
Needs Inventory Committee, 1967, p. 35-36; U.S. Soil Conservation Service, s
1967, p. 32-33).
27
-------
The annual tonnages of nitrogen and phosphorus contributed by animal
wastes (Ta) computed for each county by multiplying animal densities times
average animal nutrient-production factors, are listed below (Omernik 1976,
P. 13).
Total N Total P
Animal (tons/animal)/yr (tons/animal)/yr
Cattle 6.34 x 10"^ 1.94 x 10~*
Hogs 1.07 x 10 j 3.56 x 10~^
Sheep 1.11 x 10 1.62 x 10~J
Chickens , ,
Layers 4.63 x 10~7 1.76 x 10~g
Broilers 4.30 x 10~* 9.92 x 10~J
Animal densities were obtained from agricultural census (U.S. Dept. of Com-
merce, 1972b, 1972c, tables 8 - 11). Annual tonnages (T±) of nitrogen and
phosphorus estimated for chemical fertilizer plus animal wastes are tabulated
for each county in the study area in appendix 4.
MULTIPLE-REGRESSION ANALYSIS
The linear-regression model (eq. 1) and the log-transform model (eq. 3)
were initially tested for four water-quality characteristics (SEDYLD, DSYLD,
NAVE, and PAVE). The linear model for PAVE was considered unsuccessful.
Moreover, by comparison of the linear and log-transform models it was found
for SEDYLD, DSYLD, and NAVE that the log-transform model provided lower stan-
dard errors (1 to 7 percent lower) and higher explained variance (6 to 18 per-
cent) . In addition, the residuals (differences between the observed and cal-
culated values) were generally more randomly distributed for the three log-
transform models. Therefore, the log-transform form of model was used to
develop regressions for all the water-quality characteristics. The results
of regression analyses for 17 water-quality characteristics are listed in
table 1. To demonstrate this table, the regression model for sediment yield
(SEDYLD) is
SEDYLD - (3.24xl06)(PHA)~6-66(LU2+l)°'288 (19)
The accuracy of an estimate computed by this equation is indicated by the
standard error of estimate (table 1), which implies that approximately two-
thirds of the sediment yields computed for the 28 stream sites used in this
regression have an error within + 40 percent when compared to measured yields.
The percent of variation explained, shown in table 1, is calculated as the
square of the multiple-correlation coefficient times 100 (Afifi and Azen,
1972, p. 117). One hundred percent of variation explained would indicate a
perfect regression model with no error. Zero percent indicates that the vari-
ation about the regression model is equivalent to the variation about the mean
of the water-quality characteristic, in which case, the model serves no pur-
pose.
As previously explained, a value of one was added to several of the Inde-
pendent variables to avoid taking logarithms of zero. In some cases, a number
28
-------
TABLE 1.—Results of maltiple-linear-reyression analysis of logarithmic-transformed variables
Water-quality
. characteristics1'
yl/
SEDTLD-Sediment yield
in (tons/ral2)/yr
SEDCONC-Sedinent concen-
tration in og/L
DSYLD-Dissolved-solids
yield in (t6ns/al2J/yr
DSCONC-Dissolved-Solids
concentration in mg/L
VSEXP-V
DSCOEF-1/
HAVE— average nitrogen
concentration in ng/L
NSD-NItrogen standard
deviation in mg/L
N03AVE-Average nitrate
concentration in ng/L
H03SD-Nitrate standard
deviation in mg/L
N03YLD-Nitrate yield
in (tons/«i2Vyr
NH4AVE-Average ammonia
concentration in ng/L
PAVE-Average phosphorus
concentration in mg/li
PSD-Phosphorus standard
deviation in ng/L
PYLD-Phosphorus yield
in (tons/ni2yyr
PO4AVE-Average phosphate
concentration in mg/L
P04SD-Phosphate standard
deviation In ng/L
flmlpncnrfpnr varlahlPK v' ^2/.>\RfiBression coefficients, b «S3/
fmn~6-66 miTiiV288
mur6'47 fMAQ9N-1.29 , .255
(rlIA) ^ARE/J OU2I1)
(miiir438 mnur282 /MAQIOU.IS frOALm-i24 fxr,Tn-333
/i in ji ^ "^82 /tin 1.1 \ -296 /r-rt^t » i \ • ^-^^ /vr-A-rv^ -272
No significant regression relationship established
No significant regression relationship established
ruATnAn1-55 (*** uV68^ r-inrrv150 ruiinr378
(, WAI LAI',; V AREA 7 CoLUrJiJ (LUrrLJ
(lfA«)— ^CAP)-^
fuATrjvM3'29 n-wnnAr4-89 TAGN nV640 fruiTAi1'63
(WAIUUV ti.T2OOAj \AREA / ^CLAYA;
/AGN .A-676 /rnFrrrl2-19 fuA-rrirV913 nmiir317
VAREA / tnctcirj (WATCAFJ (LUIII;
No significant regression relationship established
mrrTrm-592 ^AGP +iV937 rriiA^6-^4 MTinoAl-3-67
(ntllutlj VAREA y IFUAJ tLTZUOA;
(pAVE)Ml
ii+n.^-eg,*^
fWATCAP)~4'76 rpFRMA^1-16 f^P ,:\l.2i, .834
Regression
constant
a2/
3.24xl06
2.92xl06
8.43
l.lSxlO1
7.02
2.30xl05
1.94X101
1.29xl09
6.25xlO~4
4.51X10'1
B.UxlO"1
3.10xlO~2
1.85xlO~8
2.98xlO~2
Standard
error of
estimate
in percent
40
40
24
17
17
26
50
56
31
75
47
36
44
69
Percentage
of variation
explained'"
63
72
82
89
77
68
76
71
89
58
84
68
74
56
Number
of
stream
stations
28
28
26
26
48
48
27
27
58
58
24
46
49
49
20
20
20
N>
VO
Jftefined in section entitled "Water-quality characteristics".
Defined in section entitled "Basin characteristics".
. . . X
-^According to equation 3: Y = a X l X2 2
, where R is the multiple-correlation coefficient (Afifi and Azen, 1972, p. 115-117).
-------
other than one was tested in an attempt to improve the linear fit of the mod-
el; however, no improvements were achieved in the standard error or the per-
cent of variation explained.
Sensitivity of Independent Variables
From the standpoint of applying the regression models in table 1, it is
useful to evaluate the relative effect (sensitivity) of each independent vari-
able on a water-quality characteristic. The relative magnitudes of regression
coefficients may not be proportional to the relative sensitivity of each inde-
pendent variable because the coefficients are dependent on both the magnitude
and variation of that independent variable. The relative sensitivity of each
independent variable in a particular regression equation can, however, be
approximated by comparing the regression weights of each independent variable.
Regression weights are similar to coefficients except that they are computed
by first standardizing the dependent and independent variables. Standardized
variables are obtained by subtracting the mean and dividing by the standard
deviation. These variables have a mean of zero and a standard deviation of
one. On the basis of this approach, table 2 shows the computed regression
weights for each independent variable. The observed range of each variable
is also shown. Independent variables are listed from left to right in rela-
tive order of decreasing absolute values of regression weights. It is note-
worthy that in five of the regression models the water-quality characteristics
are most sensitive to the land-use related variables.
Validity of Regression Models
The acceptability of regression models should not be based entirely on
statistical tests. The independent variables and regression coefficients of
each equation also must be evaluated from the standpoint of conceptual knowl-
edge of the water-quality processes. In this section, two basic questions are
considered. (1) Is each of the independent variables related directly or
indirectly to the water-quality characteristic? (2) Is the sign of each re-
gression coefficient realistic in terms of intuitive understanding? In the
first consideration, it is essential to know if any of the independent vari-
ables are surrogates that indirectly explain some other effect on water qual-
ity. For example, percent urbanization indirectly represents the effect of
sewage effluent on the stream load of total nitrogen. In this case, percent
urbanization is used as a surrogate. Second, the sign of a regression coeffi-
cient indicates a direct (positive sign) or inverse (negative sign) relation-
ship between the dependent and independent variable. If the sign of a regres-
sion coefficient is contrary to intuitive understanding of the process in-
volved, one of the following causes could be indicated:
1. The process involving the effect of an independent variable on a water-
quality characteristic is not well understood.
2. The independent variable is a surrogate for another variable.
3. A large error occurred during compilation of a dependent or independent
variable. .
30
-------
TABLE 2.—Ranges of observed variables, anil regression weights and selected correlation coefficients of independent variables
Mater-quality
characteristic
•inlAUB~Baximm
SEDYLD
21.3-299.
SEDCONC
13.3-295.
DSYLD
33.4-308.
DSCOHC
29.0-282.
HAVE
.40-1.59
HSD
.18-. 98
H03AVE
.15-7.45
, H03SD
-. .07-4.14
N03YLD
.27-8.98
PAVE
.02-1.24
PSD
.01-1.18
PYID
,03-. 35
P04AVE
.O1-.20
PO4SD
.01-.19
Independent variables
minimum-maximum/ regression
PHA
4.9-6. 3/-. 71
PHA
4.9-6.3/-.61
LU1
0-13.9/.60
. LU1
0-13. 9/. 58
WATCAP
.06-.13/1.0
WATCAP
.06-.13/.72
AGN/AREA
0-41. 5/. 61
WATCAP
.06-.16/.83
AGN/AREA
0-36.8/.63
METIG
0-67. 11. 63
PAVE
.02-1. 24/. 92
LU1
0-12. 7/. 80
WATCAP
.06-.13/-1.30
XCATG
7.3-18.4/.93
LU2
0-81.0/.42
MAQ9/AREA
. 94-1. 8/-. 37
L02
0-64. 3/. 48
UI2
0-64.3/.56
AGN/AREA
.28-7.50/.92
PRECIP
33.6-42.S/-.43
WATCAP
.06-.16/.46
LT200A
42. 1-78. 5/-. 68
PRECIP
33.6-46.0/.28
AGP/ AREA
0-13.2/.54
AGP/AREA
0-H.8/.31
PEEMA
.73-6.03/1.04
BSLOPE
60.-150./-.68
UJ2
0-81. 01. 33
MAQ10/AREA
.98-1.99/.45
COAL
0-81.0/.33
SLOPE
1.8-289.0/.59
LU5
0-4.4/-.42
AGN/AREA
0-41. 5/. 55
LU1
0-12. 9/. 25
PHA
weight3
COAL XCATG
0-81. O/. 35 4. 9-34. 8/. 30
XCATG
4.9-34. 8/. 27
LU1
0-3. 11. 58
LU1
0-3. 11. 26
CLAYA
10. 4-21. 8/. 25
WATCAP
.07-.14/.24
LT200A
4. 9-6. 6/. 50 42.1-78.5/-.47
AGP/AREA Till
0-11. 8/. 75
HATCAP
.06-.13/-.65
0-12. 11. 66
Pairs of independent variables
bivariate correlation coefficients of logarithms'*
/MAg9\
(LU2-U)v.VAREA/
-.52
/MAQ10N
(LU2+l)v. (COAL+l) (LU2+l)v.V AREA/
-.62 -,54
(L02+l)v. (COAL+1)
-.62
(f RECIP)v. (WATCAP)
.51
(LT200A)v. (WATCAP) lCLAYA)v. (WATCAP) (CLAYA)v. (LT200A)
.84 .58 .50
(PRECIP)v. (WATCAP)
•69 / x
fAGP /\ (AGt \
(LT200A)v. (METIG+1) (LT200A)v. \AREA V (METIG+l)v. \AREA J
.70 .57 .56
(^+l)
(LUI+DV.VAREA J
.76
/AGP \
(PERMA) V. (WATCAP) (UU+DvAAREA /
.84 .58
(XCATG)v. (WATCAP)
.72
Defined in section entitled "Water-quality characteristics".
2 Defined in section entitled "Basin characteristics".
Refer to documentation of U.S. Geological Survey computer program DO095 "General
Regression (Step Backward) — STATPAC" -(written coumraication, Gary I. Seiner, 1975).
Only those correlation'coefficients which exceed 0.5 are shown.
-------
4. Significant cross-correlations between independent variables may cause
the regression coefficients to be inaccurate,
5. The relation may be spurious. That is, the apparent significance of an
independent variable may be due to chance.
These aspects were considered in the selection of independent variables and
for each water-quality model and are discussed in the following sections.
Sediment Models
The suspended-sediment yield (SEDYLD) and concentration (SEDCONC) models
both had a standard error of estimate of 40 percent (table 1). This level of
error is not significantly larger than estimates of the errors in sediment
loads and concentrations computed by the transport-curve method. (See
"Accuracy of the generated sediment loads.") The percent of drainage area
under agriculture (LU2) is significant in both sediment models. Agricultural
land use is considered generally to be a major source of. sediment. The 9-year
mean-annual discharge (MAQ9/AREA) is inversely related to SEDCONC as shown by
the negative sign of the regression coefficient in table 1. This indicates
that discharge-weighted sediment concentrations are more dilute in areas of
higher average flows. Variations in sediment yields, however, are apparently
not affected significantly by the average flow level.
The inverse relationship with soil pH (PHA), as indicated by the negative
sign of the regression coefficient shown in table 1, is difficult to explain.
Soil pH may be explaining a closely related soil property or a land use. It
should be noted that the range in soil pH is 4.9 to 6.3, indicating relatively
acidic soils. Correlation coefficients between independent variables do not
exceed 0.52. (See table 2.)
Dissolved-Solids Models
The regression models for dissolved-solids yields (DSYLD) and concentra-
tions (DSCQNC) explain 82 and 89 percent of the variation, respectively, and
the standard errors of estimate are 24 and 17 percent, respectively. (See
table 1.)
Four of the five independent variables found significant in the dissolved-
solids yield (DSYLD) and concentration (DSCONC) models define realistic
sources of dissolved constituents. These are (1) percent urban (LU1), (2) per-
cent agriculture (LU2), (3) extractable cations in soil (XCATG), and (4) per-
cent of basin overlying coal formations (COAL). The user of these models
should recognize that LU1 may be a surrogate defining the effects of domestic-
sewage effluents. Also, the characteristic, COAL, may represent the effect of
acid-mine drainage, which is primarily a result of exposing coal formations to
air and water. Therefore COAL may represent the effect of land use rather
than geology. The 10-year mean-annual discharge, (MAQ10/AREA), relates to
increased yields of dissolved solids in areas of high average flows; however,
the effect of flow on concentrations is not indicated. Correlation coeffi-
cients between independent variables do not exceed 0.62. Regression models
32
-------
for DSEXP and DSCOEF did not appear to be meaningful. These models could
only explain about 25 percent of the variation. The standard errors of esti-
mate of the DSCOEF and DSEXP models were about 90 and 11 percent, respectively.
Nitrogen Models
Five of the six nitrogen models were successfully calibrated with real-
istic results. Standard errors of estimate ranged from 17 to 56 percent, and
the percent of variation explained ranged from 68 to 89 percent. (See
table 1.) A sixth model, the average ammonia concentration (NH4AVE), was
considered to be of minimal value based on the small proportion of explained
variation (about 27 percent). The difficulty in deriving a useful ammonia
model may be due in part to the biochemical instability of ammonia and per-
haps in part to laboratory analytical error.
Four independent variables found significant in various combinations in
the NAVE, N03AVE, and N03YLD models describe possible sources of nitrogen.
These independent variables are: (1) agricultural nitrogen (AGN), (2) percent
urbanization (LU1), (3) mean annual precipitation (PRECIP), and (4) water
capacity of soil (WATCAP). LU1 may be a surrogate defining the effects of
domestic sewage effluents. Water capacity explains the ability of the soil
to support vegetation, which indirectly relates to the occurrence of nitrogen
in the soil. A fifth variable found in the total nitrogen model (NAVE), chan-
nel slope (SLOPE), indicates lower concentrations of nitrogen as a function of
lower slopes. This may be the result of increased biological uptake of nitro-
gen occurring in the more sluggish streams which are characterized by lesser
slopes. The cross correlations between independent variables in each of the
three models were relatively small (less than 0.69).
It is difficult to explain the cause and effect of characteristics defin-
ing the standard deviation models for total nitrogen (NSD) and nitrate (N03SD).
As shown in table 1, the independent variables (AGN, LU1, PRECIP, and WATCAP)
that define sources of NAVE and N03AVE also explain the standard deviations,
NSD and N03SD. The significance of LU5 in the NSD model indicates that smaller
variations in total nitrogen are associated with greater parts of the drainage
area covered by water. A possible explanation is the biological uptake of ni-
trogen occurs more readily in lakes, ponds, and wide sluggish channnels than
in rapidly flowing streams, therefore tending to dampen seasonal variations.
There is no apparent explanation for the association of N03SD to CLAYA (the
percent of CLAY in the A soil horizon) and LT200A (the percent soil passing the
No. 200 sieve). It is possible that CLAYA and LT200A may be surrogates for
other regional parameters.
Phosphorus Models
The standard errors of estimate of the five phosphorus models ranged from
36 to 75 percent, and the percent of variation of the dependent variable ex-
plained ranged from 56 to 84 percent (table 1). These results indicate lower
model accuracies than those for the nitrogen models.
The three primary phosphorus models (PAVE, PYLD, and P04AVE) incorporate
the effects of agricultural phosphorus (AGP) and urbanization (LU1), which
33
-------
define possible man-induced sources of phosphorus. LU1 may be a surrogate
defining the effects of domestic-sewage effluents.
The association between metamorphic rocks-igneous rocks (METIG) and
phosphorus concentrations is consistent with results found by Omernik (1976,
p. 62-63) in the Eastern United States, where forested streams overlying
metamorphic and igneous rocks were shown to have higher phosphorus concentra-
tions than streams draining sedimentary rocks. The effect of the combination
of water capacity (WATCAP) and permeability (PERMA) of soils on orthophosphate
is difficult to define. The relationship of WATCAP to P04AVE is inverse,
whereas that of PERMA is direct. The cross-correlation between WATCAP and
PERMA is high (0.84); therefore, the effect of these variables on P04AVE
should not be evaluated separately.
The remaining variables that affect phosphorus (PAVE) represent chemi-
cal processes rather than sources of phosphorus. The association of low soil
pH (PHA) to decreased total phosphorus concentrations (PAVE) may be a result
of increased anion-adsorption capacity as water passes through the soil
co.lumn, which permits less phosphorus to reach the ground water (Barrow, 1970).
The inverse relationship of PAVE to the percent of soil passing a No. 200
screen (LT200A) is similar. As the soil becomes finer, the surface area of
soil particles increases, causing increased phosphorus adsorption in the soil
horizon.
Accuracy of Regressj.on Models
As mentioned earlier, the accuracy of a regression model is often judged
on the basis of the standard error of estimate (SEE). (See table 1.) The
apparent SEE of any regression model is comprised of both model error and
sampling error. True model error is introduced by nonlinear relationships,
incorrect choice of independent variables, or errors in the compilation of the
dependent or independent variables. Sampling error involves temporal and
spatial sampling errors that result from relatively short records and sparse
distribution of stream-sampling sites.
The true error of a regression model is approached as the length of water-
quality records and the number of subbasins used for calibration approach
infinity. True error can be estimated indirectly for a particular regression
model by a statistical procedure described by Moss (1976). This procedure is
based on a Monte Carlo simulation of probable standard errors for a selected
regression model by statistically representing a large number of stream sites
and long periods of water-quality records. Estimates of true model error
34
-------
were made for two regression models defining suspended-sediment yield (SEDYLD)
and dissolved-solids yield (DSYLD). Computations were made using computer
programs (M. E. Moss, written commun., 1977) available on the USGS computer
system. Results are shown below:
Apparent standard
error in percent Simulated true
Model (from table 1) model error in percent
SEDYLD 40 38
DSYLD 24 24
A comparison of the apparent standard error and the simulated true model
error shows little difference for either model. This indicates that the
apparent standard error is predominantly model error and is not significantly
affected by temporal and spatial sampling errors. Consequently, development
of more appropriate models and independent variables, and improvement of the
accuracy of variables, are possible means of improving the standard errors of
estimate.
Independent Testing of Regression Models
It is desirable to assess the usefulness of the regression models by
comparing model results with observed water quality for several independent
subbasins that were not used in model calibration. However, all available
data for the study period were used for model calibration. Consequently,
model testing is based on limited new data collected for 23 subbasins during
water year 1976 and part of 1977. Ten of these subbasins were used (or, if
not, their drainage areas are nearly equivalent to those used) for model cali-
bration. The 13 additional basins were not used in deriving the models. Part
of the nutrient data available for verification was collected by the Pennsyl-
vania Department of Environmental Resources.
Table 3 is a tabulation of observed water-quality characteristics and
corresponding characteristics simulated by eight of the 14 regression models
given in table 1. Adequate data were not available to define sediment-
transport curves, and consequently sediment models are not included in table 3.
A comparison of observed versus simulated characteristics indicates generally
that the dissolved-solids and nutrient models provide useful estimates of
water quality.
To summarize table 3, the differences between the observed and simulated
values were computed as a percentage of the observed, and then averaged for
each water-quality characteristic. These average errors, except for the
P04AVE model, are less than or in close agreement with the standard errors of
estimate of the regression models shown in table 1. The large average error
of the P04AVE model (95.8 percent) is due mostly to the fact that only four
stations are represented and that one of these has a large difference between
the observed and simulated values (table 3). The occasional large deviations
between the observed and simulated values of some nutrient characteristics
may be due to the uncertainties of the estimated agricultural phosphorus or
nitrogen characteristics (AGP and AGN) for small basins.
35
-------
Table 3.--Testing of Regression Models
Station
number
Station name
Value
status
Observed and simulated values of water-quality characteristics
DSYLD
DSCONC
NAVE
N03AVE
N03YLD
PAVE
PYLD
POAAVE
1502770
1509150
1515050
1516350
1518000
1518550
1519500
1520500
1531000
1534300
1545600
1553500
1555210
1555860
1556480
1557550
1560510
1563500
1564995
1565510
1566010
1571197
1571505
Susquehanna R. nr Great Bend, Pa.
Crldley Cr. above East Virgil, NY.
Susquehanna R. at Say re. Pa.
Tioga R. nr Mansfield, Pa.
Tioga R. at Tioga, Pa.
Crooked Cr. at Tioga, Pa.
Cowanesque R. at Cowanesque, Pa.
Tioga ft. at Lindley, NY.
Chemung R. at Chemung, NY.
Lackawanna R. nr Forest City, Pa.
Young Woman's Cr. nr Renovo, Pa.
West Branch Susquehanna R. at Lewisburg
Middle Cr., Pa.
Beaverdan Branch Juniata R. , Pa.
Little Juniata R. on Rt 220, Pa.
South Bald Eagle Cr, on Rt 350, Pa.
Dunning Cr. off T-477 nr mouth, Pa.
Juniata R. at Maple ton Depot, Pa.
Honey Cr. at Reeds ville, Pa.
Kishacoqulllas Cr. at Lewis town. Pa.
Tuscarora Cr. at Port Royal, Pa.
Mountain Cr. at Jet. to Yellow Breaches
Pa.
Yellow Breeches Cr., Pa.
Observed --139.
Simulated 128.
Observed
Simulated
Observed -/154.
Simulated 149.
Observed
Simulated
Observed —112.
Simulated 133.
Observed 1/73.5
Simulated 60.9
Observed 1/92.4
Simulated 74.3
Observed -'<)t,t>
Simulated 84.2
Observed -120:
Simulated 91.6
Observed
Simulated
Observed —40.4
Simulated 45.5
. Pa. Observed 1/160 .
Simulated 136.
Observed
Simulated
Observed
Simulated
Observed
Simulated
Observed
Simulated
Observed
Simulated
Observed 140.
Simulated 172.
Observed
Simulated
Observed
Simulated
Observed
Simulated
Cr., Observed
Simulated
Observed
Simulated
Average absolute error as percent of observed: 15.0
'87.6 -0.92
84.5 .88
^96.2 i'l.14
94.3 .71
in!
1/74.8
74.1
1/95.7
91.0
^88.6
92.1
-113. 1.03
98.4 .82
•^25.1
28.4
1/98.9 1/1.09
84.4 .82
111.
137.
10.6 21.8
i'0.57
.75
±'.54
.96
.51
.62
.41
.43
.58
.68
.55
1.32
1.22
1.87
1.06
1.47
.87 .
.79
.48
1.76
1.10
.93
1.30
2.46
1.88
1.06
1.04
1.08
.94
1.95
2.21
26.0
0.90 -0.07
.88 .06
.82 -'.II
.94 .06
.05
.05
.62 l^.OS
.61 .07
1.26 .06
.82 .06
i'.02
.02
1/.03
.02
.10
.12
.62
.14
.83
.10
.11
.14
.13
.11
.08
.11
,15
.13
.07
.18
.12
.51
.11
.42
13.3 67.6
0.11 0.02
.03 .02
.18 1^02
.09 .03
.01
.04
.08 -'.03
.09 .04
.11
.10
.03
.03
.05
.07
23.1 95.8
•Station was used (or equivalent to station used) in calibration of regression model.
36
-------
Although the data used for testing have a very limited range, table 3 is
a reasonable representation of the accuracies of the models that may be ex-
pected if they are applied to previously unsampled streams.
APPLICATIONS OF REGRESSION MODELS
The multiple-regression models given in table 1 can be applied in a
generalized manner or on a site-specific basis. Examples of these applica-
tions and their limitations are discussed in the following sections.
Generalized Applications
The multiple-regression models can be used to estimate background water-
quality conditions by hypothetically removing the culturally induced effects
of land use. In this approach, land-use variables such as percent urbaniza-
tion (LU1) and percent agriculture (LU2) are set equal to zero. By doing so,
the effects of these given land uses are removed mathematically from the
model. By this method the equations in table 1 are used to estimate hypo-
thetical ranges of minimum and maximum values for each water-quality charac-
teristic. The estimated background ranges are compared to the observed ranges
of water-quality characteristics in table 4. These comparisons suggest that
the impact of land use on certain water-quality characteristics is consider-
able. For example, the maximum of the observed range of nitrate yields
(N03YLD) and phosphorus yields (PYLD) is greater than 10 times the estimated
background range. The ranges shown in table 4 are for a selected set of
stream stations that were used to calibrate each model. Actual ranges for all
possible stream sites in the Susquehanna River basin may differ from those
shown. Considering the broad areal coverage of the stream stations used for
each model (fig. 3), it is reasonable to assume, however, that these ranges
are representative of the study region.
Similar general applications of the regression models can be used to
evaluate the generalized effects of any independent variable. However, con-
sideration must be given to the limitations and cautionary aspects discussed
under "Limitations of the regression models."
Specific Applications
Regression models can be used to estimate water-quality characteristics
for specific stream sites in the study region. These estimates are based on
regression models given in table 1 and coupled with estimates of the specified
independent variables. Moreover, the independent variables can be hypotheti-
cally adjusted to evaluate the effects of changing land-use conditions. This
procedure is similar to the approach described above.
Limitations of the Regression Models
Application of the regression models and interpretation of results is
subject to a number of limitations. Each application should be evaluated on
the basis of the following five considerations.
1. The regression models developed in this study are limited to conditions
37
-------
TABLE 4.—Observed ranges of water-quality yields and concentrations and
background ranges simulated by regression models
Water-quality
characteristics1
SEDYLD-Sediment yield
in (tons/mi2)/yr
SEDCONC-Sediment concen-
tration in mg/L
DSYLD-Dissolved-solids
yield in (tons/ 'mi2)/ 'yr
DSYLD-Dissolved-solids
yield in (tons/mi 2J/yr
DSCONC-Dissolved-solids
concentration in mg/L
DSCONC-Dissolved-solids
concentration in mg/L
NAVE-Average nitrogen
concentration in mg/L
NSD-Nitrogen standard
deviation in mg/L
N03AVE- Average nitrate
concentration in mg/L
N03SD-Nitrate standard
deviation in mg/L
NOSYLD-Nitrate yield
in (tons/mi2)/yr
PAVE-Average phosphorus
concentration in mg/L
PSD-Phosphorus standard
deviation in mg/L
PYLD-Phosphorus yield
in (tons/mi2J/yr
P04AVE-Average phosphate
concentration in mg/L
P04SD-Phosphate standard
deviation in mg/L
Observed
range
Minimum Maximum
21.3 299.
13.3 295.
33.4 308.
33.4 308.
29.0 282.
29.0 282.
.40 1.59
.18 .98
.15 7.45
.07 4.14
.27 8.98
.02 1.24
.01 1.18
.03 .35
.01 .20
.01 .19
Simulated
background range
Minimum
16.2
13.1
16.9
16.9
17.4
19.3
.15
.25
.13
.06
.12
.01
".01
.03
.00
.01
Maximum
83.0
102.
36.0
60.7
29.6
33.2
.46
.75
.69
.46
•43
.14
".11
.03
.01
. .13
Culturally affec'
variables2
held constant
at zero
LU2
LU2
LU1.LU2.COAL
LU1.LU2
LU1.LU2.COAL
LU1.LU2
AGN
AREA'LU1
LU1.LU5
AGN
AREA
AGN
AREA
HA-LUI '
AGP
AREA
Tin AGP
LU1'AREA
Till AGP
LU1'AREA
ted Variables3
assumed
to be
natural
PHA
PHA'AREA
'^•S2
XCATS.^.COAL
XCATG
XCATG, COAL
SLOPE, WATCAP
PRECIP.WATCAP
WATCAP
CLAYA .WATCAP , LT200A
PRECIP.WATCAP
PHA.METIG.LT200A
PERMA .WATCAP
XCATG , BSLOPE .WATCAP
Defined in section entitled "Water-quality characteristics".
2Variables explained in section entitled "Basin characteristics".
Includes only those variables affected significantly by man.
'Variables explained in section entitled "Basin characteristics".
14Based on simulated background range of PAVE.
38
-------
in the Susquehanna River basin and in adjacent areas having similar
physiographic and hydrologic properties.
2. The regression models can only define the effects of the independent vari-
ables found significant for each model. These models do not include
basin characteristics that define the effects of major industrial point
sources of pollution or localized nonpoint sources. Consequently, con-
tributions by additional variables for each model should be considered
by the user.
3. The estimates of background water quality discussed earlier in "General-
ized applications," must be qualified as quasi-natural. The present
water quality of the least developed streams may be affected substan-
tially by air pollution, rainfall, and the after-effects of a previous
land use. The first two qualifications pertain primarily to nutrients
and the latter particularly to suspended sediment. Consequently, the
estimates of quasi-natural water quality should not be equated to pris-
tine conditions.
4. Interpretations of the causal effects of independent variables should be
judged carefully. Variables that indirectly explain the effect of
another variable can be misleading. These variables, referred to as
surrogates, are discussed in the section entitled "Validity of regres-
sion models." Although the inclusion of surrogates may be useful, the
user should be aware of their limitations before using these models in
decisionraaking processes.
5. Expected errors in predicted water-quality characteristics are indicated
by the standard errors of estimate listed in table 1. In cases where
the regression models are used to evaluate specific effects of one or
more independent variables, attention should be given to the cross-
correlations between variables. If two independent variables in a
regression model are highly correlated, the resulting regression coef-
ficients for these variables may be improperly defined. Consequently,
if either variable is held at a constant value while the other is hypo-
thetically varied, the resulting computation of the water-quality charac-
teristic may be significantly in error. Improper distribution of regres-
sion coefficients may occur, with cross-correlation coefficients as low
as 0.5; however, significant errors may not occur unless correlation
coefficients are 0.8 or larger. Correlation coefficients between inde-
pendent variables that exceed 0.5 are listed in table 2. Cross-
correlating independent variables will not have a large effect on the
accuracy of the regression model unless the effect of one of these vari-
ables is evaluated in the manner just described.
39
-------
DISCUSSION AND CONCLUSIONS
Multiple-regression analysis was found to be a useful technique for
assessing regional variations in water-quality characteristics in the Susque-
hanna River basin. The method was specifically structured to define those
basin characteristics that control nonpoint sources of pollution. The
multiple-regression models developed in this study are applicable only to the
Susquehanna River basin and hydrologically similar adjacent areas. The gen-
eral approach, however, should be potentially applicable to other regions. In
most regions, the most limiting factor is the availability of land-use and
water-quality data. Land-use maps are becoming more widely available as a
result of newly developed remote-sensing techniques. Deficiencies in water-
quality data, however, can be overcome only by comprehensive data-network
planning, sampling, and analysis.
Methods for compiling 17 water-quality characteristics and 57 basin
characteristics from available data sources are described in detail. Selec-
tion of basin characteristics for each regression was based on statistical
significance and from knowledge of the hydrologic processes involved. Eighteen
of the 57 basin characteristics were selected for use in 14 successful regres-
sion models (table 1).
The 14 multiple-regression models, relating water quality to basin char-
acteristics, explained from 56 to 89 percent of the variation of the water-
quality characteristics, with standard errors of estimate ranging from 17 to
75 percent. The principal sources of error were coarseness in the model struc-
ture and errors inherent in the data and methods of data compilation. It is
particularly important that the limitations described in this report be under-
stood by the user to avoid misuse of the model results.
The regression models developed in this study can be used to make gener-
alized conclusions about nonpoint sources of pollution. For example, regres-
sion models are used to estimate ranges of background water quality by mathe-
matically removing the effect of land-use variables from each model. Compari-
son of ranges of observed water-quality characteristics to the estimated
background ranges (table 4) shows that land use has a significant impact on
12 of the investigated water-quality characteristics. The greatest impact is
indicated for nitrate yields where the maximum observed value is 20 times
greater than the maximum estimated background value. This difference is
indicated to be the result of chemical fertilizer, animal wastes, and urbani-
zation. In view of this contrast, the standard error of estimate of the ni-
trate-yield model (+ 24 percent) is very good. By the same comparisons, the
standard errors (ranging from 17 to 75 percent) of the 14 models range from
acceptable to poor for making generalized estimates of background water qual-
ity. The models can also be used for estimating water quality at specific
sites where water-quality data are lacking. The effect of individual land
uses or other basin characteristics can be evaluated for a specific site in
a manner similar to the generalized example.
The use of the regression models should be tempered by the limitations
specified and by the scope of the general method used. It is particularly
40
-------
important to realize that the effects of land use explained by the regression
models represent generalizations of the prevailing management practices dur-
ing water years 1966 to 1975 for sediment and dissolved solids and during
1970 to 1975 for nitrogen and phosphorus. This methodology should be consid-
ered a "first-cut" approach for evaluating water quality on a regional basis.
Based on this type of study, the need for more detailed data collection and
areal investigations can be planned according to regional needs and problems.
41
-------
SELECTED REFERENCES
Afifi, A. A. and Azen, S. P., 1972, Statistical analysis: a computer oriented
approach: New York, Academic Press Inc., 366 p.
Anderson, J. R., Hardy, E. E., Roach, J. T., and Witmer, R. E., 1976, A land
use and land cover classification system for use with remote sensor data:
U. S. Geol. Survey Prof. Paper 964, 28 p.
Anderson, P. W., 1963, Variations in the chemical character of the Susquehanna
River at Harrisburg, Pennsylvania: U. S. Geol. Survey Water-Supply Paper
1779-B, 17 p.
Arnold, R. W., Kick, L. W., Marshall, R. L., Pearson, C. A., Cline, M. G., and
field soil scientists, 1970, Generalized soil map of New York: National
Cooperative Soil Survey in New York State.
Barrow, N. J., 1970, Comparison of the adsorption of molybdate, sulfate, and
phosphate by soils: Soil Science, v. 109, no. 5, p. 282-288.
Bailey, J. F., Patterson, J. L., and Paulus, J. L. H., 1975, Hurricane Agnes
rainfall and floods, June - July 1972: U. S. Geol. Survey Prof. Paper
924, 403 p.
Benson, M. A., 1962, Factors influencing the occurrence of floods in a humid
region of diverse terrain: U. S. Geol. Survey Water-Supply Paper 1580-B,
64 p.
Benson, M. A., and Carter, R. W., 1973, A national study of the streamflow
data-collection program: U. S. Geol. Survey Water-Supply Paper 2028,
44 p.
Biesecker, J. E., and Leifeste, D. K., 1975, Water quality of hydrologic bench
marks—an indicator of water quality in the natural environment: U. S.
Geol. Survey Circ. 460-E, 21 p.
Branson, F. A., and Owen, J. R., 1970, Plant cover, runoff, and sediment yield
relationships on Mancos shales in western Colorado: Water Resources
Research, v. 6, no. 3, p. 783-790.
Ciolkosz, E. J., Ranney, R. W., Petersen, G. W., Cunningham, R. L., and
Matelski, R. P., 1972, Characteristics, interpretations, and uses of
Pennsylvania soils: Bedford County: Pennsylvania State Univ. Prog.
Rept. 323, 46 p.
Ciolkosz, E. J., Petersen, G. W., Cunningham, R. L., Matelski, R. P., and
Pennock, R., Jr., 1974, Characteristics, interpretations, and uses of
Pennsylvania soils developed from Cherty limestone materials: Pennsyl-
vania State Univ. Prog. Rept. 341, 108 p.
42
-------
Crowder, Robert, 1974, Land use and natural resource inventory of New York
State: N. Y. State Office of Plan. Services, 16 p.
Cunningham, R. L., Petersen, G. W., Ciolkosz, E, J., Ranney, R, W., and
Matelski, R. P., 1972, Characteristics, interpretations, and uses of
Pennsylvania soils: Butler County: Pennsylvania State Univ. Prog.
Rept. 326, 57 p.
Cunningham, R. L., Petersen, G. W., Matelski, R. P., Ranney, R. W., and
Ciolkosz, E. J., 1972, Laboratory characterization data and field descrip-
tions of selected Pennsylvania soils: Pennsylvania State Univ. Ser. 25,
805 p.
Darmer, K. I., 1970, A proposed streamflow-data program for New York: U. S.
Geol. Survey open-file rept. 47 p.
Dethier, B. E., 1966, Precipitation in New York State: New York State Coll.
Agriculture Bull. 1009, 78 p.
Draper, N. R., and Smith, H., 1966, Applied regression analysis: New York,
John Wiley & Sons, Inc., 407 p.
Fenneman, N. M., 1928, Physiographic divisions of the United States (3d ed.):
Assoc. Am. Geographers Annals, v. 18, no. 4, p. 261-353.
Flaxman, E. M., 1972, Predicting sediment yield in western United States: Am.
Soc. Civil Eng. Proc., Jour. Hydraulics Div., v. 98, no. HY 12, p. 2073-
2085.
Flippo, H. N., Jr., 1977, Floods in Pennsylvania: A manual for estimation of
their magnitude and frequency: Dept. Environmental Resources Bull., in
press.
Goolsby, D. A., Mattraw, H. D., Lamonds, A. G., Maddy D. V., and Rollo, J. R.,
1976, Analysis of historical water quality and description of plan for a
sampling network in central and southern Florida: U. S. Geol. Survey
Water Resources Inv. 76-52, 124 p.
Hindall, S. M., 1976, Measurement and prediction of sediment yields in Wiscon-
sin streams: U. S. Geol. Survey Water-Resources Inv. 54-75, 27 p.
Hollyday, E. F., 1969, Geologic map of New York, An appraisal of ground-water
resources of the Susquehanna River in New York State: U. S. Geol. Survey
open-file rept.
Jansen, J. M. L., and Painter, R. B., 1974, Predicting sediment yield from
climate and topography: Jour. Hydrol., v. 21, p. 371-380.
Johnson, E. C., 1960, Climates of the states, New York, in climatography of
the United States: U. S. Dept. Commerce, Weather Bur., no. 60-30, 20 p.
43
-------
Kauffman, N. M., 1960, Climates of the states, Pennsylvania, in climatography
of the United States; U. S. Dept. Commerce, Weather Bur,, no. 60-36,
20 p.
Ku, Henry F. H., Randall, A. D., and MacNish, R. D., 1975, Streamflow in the
New York part of the Susquehanna River basin: U. S. Geol. Survey Bull.
71, 130 p.
Lystrom, D. J., Rinella, F. A., and Knox, W. D., 1978, Definition of regional
relationships between dissolved solids and specific conductance, Susque-
hanna River basin, Pennsylvania and New York: U. S. Geol. Survey, Jour.
Research v. 6, no. 4, p. 541-545.
Mansue, L. J., and Commings, A. B., 1974, Sediment transport by streams drain-
ing into the Delaware estuary: U. S. Geol. Survey Water-Supply Paper
1532-H, 17 p.
Meade, R. H., and Trimble, S. W., 1974, Changes in sediment loads in rivers
of the Atlantic drainage of the United States since 1900: IAHS-AISH Pub.
no. 113, p. 99-104.
Mendenhall, William, 1971, Introduction to probability and statistics: Bel-
mont, Ca., Duxbury Press, 466 p.
Miller, R. A., 1974, Hydrologic data of the June 1972 flood in Pennsylvania:
Dept. of Environmental Resources and U. S. Geol. Survey Bull. 9, 97 p.
Miller, C. R., 1951, Analysis of flow-duration, sediment-rating curve method
of computing sediment yield: U. S. Bur. Reclamation, Hydrology Br.,
Proj. Plan. Div., Denver, Colo., 55 p.
Moss, M. E., 1976, Design of surface water-data networks for regional informa-
tion: Hydrol. Sci. Bull. v. 21, no. 1, p. 113-127.
New York State Department of Agriculture and Markets, 1969, Ann. fertilizer and
lime tonnage rept: N. Y. State Dept. Agriculture and Markets, 8 p.
Omernik, J. M., 1976, The influence of land use on stream nutrient levels:
U. S. Environmental Protection Agency, 106 p.
Page, L. V., 1970, A proposed streamflow-data program for Pennsylvania: U. S.
Geol. Survey open-file rept., 57 p.
Petersen, G. W., Cunningham, R. L., and Matelski, R. P., 1968, Characteristics,
interpretations, and uses of Pennsylvania soils: Dauphin County: Penn-
sylvania State Univ. Prog. Rept. 290, 40 p.
Petersen, G. W., Ranney, R. W., Ciolkosz, E. J., Cunningham, R. L., and
Matelski, R. P., 1972, Characteristics, interpretations and uses of
Pennsylvania soils: Bucks County: Pennsylvania State Univ. Prog.
Rept. 324, 56 p.
44
-------
Ranney, R. W., Ciolkosz, E. J., Matelski, R. P., Petersen, G. W., and Cunning-
ham, R. L., 1970, Characteristics, interpretations, and uses of Pennsyl-
vania soils: Huntingdon County: Pennsylvania State Univ. Prog. Rept. 300
48 p.
Ranney, R. W., Petersen, G. W., Matelski, R. P., Cunningham, R, L., and
Ciolkosz, E. J., 1972, Characteristics, interpretations, and uses of
Pennsylvania soils: Bradford County: Pennsylvania State Univ. Prog.
Rept. 320, 64 p.
Reich, B. M., McGinnis, D. F., and Kerr, R. L., 1970, Design procedures for
rainfall-duration-frequency in Pa.: Inst. for Research on Land and Water
Resources, Pennsylvania State Univ., 60 p.
Riggs, H. C., 1968, Some statistical tools in hydrology: U. S. Geol. Survey
Tech. Water-Resources Inv., book 4, chap. Al, 39 p.
1973, Regional analyses of streamflow characteristics: U. S. Geol.
Survey Tech. Water-Resources Inv., book 4, chap. B3, 15 p.
Ritter, J. R., 1974, The effects of the hurricane Agnes flood on channel geo-
metry and sediment discharge of selected streams in the Susquehanna River
basin, Pennsylvania: U. S. Geol. Survey, Jour. Research v. 2, no. 6,
p. 753-761.
Rudisill, S. E.,.1976, Assessment of the water quality of streams in the Sus-
quehanna River basin: Susquehanna River Basin Comm., Mechanicsburg, Pa.,
120 p.
Seaber, P. R., 1968, An appraisal of the ground-water resources of the upper
Susquehanna River basin in Pennsylvania: U. S. Geol. Survey open-file
rept., 75 p.
Seaber, P. R., and Hollyday, E. F., 1965, An appraisal of the ground-water re-
sources of the lower Susquehanna River basin—an interim report: U. S.
Geol. Survey open-file rept., 75 p.
1966, An appraisal of the ground-water resources of the Juniata River
basin: U. S. Geol. Survey open-file rept., 57 p.
Socolow, A. A., 1960, Geologic map of Pennsylvania: Dept. of Environmental
Resources.
Sower, F. B., Eicher, R. N., and Seiner, G. I., 1971, The STATPAC system:
U. S. Geol. Survey computer contr. no. 11, 36 p.
State Conservation Needs Inventory Committee, 1967, Pennsylvania soil and
water needs inventory: State Conserv. Needs Inventory Comm., 257 p.
Steele, T. D., and Jennings, M. E., 1972, Regional analysis of streamflow
chemical quality in Texas: U. S. Geol. Survey Water-Resources Research,
v. 8, no. 2, p. 460-477.
45
-------
Steele, T. D., 1972, The syslab system for data analysis of historical water-
quality records (basic programs): U. S. Geol. Survey interim rept.
computer contr. 19, 49 p.
Takita, C. S,, 1975, Nonpoint source pollution assessment of the Chemung and
Susquehanna River subbasins: Susquehanna River Basin Comm., 65 p.
1975, Nonpoint source pollution assessment of the lower Susquehanna
River basin: Susquehanna River Basin Comm. draft copy, 50 p.
Thomas, D. M,, and Benson, M. A., 1970, Generalization of streamflow charac-
teristics: U. S. Geol. Survey Water-Supply Paper, 1975, 55 p.
U. S. Dept. of Agriculture, 1973, Commercial feed, fertilizer, and liming
materials, repts. for the period July 1, 1972 to June 30, 1973. Common-
wealth of Pennsylvania Dept. of Agriculture, p. 208-211.
U. S. Department of Commerce, 1972a, 1969 Census of agriculture: N. Y. county
data, pt. 7, sec. 2, table 19, 464 p.
1972b, 1969 Census of agriculture: N. Y. county summary data for selec-
ted items, pt. 7, sec. 1, chap. 2, tables 8-11, 343 p.'
1972c, 1969 Census of agriculture: Pa. county summary data for selec-
ted items, pt. 9, sec. 1, chap. 2, tables 8-11, 352 p.
U. S. Geological Survey, 1966-75, Water-quality records, Pt. 2 of Water re-
sources data for Pennsylvania: U. S. Geol. Survey (yearly repts.).
1966-75, Water-quality records, Pt. 2 of Water resources data for New
York: U. S. Geol. Survey (yearly repts.).
U. S. National Oceanic and Atmospheric Administration, 1972, Climates of the
states, New York, no. 60-30, 29 p.
U. S. Soil Conservation Service, 1967, Soil survey laboratory methods and pro-
cedures for collecting soil samples: U. S. Dept. of Agriculture Soil
Survey Inv., rept. no. 1, 50 p.
1967, New York State inventory of soil and water conservation needs:
U. S. Dept. of Agriculture, 287 p.
1971, Guide for interpreting engineering uses of soils: U. S. Dept.
of Agriculture, 87 p.
1972, General soil map of Pennsylvania: U. S. Dept. of Agriculture,
August 1972.
1974a, Soil survey laboratory data and descriptions for some soils of
Pennsylvania: U. S. Dept. of Agriculture Soil Survey Inv. rept. 27,
81 p.
46
-------
U. S. Soil Conservation Service, 1974b, Soil Survey laboratory data and des-
criptions for some soils of New York: II. S. Dept. of Agriculture Soil
Survey Inv., rept. 25, 107 p.
U. S. Water Resources Council, 1976, Guidelines for determining flood flow
frequency: U, S« Water Resources Council, Bull. 17 of the Hydrol. Comm.
195 p.
U. S. Weather Bureau, 1964, Climatography of the United States - Pennsylvania,
no. 86-32.
1961, Rainfall frequency atlas of the United States: U. S. Weather
Bur. Technol. Paper 40, 115 p.
Williams, K. F., and Reed, L. A., 1972, Appraisal of stream sedimentation in
the Susquehanna River basin: U. S. Geol. Survey Water-Supply Paper
1532-F, 24 p.
Wischmeier, W. H., 1959, A rainfall erosion index for a universal soil-loss
equation: Proc. Am. Soc. of Soil Sci., v. 23, no. 3, p. 246-249.
1974, New developments in estimating water erosion: Proc. of the 29th
Ann. Mtg. of the Soil Conserv. Soc. of America, Aug. 11-14, p. 179-185.
Wischmeier, W. H. and Smith, D, D., 1965, Predicting rainfall-erosion losses
from cropland east of the Rocky Mountains: Agr. Research Service, Agr.
Handbook 282, 47 p.
47
-------
APPENDIXES
48
-------
APPENDIX' 1.—Mater-quality
Station
number Station name SEDYLD SEDCONC DS
1500500 SUSOUEHnNNA H. AT UNA01LLA, N.Y. 112.6 64. 1 129
1502000 BUTTERNUT CR. AT MORRIS. N.Y. « " 89
1502500 UNAOILLA R. AT ROCKpAU" N.I. 110.4 69. V
1503000 SUSOUEHANNA k. AT CONKLIN. H.I. 1*3.6 90. 5 129
1507500 GENEGANTSLET CR. AT SMITHVILLF. FLATS, N.Y.
1508800 FACTORY BfOOK AT HOMER. N.Y.
1508803 ». BR. TIOUGHN10GA R. AT HOMfM. N.Y.
1509150 GRIDLEY CR. ABOVE EAST VIRGIL. N.Y.
1513107 SUSOUEH4NNA R. AT JOHNSON CITY. N.Y.
1611)000 OVCGO CR. NEAR OUFGO. N.Y. 90. * 61 .5
1515064 SUSOUEHANNA ft. NEAfl UVERLV. »J.Y. 145.2 114. d
1515050 SUSOUEHANNA ft. AT SAYRE, HA. -- — 138
1516820 TIOGA R. »T LAMBS CR.. PA.
1517000 ELK RUN NEAR MAlNESfUJOG. PA. 145.0 Ul.S
1517500 MILL CR. NEAR TIOGA. PA.
fLD DSCONC DSEXP DSCOEF
.0 82. 0 .875 .585
.7 S3. 7 .890 .256
.0 B1.5 .834 .918
.903 .172
.- .846 .737
.800 1.167
.710 .546
.0 66.4 .752 2.369
.638 2.670
.869 .431
1518400 CROOKED CR. AT MIDOLFBURY CfNTER. PA. -- " — " .842 .486
1618500 CROOKED CP. AT TIOGA. PA. 98.9 106.4 33.4 82. 9 .880 .437
1518700 TIOGA R. AT TIOGA JUNCTION. PA. -• -- « " .739 1.522
1SU850 COXANESOUE H. AT tlESTFIELD. PA. " " — — .845 .358
l$Ie8eO"MlL'L CREEK AT »ESTFIFLD. PA, — — — — .759 .558
1518870 COKANES8UE ». AT COWANESOUe. PA. " " — — .636 1.630
1519000 TROUPS CR. AT KNOXVILLE, PA. " " -- " .874 .461
1520000 COrfANESOUE R. NEAR LAHRENCEVILLE, PA. " — 98.7 102.0 .700 1.8*8
1520500 TI06A R. AT LINDLEY, N.Y. 237.0 237.9 98.1 93.8 .733 1.791
1526500 TIOGA R. Nt'AH tHxINS, N.Y. 2»9.4 i94.9
1528000 FIVEMILE CH. NEAR KANONA, N.Y. '" " 126.0 107.0 .823 .724
1531000 CHEMUNG 9. AT CMEMUNG. N.Y. 217.9 J14.7 137.0 129.0 .787 2.127
1533205 SUSOUEHANNA R. AT L.P. 65041, PA.
1534000 TUNKHANNOCK CH. NEAR TUNKHAMNOCK. PA. 84.2 59.5
153409* AUSOUEHANNA R. AT FALLS. PA. " -- — 123.4 SA.S .752 2.810
1534500 LACKAHANN* R. AT ARCHBALO. PA. -- -- 286. 0 1*6.0 .671 Z.SZ9
1536006 LACKAWANHA R. AT OLD FORGE. PA. " — 181.0 136.0 .794 1.400
1539000 fISHJNG CH. NEAR BLOOMSBUR6. PA. 214.0 121.9
1541000 ». BR. SUSOUEHANNA R. A1 HOMER. PA. 90. 5 51.4
15*3000 0«IFT*000 (W.SINNEMAMONlNG CR.STE'LING HON» PA. 71,5 40.4 " " — "•
15*3500 SINNEMAHONINO CH. AT SINNEMAHON1NG. PA. " -- 110.0 A4.4 .675 2.007
1544500 KETTLE CH. AT CHOSS FORK. PA. 21.3 13.3
15*5500 ». BR. SUSOUtnANNA R. AT REMOvO. PA. 55.8 32. 5 275.0 157.0 .720 S.1T2
154S600 YOUNG WMAN'S CD. NfAR RENOVO. PA. 76.* 49.5 46t7 29.0 .986 .0)14
1546500 SPRING C«. NEA* AxtvAM, PA. " — 277.0 282.0 .965 .898
15*7500 BALD EAGLE CR. AT BLANCHARDi PA. — — 200.0 152.0 .776 1.708
1547950 BEECH CR. AT MONUMENT, PA. " " -- — .652 2.187
15*8500 PINE CR. AT CEOAK RUN. PA. -- — 81.9 56.1 .676 .377
15*9500 ULOCKMPU5F CH. NEAR ENGLISH CENT", PA. ?76.7 169.7 -• " -_-. _-.-....
1553500 5. BR. s5s4uEHAf.NA p. AT LEnlSBURO. PA: 5«.ft St.B 158.0 97.6 .706 4..513
15SSOOO PENNS C*. AT PENNS CREEK, PA. — -- 138.0 92.4 .879 .55?
1555500 E. MAHANTANGO CR. NEAR DALMATIA. PA. " " 147.0 98.6 .823 .796
1555600 HISCONISCO CH. AT MIlLEHSSUWtt, PA.
1556010 "FRANKSTOMN BM. JUNlATA ». NFAR CLOVFH CH.. PA. •• -- -- -- «
1559000 JUNlATA R. AT HUNTINGDON, PA. 65.6 52.2 227.0 164.0 .S15 1.7*1
1559920 BOBS C». AT MEYNOLOSOALE. PA.
1560000 DUNNING CH. AT BELOCN, PA. *7.5 36.2
1561000 BRUSH C«. »T OAPSVILlf, PA.
1562000 RAYSTOHN Bf). JUNIATA R. AT bAXTON. PA. f>4.5 55.2 130.0 104.0 .760 1.M3
1562010 SHOUP HUN AT SAX70N« PA.
1562200 SHY BEAVER CH. NEAR ENTr»IK£N. PA. ..
1562250 1ATMAN HUN NEAR ENTPIKEN. PA. .....
1562350 COFfEE RUN NtAR ENTRIHEM, PA. .....
1562500 GREAT THOUGH CR. NEAR MARHLCSRUDG. PA.
1563000 RAYSTOMN 6R. JUNIATA B. NEAH HUNTINGOON. PA.
1563210 RAYiTOfcN BH. JUNIATA R. AT AROINxEIM, fA.
1564515 AUGHNICK CR. AT AUGHHICK MILLS, PA.
1565300 KISHACCOU1LLAS CH. AT U.S. 4400?, »A, —
1665515 JACKS CH. AT LExISTOKN, PA. " "_ _. •
•• .571 2.150
.764 .569
.790 .2*6
.718 .60*6
.888 .216
— ~ .857 .904
1.025 .256
IliJjJj 'JUNliTA A. AT NtuPOHT. PA. 65.6 54.8^ 161.0 124.0 .746 3.138
1567500 BULEH RUN NEAR LOYSVIUE, PA. 67,4 59.3
1568000 SHERMAN CR. AT SHERMANS DALE. PA. 44.7 31. A
1568200 SHERMANS CR,, PA. -- -- -- — -- -••
1569320 MIDDLE SPRING CH.. PA. -- — " — -- — •
1569900 CONOD06UINET M. PA.
1573205 OUITTAPAMILL* C«. AT SYNtH, P»,
1574000 CONEWAGE CR. NEAR MANCHESTER, PA. 133.0 110.2
1575000 S. BR. CODIWUS CH. NFAR YOHK. PA. — — '
1575990 CH1CKIES CR.. PA. .. .
.
1 5 76564 CON«f06A *. AT LANCASTER, PA. 156,6 129,4 308.0 234.0 .941 .931
1576515 MILL CR. AT L.R. 36009. PA. —
1576600 CONESTOG* CH. NEAR C"NESTOGA, P«.
15767B9 PEOUEA CR.. PA.
1577500 MUOOY CR. AT CASTLE FIN« PA, « -- — -- 1.00* .167
49
-------
Characteristics
NAVE
.49
.70
mm
__
1.3*
.87
.72
.*?,
.62
.82
.89
.75
1.13
1.18
1.25
.90
.00
1.25
--
—
.-
• •
NSD
.44
.18
22
..
__
.36
.46
.41
.32
.38
.42
.53
.30
.35
.41
.66
.48
.39
.88
—
«
-
-
N03AVE
.56
.41
.21
-^mi"'
1.4S
.57
.85
.59
.5»
.41
— ;5B'
.47
.46
.57
.40
.59
.46
.84
.56
.7*
.7?
.5*
.71
• 6S
.37
.37
•17
N03YLD
8R
65
__
*
»
-
96
„
- .74
.45
—
.54
.87
.74
.75
1.04
.67
.67
.65
•Z7
N03SD
.21
.24
.17
.58
.30
.18
.37
.28
.30
.34
.31
.34
.32
.52
.31
.30
.29
.58
.3H
.31
.49
.22
.38
.63
.67
f*
.21
.18
•"
NH4AVE
.09
vw
.10
--
•-
— ^
.06
.11
.06
.10
.10
.07
.10
.08
.16
.29
.07
.08
.08
.12
.08
.20
.42
"
.56
^~
PAVE
.04
.04
«•
--
~"
.07
.09
.07
.02
.06
.07
.08
.06
.07
.10
.07
.03
.05
.05
.06
.12
.06
.10
--
.02
PSD
.02
.03
••
"•
*•
.09
.07
.07
.02
.07
.10
.05
.08
.16
.04
.05
.04
.04
.03
.04
.Oft
.08
.II
—
.02
PVLD
.06
• •
.06
*•
~~
~™
""
.15
—
• BB
.08
..
.05
.05
.07
.12
.08
.15
•><•
.03
POMAVE
.01
—
<•*
"~
""*
<••
.03
.04
.01
.05
.05
.04
Tor
.03
.03
.03
.03
.07
..
• •
il
.01
P01SD
.01
—
"*
~™
~™
..
.03
.03
.01
.10
.0*
.»h
.04
.02
.03
.02
.01
.04
..
mm
H
.02
.07
.78
1.5*
17T?
1.12
1.01
1.45
.67
iTti-
1.43
1,30
«*
V*
.30
.57
— :ii —
.98
.63
.53
.63
.75
.79
• •
.37
—
—
.51
2.58
1.62
1.11
1.58
1.75
2.8A
1.32
— .54
.89
.62
.90
.34
1.02
1.09
2.59
2.H9
1.78
\86
2.34
3.01
•|1|"*1;62"""
5.72
3.66
4.29
6.83
4.1?
7.45
7.3*
.83
3.83
• V
2.1*
1.65
.39
1.0*
• V
1.11
•*
4.26
8.98
.18
1.49
.46
.89
,t>l
1.32
1.90
.72
.15
.99
.42
.31
.23
.43
.77
4.1*
1.42
1.27
1*7
2.36
1.31
3. IS
2.81
1.51
3.03
2.16
3.72
2.35
1.37
.06
.20
.SS
.13
1.01
.11
.04
.08
.13
.13
.08
.12
.11
.22
.60
.14
.20
.58
.4(1
.23
.28
.18
.75
.3*
.1*
.0*
.08
.10
.65
.08
l.Ofl
.07
.0*
.03
.03
.04
.03
.06
.04
.29
.13
.1)
.11
.18
1.24
.38
.12
.18
.27
1.12
.58
.2S
.02
.05
.07
.79
.06
.93
.07
" .04
.01
.03
.01
.02
.06
.03
.25
.10
.0*
.10
• •
.29
1.18
.15
• •
.05
.13
.13
1.16
.43
.23
.06 .
.12
.09
*•
.03
.96
.14
.14
.35
•>•
-&
•
•
02
-
02
•
"51
•
m
r?o
••>
31"
.02
--
• 01
z
*•>
—
,19
50
-------
APPENDIX 2.—Basin
Station
number
Climatic characteristics
PRECIP
124, J
R
SN
HINJAN
Topographic characteristics
AREA
CONTDA
SLOPE
BSLOPE
SLOT20
STOR
1500500
1502000
1502500
1503000
1507500
1S6»JOO
1508803
1509150
1513197
1514000
mms ""
1515050
1516620
1517000
151 7500
1518000"
1518400
1518500
1518700
1518850
miffs —
1518870
1519000
1520000
1570500
1526506
1528000
1531000
1533205
153*000
isisow •
153*500
1536000
153*000
is*ioo(i
15*3000
15*3500
15*4500
15*5500
15*5600
il*6»0
15*7500
15*7950
15*8500
15*9500
1553500
1555000
1555500
1555600
1SS6010
15i«M6
1559920
15*0000
1561000
1562000
Ifiloio •
1562200
1562250
1562350
1562500
1543686
1563210
156*515
1565300
1565515
1S67000 ~
1567500
1568000
15*8200
1569320
IUHH
1S7320S
157*000
1575000
1575990
1176500
1576515
1576600
1576789
1577500
39.7
38.7
39.1
39.*
*O.S
36.6
36.0
36.0
M.O
38.2
II. »
*1.0
36.0
35.0
3*.0
35.0
35.0
36.*
3«.0
3*.0
- 3s. o"
35.0
3*.0
36.5
3*.0
35. »
33.6
3*. 2
36.0
*2.0
Hlo
**.5
*2.5
*3.0
«*.s
*?!o
*s.s
*3.0
**.o
*0.3
— So —
39.2
*0.0
37.0
37.8
*2.0
39.0
*6.0
*s.o
*2.0
*2.
93.7
99.1
99.0
76.5
99.8
96.2
97.0
97.4
94.5
94.0
94.0
93.0
97.0
97.0
97.0
87.2
94.7
10S.6
11S.O
9*.5
113.8
120.7
135.5
115.2
106.9
109.4
123.0
111.2
104.3
iuU
114.7
111.5
107.]
104.7
11* lo
UK. 4
156.3
150.5
120.0
— rtr.T-
11B.T
119.0
126.5
125.0
ils.'i
124.0
124.5
124.0
IPl?
If*:}
124.3
130.4
120.4
123.0 ..
130.8
130.3
132.0
1*2.8
137.5
W.I
15*. 3
1*5.6
151.0
m.6
.»
165.6
162.1
167.5
151.4
60.0
80. 0
80.0
60.0
70.0
BO. A
80.0
80.0
80.0
80.0
/0..6
70.0
60.0
59.0
52.5
57.0
52.5
50.0
52,5
49.0
$3.6 "
54.0
50.0
52.0
54.0
55.0
45.0
53.0
60.0
34.0
54. a
62. 5
41.0
36.5
67. fl
53.6
54.0
53.0
58.0
' 52.5
1 ' A*.*
38.0
bb.O
59. A
57.*
Sols
«6.0
36.0
3S.O
56.0
' 34.0
39.*
55.0
53.0
53.0
M.i)
50.0
44.0
38. 0
46.0
*7.4
44.0
»i.O
40.0
44.0
50.5
38.0
42.0
42.0
39.0
*9.r
24.9
29.0
31.0
25.0
42.0
24. A
24.0
2S.O
33.5
15.0
13.9
13.5
U.8
13.2
17.6
17.0
17.0
' 14.0
is. 2
Jj.o
13.0
18.0
18.0
18.0
18.0
17.0
17.0
18.0
16.0
16.0
16.7
16.0
15.9
16.7
16.7
19.0
17.0
17.0
16.0
"" 19.6
16.0
18.0
18.0
' 18«
• •
" 3354.6' "
15.0
200.0
*•
510.0
74.0
• •
324*0
***
130.8
2.8
27.8
4.8
3.6
40.6
e«.o
9.8
26.3
8.8
14.3
1.5
1.5
10.5
47.3
33.3
44.6
7».4
27.8
19.7
35.0
44.4
90.9
20.1
?4.4
7.1
12.6
7.2
.6
21.3
1 3.9
38.9
21.9
39.6
49.6
44.5
' 7.6
7.6
152.7
S.3
6.8
5. S
16.2
8.5
7.4
50.0
6.1
9.2
17.6
110.0
90.0
80.0
120.0
50.0
I30TO
130.0
110.0
100.0
100.0
koo.b
95.0
130.0
110.0
110.0
130.0
100.0
88.0
100.0
110.0
130.0
100.0
80.0
100.0
o«-o
120.0
120. 0
100.0
90.0
90.0
100.0
110.0
120.0
120.0
90.0
0.0
120.0
170.0
120.0
90.0
60.6
110.0
100.0
130.0
160.1
130.0
120.0
130.0
95.0
1*0.0
loo. 6
105.0
100.0
130.0
m.O
.0
210.0
130.0
160.0
m,0
.6
130.0
110.0
100.0
110.0
156.1
80.0
' 90.0
90.0
65.0
SO. 6
30.0
60.0
110.0
50.0
60.0
*0.0
*0.0
90.0
80.0
4.00
1.00
1.00
1.00
1.00
5.00
1.00
7.0(1
13.no
6.00
15.00
5.50
1.00
7.00
1.00
1.00
5.00
5.00
^.6TS
5.00
1.00
5.00
6.00
iJ.f}
18.00
4.00
7.50
1.00
4.70
S.3n
19.00
1.00
7.00
4.70
33.00
1.00
5.00
1.00
1.00
7.00
6.00
*.oo
5.00
4.00
10.50
1.00
13.00
1.00
1.00
i.oo
9.00
!•!*
5.00
33.00
10.00
11.00
• «i'88 ••
1>.4T
8.00
12.00
i.oo
5.00
16.46
1.00
I.OO
i.oo
- ?'X2 -
1.00
i.oo
1.00
i.oo
f-SS
I.oo
1.00
1.00
- l.Oo
.62
1.3*
.15
.10
.71
.5*
• *
*0
••
.12
• *
.01
• •
.0
•>«
.0
.0
7W
.03
.53
2.50
1.67
.56
•*l
.01
.01
.0
.02
.0
.0
.0
.0
.02
.0
.0
.0
.0
.01
mm
«••
~0
'-
~ ~ ilO
' 6.00
.0
.0
• •
.21
.21
.61
.0
51
-------
characteristics
ELEV
1.1 1"
l.3b?
1.1 09
1.012
1 .
ftS.1
.0
.'>
26.7
S<>,0
.n
.0
47.1)
60.1)
73.3
?7.0
.0
36.0
.0
33.0
.0
100.0
.0
2A.O
4•«
C.94
.93
1.37
.46
1.41
1.77
.63
1.16
.46
.49
1.19
1.H4
.4J
1.63
.Tb
1.2b
1.60
1.67
I.U
4.97
2.92
5.2?
.75
.80
?'??
1.11
.90
2.39
1.06
.90
52
-------
APPENDIX 2.—Baa In
Station
number
Soil characteristics
CLAYA
SILTA
SOILNA
SOILNO
XACIDA
XACIDO
XOATA
XOATO
OECA
CECQ
1500500
1502000
1502500
1503000
1507500
1508800
1508803
I5091SO
1513107
1514000
1515000
1515050
1 SI 6820
1517000
1517SOO
1518000
15184-00
1518500
1518700
1516850
1516860
1518870
1519000
1520000
1520500
1526500
1528000
1531000
1533205
153*000
153*090
153*500
1536000
1539000
ISM 000
U*304»
15*3500
15". 4500
1545500
15*5600
15*6500
15*7500
15*7950
1548500
15*9500
15S3SOO
1S5SOOO
1S5SSOO
1555*600
1556010
1559000
1559920
1560000
1661000
1 562000
1562010
1562200
1562250
1S623SO
1562500
liSSooo
1563210
1 56*515
1565300
1565515
1567000
1567500
1568000
1568200
1569320
1569900
1573205
157*000
157SOOO
1575990
1576500
1576515
1576600
1S767B9
1577500
12.7
11.3
13.3
13.*
13.7
14.5
13.2
1*.0
13.9
13.*
13.4
13.9
16.1
18.2
17.8
15.6
16.6
16.3
17.3
15.5
16.2
14.7
15.3
15.9
15.0
10.*
13.3
14.2
17.1
1*.*
15.3
15.3
18,1
15.9
17. S
16.9
16.6
16.1
17.0
21.1
18.3
16. 6
19.0
19.9
I7ll
19.3
16.1
15.9
20.2
17.8
18.9
18.6
16.2
19.*
16.*
20.6
16.6
21.8
15.6
t9.0
17.9
16.9
20.8
19.0
18.7
22.9
19.*
19.1
18.7
I"*
18.*
I*. e
la. e
17.3
16.0
19.7
16.6
17.7
18.3
39.6
33.9
40.4
42. b
45.1
**.9
43.1
45.6
45.7
45.4
44.6
45.6
46.4
45.6
45.9
*7.5
46.2
4S.2
44.6
46.0
45.4
43.4
44.0
45.0
45.8
46.6
*1.7
45.1
46.5
44.1
42.3
38.*
45.8
S6.6
44.1
45.3
48.8
49. ft
40.9
S3. 2
48.4
47.4
45.7
40.9
48.1
48.9
49.4
**. 9
48.6
47.0
50.0
49.0
44.0
49.2
49.1
53.6
42.6
55.8
55,7
49.4
49,2
44.5
51.4
58.4
46.5
50.1
42.1
42. S
60.7
s4.»
57.4
53.5
ss.s
59.2
57.2
66.7
58.8
S*.«
*9.5
.128
.139
.1*9
.141
.1*6
.138
.142
.1*7
.151
.166
.1*9
.1*9
.155
.163
.161
.151
.155
.157
.157
.1*9
.152
.1*2
.153
.156
.154
.15*
.1*5
.155
.168
.150
.15*
.1*9
.176
.176
.122
.125
.1*2
.1*6
.120
.1*5
.148
.138
.161
.168
.151
.136
.1*4
.130
.169
.147
.170
.170
.138
.161
.153
.172
.154
.170
.170
.162
.148
.1*7
.170
.32*
.161
.2*6
.169
.168
.119
.16)
.143
.14*
.128
file
.106
.095
.104
.099
.14*
.26(1
.208
.391
.290
.216
.200
.202
.212
.300
.217
.277
.277
.386
.521
.482
.290
.389
.434
.358
' .206
.301
.199
.321
.395
.122
.215
.275
.323
1.097
.366
.990
.686
.991
.236
.290
.237
.246
.213
.16*
.265
.221
.187
.550
.505
.332
.163
.207
.193
•zn
.2J2
.200
.145
.186
.199
.207
.245
.187
.257
.231
.201
.199
.182
.206
.054
.200
.132
.156
.15A
.209
.229
.226
.263
.225
.201
.161
.128
.155
.155
.267
13.59
17.08
14.74
13.82
12.80
12.09
13.33
13.96
13.30
13.40
13.27
13.27
13.49
14.20
13.98
12.94
13.37
13.71
14.49
13.61
13.64
14.73
1*.**
14.09
13.34
14.27
12.90
13.48
14.06
13.37
14.24
13.37
11.37
12.56
13.16
7.47
11.36
11.65
5,00
7.18
12.77
14.30
16.58
11.47
7,97
7.04
8.2ft
7.0*
7.76
8,74
9.10
7.15
7.67
9.34
6.97
8.55
6.29
8.23
7.64
8.62
8.23
6.68
11.60
7.96
8.44
7.94
7.90
7.39
8.8?
7. 62
8.82
11.47
9.02
9.06
10.09
9.30
11.40
12.87
18.27
23.19
20.29
18.47
18.24
15.76
17.35
18.31
18.07
16.82
18.05
18.05
17.78
17.02
17.00
16,93
16.80
18,08
17.81
17.72
17.31
18.87
19.77
19.01
18.17
19.00
17.3*
18.21
16.09
15.73
17.22
16.**
18.90
28. 7»"
i!*.97
13.92
19.38
16.09
10.55
12.78
18.83
17.66
18.80
17.22
11.52
13.93
16.3*
11.28
I*. 2*
15.14
15.40
11.94
12.57
15.91
11.67
12.83
10.57
($•21
Iz.tt
14.95
12.83
10.16
19.8*
12.6*
13.18
10.85
11.04
14.69
14.44
13.42
19.86
20.92
18. jl
16.75
17.98
17.03
20.96
23.78
3.52
3.65
4.87
4.2*
*.35
5.11
4.25
*.35
4.64
5.9*
*.67
4.67
4.57
4.42
4.45
4.60
4.47
4,50
4.41
4.53
4.45
4.65
4.56
4,56
4.56
4.25
4.31
*.72
4,42
4.14
3,*2
5.41
7.49
3.64
3.18
8.42
5.59
3.97
15.77
11.39
4.19
4.41
4.15
6.18
11.74
8.49
7.1*
15.8*
11.57
12.30
11.49
8.36
13.**
7.9*
14.77
8.85
16.11
?'?!
12.87
10.94
9.16
12.98
{.75
rJ.n
18.43
12.S3
12.25
8.17
16.61
9.73
5.04
5.36
5.84
6.16
6.23
6.10
4.S4
4.25
7.29
8.82
9.75
8.**
8.61
10.04
8.50
8.89
8.88
9.96
8.87
8.87
8.19
7.11
7.37
8.65
7.88
7.82
.77
.81
.20
.99
.65
.22
8.63
7.39
7.80
8.54
6.76
6.58
4.94
6.88
13.10
l.fl
8.35
10.16
10.24
7.61
34.80
21.03
9.04
6.90
*'J2
10.39
16.67
10.17
9.46
21.91
19.38
16.10
15.50
10.50
18.43
11.92
28.35
12.60
32.S1
1?.{7
17.77
17.47
13.19
25.98
8.94
18.21
21.80
15.21
14.88
25.06
21.4*
25.02
12.30
12.71
14.26
l5.il
18.45
15.92
13.62
7.18
17.11
20.74
19.59
18.0*
17.15
17.20
17.58
18.33
17.93
19.33
17.94
17.9*
17.92
18.61
18.43
17.53
17.84
18.00
18.91
18.15
18.10
19.38
18.71
18.38
17.59
18.51
17.10
18.13
18.37
17.37
17.55
18.43
18.82
14.22
16.3%
15.90
16.94
15.67
20.78
18.58
16.97
18.70
20.75
17.62
19.73
15.53
15.40
22.87
19.33
21.03
20,58
15.54
21,12
17.27
21.73
17.41
22.39
>7.}[
20. Si
19.55
17.40
19.68
18.31
20.05
26.88
20.51
20.18
15.54
14.48
17.35
13.88
16.83
14.86
is. 22
16.33
15.40
15.92
17.12
24. 3d
31.46
28.00
2*. 92
23.93
22,9*
23.45
2*. 54
24.69
2*. 37
24.49
24.49
22.73
21.42
21.55
22.17
21.61
22.82
22.91
23.36
22.51
25.66
25.32
2*. 10
23.45
25.28
22.89
2*. 37
20.47
20.24
20.**
21.07
?*.26
31.52
27.35
21.11
23.60
ViTS
41.19
28.91
22.12
21.67
23. »9
23.10
22.98
21.27
22.59
28.02
28.29
25.85
25.22
18.96
25.96
22.22
33.45
19.S7
37.09
Z2.42
25.37
26.99
20.36
32.71
27.25
26.22
31.45
22.86
22.71
33.49
28.30
31.06
26.52
28.81
25.59
24.58
26.35
24.83
27.21
28.76
53
-------
characteristics—Continued
Soil characteristics
PHA
PHL
KA
PERMA
PERM,
HSO
WATCAP
BDRK
LT200A
axAVA
STONEA
5.2
S.I
5.3
5.3
5.*
5.4
5.1
5.3
5.5
5.6
5.S
5.5
5.4
5.3
5.3
5.3
5.4
5.3
5.3
5.3
5.3
5.3
5.3
5.3
5.3
5.2
5.2
4.9
5.4
5.3
5.3
5.0
4.4
5.4
5.6
5.3
5.0
6.1
5.3
5.3
6>4
6.1
5.1
5.3
5.1
5.4
6.0
6.1
5.9
6.3
6.0
6.1
6.0
6.1
6.3
5.8
6.3
6.0
6.4
6.1
6.3
5.9
5.9
6.3
S.7
6.1
6.2
6.1
6.1
6.0
6.6
6.0
S.S
5.3
5.6
s.s~
s.s
5.5
5.2
5.1
4.5
4.1
4.5
4.7
4.8
4.9
4.5
4.8
4.9
5.0
4,9
4.9
4.9
4.9
4.9
4.9
4.9
4.8
4.8
4.S
.8
.8
.6
.8
.8
.7
4.7
4.4
4.8
4.8
4.7
4.4
4.1
4.8
4.4
4.4
4.3
5.0
4.4
4.5
5.3
4.9
4.3
4.8
4.7
4.6
4.9
4.9
4.8
S»2
4.9
5.1
S.O
4.8
5.2
4.7
5.1
4.7
5.2
4.7
5:1 '
4.9
4.7
S.I
4.9
5.0
5.6
S.I
S.I
4.7
sli
4.6
4.5
4.5
4.4
4.4
4.5
4.4
4.3
4.S
.22
.21
.21
.21
.21
.24
•2?
.22
.22
.21
.21
.21
.23
.24
.23
.23
.23
.23
.22
.23
.22
.22
.22
.?:
.22
.23
.19
.20
.21
.24
.21
.22
.21
.24
.31
.29
.30
.25
.29
.27
.29
.27
.29
.25
.24
.28
.28
.25
.25
.27
.26
.27
.27
.25
.27
.27
.29
.27
.30
.30
.?7
.28
.27
.28
.26
.28
.26
.26
.26
.30
.31
.30
.29
.34
.35
.30
.35
.31
.32
.34
1.40
1.52
1.57
1.29
.94
1.03
.88
.82
1.26
1.29
1.23
1.23
.93
1.00
.95
1.06
.73
.84
1.04
.96
.78
.84
.74
1.10
1.09
1.14
1.54
1.13
1.23
1.07
1.14
1.08
1.27
2.15
1.28
2.84
2.42
4.44
2.28
6.03
3.64
4.36
3.07
1.56
1.30
2.41
5.12
4.33
5. 15
3.91
3.94
3.07
3.06
6.43
4.26
3.91
3.06
5.81
2.98
2.62
4.24
2.62
5.38
5.14
?•!*
4.90
3.06
6.57
6.47
2.35
4.51
2.96
1.80
1.46
1.70
2.25
1.43
2.09
2. OS
1.30
1.10
1.4S
1.19
.98
.70
.48
.60
.58
.91
1.19
.93
.93
.68
.71
.60
.80
.51
.59
.78
.76
.62
.64
.77
.94
.86
.85
1.12
.90
.96
-.54
.86
.61
.82
1.68
.96
2.24
1.94
4.03
1.88
5.39
2.39
3.66
2.65
1.23
1.11
1.98
4.64
4.03
«.75
3.46
3.40
2.92
2.91
5.82
3.85
3.59
2.38
5.37
2.07
2.17
3. 94
2.28
4.96
3.93
1.61
4.31
2.65
5.84
5.75
1.45
3.92
2.26
1.67
1.31
1.42
2.10
1.30
1.95
1.95
1.30
2.7
2.7
2.6
2.7
2.8
2.7
3.0
3.0
2.7
2.6
2.7
2.7
2.9
3.0
3.0
2.9
3.0
3.0
2.9
3.0
3.0
3.0
3.0
2.9
2.9
2.8
2.5
2.6
2.7
3.0
2.7
2.7
2.6
2'.7
2.8 _
2.6
2.5
2.7
2.6
2.7
2.8
2.9
2.6
2.9
3.0
2.7
2.7
2.5
2.6
3.0
2. a
2.8
2.8
2.9
2.9
2.7
3.0
3.0
3.0
2.9
2.9
2.7
2.9
2.9
2.4
2.9
2.9
3.0
3.0
2.5
3.2
2.8
2.4
2.4
2.5
2.3
2.1
2.2
2.1
2.5
.099
.109
.095
.086
.067
.098
.076
.071
.082
.071
.078
.078
.068
.076
.074
.072
.064
.069
.070
.078
.669
.071
.085
.076
.073
.072
.084
.071
.075
.079
..07S
.076
.082
.088
.115
.106
.105
.107
.108
.098
.V42
.113
.105
.090
.098
.103
.114
.103
.104
.104
.111
.102
.098
.098
.107
.103
.116
.087
.127
.115
.106
.112
.094
.126
.107
.105
.103
.097
.097
-.148
.118
.130
.123
.145
.132
.130
.157
.135
.135
.136
44
37
47
47
52
S2
50
50
52
51
51
51
SO
46
47
48
52
49
48
45
49
47
44
47
48
4R
52
44
50
51
48
47
41
45
42
52
49
45
45
45
44
40
45
45
35
44
42
44
45
37
41
40
38
41
40
41
34
34
37
43
30
41
36
43
50
40
42
41
41
47
37
37
48
51
46
4A
48
47
47
54
50.7
50.6
50.4
49.8
50.3
sl.4
51.1
50.1
50.9
48.8
56.1
50.1
50.0
50.6
50.6
4?,7
50.4
50.0
49.2
50.0
49.9
49.6
52.1
49.2
49.4
50.0
48.5
46.3
49.4
53.6
49.0
49.5
45.3
53.4
72.1
61.5
62.9
53.3
64. 5
53.4
72.0
56.8
63.1
53.9
49.6
59. 3
57.6
54.7
52.8
55.3
Sflt 1
54.2
S2.4
47.1
55.0
57.0
59.6
42.1
65.7
64.7
54.2
58.4
45.9
63.0
51.9
53.2
54.6
48.7
48.7
71.3
57.3
62.1
62.3
70.0
65.0
63.2
78.5
65.9
70.1
64.6
32.0
33.8
32.9
33.5
34.2
31.6
34.1
35.9
34.6
36.3
34.7
34.7
35.6
35.7
35.7
35.7
35.6
35.3
35.3
37.0
36.8
36.1
36.7
36.7
36.1
32.7
33.8
31.8
35.0
30.4
34.0
28.7
28.0
29.6
13.9
11.9
11.9
27.4
IS.O
24.5
12.0
25.6
15.7
31.3
38.6
23.1
25. 5
25.6
26.7
..?B.8. _
22.2
28.6
29.8
34.1
29.3
24.9
28.2
40.7
23.2
21.5
30. Z
23.0
36.2
25.1
43.0
30.2
33.6
35.0
34.9
19.0
35.7
23.9
21.1
13.0
13.5
18.3
7.7
16.1
10.9
15.3
7.2
8.5
7.6
7.6
ft. 5
B.I
8.7
7.6
7.0
7.7
7.7
8.0
7,8
7.9
7.8
8.4
8.0
7.7
7,9
8.3
A.I
7.5
7.7
7.8
7.3
7.5
7.0
7.8
7,4
7.7
6.9
6.2
10.2
..3i»
4.8
4.4
ft. -6
4.4
I'l-
5.6
8.0
4.4
6.4
7.3
5.8
7.1
9.9
6.8
7.0
6.2
6,7
7.3
9.1
7.1
ft. 6
9.1
11.6
8.1
5.4
7.5
5.9
9,9
7.8
10. f,
6.3
5.0
R.I
8.1
4.6
IV. 8
7.9
3.1
2.2
2.3
4.2
1.2
3.6
2.2
2.4
54
-------
APPENDIX 2.— Basin characteristics—Continued
Station
number
1500500
1502000
1503500
1S03000
ISO 7500
1508803
1509150
1513107
1SUOOO
1515050
1516820
IS17000
}Him
1518*00
1S18SOO
1518700
1518850
1518870
1519000
1520000
1520500
1528000
1531000
1533205
153*000
153*500
1536000
1539000
15*1000
15*3500
154*500
15*5500
15*5600
15*7500
1547950
15*8500
l|*2|fi|
1555000
1S5SSOO
1555600
IffiSlfi
1559000
1559920
1560000
1561000
1562000
1562200
1562250
1562350
}8$m —
1563210
156*515
1565300
gftfif
1567500
1568000
1568200
Jfffifjj
1573205
157*000
1575000
1576515
1576600
1576789
1577500
Land-use characteristics
LU1
l.S
1.9
1.0
1.8
.*
.9
2.2
2.8
.6
2.5
2.0
.0
JS—
1.3
.0
.6
1.0
.0
1.6
.2
.9
"I**
1.7
3.2
2.2
2.0
..3.7 ,
1.7
6.7
12.9
.5
1.7
1.3
.2
l.S
.0
6.3
.5
.6
— w
1.7
.8
1.1
2.3
7.5
*.6
1.0
1.0
.0
— H —
2.5
.0
.0
-. .'3
2.9
1.2
2.0
Iff
1.9
.0
.1
.1
2.8
13.9
*.3
6.3.
s j
1.3
1*.2
3.9
10.0
LU2
29.7
16.*
3*. 6
30.7
15.5
2S.2
61.2
3*. 8
38.5
33.1
36.0
81.0
60.8
*2.*
58.5
51.6
**.8
31.3
50.*
36.2
6*. 3
tM •
39.7
38.3
37.0
_. *3.0
36.3
21.0
16.7
35.2
25.0
3.*
*.S
a.i
12.3
.0
36.0
.5
18.3
3J.3
16.6
33.1
52.9
36.3
2».2
31.1
35.*
*3.0
23.6
_J7.?
10.0
23.3
31.6
«r|
32.1
26.6
39.0
35.0
30.8
55.2
33.0
3*. 2
*lt7
6*. 8
75.6
70.*
66.2
^J8,5
6*. 2
75.3
69.8
77.6
61.0
LU1
67.0
81.2
63.7
62.8
83.*
56.*
73.3
36.5
62.*
60.7
62.9
62.9
60.1
19.0
38.8
5*. 9
*1.2
*7.*
53.2
68.7
51.9
*8.0
63.*
3*.*
55.9
58.5
56.7
58.1
60.0
51.7
61.0
58.2
5*. 6
63.7
67.0
9*. 6
91.9
91.8
82.8
100.0
36.*
56.6
95.9
80.8
66.7
78.7
66.0
*3.*
59.7
67.0
63.3
63.0
56.0
76.3
59.3
89.*
87.5
76.7
68.*
72.9
63.1
63.5
72.1
58.9
6*.0
65.9
**.8
66.8
65.6
55.5
30.1
9.1
2*.0
25.2
16.0
22.1
23.*
15.8
18.2
29.0
LU5
1.6
.6
.5
*.*
.7
.0
.1
.0
.8
.2
1.3
1.3
.3
.0
.3
.3
.0
.0
.2
.0
.0
.2
.1
.2
.1
.3
.1
.6
.»
1.?
.9
1.9
1.0
.6
.0
.11
.0
.0
.2
.0
.0
.6
.0
.0
.0
.5
.1
.0
.0
.2
.0
.0
.0
tg
.0
.0
.0
.0
.0
.9
.6
.1
.0
.0
• 6
.0
.0
.0
.0
.0
.5
.6
2.0
.0
.0
.3
.0
• 0
LU7
.2
.0
'.2
.0
.0
.0
.0
.1
.2
1.1
.0
.0
.8
.0
.0
.6
.0
.0
.2
.0
.1
.1
.1
!i
.1
.1
12)1
i*.*
.0
6.3
.4
2.1
.0
3.1
.0
.0
.2
3.2
.3
.0
Z.«
.0
2.*
1.7
.9
.5
.5
.2
.0
.3
7.1
.0
.0
.0
2.8
.7 '"
.6
.0
.2
.0
.*
.0
.0
.1
.0
.4
.8
.7
.0
.*
.0
.3
.2
.0
c
.018
.010
.019
.021
.009
.0*1 —
.022
.051
.031
.023
.ozi—
.021
.037
.056
.0*3
' ~o39 —
.0*1
.037
.038
.023
:TO —
.038
.026
.0*6
.031
V535
.031
.029
.028
.027
.1*2
.162
.033
.081
.irre —
.026
.009
.0*1
.002
.63* —
.033
.03*
.021
.029
;o37 —
.031
.073
.051
.029
.030
.032
.031
.019
.032
.675
.009
.019
.02*
.047
;»33 —
.031
.025
.037
.032
"~753l
.0*9
.030
.032
.036
"Jos?
.073
.067
.057
-roil —
.06*
.063
.068
.053
AGP
1272
2760
55 —
103
36
5518
6227
250
31
176
454
165
239
792
91
Z2
226
93
736
1325
104
4068
12016
13322
209
- ' 1,3' --
1604
0
586
2
— 5332
565
257
1466
185
328
35
1204
4
7
6
83
— me —
3035
339
428
157
—5533
365
|7*
— IS08
771
2155
457
— 35T5
223
46*0
1730
*79
AQN
*011
8908
139
321
113
17580
197S2
7*4
93
5*0
" 1348
*90
711
235*
65
650
288
2230
3926
321
12*18
37*01
*1SS6
675
93
*28*
0
1527
6
1*671
1620
730
*28* ""
5*5
969
too
6-
12
19
17
— 3BJ5
8784
939
1256
15948
983
*lti
209*
6*82
132*
11911
706
1*5*7
5465
1387
55
-------
APPENDIX 2.—Basin characteristics—Continued
Station
number
1500500
1502000
1502500
1S03000
1507500
1SOB800
1508803
1509150
1513107
1514000
1515000
1515050
1516820
1517000
1517500
1518000
1518400
1518500
1518700
1518850
151B660
1518870
1519000
1520000
1520500
1526560'
1528000
1531000
1533205
1534000
153*040
1534500
1536000
1539000
1541000
1543000
1543500
1544500
1S4SSOO
1545600
1546500
1S47500
1547950
1548500
1549500
1553500
1555000
1555500
1555600
1S56010
1559000
1559920
1560000
1561000
1562000
1562010
1562200
1562250
15623SO
1562500
1563606
1563210
1564515
1565300
1565515
1567000
1567500
1568000
1568200
1569320
1569900
1573205
1574000
1575000
J575990
Streamflow characteristics
KAQ10
1568.0
101.0
849.0
3585.0
285.0
7758.0
7758.0
11.3
364.0
122.0
292,0
817.0
1444.0
79.6
2689.0
11920.0
569.0
14098.0
214.7
449.0
510. 0
589.0
500.0
1187.0
228.0
5277.0
75.3
86.7
452.0
895.0
62.9
11264.0
455.0
244.0
1119.0
244.0
958.0
96.0
995.0
4405.0
19.9
311.0
mf
673.0
138.0
llillltt -"' 432.0
1576515
1S76600
1576769
1577500
MA<59
1520.0
830.0
3477.0
276.0
7488.0
10.6
340.0
115.0
262.0
779.0
1339.0
2522.0
550.0
488.0
563.0
488.0
219.0
5102.0
72.1
862.0
59.7
16)35.0
427.0
" 1646.6
229.0
896,0
• *
4071.0
17.3
285.0
624.0
108.0
396.0
PK10
12400
2580
9700
32100
14200
121000
121000
3940
59000
21000
40500
128000
190000
5110
184000
21200
364000
30900
27500
32000
60800
14300
181000
5370
5410
66000
6260
300000
34600
69900
57000
12000
40200
--
187000
5670
27500
81700
26700
88300
P2
12500
1910
8562
31100
2640
5910
64700
64700
593
10400
3910
9*80
21500
32060
1500
46500
13000
123660
7540
7650
8460
17600
3520
58300
753
64j
11600
1880
1165AO
S130
4180
13900
3860
13200
1670
15200
45300
770
6590
15200
2308
6340
4880
P25
23500
3540
16100
55800
5150
14900
122300
122300
1860
35600
11400
28400
70200
B930U
3480
113000
_.., 3?*OS_
241466
23200
17600
27180
52600
10800
135100
3020
2097
33800
5430
221000
16000
17085
39600
8320
34406
4880
24000
111000
4315
20200
43900
9620
23400
13400
PK10/P10
.62
.87
.72
.68
'" 1
1.24
1.16
1.18
2.92
2.35
2.49
1.92
2.56
2.8S1
1.86
2.15
f-ff
1.62
1.84
1.99
1.64
1.58
1.82
1.70
2.63
3.63
2.67
1.56
1.66
3.02
6.29
.96
1.79
1.53
"
••
2.18
2.16
1.89
2.59
4.28
5.66
*••»
56
-------
APPENDIX 3.—Average soil characteristics of the principal
associations CLAYA
Pennsylvania 2/
MA 20.0
MB (6.1
MC 13. 8
AID 9.9
A1E 16. 5
A2A 17.1
» 11.8
402 11.8
404 17.4
416 12.7
417 13.3
418 12.1
419 13.7
420 20.2
427 16.8
438
Soil characteristics
SILTA SOILNA
68.1 .110
54.9 .148
52.1 .167
44.3 .098
48.? .140
47.2 .178
42.5 .110
42.5 .110
36.5 .172
34.7 .100
61.1 .194
43.0 .113
52.5 .163
5«.2 .192
66.7 .095
67.2 .123
61.6 .165
61.1 .127
67.9 .090
44.3 .206
19.7 .059
58.4 .365
53.1 .115
54. 5 ,\3J
36.7 .092
43.1 .153
54. 4 .123
67.5 .113
37.9 .157
43. S .177
46.6 .180
49.3 .188
S4.8 .176
40.7 .117
46.1 .146
40.5 .194
49.2 .138
59.0 .142
33.3 .181
29.0
46.8 .240
41.3 .207
63.5
30.7 .137
33.5 .160
46.8 .240
26.8 .110
49,3 ,262
46.0 .120
46.7
55.3
4».2 .190
SO. 5 .140
66.1 .090
53.2 .260
49.2 .194
53.5 .245
44.0 .140
51.7 .151
47.3 .147.
49.2 .163
47.9 .169
47.3 .221
46.8 .153
62.6 .030
•»w __
ST. 7 .160
59.6
34.0 .140
34.0 .140
47.3 .161
40.4 .201
43.2 .146
38. S .149
44.4 .147
46.6 .170
S7.8 .1ST
—
SOILNG
.160
.222
.342
.191
.580
.2119
.135
.135
.158
.175
.255
.160
.222
.222
.128
.243
.28B
.147
.110
.211
.214
.245
.243
.164
.298
.149
.148
.189
.77S
1.957
1.828
.234
.160
.211
.608
.740
.242
1.057
__
.240
.265
.247
.300
.240
.120
.345
.176
__
__
.270
.200
.090
.451
1.772
.413
.200
.219
.213
1.427
1.012
.366
.222
.100
«<•
.160
__
.200
.200
.166
.336
.213
.214
.212
.164
.228
--
XACIDA
6.50
5.14
8.52
7.79
6.60
9.01
12.55
17.55
7.94
14.05
9.83
15.80
12.79
9.77
10.09
6.53
4.48
8.40
9.60
6.37
5.72
12.22
12.70
13.13
9.13
17.54
4.81
8.48
17.76
15.50
14.57
14.21
7.23
17.20
12.51
13.18
10.76
17.3?
17.63
„
9.90
9.69
--
7.57
9.40
9.90
4.40
14.64
6.80
__
__
12.20
17.25
5.10
19.84
15.01
1H. 25
14.39
10.82
13.02
10.82
11.74
19.67
13.74
20.60
_•.
12.20
— ••
20.10
20.10
21.51
21.96
15.41
18.13
14.70
12.89
11.51
--
XACIDO
13.80
13.91
22.73
15.74
45.70
14,96
17.17
17.17
10.01
27.75
17. 3D
22.07
20.47
17.67
17.98
11.87
7.64
15.8?
17.00
8.64
26.72
22.13
19.75
25.56
23.36
21.53
7.00
14.14
20.79
17.12
14.95
14.28
9.92
24.97
16.90
27.19
17.75
17.38
37.66
..
9.90
16.40
22.73
30.30
9.90
6.70
17.07
9.20
„_
13.90
20.75
12.10
25.16
15.01
22.81
18.66
15.32
17.40
11.70
12.96
24.46
18.11
20,70
•»
16.20
— _
24.lt
24.10
24.04
27.35
19.67
22.27
18.99
14.70
16.01
~
l/
XCATA
7.20
10.75
4.87
3.93
9.00
10.76
1.52
1.52
6.32
1.20
9. 3D
.83
5.93
8.82
6.23
9.83
19.67
7.45
6.90
29.41
1.37
6.15
3.80
3.39
3.12
5.34
6.89
8.72
4.04
4.23
4.32
4.42
8.67
.93
4.66
6. 55
6.02
7.84
S.I 3
..
11.40
7.98
3.53
3.20
11.40
7.50
6,66
6.90
__
16.60
2.90
3.40
6.46
3.98
10.20
4.71
4.66
4.50
6.50
6.16
7.27
».03
7.00
4.00
__
3.70
3.70
2.57
3.27
4.05
3. 53
4.23
5.80
4.21
—
XCATG
15.20
11.99
12.35
9.11
14.40
15.84
4.47
4.47
8.30
5.85
14.51
6.30
12.22
14.07
18.45
34.53
43.63
12.47
18.20
33.82
5.62
8.46
11.15
7.51
7.34
6.74
15.73
18.05
7.40
5.42
5.48
S.48
10.59
2.93
9.18
8.38
7.58
8.26
7.84
11.40
9.02
5.90
5.70
11.40
7.90
12,94
12.40
__
23.60
6.55
7.00
19.48
4.82
20.38
10.28
8.38
8.92
6.48
8.01
16.65
7.76
16.60
30.00
v«
9.60
9.80
3.27
12.16
8.63
8.29
8.87
7.47
7.26
--
CECA
13.70
15.09
13.4}
11.72
15.60
19. 7S
14.12
14.12
14.32
15.20
19.15
16.67
18.68
11.58
16.33
16.37
24.15
15.85
16. SO
35. 7S
7.06
18.41
16.50
16.51
12.19
I7.8H
11.70
17.20
21.86
19.80
18.65
If. 40
15.90
11.13
17.16
18.26
15.31
23.34
19.53
21.30
17.67
11.10
12.60
21.30
11.80
23.30
13.70
__
29.00
20.15
8.40
28.30
18.77
26.45
19.11
15.49
17.52
17.16
17.81
2 A. 94
17.77
27.60
t».20
23.80
23.80
24.13
25.23
19.45
21.66
16.93
18.75
15.72
CECG
24.20
22.98
27.88
19.78
51.60
22.53
18.50
18.50
15.65
29. 31)
23.92
23.03
24.91
23.02
26.35
39.37
46.79
20.65
24.90
37.32
29.05
29.21
30.25
29.61
29.02
27.68
19.67
26.25
26.87
20.60
18.85
18.53
17.57
25.53
22.42
33.25
22.16
?6.36
41.40
21.30
26.55
26.43
33.90
21.30
12.00
26.59
17.60
__
37.70
26.85
19.10
36.17
16.77
36.07
25.53
19.70
23.12
16.18
18.94
36.84
23.67
27.60
30.30
33.90
33.90
26.06
35.54
26.59
30.56
25.61
20.14
20.42
i/ Defined In section entitled "Basin characteristics".
£/ According to general soil association nap of Pennsylvania (U.S. SCS, 1972).
2/ According to general aoll association map of New York (Arnold and others, 1970).
57
-------
Soil associations in the Susquehanna River basin
Soil characteristics I/
PHA
6.2
6.5
5.5
5.2
6.7
b.O
4. »
4.8
5.9
4.6
S.9
4.S
S.3
5.9
5.5
6.2
6.7
6.0
5.7
6.8
5.1
5.6
5.1
4.9
5.2
S.3
6.1
5.8
5.0
5.1
5.0
5.4
5.9
3.8
S.i.
5.4
5.S
5.2
5.1
6.2
6.S
6.1
6.6
5.8
5.5
6.5
7.0
5.7
5.6
6.9
7.2
5.6
4.9
5.3
5.0
5.4
S.2
5.3
6.5
S.4
S.6
5.S
S.O
S.3
5.4
__
S.O
4.9
4.9
4.9
4.3
*. a
5.2
S.O
5.3
S.3
S.I
PHL 1 KA
4.5 .43
5.2 .24
4.2 .30
3.9 .23
4.8 .32
4.7 .28
4.4 .31)
4.4 .30
4.6 .26
4.0 .26
4.5 .34
4.2 .28
4.4 .29
4.6 .39
4.5 .35
4.7 .32
5.5 .31
4.9 .29
4.4 .32
6.3 .27
4.2 .18
4.9 .25
4.7 .32
4.3 .32
4.6 .25
4.8 .36
S.7 .22
5.0 .36
4.S .22
4.8 .25
4.6 .25
4.7 .2S
4.8 .25
3.4 .19
4.9 .22
4. a .17
S.O .43
4.8 .43
4.9 .17
6.2 .24
S.9 .17
S.3 .17
6.6 .17
4.5 .17
4.4 .17
5.9 .17
5.3 .17
5.3 .18
5.3 .30
6.9 .32
7.2 .10
5.3 .32
4.4 .18
4.9 .49
4.7 .22
4.7 .26
4.8 .22
4.6 .22
S.2 .22
4.9 .22
4.6 .21
4.7 .21
4.S .21
4.9 .21
S.3 .25
.49
4.7 .49
4.9 .49
4.0 .20
4.0 .22
4.2 .22
4.1 .20
4.6 .21
4.3 .20
4.T .21
4.8 .21
4.9 .30
.28
PERMA
1.30
3.26
1.97
3.30
1.30
3.31
1.97
1.97
9.12
3.65
.59
3.10
1.95
1.01
1.43
2.63
2.75
1.30
1.30
2.34
4. SI
1.92
1.30
1.30
2.67
1.30
1.72
1.30
1.30
1.30
1.30
1.30
3.15
1.30
.64
3.52
2.11
1.22
3.86
2.73
3.30
3.30
1.30
2.63
2.01
2. OS
2.77
2.36
1.30
1.30
1.30
1.30
1.30
1.30
1.30
1.30
1.30
.68
.57
.72
1.30
1.30
1.30
.94
1.30
1.30
1.10
.40
1.30
1.30
1.30
1.30
.9*
1.30
.89
1.30
.99
1.3»
PERML
.40
3.06
1.82
1.30
.13
3.31
1.53
1.53
n.ll
1.65
.18
2.80
1.75
.17
1.30
1.30
1.25
.91
1.30
2.22
4.51
1.28
.85
1.30
2.67
1.30
1.55
.55
1.30
.94
.42
.22
3.06
.26
.44
3.18
1.90
.83
3.45
2.73
3.30
3.17
1.30
2.43
1.91
2. OS
2.77
2.08
.13
.69
.71
.34
.71
.15
.13
.22
.13
.71
.20
.47
.41
.59
.49
.47
.06
.13
.06
.13
1.30
1.30
1.30
.88
.71
.99
.65
.86
.36
.94
HSQ
3.0
2.6
2.5
2.4
3.0
3.0
2.5
2.5
3.0
2.0
3.0
2.3
2.6
3.0
2.1
2.7
3.0
2.2
2.0
3.2
1.9
2.4
2.5
2.4
2.0
2.6
2.0
2.5
3.0
3.0
3.0
3.0
2.2
3.0
3.0
1.3
2.0
2.8
1.5
1.8
1.0
1.0
1.3
1.6
1.6
1.5
1.7
2.0
2.0
2.5
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.2
3.0
4.0
4.0
3.0
3.0
3.2
3.0
3.0
3.0
3.0
3.0
3.4
3.3
WATCAP
.110
.116
.122
.104
.120
.083
.107
.107
.092
.095
.122
.100
.107
.125
.157
.170
.156
.122
.170
.105
.098
.109
.140
.133
.103
.139
.112
.125
.107
.090
.088
.091
.083
.097
.059
.068
.126
.156
.083
.086
.050
.050
.080
.083
.121
.119
.094
.064
.140
.139
.120
.103
.115
.076
.056
.091
.055
.082
.040
.063
.091
.092
.078
.063
.050
.140
.160
.124
.130
.121
.099
.105
.082
.104
.077
.094
.076
.123
BDRK | LT200A
40 62. 5
46 56,3
42 54.6
49 43.0
60 62.5
2B 41.5
48 5A.2
48 56.2
42 42.9
45 56.2
41 74.6
48 M.5
43 70,2
49 HO. 6
48 78.5
43 SO. 8
43 SI. 8
60 55.0
48 77.5
37 62.5
53 39.4
52 48.3
48 70.0
55 65.0
46 58.4
52 64.1
59 77.4
SO 71.4
30 48.2
40 50.8
55 57.5
60 AO.B
40 60.5
60 46.7
54 50.4
54 41.3
49 65.5
50 6ft. 0
50 37.5
60 44.3
60 40. 0
60 43.1
60 42. S
60 52. 5
60 S8.7
60 55.6
60 46.8
60 42.9
60 53.3
60 64.4
37 55.0
60 58.7
51 52.5
60 85.0
60 49.0
60 63.9
60 48.7
46 52.5
60 48.7
$3 SO.O
5S 44.4
49 44.1
51 50.0
52 48.1
60 69.2
60 67. S
60 7S.O
60 80.0
30 52. S
30 50.9
. 23 45.0
41 51.6
46 49.8
39 49.6
4a 50.1
41 43.1
48 19.2
28 58.6
ORAVA
10. a
24.6
19.6
24.3
10.0
43.0
22.5
22.5
37.7
16.2
IS. 8
in."
11.9
10.6
7.7
9.2
10.0
37. S
7.S
23.5
4.7
49.2
13.7
12.'.
30.6
18.9
35.8
30.8
»r. a
35.8
23.7
19.4
22.7
14.2
35. 5
34.6
14.7
«,0
35.2
10.1
37.5
34.4
35.0
20.8
18.7
21.9
29.5
36.9
22.7
12.6
17.5
23.7
33.7
2.5
36.0
16.4
36.3
36.2
35.0
35.6
37.6
39.9
36.5
35.6
20. a
5.0
5.0
2.5
37.5
37.5
43.8
36.6
36.2
36. a
16.0
44.1
21.0
17.5
STONE A
2.5
5.?
2.6
6.0
S.O
12.1
6.?
6.2
10.8
4.5
4.1
4.2
3.3
2.5
1.2
S.O
5.5
2.5
0.0
1.9
?.5
7.7
2.5
2.9
6.0
1.8
4.4
5.0
7.b
7.1
6.7
6.3
24,. 0
6.7
8.6
5.4
.7
1.0
4.0
.6
2.5
5.6
2.5
6.7
3.6
.9
S.5
3.3
1.2
1.2
2.9
7.5
1.0
7.5
6.1
7.5
7.S
9.4
8.8
7,5
7.5
7.5
9.7
6.7
0.0
0.0
0.0
7.5
9.1
7.5
7.5
a. a
a. 9
a. 7
7.5
6.2
i.a
58
-------
APPENDIX 4.—Annual tonnages, by county, of commercial fertilizer and animal
wastes expressed as nitrogen and phosphorus in (tons/mi2)/yr
New York Counties
County
Allegany
Broome
Cayuga
Chemung
Chenango
Cortland
Delaware
Herkiner
Livingston
Madison
County
Adams
Bedford
Berks
Blair
Bradford
Cambria
Cameron
Centre
Chester
Clearfield
Clinton
Columbia
Cumberland
Dauphin
Elk
Franklin
Fulton
Huntingdon
Indiana
Jefferson
Juniata
Phosphorus
3.2
2.8
4.6
3.5
3.3
5.7
5.2
3.7
4.1
5.1
X
Phosphorus
6.1
4.0
7.0
6.8
3.7
5.4
5.6
4.8
5.6
3.6
5.0
5.6
5.4
6.5
2.0
6.8
4.1
4.1
3.8
5.1
7.0
Nitrogen
10.1
8.9
14.2
10.9
10.4
17.8
16.5
11.9
12.5
15.9
County
Oneida
Onandaga
Ontario
Otsego
Schohorie
Schuyler
Steuben
Tioga
Tompkins
Yates
Pennsylvania Counties
Nitrogen
19.0
11.7
19.7
20.6
10.7
16.0
10.5
12.5
17.3
9.8
15.0
15.3
16.8
18.9
6.8
18.9
11.4
11.1
10.1
13.1
21.0
County
Phosphorus
3.9
5.0
4.2
4.2
3.5
2.8
3.9
4.6
3.8
4.0
Phosphorus
Lackawanna 3.9
Lancaster
Lebanon
Luzerne
Lycoming
McKean
Mifflin
Montour
17.4
10.2
3.3
5.2
3.3
7.6
4.5
Northumberland 6.0
Perry
Potter
5.1
6.0
Schuylkill 6.8
Snyder
Somerset
Sullivan
6.6
5.9
4.0
Susquehanna 3 . 5
Tioga
Union
Wayne
Wyoming
York
3.8
6.0
3.8
4.1
5.9
Nitrogen
12.4
15.4
13.0
13.6
11.0
9.0
12.1
14.4
12.0
12.0
Nitrogen
12.5
55.0
27.7
9.1
14.0
9.2
22.3
13.3
18.0
13.5
16.8
19.2
19.5
16.0
13.0
11.9
11.3
17.1
12.3
12.1
17.1
59
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/7-78-198
2.
3. RECIPIENT'S ACCESSIOC+NO.
4. TITLE AND SUBTITLE
MULTIPLE REGRESSION MODELING APPROACH
FOR REGIONAL WATER QUALITY MANAGEMENT
5. REPORT DATE
October 1978 issuing date
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
D.J. Lystrom, F.A. Rinella, D.A. Rickert, L. Zimmermanr
8. PERFORMING ORGANIZATION REPORT NO.
WRI 78-12
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Geological Survey
U.S. Department of the Interior
Portland, Oregon 97232
10. PROGRAM ELEMENT NO.
1HE775
11. CONTRACT/GRANT NO.
EPA-IAG-D5-0792
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Research Laboratory—Athens, Ga.
Office of Research and Development
U.S. Environmental Protection Agency
Athens, Georgia 30605
13. TYPE OF REPORT AND PERIOD COVERED
Final fi/75 t-n 1?/77
14. SPONSORING AGENCY CODE
EPA/600/01
15. SUPPLEMENTARY NOTES published by Geological Survey as Regional Analysis of the Effects
of Land Use on Stream-Water Quality, Methodology and Application in the Susquehanna
River Basin, Pennsylvania and New York. NTIS accession no. PB284 185/AS
16. ABSTRACT
A framework is presented for compiling available data for assessing statistical
relationships between water quality and several factors of climate, physiography, and
land use. Seventeen water quality characteristics studied represent annual mean con-
centrations or calculated annual yields of suspended sediment, dissolved solids and
various chemical species of nitrogen and phosphorus. Usable multiple-linear regressior
were developed relating water quality characteristics to basin characteristics for 14
of 17 water quality characteristics with standard errors of estimate ranging from 17
to 75 percent. These models can be used to estimate water quality at specific stream
sites or to simulate the generalized effect of land use characteristics on water qual-
ity. For example, observed nitrate yields were up to 20 times greater than the simula-
ted background yields. This increase is indicated to be the result of chemical ferti-
lizers, animal wastes, and urbanization. It was concluded that this was a viable
method of assessing the relationships between water quality and basin characteristics
on a regional basis.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Water quality
Regional planning
Statistical analysis
Water pollution
Regression analysis
Sediment transport
Land use
Susquehanna River basin
Water quality character-
istics
Basin characteristics
Regression models
Nonpoint sources
Soil characteristics
02C
07C
12A
68D
91A
13. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
UNCT.ASSTFTF.D
21. NO. OF PAGES
68
20. SECURITY CLASS (This page)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
60 :,- U. S. GOVERNMENT PRINTING OFFICE: 1978-657-060/1533 Region No. 5-11
------- |