Water Quality and Quantity Inputs for the Urban Creeks
Future Needs Assessment
By
Michael E. Barrett, Ann M. Quenzer, and David R. Maidment
January 15, 1998
Center for Research in Water Resources
Bureau of Engineering Reseach
The University of Texas at Austin
2
TABLE OF CONTENTS
INTRODUCTION............................................................................................................. 4
SCOPE ............................................................................................................................... 5
RUNOFF COEFFICIENTS............................................................................................. 5
SINGLE LAND USE RUNOFF DATA ................................................................................... 7
LARGE WATERSHED BASEFLOW DATA............................................................................ 9
WATER QUALITY........................................................................................................ 10
SINGLE LAND USE WATER QUALITY DATA................................................................... 11
BOD Concentration .................................................................................................. 14
COD Concentrations ................................................................................................ 15
Copper Concentrations............................................................................................. 16
Dissolved Phosphorus Concentrations..................................................................... 17
Ammonia Concentration ........................................................................................... 18
Nitrate Concentrations ............................................................................................. 19
Lead Concentration .................................................................................................. 20
TKN Concentration................................................................................................... 21
TOC Concentration................................................................................................... 22
Total Phosphorous Concentration............................................................................ 23
TSS Concentration .................................................................................................... 24
Zinc Concentrations.................................................................................................. 25
BASEFLOW WATER QUALITY DATA .............................................................................. 26
LARGE WATERSHED WATER QUALITY DATA................................................................ 26
SEASONAL VARIATIONS IN CONSTITUENT CONCENTRATIONS IN AUSTIN CREEKS......... 31
Procedure.................................................................................................................. 31
Results....................................................................................................................... 32
Conclusions............................................................................................................... 34
BIBLIOGRAPHY........................................................................................................... 36
3
4
INTRODUCTION
To develop estimates of pollutant loads for urban creeks in the Austin area it is
necessary to determine the quantity and quality of stormwater runoff which might be
expected from all sites within the watersheds of these creeks. The City of Austin has
undertaken the most comprehensive water quality monitoring program in the United
States. Almost 20 years of stormwater data collected by the City in this program form
the basis for this study.
The annual pollutant loads derived from stormwater runoff from an area are
commonly calculated using the formula suggested by the EPA (1992):
)72.2)()((
12
))()((
ii
i
i
AC
RvCFP
L
?
?
?
?
?
?
=
where:
L
i
= Annual pollutant load (lb/ac/yr)
P = Annual precipitation (in/yr)
CF = Correction factor that adjusts for storms without runoff
Rv
I
= Weighted average runoff coefficient
C
i
= Average event mean concentration of the pollutant
A
i
= Catchment area (acres)
The equation is normally applied to gauged watersheds where stormwater runoff
quantity and quality has been measured during selected events and an estimate of average
annual load is required. The concept can be extended to estimate loads from ungauged
watersheds when estimates can be made of the values of the variables for all areas in the
contributing watershed. For many of the variables, such as area and average rainfall this
is a trivial task; however, estimating runoff coefficients and water quality for all areas is
more daunting.
The City of Austin?s stormwater monitoring program was designed to
characterize the quantity and quality of runoff associated with various land uses such as
single family residential, commercial, industrial, etc. Consequently, it was felt that a
5
correlation between land use and stormwater runoff characteristics could be used to
predict current pollutant loads in ungauged watersheds as well as future loads city-wide
based on various development scenarios. This report presents the results of the analysis
of this single land use monitoring data. This analysis is intended to develop estimates of
runoff coefficients and average stormwater constituent concentrations for the Austin area.
SCOPE
The time and funds available for this study limit the detail with which the
available data can be analyzed. Consequently, certain assumptions have been made about
the reliability and accuracy of the data collection and processing. In particular, it has
been assumed that each of the field sites was equipped to accurately measure flow and to
collect samples representative of the runoff, and that samples were handled and analyzed
in accordance with generally accepted protocols. The raw data was not examined for
errors in transcription, laboratory reporting or for the methodology used to deal with
censored data (i.e., values reported as below detection limit).
For each storm sampled, an event mean concentration (EMC) has been calculated
by City staff as a flow weighted concentration based on the relative volumes associated
with each discrete sample. It is assumed that these calculations were done correctly. The
area and impervious cover of each of the monitored watersheds is assumed to accurately
characterize the drainage area. The data used in this analysis consists of the runoff
coefficients and EMCs for individual events for each of the monitored watersheds.
Because of these limitations, the values developed during this initial review should not be
considered final, but they are appropriate for planning purposes and model development.
RUNOFF COEFFICIENTS
To accurately characterize the impact of increases in impervious cover on the
water quantity in urban creeks it is necessary to determine the amount of flow contributed
by all portions of the watershed. This flow is derived from direct surface runoff during
storm events as well as baseflow which originates from rainfall infiltrating on pervious
areas of the watershed. Data from the single land use monitoring program will be used to
develop estimates of the amount of surface runoff; however, none of these sites has dry
6
weather flow. Therefore, the relationship between baseflow and land use will be derived
from USGS data for large watersheds.
The runoff coefficient for a watershed is a statistical measure which attempts to
express the relationship between rainfall and direct runoff as a constant value. It is well
known that the runoff coefficient is not a constant, but depends on factors such as the
antecedent soil moisture, rainfall intensity, and rainfall volume. In addition,
extrapolation of rainfall depth measured at a single point to uniform coverage of the
entire watershed also introduces errors into the estimate. For watersheds with a high
degree of impervious cover, this extrapolation may result in an apparent contradiction for
some storms as runoff depth exceeds the rainfall depth. Despite these shortcomings, the
concept of a single runoff coefficient value for a watershed has found widespread
acceptance among engineers for estimating stormwater volumes. Consequently, in this
study, only a single value will be developed for a given area.
The term ?runoff coefficient? also is commonly applied to one of the terms in the
rational equation. In this usage, it is a coefficient which relates runoff rate to rainfall
intensity. Although the values for the different applications are similar, the values are
not interchangeable and should not be confused.
Although a runoff coefficient can be calculated for individual storms, it can also
be defined to be the ratio of runoff to rainfall over a given time period. Since one of the
goals of this research is to predict annual pollutant loads, an estimate of annual
stormwater runoff is required. The long term average runoff is required for this
calculation; therefore, the runoff coefficient for a site will be defined as suggested by
Chow et al. (1988):
?
=
=
M
m
m
d
v
R
r
R
1
where:
R
v
= the watershed runoff coefficient
7
?
=
M
m
m
R
1
= the total rainfall for all monitored events
r
d
= the corresponding depth of runoff
There are other methods for calculating the average runoff coefficient for a site
based on the underlying distribution of the data, area climate factors, or size of monitored
events. These methods are much more complex and not routinely used by engineers for
design purposes. In addition, because of the scatter in the data collected at each of the
sites, the other methods do not significantly increase the accuracy of the estimate.
Further refinement of the estimate of are runoff coefficients should include examination
of the field sites to verify that each has been equipped to accurately measure flow and
rainfall.
Single Land Use Runoff Data
Rainfall/runoff data is available for 18 watersheds in the Austin area, which have
an impervious cover which ranges from near zero to approximately 100%. A list of the
watersheds used in this analysis and their characteristics are shown in Table 1.
The runoff coefficient for an ungauged watershed is normally estimated by
developing a relationship between impervious cover and runoff coefficient for other area
watersheds. A perfect correlation between these two variables does not exist because of
other factors which vary between the monitored watersheds such as slope, soil type,
geology and other factors. Fortunately, these other factors are of secondary importance
and most previous researchers have successfully predicted runoff coefficients based on
impervious cover alone. A linear relationship was suggested by Shelley and Gaboury
(1986), while Urbonas et al. (1990) fit their data with a 3
rd
order polynomial.
The use of a linear relationship is attractive because of the well developed
statistical foundation for estimating parameter uncertainty. However, one might expect
that the effect of an increase in impervious cover of a watershed would depend on the
amount of existing impervious cover. In the extreme case, it is unlikely that all of the
first 5% of impervious cover in a watershed would be directly connected to the receiving
water so its effect would be diminished by flowing across surrounding pervious areas.
8
Conversely, the last 5% of impervious cover would all be directly connected to the
drainage system and receiving water. In addition, the best fit linear relationship predicts
a negative runoff coefficient for areas with low impervious cover. Consequently, the
rainfall/runoff data for Austin area watersheds were fit with a 2
nd
order polynomial as
shown in Figure 1.
Table 1 Watersheds Used to Estimate Rainfall/Runoff Relationship
Watershed Impervious
Cover
Area
(ac)
Number of
Observations
Total
Rain (in)
Runoff
Coefficient
Alta Vista 0.62 0.7 18 14.3 0.42
Brodie Oaks 0.95 30.9 10 14.0 0.91
Hwy 6 BMP 0.58 4.9 57 37.5 0.36
Barton Ridge Plaza 0.80 3.0 37 22.9 0.77
Hwy 5 BMP 0.64 4.6 38 26.3 0.68
Holly @ Anthony 0.43 51.3 23 15.9 0.36
Airport 0.46 99.1 15 13.5 0.38
Bear @ 1826 0.01 3563 29 31.2 0.04
Windago Way 0.01 50 78 48.9 0.03
Jollyville Rd. 0.94 7.0 29 23.6 0.77
Lost Creek 0.22 210 25 27.3 0.14
Metric Blvd. 0.6 203 45 28.0 0.48
Spy Glass 0.88 3 29 21.1 0.67
Barton Creek Mall 0.86 47 52 55.6 0.80
St. Elmo East 0.60 16.4 21 15.9 0.60
St. Elmo West 0.84 5.8 21 15.9 0.72
Tar Branch 0.45 49 18 15.0 0.26
Travis Country 0.42 42 43 40.0 0.19
Since soil type and geology affect the value of the runoff coefficient, one would
expect to see consistent differences in Austin area values related to these factors.
Potentially, one of the more important regional differences in this area is related to the
presence of the recharge zone of the Edwards aquifer. This is the area underlain by
porous Edwards limestone and which might be expected to exhibit a lower runoff
coefficient than those areas underlain by clays and relatively impermeable limestone.
The runoff coefficients for sites located on the recharge zone have been plotted in Figure
9
1 using square symbols. As many of these sites fall above the best fit regression line as
below indicating that the line applies equally well to areas on and off the recharge zone.
y = 0.3428x
2
+ 0.5677x + 0.0125
R
2
= 0.9155
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Impervious Cover
Rv
Non-Recharge Recharge
Figure 1 Relationship Between Runoff Coefficient and Impervious Cover
Large Watershed Baseflow Data
The effect of increased impervious cover on the amount of baseflow in the Austin
area creeks was estimated with stream flow data collected by the USGS. The observed
hydrograph was divided into baseflow and direct runoff components with a computer
program developed City of Austin staff. The separation of these two components is by
nature an arbitrary process, and many different algorithms are commonly used.
Consequently, there is no unique ?correct? solution. The City program was not analyzed
for methodology; however, the output of the flow separation program was reviewed for
several monitoring stations and the results appear to be reasonable.
The amount of baseflow (as a fraction of the rainfall) was calculated for several
creeks with different levels of impervious cover. A subsurface runoff coefficient, R
s
, can
be defined in a manner analogous to the conventional surface runoff coefficient as the
total baseflow divided by the total rainfall for the corresponding period. A list of the sites
and amount of impervious cover is contained in Table 1. The amount of impervious
cover for each watershed was estimated from maps of land use provided by the City
10
Planning Department. The department also provided a table which related land use to the
degree of impervious cover.
Table 2 USGS Sites Used to Estimate Baseflow
Site Impervious Cover R
s
Barton Creek @ Lost Creek 0.09 0.14
Williamson Creek @ Oak Hill 0.16 0.12
Shoal Creek @ 12
th
0.54 0.03
Walnut Creek @ Webberville 0.30 0.09
Bear Creek @ 1826 0.07 0.16
Slaughter Creek @ 1826 0.13 0.18
Bull Creek @ Loop 360 0.14 0.15
Boggy 0.53 0.02
To determine the effect of impervious cover on the amount of baseflow, a linear
regression was performed on the two variables as shown in Figure 2. A perfect
correlation would not be expected because of differences in soil type, bedrock geology
and other factors that influence the amount of rainfall which reappears in a given stream.
The regression line intercepts the x-axis at an impervious cover of 0.87, indicating that no
baseflow would be generated from land uses with greater impervious cover.
WATER QUALITY
The total constituent load delivered by each creek is a function of the quality and
quantity of both baseflow and direct storm runoff. Consequently, it is necessary to
estimate the average water quality for both flow regimes and to relate that quality to the
land use, impervious cover or other characteristics of the watershed. The water quality of
stormwater discharges can be estimated from the single land use monitoring data
collected by the City; however, since there is no baseflow at these sites, the relationship
between baseflow quality and watershed characteristics must be estimated from the large
watershed data collected by the USGS.
11
y = -0.296x + 0.1837
R
2
= 0.9315
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0 0.1 0.2 0.3 0.4 0.5 0.6
IC
Rs
Figure 2 Relationship Between Impervious Cover and Baseflow
Single Land Use Water Quality Data
It is normally assumed that the quality of stormwater runoff from an area is
largely a function of the land use. Consequently, most stormwater monitoring programs
(including those mandated by the EPA) are developed to assess the quality associated
with specific land uses. If a municipality were to develop a program based on sampling
from watersheds which each have a unique land use, it is likely that each land use would
have a unique value associated with it. In all subsequent calculations it would be
assumed that all other areas with the same land use would have a similar average water
quality. The City of Austin monitoring program was large enough to support the
monitoring of several sites with similar land uses. In particular, there are multiple sites
representing single and multi-family residential, commercial, industrial and other land
uses. There are then two main objectives. The first is to characterize the ?average?
quality of stormwater runoff for the monitored site (i.e., that value when multiplied by the
annual runoff produces the total annual load from the watershed). The second is to
extrapolate those data to develop ?average? water quality values for ungauged watersheds
based on land use or other considerations.
12
Event mean concentrations have been developed by City staff for individual
storms at the monitoring sites based on a flow weighted average concentration of discrete
samples. There are several possible methodologies which use these values to calculate
the concentration which represents the long term average quality. The choice of
methodology might depend on assumptions about the underlying distribution of the data,
the types of storms sampled relative to those that occur most commonly in the area, or
other factors. It is also helpful to select a methodology which is widely accepted by the
engineering and scientific communities. Consequently, the method recommended by the
Driscoll (1983) and Gilbert (1987) for calculation of average concentrations for
constituents which exhibit a lognormal distribution will be used. This method calculates
the average concentration of a constituent at a site as:
)
2
(
2
w
u
eM
+
=
where:
M = average concentration
u = mean of the log transformed EMCs
w
2
= variance of the log transformed EMCs
Once the average concentration for a site has been determined, it is necessary to
estimate concentrations for other, ungauged watersheds. It is commonly assumed that the
type of land use is the major factor controlling the quality of stormwater runoff; however,
these data did not strongly support that assumption. Within each land use category, it
was found that concentrations varied widely for all constituents, so that the average
concentration for each land use was not statistically different from those calculated for
other land uses with multiple sites. This was also the conclusion of the EPA based on the
analysis of water quality from sites across the country as part of the National Urban
Runoff Program (1983). However, a fairly strong linear correlation with impervious
cover was evident for many constituents and these relationships were used to estimate the
water quality derived from ungauged watersheds. Concentrations of constituents
13
correlated with impervious cover at a confidence level of less than 85% were estimated
based on the arithmetic mean of all monitored sites.
There were several sites which were part of the monitoring program were
excluded from this analysis. The list of sites and the reason for not using the data are
listed in Table 3.
Table 3 Monitoring Sites Excluded from Analysis
Site Comments
Airport Water quality not representative of most urban land uses
Alta Vista Samples collected after runoff flows across grassy swale
Old Bear Creek Monitoring data challenged in court
Travis Country Ditch
Concentrations much lower than adjacent site, represents
water quality after grassy swale
Holly @ Anthony
Concentrations far higher than at any other site for all
constituents. May include in future analyses pending field
check for illegal connections or dumping.
14
BOD Concentration
Table 4 Sites Used to Estimate BOD Concentrations
Site Land Use Drainage Imperv. # EMCs Average EMC
(mg/L)
Barton Ck. Sq. Mall COMM 47.0 0.86 22 13
Barton Ridge Inflow COMM 3.0 0.80 14 13
Lavaca COMM 13.7 0.97 21 19
5th Street COMM 4.0 0.95 18 22
St. Elmo East INDU 16.4 0.60 6 7
East 7th Street INDU 29.3 0.70 2 14
Metric Blvd. INDU 202.9 0.60 21 14
Highwood Apts. MFR 3.0 0.50 24 9
Burton Road MFR 12.0 0.82 17 20
Spy Glass OFFI 1.5 0.86 13 14
Rollingwood SFR 62.8 0.21 8 6
Lost Creek SFR 209.9 0.23 18 7
Maple Run SFR 27.8 0.36 24 8
Hart Lane SFR 371.0 0.39 20 10
Travis Country Pipe SFR 41.6 0.41 15 12
Jollyville Road TRAN 9.5 0.81 24 8
Windago Way UNDEV 50.0 0.01 8 4
y = 13.859x + 3.5014
R
2
= 0.6202
0
5
10
15
20
25
0.00 0.20 0.40 0.60 0.80 1.00
Impervious Cover
BO
D EM
C,
m
g
/
L
Figure 3 Relationship Between BOD and Impervious Cover
15
COD Concentrations
Table 5 Sites Used to Estimate COD Concentrations
Site Land Use Drainage Imperv. # EMCs Average EMC
(mg/L)
Barton Ridge Inflow COMM 3.0 0.80 15 80
Barton Ck. Sq. Mall COMM 47.0 0.86 23 106
Lavaca COMM 13.7 0.97 22 115
5th Street COMM 4.0 0.95 25 154
St. Elmo East INDU 16.4 0.60 6 51
Metric Blvd. INDU 202.9 0.60 21 78
East 7th Street INDU 29.3 0.70 3 124
Highwood Apts. MFR 3.0 0.50 25 39
Burton Road MFR 12.0 0.82 17 120
Spy Glass OFFI 1.5 0.86 15 85
Maple Run SFR 27.8 0.36 25 34
Hart Lane SFR 371.0 0.39 29 44
Lost Creek SFR 209.9 0.23 19 49
Rollingwood SFR 62.8 0.21 20 51
Travis Country Pipe SFR 41.6 0.41 19 71
HWY BMP 5 Inflow TRAN 4.6 0.64 5 63
Jollyville Road TRAN 9.5 0.81 28 76
Windago Way UNDEV 50.0 0.01 8 39
y = 97.72x + 18.254
R
2
= 0.6178
0
20
40
60
80
100
120
140
160
180
0.00 0.20 0.40 0.60 0.80 1.00
Impervious Cover
CO
D EM
C,
m
g
/
L
Figure 4 Relationship Between COD and Impervious Cover
16
Copper Concentrations
Table 6 Sites Used to Estimate Copper Concentrations
Site Land Use Drainage Imperv. # EMCs Average EMC
(mg/L)
Barton Ridge Inflow COMM 3.0 0.80 5 0.007
5th Street COMM 4.0 0.95 9 0.025
Lavaca COMM 13.7 0.97 12 0.029
St. Elmo East INDU 16.4 0.60 7 0.014
East 7th Street INDU 29.3 0.70 2 0.016
Metric Blvd. INDU 202.9 0.60 9 0.030
Highwood Apts. MFR 3.0 0.50 25 0.009
Burton Road MFR 12.0 0.82 11 0.022
Spy Glass OFFI 1.5 0.86 4 0.011
Travis Country Pipe SFR 41.6 0.41 11 0.007
Maple Run SFR 27.8 0.36 25 0.007
Rollingwood SFR 62.8 0.21 10 0.009
Lost Creek SFR 209.9 0.23 11 0.013
Hart Lane SFR 371.0 0.39 18 0.015
Jollyville Road TRAN 9.5 0.81 28 0.018
Windago Way UNDEV 50.0 0.01 3 0.008
y = 0.0156x + 0.006
R
2
= 0.3308
0.000
0.005
0.010
0.015
0.020
0.025
0.030
0.035
0.00 0.20 0.40 0.60 0.80 1.00
Impervious Cover
Cu
EM
C,
m
g
/
L
Figure 5 Relationship Between Copper and Impervious Cover
17
Dissolved Phosphorus Concentrations
Table 7 Sites Used to Estimate Dissolved Phosphorus Concentrations
Site Land Use Drainage Imperv. # EMCs Average EMC
(mg/L)
Barton Ridge Inflow COMM 3.0 0.80 12 0.138
5th Street COMM 4.0 0.95 18 0.252
Lavaca COMM 13.7 0.97 11 0.427
St. Elmo East INDU 16.4 0.60 7 0.079
Metric Blvd. INDU 202.9 0.60 17 0.168
East 7th Street INDU 29.3 0.70 2 0.207
Burton Road MFR 12.0 0.82 8 0.318
Spy Glass OFFI 1.5 0.86 13 0.138
Lost Creek SFR 209.9 0.23 7 0.130
Travis Country Pipe SFR 41.6 0.41 13 0.196
Windago Way UNDEV 50.0 0.01 5 0.044
y = 0.2371x + 0.0405
R
2
= 0.4421
0.000
0.050
0.100
0.150
0.200
0.250
0.300
0.350
0.400
0.450
0.00 0.20 0.40 0.60 0.80 1.00
IC
DP EM
C,
m
g
/
L
Figure 6 Relationship Between Dissolved Phosphorus and Impervious Cover
18
Ammonia Concentration
Table 8 Sites Used to Estimate Ammonia Concentrations
Site Land Use Drainage Imperv. # EMCs Average EMC
(mg/L)
Barton Ridge Inflow COMM 3.0 0.80 15 0.299
Lavaca COMM 13.7 0.97 13 0.367
5th Street COMM 4.0 0.95 25 0.379
St. Elmo East INDU 16.4 0.60 7 0.299
Metric Blvd. INDU 202.9 0.60 16 0.302
East 7th Street INDU 29.3 0.70 3 0.319
Highwood Apts. MFR 3.0 0.50 25 0.223
Burton Road MFR 12.0 0.82 8 0.300
Spy Glass OFFI 1.5 0.86 15 0.222
Rollingwood SFR 62.8 0.21 12 0.179
Maple Run SFR 27.8 0.36 25 0.205
Lost Creek SFR 209.9 0.23 11 0.207
Hart Lane SFR 371.0 0.39 20 0.214
Travis Country Pipe SFR 41.6 0.41 16 0.306
Jollyville Road TRAN 9.5 0.81 28 0.400
Windago Way UNDEV 50.0 0.01 7 0.074
y = 0.2446x + 0.1273
R
2
= 0.695
0.000
0.100
0.200
0.300
0.400
0.500
0.00 0.20 0.40 0.60 0.80 1.00
Impervious Cover
N
H
3
E
M
C
,
m
g
/L
Figure 7 Relationship Between Ammonia and Impervious Cover
19
Nitrate Concentrations
Nitrate concentrations in stormwater runoff or not correlated with either land use
or impervious cover, so the average concentration for all sites of 0.82 mg/L-N was used.
Table 9 Sites Used to Estimate Nitrate Concentrations
Site Land Use Drainage Imperv. # EMCs Average EMC
(mg/L)
Barton Ck. Sq. Mall COMM 47.0 0.86 23 0.422
Lavaca COMM 13.7 0.97 21 0.676
Barton Ridge Inflow COMM 3.0 0.80 15 0.690
5th Street COMM 4.0 0.95 20 0.839
Metric Blvd. INDU 202.9 0.60 20 0.692
St. Elmo East INDU 16.4 0.60 6 1.490
East 7th Street INDU 29.3 0.70 2 2.184
Highwood Apts. MFR 3.0 0.50 25 0.293
Burton Road MFR 12.0 0.82 15 0.713
Spy Glass OFFI 1.5 0.86 16 0.889
Maple Run SFR 27.8 0.36 25 0.427
Lost Creek SFR 209.9 0.23 16 0.630
Travis Country Pipe SFR 41.6 0.41 18 0.662
Rollingwood SFR 62.8 0.21 20 0.919
Hart Lane SFR 371.0 0.39 30 1.148
HWY BMP 5 Inflow TRAN 4.6 0.64 2 0.430
Jollyville Road TRAN 9.5 0.81 27 0.472
Windago Way UNDEV 50.0 0.01 9 1.230
0.000
0.500
1.000
1.500
2.000
2.500
0.00 0.20 0.40 0.60 0.80 1.00
IC
N
O
3
E
M
C
,
m
g
/L
Figure 8 Relationship (or lack of) Between Nitrate and Impervious Cover
20
Lead Concentration
Table 10 Sites Used to Estimate Lead Concentrations
Site Land Use Drainage Imperv. # EMCs Average EMC
(mg/L)
Barton Ridge Inflow COMM 3.0 0.80 5 0.013
5th Street COMM 4.0 0.95 9 0.036
Lavaca COMM 13.7 0.97 10 0.075
St. Elmo East INDU 16.4 0.60 6 0.010
East 7th Street INDU 29.3 0.70 2 0.024
Metric Blvd. INDU 202.9 0.60 9 0.031
Highwood Apts. MFR 3.0 0.50 25 0.011
Burton Road MFR 12.0 0.82 11 0.024
Spy Glass OFFI 1.5 0.86 3 0.015
Maple Run SFR 27.8 0.36 25 0.007
Lost Creek SFR 209.9 0.23 11 0.007
Rollingwood SFR 62.8 0.21 10 0.014
Travis Country Pipe SFR 41.6 0.41 12 0.015
Hart Lane SFR 371.0 0.39 18 0.044
HWY BMP 5 Inflow TRAN 4.6 0.64 2 0.038
Jollyville Road TRAN 9.5 0.81 28 0.049
Windago Way UNDEV 50.0 0.01 3 0.007
y = 0.0383x + 0.0025
R
2
= 0.328
0.000
0.010
0.020
0.030
0.040
0.050
0.060
0.070
0.080
0.00 0.20 0.40 0.60 0.80 1.00
Impervious Cover
P
b
E
M
C
,
m
g
/L
Figure 9 Relationship Between Lead and Impervious Cover
21
TKN Concentration
Table 11 Sites Used to Estimate TKN Concentrations
Site Land Use Drainage Imperv. # EMCs Average EMC
(mg/L)
Barton Ck. Sq. Mall COMM 47.0 0.86 23 1.77
Barton Ridge Inflow COMM 3.0 0.80 15 1.85
Lavaca COMM 13.7 0.97 22 2.50
5th Street COMM 4.0 0.95 23 3
St. Elmo East INDU 16.4 0.60 7 1.06
Metric Blvd. INDU 202.9 0.60 21 1.83
East 7th Street INDU 29.3 0.70 3 2.15
Highwood Apts. MFR 3.0 0.50 22 0.69
Burton Road MFR 12.0 0.82 18 2.01
Spy Glass OFFI 1.5 0.86 15 1.58
Maple Run SFR 27.8 0.36 25 0.84
Hart Lane SFR 371.0 0.39 20 0.97
Rollingwood SFR 62.8 0.21 12 1.03
Lost Creek SFR 209.9 0.23 18 1.45
Travis Country Pipe SFR 41.6 0.41 17 1.90
Jollyville Road TRAN 9.5 0.81 27 1.09
HWY BMP 5 Inflow TRAN 4.6 0.64 2 1.21
Windago Way UNDEV 50.0 0.01 9 0.88
y = 1.4104x + 0.6852
R
2
= 0.4419
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
0.00 0.20 0.40 0.60 0.80 1.00
Impervious Cover
T
K
N
E
M
C
,
m
g
/L
Figure 10 Relationship Between TKN and Impervious Cover
22
TOC Concentration
Table 12 Sites Used to Estimate TOC Concentrations
Site Land Use Drainage Imperv. # EMCs Average EMC
(mg/L)
Barton Ridge Inflow COMM 3.0 0.80 14 6
Lavaca COMM 13.7 0.97 14 11
5th Street COMM 4.0 0.95 21 21
Barton Ck. Sq. Mall COMM 47.0 0.86 23 25
St. Elmo East INDU 16.4 0.60 7 9
East 7th Street INDU 29.3 0.70 3 9
Metric Blvd. INDU 202.9 0.60 15 13
Highwood Apts. MFR 3.0 0.50 25 10
Burton Road MFR 12.0 0.82 7 14
Spy Glass OFFI 1.5 0.86 13 18
Lost Creek SFR 209.9 0.23 10 7
Travis Country Pipe SFR 41.6 0.41 18 9
Hart Lane SFR 371.0 0.39 32 10
Maple Run SFR 27.8 0.36 25 12
Rollingwood SFR 62.8 0.21 19 20
HWY BMP 5 Inflow TRAN 4.6 0.64 5 7
Jollyville Road TRAN 9.5 0.81 24 26
Windago Way UNDEV 50.0 0.01 8 8
y = 8.5842x + 8.0139
R
2
= 0.1484
0
5
10
15
20
25
30
0.00 0.20 0.40 0.60 0.80 1.00
Impervious Cover
T
O
C EM
C,
m
g
/
L
Figure 11 Relationship Between TOC and Impervious Cover
23
Total Phosphorous Concentration
Table 13 Sites Used to Estimate Total Phosphorous Concentration
Site Land Use Drainage Imperv. # EMCs Average EMC
(mg/L)
Barton Ck. Sq. Mall COMM 47.0 0.86 23 0.249
Barton Ridge Inflow COMM 3.0 0.80 15 0.391
Lavaca COMM 13.7 0.97 22 0.552
5th Street COMM 4.0 0.95 23 0.617
St. Elmo East INDU 16.4 0.60 7 0.307
Metric Blvd. INDU 202.9 0.60 22 0.470
East 7th Street INDU 29.3 0.70 3 1.104
Highwood Apts. MFR 3.0 0.50 24 0.211
Burton Road MFR 12.0 0.82 17 0.592
Spy Glass OFFI 1.5 0.86 15 0.216
Maple Run SFR 27.8 0.36 25 0.249
Rollingwood SFR 62.8 0.21 19 0.261
Hart Lane SFR 371.0 0.39 33 0.295
Lost Creek SFR 209.9 0.23 18 0.307
Travis Country Pipe SFR 41.6 0.41 18 0.414
Jollyville Road TRAN 9.5 0.81 28 0.222
HWY BMP 5 Inflow TRAN 4.6 0.64 2 0.301
Windago Way UNDEV 50.0 0.01 9 0.153
y = 0.3177x + 0.1944
R
2
= 0.1546
0.000
0.200
0.400
0.600
0.800
1.000
1.200
0.00 0.20 0.40 0.60 0.80 1.00
IC
T
P
E
M
C
,
m
g
/L
Figure 12 Relationship Between Total Phosphorus and Impervious Cover
24
TSS Concentration
A TSS concentration of 190 mg/L was used based on the average of all monitored
sites.
Table 14 Sites Used to Estimate TSS Concentrations
Site Land Use Drainage Imperv. # EMCs Average EMC
(mg/L)
5th Street COMM 4.0 0.95 18 142
Lavaca COMM 13.7 0.97 23 179
Barton Ck. Sq. Mall COMM 47.0 0.86 23 231
Barton Ridge Inflow COMM 3.0 0.80 15 289
St. Elmo East INDU 16.4 0.60 6 155
East 7th Street INDU 29.3 0.70 3 193
Metric Blvd. INDU 202.9 0.60 22 268
Highwood Apts. MFR 3.0 0.50 25 116
Burton Road MFR 12.0 0.82 17 296
Spy Glass OFFI 1.5 0.86 13 66
Lost Creek SFR 209.9 0.23 18 110
Travis Country Pipe SFR 41.6 0.41 18 128
Hart Lane SFR 371.0 0.39 33 156
Rollingwood SFR 62.8 0.21 19 206
Maple Run SFR 27.8 0.36 25 305
HWY BMP 5 Inflow TRAN 4.6 0.64 5 128
Jollyville Road TRAN 9.5 0.81 28 335
Windago Way UNDEV 50.0 0.01 10 95
0
50
100
150
200
250
300
350
400
0.00 0.20 0.40 0.60 0.80 1.00
Impervious Cover
T
SS EM
C (
m
g
/
L
)
Figure 13 Relationship Between TSS and Impervious Cover
25
Zinc Concentrations
Table 15 Sites Used to Estimate Zinc Concentrations
Site Land Use Drainage Imperv. # EMCs Average EMC
(mg/L)
Barton Ridge Inflow COMM 3.0 0.80 4 0.094
5th Street COMM 4.0 0.95 10 0.234
Lavaca COMM 13.7 0.97 9 0.271
St. Elmo East INDU 16.4 0.60 6 0.098
East 7th Street INDU 29.3 0.70 2 0.108
Metric Blvd. INDU 202.9 0.60 9 0.174
Highwood Apts. MFR 3.0 0.50 25 0.045
Burton Road MFR 12.0 0.82 11 0.131
Spy Glass OFFI 1.5 0.86 4 0.099
Maple Run SFR 27.8 0.36 25 0.022
Rollingwood SFR 62.8 0.21 10 0.039
Travis Country Pipe SFR 41.6 0.41 10 0.045
Lost Creek SFR 209.9 0.23 10 0.047
Hart Lane SFR 371.0 0.39 18 0.051
HWY BMP 5 Inflow TRAN 4.6 0.64 6 0.145
Jollyville Road TRAN 9.5 0.81 28 0.170
Windago Way UNDEV 50.0 0.01 3 0.065
y = 0.1877x
R
2
= 0.5835
0.000
0.050
0.100
0.150
0.200
0.250
0.300
0.00 0.20 0.40 0.60 0.80 1.00
Impervious Cover
Z
n
E
M
C
,
m
g
/L
Figure 14 Relationship Between Zinc and Impervious Cover
26
Baseflow Water Quality Data
The relationship between the quality of baseflow and land use was determined
from an analysis of the large watershed water quality data. The USGS sites on the large
watersheds sample water derived from a wide range of land uses and impervious covers;
consequently, it is essentially impossible to associate water quality with specific land
uses or degree of impervious cover. Therefore baseflow quality was determined for only
two cases, developed and undeveloped. Watersheds selected to represent largely
undeveloped watersheds include Barton Creek at Highway 71, Onion Creek near
Driftwood, and Slaughter Creek at 1826. Developed watersheds selected include Shoal
Creek at 12
th
Street, Waller at 38
th
, Waller at 23
rd
, and Boggy Creek. The average
baseflow quality at each site was calculated by City staff as the arithmetic mean of all
samples collected during dry weather flow. The average for each category was calculated
as the arithmetic mean of the appropriate site concentrations. Since baseflow is derived
from groundwater, it was assumed that the TSS concentration was essentially zero and
that sediment present during baseflow was derived from channel erosion or growth of
algae in the stream. The results of these calculations are presented in Table 16. The
quality of direct runoff for each parameter as a function of impervious cover (IC,
expressed as decimal fraction) also is included in the table. There was no data available
on total copper, lead, or zinc concentrations in baseflow.
Large Watershed Water Quality Data
The annual constituent load for the large watersheds is calculated as the product
of annual runoff and average constituent concentration and is required for calibration of
the GIS model. The load can be estimated by dividing the flow into direct runoff and
baseflow and characterizing the quality of each of these separately. The data collected as
part of the COA/USGS joint monitoring program provides the basis for these estimates.
The amount of baseflow and direct runoff at each of the monitored sites were
calculated by City staff using a computer program to separate these components of the
stream hydrographs. This is the same program that was used to develop a relationship
between impervious cover and the amount of baseflow. Although the program
27
algorithms were not reviewed as part of this study, the results appear reasonable and are
in general agreement with estimates made by CRWR researchers.
Table 16 Water Quality Model Inputs
Constituent Storm Conc.
(mg/L)
Undev.
Baseflow
Dev.
Baseflow
TSS 190 0 0
BOD C=14(IC)+3.5 0.45 0.8
COD C=98(IC)+18 12 20
TOC C=8.6(IC)+8 2 5
DP C=0.24(IC)+0.04 0.014 0.06
TP C=0.32(IC)+0.19 0.02 0.12
NH
3
C=0.24(IC)+0.13 0.02 0.06
TKN C=1.53(IC)+0.13 0.28 0.46
NO
3
0.82 0.15 0.6
Cu C=0.016(IC)+0.006 NA NA
Pb C=0.038(IC)+0.003 NA NA
Zn C=0.19(IC) NA NA
Water quality measurements at these sites consists of discrete samples collected
periodically during dry weather to characterize the quality of baseflow. In addition, more
intensive sampling has been conducted during storm events, usually consisting of 4 to 6
samples per event. City staff has proposed several ways to analyze these data to calculate
average concentrations in the creeks; however, it is advantageous to adopt a standard
methodology so the calculated values can be readily compared with those reported in
other studies.
Event mean concentrations (EMC) for each monitored direct runoff event were
calculated by City staff based on a flow weighted average during the events. EMC?s are
known to generally exhibit a lognormal distribution; consequently, the long term mean
concentration can be estimated with the equation suggested by Driscoll (1986) and Huber
(1992).
)CV(MedianMean
2
1+?=
28
where CV is the coefficient of variation of the measured values. The mean
concentrations calculated for direct runoff are shown in Table 17. The column labeled
weight indicates the fraction of the total runoff that is composed of direct storm runoff at
each of the sites. The shaded entries in the table are estimated values and were not
calculated from concentrations measured at that location.
The measured baseflow concentrations may follow either a normal or lognormal
distribution; however, the range of measured values is relatively small so the calculated
mean is not very sensitive to the method selected for the estimate. Therefore, the average
concentration during dry weather flow was estimated as the arithmetic mean of the
measured values. The calculated concentrations are shown in the second part of Table
17.
The long term average concentration in the creeks was calculated as the weighted
average of baseflow and direct runoff concentrations. These average concentrations are
shown in the final portion of Table 17.
Storm conditions
Site Weight BOD
(mg/L)
COD
(mg/L)
TSS
(mg/L)
VSS
(mg/L)
NH3
(mg/L)
TKN
(mg/L)
NO23
(mg/L)
TN
(mg/L)
TP
(mg/L)
DP
(mg/L)
TOC
(mg/L)
TCD
(?g/L)
TPB
(?g/L)
FCOL
(?g/L)
FSTR
(?g/L)
Bull Ck 53.70 5.0 76 2145 177 0.08 2.85 0.51 3.25 0.29 0.06 38.4 0.05 15.50 52705 49704
BC @ Hwy 71 53.71 2.9 37 486 43 0.03 0.70 0.18 1.00 0.09 0.03 12.6 1.00 9.00 13054 24543
BC@Lost C. 52.87 2.9 34 323 31 0.03 0.67 0.22 0.89 0.11 0.04 13.1 1.99 6.01 12772 20808
BC@Loop 360 65.81 3.9 41 863 83 0.06 1.38 0.31 1.73 0.15 0.05 21.9 . 8.53 22222 27102
Shoal C. 91.61 13.8 90 1534 206 0.17 3.64 0.57 4.11 1.17 0.30 38.2 . 38.90 162762 156482
Waller 38 69.08 11.0 81 586 80 0.21 2.28 0.91 3.07 0.64 0.22 9.5 1.05 114.00 66599 84419
Waller 23 76.56 12.9 92 488 96 0.24 2.36 0.83 3.13 0.71 0.24 11.9 0.86 121.00 78881 88845
Boggy 95.13 14.6 89 1874 205 0.17 3.87 0.49 4.27 1.67 0.12 44.0 . . 250895 269771
Walnut 71.22 8.5 79 1412 151 0.18 1.88 0.68 2.62 0.60 0.18 20.4 1.28 27.30 20522 77693
Williamson 66.54 9.0 . 625 91 0.09 2.92 0.40 3.28 0.63 . 29.3 . . 81875 144789
Baseflow conditions
Site Weight BOD
(mg/L)
COD
(mg/L)
TSS
(mg/L)
VSS
(mg/L)
NH3
(mg/L)
TKN
(mg/L)
NO23
(mg/L)
TN
(mg/L)
TP
(mg/L)
DP
(mg/L)
TOC
(mg/L)
TCD
(?g/L)
TPB
(?g/L)
FCOL
(?g/L)
FSTR
(?g/L)
Bull Ck 46.30 0.8 12 4 3 0.03 0.35 0.21 0.56 0.02 0.07 3.2 1.00 1.00 567 1166
BC @ Hwy 71 46.29 0.4 12 3 3 0.02 0.25 0.10 0.37 0.02 0.02 2.3 1.00 2.86 60 530
BC@Lost C. 47.13 0.5 11 4 4 0.02 0.22 0.17 0.40 0.03 0.02 2.2 1.00 1.18 82 154
BC@Loop 360 34.19 0.4 25 4 5 0.02 0.43 0.16 0.62 0.01 0.02 2.3 1.00 1.10 38 109
Shoal C. 8.39 0.8 15 6 3 0.05 0.46 0.60 1.06 0.04 0.04 3.9 . 1.29 7822 3364
Waller 38 30.92 0.8 10 5 2 0.05 0.59 0.80 1.78 0.25 0.04 . . . 437 391
Waller 23 23.44 0.8 10 5 2 0.10 0.59 1.19 1.78 0.25 0.04 . . . . .
Boggy 4.87 0.9 34 9 3 0.03 0.35 0.47 0.82 0.05 0.09 5.4 . . 3023 1311
Walnut 28.78 0.8 15 5 3 0.04 0.48 0.55 1.05 0.03 0.04 3.8 . . 533 598
Williamson 33.46 0.6 10 3 2 0.03 0.34 0.26 0.56 0.17 0.08 3.0 . . 251 598
30
All conditions
Site Weight BOD
(mg/L)
COD
(mg/L)
TSS
(mg/L)
VSS
(mg/L)
NH3
(mg/L)
TKN
(mg/L)
NO23
(mg/L)
TN
(mg/L)
TP
(mg/L)
DP
(mg/L)
TOC
(mg/L)
TCD
(?g/L)
TPB
(?g/L)
FCOL
(?g/L)
FSTR
(?g/L)
Bull Ck 100.00 3.1 46 1154 96 0.06 1.69 0.37 2.00 0.16 0.07 22.1 0.49 8.79 28567 27233
BC @ Hwy 71 100.00 1.8 25 262 24 0.02 0.49 0.14 0.71 0.05 0.02 7.8 1.00 6.16 7039 13427
BC@Lost C. 100.00 1.8 23 172 19 0.03 0.46 0.20 0.66 0.07 0.03 7.9 1.52 3.73 6792 11075
BC@Loop 360 100.00 2.7 36 569 56 0.04 1.06 0.26 1.35 0.11 0.04 15.2 . 5.99 14636 17872
Shoal C. 100.00 12.7 84 1406 189 0.16 3.37 0.57 3.85 1.08 0.28 35.3 . 35.74 149759 143632
Waller 38 100.00 7.8 59 406 56 0.16 1.76 0.87 2.67 0.52 0.17 . . . . .
Waller 23 100.00 10.1 73 375 74 0.21 1.95 0.91 2.81 0.60 0.19 . . . . .
Boggy 100.00 13.9 86 1783 195 0.17 3.70 0.49 4.10 1.59 0.12 42.1 . . 238827 256701
Walnut 100.00 6.3 60 1007 108 0.14 1.48 0.64 2.17 0.44 0.14 15.6 . . 14769 55504
Williamson 100.00 6.2 . 417 61 0.07 2.06 0.35 2.37 0.47 . 20.5 . . 54566 96547
Table 17 Average Concentrations for Large Watersheds
Seasonal Variations in Constituent Concentrations in Austin Creeks
Procedure
The constituent concentrations measured from the Creeks were separated into
those measurements made during base flow and those made during storm flow. Three
constituents are analyzed here; total nitrogen, total phosphorus and BOD; because of the
importance of these constituents in the WASP water quality model. A spreadsheet was
set up using Microsoft Excel 5.0. Columns were made for the date that the measurement
was made (day), the concentration of the constituent measured in the sample (mg/l), and
the flow of the creek at the time of the sample (m
3
/s). If there was more than one sample
on a particular day, a flow weighted average was calculated using Equation 1.
Average Concentration = ? (Flow * Concentration) / ? Flow (1)
Finally, all days were converted to the month of the year. Therefore, each day a sample
was taken represented a typical constituent concentration found in that month. For
example, January 1, 1985, is the same as January 15, 1995.
Using Fourier Series and a multiple regression analysis, the measured
concentrations were analyzed for seasonal variations. Equation 2 was used to set up the
Fourier Series.
C
(j,m)
= a
o
+ ? (a
k
sin(2k?m/12) + b
k
cos(2k?m/12)) (2)
where: C
(j,m)
= concentration measured at site j in month m (mg/l)
j = index of sampling sites
a
o
, a
k
, b
k
= intersects
k = harmonics number (1, 2, ?, 5)
m = month of the year(1, 2, ?, 12)
The sin and cos functions were calculated for each of the samples that were measured.
32
The Fourier Series is used to describe the cyclical behavior of the concentrations
where additional frequencies are added to describe the function. For example, if k = 1,
then the concentrations have a one cycle per year (12 month periodicity). If k = 2, then
the concentrations have two cycles per year (6 month periodicity), and so on.
Multiple regression uses least squares to analyze the relationship between one
dependent variable and one or more independent variables. A stepwise regression allows
control of the way the system enters and removes variable from the regression analysis.
There are two options in a stepwise regression. First, a backwards selection allows for a
stepwise analysis where all the variables are considered initially and then removed one at
a time if they are not statistically significant. Alternatively, a forward selection allows
for a stepwise analysis where no variables are considered initially and then are added one
at a time to obtain the final result.
In the monthly concentration analysis, the concentration was considered the
dependent variable with the month the measurement was taken and the sin and cos
functions being the independent variables.
A spreadsheet was set up with the concentrations in column 1, the month that the
measurement was taken in column 2, the sin and cos functions were in columns 3-12.
The spreadsheet was imported into the statistical package StatsgraphicsPlus. A multiple
regression analysis was run on the three constituents for base flow and storm flow. The
analysis included a regular multiple regression calculation, as well as, a forward and
backward selection analysis.
Results
The multiple regression tool in StatsgraphicsPlus was used to analyze the
constituents total nitrogen, total phosphorus, and BOD for storm flow and base flow. The
F-ratio and the R-squared results were used to determine if there was a seasonal variation
for constituent concentrations in the Austin Creeks. The F- ratio is a measure of variance
explained by a ratio of the mean square of the model to the mean square error. The R-
squared number reflects the extent of a linear relationship between the data sets. The F-
ratio for the data sets should be greater than four to be statistically significant showing
33
that the data is not random and has some sort of trend. For the data to have a significant
trend, the F-ratio is in the 100 to 300 range.
Tables 18 and 19 show the F-ratio and R-squared values from the multiple
regression and stepwise multiple regression analysis for the constituents; total nitrogen,
total phosphorus, and BOD for both storm flow and base flow. Figure 15 shows the
observed concentrations for each of the three constituents analyzed for the base flow and
the storm flow conditions. Both the table and the graphs illustrate very little, if any,
seasonal variation in the constituent concentrations.
Table 18 F-ratio and R- Squared Values for Storm Flow.
Storm Flow
Constituent Multiple Regression Forward Selection Backward Selection
F-Ratio R-Squared F-Ratio R-Squared F-Ratio R-Squared
Total Nitrogen 2.35 7.48 14.19 4.12 14.19 4.12
Total Phosphorus 1.93 6.27 10.15 3.00 10.15 3.00
BOD 4.28 12.91 9.40 10.37 5.98 11.51
Table 19 F-ratio and R- Squared Values for the Constituents Analyzed ? Base Flow.
Base Flow
Constituent Multiple Regression Forward Selection Backward Selection
F-Ratio R-Squared F-Ratio R-Squared F-Ratio R-Squared
Total Nitrogen 4.21 14.09 8.42 10.44 8.42 10.44
Total Phosphorus 3.08 10.71 4.60 1.55 3.37 8.65
BOD 2.44 8.69 6.16 5.99 5.23 5.13
34
Base Flow (Total Nitrogen)
0.00
1.00
2.00
3.00
4.00
5.00
0 5 10 15
Month
C
onc
e
n
t
r
a
t
i
on (
m
g/
l
)
Observed
Storm Flow (Total Nitrogen)
0.00
2.00
4.00
6.00
8.00
10.00
051015
Month
C
onc
e
n
t
r
a
t
i
on (
m
g/
l
)
Observed
Base Flow (Total Phosphorus)
0.00
0.10
0.20
0.30
0.40
0.50
0 5 10 15
Month
Co
n
cen
t
r
ati
o
n
(m
g
/
l
)
Observed
Storm Flow (Total Phosphorus)
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
0 5 10 15
Month
Co
n
cen
t
r
ati
o
n
(m
g
/
l
)
Observed
Base Flow (BOD)
0
1
2
3
4
5
6
7
8
0 5 10 15
Month
Co
n
cen
t
r
ati
o
n
(m
g
/
l
)
Observed
Storm Flow (BOD)
0
10
20
30
40
50
0 5 10 15
Month
Co
n
cen
t
r
ati
o
n
(m
g
/
l
)
Observed
Figure 15 Observed Concentrations for Base Flow and Storm Flow Condition
Conclusions
The F-ratio for the three constituents analyzed did not meet the specified criteria
of F-ratio greater than four or in the range of 100 to 300. The criteria were not met using
multiple regression and stepwise multiple regression for any of the three constituents.
The F-ratio showed that there is a marginal if any significance in the seasonal varability
of the data set. Therefore, it can be concluded that there is no seasonal variation for the
35
constituent concentrations of total nitrogen, total phosphorus, and BOD in the Austin
Creeks. The monthly load into each of the Austin Creeks can, therefore, be calculated by
multiplying the monthly flow by the mean annual load.
36
BIBLIOGRAPHY
Chow, V.T., Maidment, D.R., and Mays, L.W., 1988, Applied Hydrology, McGraw-Hill,
Inc., New York.
Driscoll, E.D., 1986, Lognormality of Point and Non-Point Source Pollutant
Concentrations, in Urban Runoff Quality ? Impact and Quality, Proceedings of an
Engineering Foundation Conference, Edited by Ben Urbonas and Larry Roesner,
Henniker, New Hampshire, June 23-27, ASCE, New York.
Gilbert, R.O., 1987, Statistical Methods for Environmental Pollution Monitoring, Van
Nostrand Reinhold Company, New York.
Huber, W.C., 1992, ?Contaminant Transport in Surface Water,? in Handbook of
Hydrology, D.R. Maidment editor, McGraw-Hill, Inc., New York.
Shelley, P.E., and Gaboury, D.R., 1986, Urban Runoff Quality, American Society of
Civil Engineers, New York.
U.S. Environmental Protection Agency, 1983, Results of the Nationwide Urban Runoff
Program Final Report, EPA Planning Division, National Technical Information
(NTIS) Service Accession No. PB84-8552.
U.S. Environmental Protection Agency, 1992, Guidance Manual for the Preparation of
Part 2 of the NPDES Permit Applications for Discharges from Municipal
Separate Storm Sewer Systems, Report # 833-B-92-002.
Urbonas, B.R., Guo, C.Y., and Tucker, L.S., 1990, ?Sizing capture volume for
stormwater quality enhancement,? Flood Hazard News, Urban Drainage and
Flood Control District, Denver, CO.
37
38
Appendix A: Impervious Cover Water Quality Relationships
39
SUMMARY OUTPUT for BOD
Regression Statistics
Multiple R 0.787538
R Square 0.620217
Adjusted R Square 0.594898
Standard Error 3.230315
Observations 17
ANOVA
df SS MS F Significance F
Regression 1 255.6163 255.6163 24.49621 0.000175
Residual 15 156.524 10.43493
Total 16 412.1403
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 90.0% Upper 90.0%
Intercept 3.501367 1.838299 1.904678 0.07618 -0.41688 7.419612 0.278735 6.724
X Variable 1 13.85932 2.800222 4.949365 0.000175 7.890785 19.82786 8.950389 18.76825
40
SUMMARY OUTPUT for COD
Regression Statistics
Multiple R 0.786025
R Square 0.617836
Adjusted R Square 0.593951
Standard Error 22.18292
Observations 18
ANOVA
df SS MS F Significance F
Regression 1 12728.6 12728.6 25.86683 0.00011
Residual 16 7873.313 492.0821
Total 17 20601.92
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 90.0% Upper 90.0%
Intercept 18.25402 12.59808 1.448952 0.166667 -8.45271 44.96075 -3.74077 40.24881
X Variable 1 97.71988 19.21371 5.085945 0.00011 56.98864 138.4511 64.17496 131.2648
41
SUMMARY OUTPUT for COPPER
Regression Statistics
Multiple R 0.575169952
R Square 0.330820473
Adjusted R Square 0.283021936
Standard Error 6.662319034
Observations 16
ANOVA
df SS MS F Significance F
Regression 1 307.205241 307.2052 6.921142 0.019755355
Residual 14 621.4109288 44.38649
Total 15 928.6161698
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 90.0% Upper 90.0%
Intercept 6.024214438 3.814984702 1.579093 0.136637 -2.158121243 14.20655012 -0.695153404 12.74358228
X Variable 1 0.156423238 0.059458286 2.630806 0.019755 0.028897784 0.283948692 0.051698809 0.261147667
42
SUMMARY OUTPUT DP
Regression Statistics
Multiple R 0.664875
R Square 0.442059
Adjusted R Square 0.380066
Standard Error 0.086256
Observations 11
ANOVA
df SS MS F Significance F
Regression 1 0.053054 0.053054 7.130741 0.025608
Residual 9 0.066961 0.00744
Total 10 0.120015
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 90.0% Upper 90.0%
Intercept 0.040489 0.061951 0.653563 0.529735 -0.09965 0.180631 -0.07307 0.154052
X Variable 1 0.002371 0.000888 2.670345 0.025608 0.000362 0.00438 0.000743 0.003999
43
SUMMARY OUTPUT NH3
Regression Statistics
Multiple R 0.833675
R Square 0.695015
Adjusted R Square 0.67323
Standard Error 0.048517
Observations 16
ANOVA
df SS MS F Significance F
Regression 1 0.075099 0.075099 31.90385 6.01E-05
Residual 14 0.032955 0.002354
Total 15 0.108054
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 90.0% Upper 90.0%
Intercept 0.127318 0.027782 4.582739 0.000426 0.067731 0.186904 0.078385 0.17625
X Variable 1 0.002446 0.000433 5.648349 6.01E-05 0.001517 0.003374 0.001683 0.003208
44
SUMMARY OUTPUT NO3
Regression Statistics
Multiple R 0.107176
R Square 0.011487
Adjusted R Square -0.0503
Standard Error 0.471023
Observations 18
ANOVA
df SS MS F Significance F
Regression 1 0.041249 0.041249 0.185921 0.672086
Residual 16 3.549807 0.221863
Total 17 3.591056
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 90.0% Upper 90.0%
Intercept 0.927395 0.267503 3.466864 0.003178 0.360315 1.494475 0.460367 1.394424
X Variable 1 -0.00176 0.00408 -0.43119 0.672086 -0.01041 0.00689 -0.00888 0.005364
45
SUMMARY OUTPUT PB
Regression Statistics
Multiple R 0.572696
R Square 0.32798
Adjusted R Square 0.283179
Standard Error 15.8718
Observations 17
ANOVA
df SS MS F Significance F
Regression 1 1844.206 1844.206 7.320773 0.016272
Residual 15 3778.711 251.9141
Total 16 5622.918
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 90.0% Upper 90.0%
Intercept 2.549932 9.074552 0.280998 0.782551 -16.792 21.89189 -13.3582 18.45808
X Variable 1 0.382659 0.141427 2.705693 0.016272 0.081214 0.684105 0.13473 0.630589
46
SUMMARY OUTPUT TKN
Regression Statistics
Multiple R 0.664751
R Square 0.441893
Adjusted R Square 0.407012
Standard Error 0.457484
Observations 18
ANOVA
df SS MS F Significance F
Regression 1 2.651381 2.651381 12.66835 0.002615
Residual 16 3.348667 0.209292
Total 17 6.000048
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 90.0% Upper 90.0%
Intercept 0.685227 0.259813 2.637381 0.017924 0.134447 1.236007 0.231623 1.138831
X Variable 1 0.014104 0.003962 3.559263 0.002615 0.005703 0.022504 0.007186 0.021022
47
SUMMARY OUTPUT TOC
Regression Statistics
Multiple R 0.385264
R Square 0.148428
Adjusted R Square 0.095205
Standard Error 5.934686
Observations 18
ANOVA
df SS MS F Significance F
Regression 1 98.22264 98.22264 2.788791 0.11437
Residual 16 563.528 35.2205
Total 17 661.7507
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 90.0% Upper 90.0%
Intercept 8.013897 3.370415 2.377718 0.030229 0.868938 15.15886 2.129543 13.89825
X Variable 1 0.085842 0.051403 1.669968 0.11437 -0.02313 0.194812 -0.0039 0.175586
48
SUMMARY OUTPUT TP
Regression Statistics
Multiple R 0.393226
R Square 0.154627
Adjusted R Square 0.101791
Standard Error 0.214388
Observations 18
ANOVA
df SS MS F Significance F
Regression 1 0.134511 0.134511 2.926551 0.106448703
Residual 16 0.735397 0.045962
Total 17 0.869909
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 80.0% Upper 80.0%
Intercept 0.194446 0.121755 1.597025 0.129819 -0.063663238 0.452555 0.031689 0.357202
X Variable 1 0.003177 0.001857 1.710717 0.106449 -0.000759831 0.007113 0.000694 0.005659
49
SUMMARY OUTPUT TSS
Regression Statistics
Multiple R 0.276352
R Square 0.07637
Adjusted R Square 0.018643
Standard Error 80.32982
Observations 18
ANOVA
df SS MS F Significance F
Regression 1 8536.891 8536.891 1.322958 0.266964
Residual 16 103246.1 6452.881
Total 17 111783
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 90.0% Upper 90.0%
Intercept 140.9873 45.62075 3.090421 0.007019 44.27565 237.6989 61.33875 220.6358
X Variable 1 0.800281 0.695776 1.150199 0.266964 -0.6747 2.275259 -0.41446 2.015025
50
SUMMARY OUTPUT Zn
Regression Statistics
Multiple R 0.764496
R Square 0.584455
Adjusted R Square 0.556752
Standard Error 0.047628
Observations 17
ANOVA
df SS MS F Significance F
Regression 1 0.047858 0.047858 21.09715 0.000352
Residual 15 0.034027 0.002268
Total 16 0.081885
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 90.0% Upper 90.0%
Intercept -0.00515 0.027231 -0.18923 0.852449 -0.06319 0.052889 -0.05289 0.042585
X Variable 1 0.194933 0.04244 4.593164 0.000352 0.104475 0.285392 0.120534 0.269333
51
52
Appendix B: Results from Seasonal Multiple Regression Analysis
B-1
B-2
B-3
B-4
B-5
B-6
B-7
B-8
B-9
B-10
B-11
B-12
B-13
B-14
B-15
B-16
B-17
B-18
B-19
B-20
B-21
B-22