-------
100
200
300
400
500
600
CA1
C01
DC1
FL1
IL2
KS1
MA1
MD1
MM
NC1
NH1
NY3
TN1
WA1
WI1
(
[R RRR
ICMRR R
I ft
IR
I R
I RR C
) 1
r~
I RR
R R 1
1
1 R
A M
R
MM IV
1 R
r
M
RR|
C
JO 21
R
R C F
1
RMR
M
MC
C
C A
JO 3
n
1
1
C
[
R
1
1
RR
R
/I
JO 4
M
R
C
C
1
H
90 5
H
M
R
ID
T
C
DO 6
791
1 j i CJ288
_Ji )973
~i r~]i24
10
Figure 6-10. Range of NO -N EMC Medians (mg/1) by Project
I
CA1
C01
DC1
FL1
111
K51
MA1
MA2
MD1
Mil
MI3
NCI
NH1
NY1
NY2
NY3
SD1
TN1
TX1
WAI
WI1
) 21
fCRRI
10 400 6
I M |
ftRR]
IB RRRR R
ICMMRR n
rr
Cu
10 8
1
1 M R RR 1
10 10
00 12
00 14
00
CM R R 1 \ 2031
1 M R M RM :? * 1726
1C
R I
1 1
1 I R RM R ? f2825R 4326
CM
IMM
MMM C
M 1 1
1 CR 1
|~RR|
(Ml
[C R R I
1~ M J € H820
CRM R
RR J
RRl
ICC R RMC R 1
f
200
400
600
800
1000
1200 1400
Figure 6-11. Range of Total Cu EMC Medians (pg/1) by Project
6-24
-------
(
CA1
C01
DC1
IL2
KS1
MAI
MA2
Mil
MI3
NY2
SD1
TN1
) 5
0 11
Pb
10 1.
iO 2(
1 M 1
10 2.
iO
[ RM R RCR 1
|R R RR R J/392R 598
314| ^396R 501
1C M C R R}/ 378
1 M R M R |
C R
MMM M C ]
|M MM I
I M |
1 M |
1C R R M |
50
100
150
200
250
Figure 6-12. Range of Total Pb EMC Medians (yg/1) by Project
Zn
CA1
C01
DC1
FL1
IL1
KS1
MAI
MA2
MD1
Mil
MI3
NC1
NH1
NY1
NY 2
NY3
SD1
TNI
TX1
WAI
WI1
1 M 1
1 R RR C R M 1
1 R RRR RR I
1 C MRMR
MR R R 1
IT C R 1
| M R M MR 1
1 C R 1
1
1 C MMM M 1
IM MMI
IR C |
IIC 1
C
RR 1
1 I
1 R RM Rf
CR R 1
IRMC R 1
M 1
1 R R 1
IRRI
ICCRRMC I
9.9
Figure 6-13. Range of Total Zn EMC Medians (yg/1) by Project
6-25
-------
TABLE 6-11. PROJECT CATEGORY SUMMARIZED BY CONSTITUENT
TSS
BOD
COD
Tot. P.
Sol. P.
TKN
NO. -N
2+3
Tot . Cu
Tot. b
Tot. Zn
iH
O
o
3
-
3
1
2
2
2
2
2
2
i-l
U
Q
1
_
1
2
3
1
1
1
1
1
1-1
ij
Pn
1
2
1
1
-
1
1
1
1
1
r-1
»J
H
2
-
3
2
-
2
_
2
2
-
t-H
W
«
3
3
3
3
3
2
_
2
1
3
i-l
<
s
3
-
2
3
2
2
3
3
2
2
i-i
Q
s
1
-
3
3
-
3
3
3
3
3
H
H
S
1
2
1
2
2
1
1
1
1
2
00
H
s
1
1
—
1
1
1
2
-
-
—
n
><
Z
2
-
1
2
-
2
_
-
1
3
1-1
2
EH
3
2
2
2
2
1
1
2
2
2
H
H
S
2
2
2
2
-
1
1
-
*^
t.
2
It must also be realized that had any particular project monitored other
local sites (or additional sites) its categorization could well change. This
can be seen qualitatively by perusing Figures 6-4 through 6-13 and mentally
dropping the highest or lowest site from each grouping. Although some loca-
tions, such as Tampa, will undoubtably and appropriately be influenced by the
relatively low EMCs and tight groupings found there in estimating probable
values for other urban sites in the area, there is little to warrant attrib-
uting similar characteristics to other locations in the same geographical
region. For the other locations it would appear that individual site differ-
ences eclipse any possible geographic ones.
Effect of Land Use Category. The data in Tables 6-1 through 6-10 were pre-
sented by land use category; residential, mixed, commercial, industrial, and
open/non-urban. The question to be addressed here is the extent to which
such site categorization can be used to assist in predicting EMC parameters
for unmonitored sites. Two approaches were used. In the first, the site
data for each project with more than three sites were normalized by dividing
the site median and its upper and lower 90 percent confidence limits by the
average project median value for the constituent in question. This procedure
simply allows all constituents to be viewed on a common scale that is
centered at unity. An example of the result is given in Figure 6-14. A
legend is provided in Figure 6-14(a) showing the lower 90 percent confidence
limit, the upper 90 percent confidence limit, and the location of the point
estimate of the median within this confidence interval for a hypothetical
constituent. Sites that fall to the right of the unity line have higher EMCs
than average for this location, while sites that fall to the left of the
unity line have lower EMCs than average. Thus, the interpretation is that
for this location, Site #1 is the "dirtiest" (has the highest EMC value) ,
Site #3 is the "cleanest", and Site #2 is in between, being somewhat
"dirtier" than average. Since the 90 percent confidence limits for these
three sites no not overlap, we know that this difference is statistically
significant.
6-26
-------
0.5 1.0 1.5 2.0
0.5
SITE » 1
SITE » 2
SITE # 3
1.5
2.0
^•LOW
1 0
L MEDIAN EMC
R 10% CONFIOEK
5 1
•UPPER 10% CO*
,£ LIMIT
IDENCE LIMIT
0 1.5 2.
(a) Significantly Different Sites
0.5
1.0
1.5
2.0
SITE # 1
SITE # 2
SITE tt 3
_^
"''i'/i^i
',,,7A
':'M
'•'7>.
'fa
V
•fa
'///
05
1.5
2.0
R-ASBURV
M-«ORTH AVE
C-VIUA IT
R-BIG DRV C.
H-116IC
R-CHERRV
d-ASBimV
M-IIORTH AVE
C-VlllA IT.
R-BIG DRV C
R-11BIC
R-CHERSY
R-ASBURV
M-IIORTH AVE
C-VILU IT
R-BIG DRV C
R-I1BIC
R-CHERRV
R-ASBURV
M-MRTH AVE
C-VIUA IT.
R-BIG DRY C
R-1I6IC
R-CKtRBV
R-ASBURV
M-IIORTH AVE.
C- VILLA IT
R-BIC DRV C
R- 11 SIC
R-CHERRV
R-ASBURV
M-IIORTH AVE
C-VILLA IT
R-BIG DRV C
R-11BC
R-CHERRV
R-AS8URV
M-IIORTH AVE
C-VILLA IT.
R-BIG DRV C
R-IIIIC
«-CHEl»V
R-ASBURV
M- NORTH AVE
C-VILLA IT
R-BIG DRV C
R-11KC
R-CHERRV
R-ASBURV
M-IIORTH AVE
C-VILLA IT
R-BIG DRV C
R-lltlC
H-CHERRV
1 — *
I
h
h
h
1
h 1
'&
%
[ **f/'
+ %
h *Y;
\ *
1 *•
) — *-
t— -
1—
y/
k
^4
*~*-7<
1-
MM
1 — *-
////
W-
*" — i
i — * —
H
K 1
H
-H
TSS
COD
TOT. P
SOLP
TKN
N02+3-N
TOT. CD
TOT. Pb
TOT. ZN
0.5
1.0 1.5
2.0
(b) Sites with No Significant
Difference
(c) EMC Data from Denver (C01)
Figure 6-14. Range of Normalized EMC Medians at Denver (CO1)
6-27
-------
The actual data for the Denver (C01) project are presented in Figure; 6-14(c).
With the exception of nitrate + nitrite, there is little to no statistically
significant difference among the majority of the sites for each constituent
examined. The lack of consistency among the sites over the various con-
stituents is apparent. One can observe that the Cherry site (residential)
tends to plot at the lowest position for all constituents, suggesting that it
is the "cleanest," the Asbury site (also residential) tends to plot at the
highest position, suggesting that it is the "dirtiest." The Big Dry
Cottonwood site, which is also residential, tends to fall between these two.
Careful examination of other site data does not provide any evidence to
explain this difference in response for sites in the same Land use category
at the same location. Thus, based on the information presented in
Figure 6-14 (c), one is forced to conclude that land use category does not
provide a useful basis for predicting differences in site EMC values, at
least for this project.
When the foregoing type of analysis was applied to the other applicable NURP
projects, the results were the same. As another example, the range of nor-
malized EMC medians at Tampa (FL1) and WASHCOG (DC1) are shown in
Figure 6-15. These are essentially similar to the Denver results just
discussed.
The WASHCOG data presented in Figure 6-15(b) suggest that there is little
consistent difference among residential land use sites at that project. The
data from Champaign/Urbana (ILl) presented in Figure 6-16 suggest just the
opposite. As a part of this project's experimental design, two site pairs
were selected. The sites of each pair were expected to respond in a similar
fashion. That they do and that the responses of the two pairs are different
from each other for most constituents is apparent in Figure 6-16. However,
there is no consistency in the pair responses. For example, the Mattis pair
has significantly higher EMC values for TSS, COD, and Total Pb, while the
John Pair is higher in Total P. The residential land use category for these
sites provides no explanation of these differences in response.
Based upon the foregoing approach, we can conclude that, while there can be
differences in the responses of different sites at a given location, signif-
icant differences do not appear to be widespread, and where they occur, the
site land use category is virtually useless in trying to understand or
predict them.
The second approach to examining the effect of land use category on the EMC
parameters of a site makes use of the observation, discussed earlier, that
geographic location has no discernible effect on site response. Since site
to site variability was shown to be very well represented by the lognormal
distribution, analysis procedures similar to those described previously for
characterizing an individual site were applied. Table 6-12 lists the median
EMCs for all sites within each land use category. The coefficient of varia-
tion quantifies the variability of site characteristics within the land use
category. To the extent that the sites included in this database provide a
"representative" sample of the land use classifications, then the information
summarized by Table 6-12 indicates the effect of land use on urban storm
runoff pollutant discharges.
6-28
-------
0.5 1.0 1.5 2.0
0.5 1.0 1 5
2.0
C-NORMA
M- WILDER
C-NORMA
M- WILDER
M- JESUIT
M -WILDER
H- JESUIT
C-NORMA
R-YOUNG
M- WILDER
M- JESUIT
C-NORMA
R-YOUNG
M- WILDER
M- JESUIT
n- YOUNG
M- WILDER
C-NORMA
v—
t—
1 —
-*
>-'&
Sf
'//
,^/X
, ^
^
i
/
»— *-<
;
i — "< '< —
*
"* — n
H
H
-4
* — *~
i— # —
i — *-
*
(-
i — * —
-* — i
W/s
&//+ -
v///.
S/7/;
'////;
s////
f
5
¥ 1
-t
^
i-
6
- '
%
ty „
^
,
,
,
— * 1
— 1
BOD
COO
TOT. P
TKN
N02 + 3N
TOT CU
TOT Pb
i TOT ZN
0.5 1.0 1.5
2.0
R-WESTLEIGH
R-FAIRIDGE
R-STEDWICK
R-STRATTON
R-OUFIEF
R-LAKERIDGE
R-WESTLEIGH
R-FAIRIOGE
R-STEDWICK
R-STRATTOK
fl-LAKERIDGE
R-WESTLEIGH
R-FAIRIOGE
D-STEDWICK
R-STRATTO«
R-LAKERIDGE
R-WESTLEIGH
R-FAIRIDGE
H-STEDWICK
R-STRATTON
R-DUFIEF
R-LAKERIDGE
R-WESTLEIGH
R-FAIRIDGE
R-STEDWICK
R-STRATTO»
R-DUFIEF
R-LAKERIDGE
R-WESTLEIGH
R-FAIRIDGE
R STEDWICK
fl-STRATTOHI
R-OUFIEF
R-LAKERIOGE
R FAIRIDCE
H-STEDWICK
R-STRATTO«
R-DUFIEF
R-LAKERIDGE
fl-WESTLEIGH
R-FAIRIDGE
R-STEDWICK
R-STRATTOH
R-DUFIEF
R-LAKERIDGE
R-WESTLEIGH
R-FAIRIDGE
R-STEDWICK
R-STRATTON
MI»H
MH
t-
I-
m
y///
//&
h i
f
h >• 4
r— * >~
%
1
|-«H |
H-*-
1— *1
H*
t-
H
* — t
I
^"7,
^•
* — 1
1
H
1 «*•
t -t
-1
^
-\
— i
— i
H
1
^
-*-^^
;•* 1
, — (
i
— * 1
* — (
h
H
-\
*
TSS
COD
TOT P
SOL P
TKN
N02 + 3N
TOT. CU
TOT Pb
H
TOT ZN
0.5 10 15 2.0
(a) Tampa Sites
(b) WASHCOG Sites
Figure 6-15. Range of Normalized EMC Medians at FL1 and DCl
6-29
-------
0.5
1.0
1.5
2.0
R MATTIS S.
M-MATTIS N.
R-JOHN N.
R-JOHN S.
R-MATTIS S.
MM ATTIC U
RlflHM N
R-JOHN S.
R-MATTIS S.
MM ATTIC U
R-JOHN N.
R-JOHN S.
R-MATTIS S.
MMATTIC M
— MAI Ho N.
Riniiu M
— junn n.
R-JOHN S.
R-MATTIS S.
MMATTIC M
— MAI llo N.
R-JOHN N.
R-JOHN S.
R-MATTIS S.
MMATTIC M
R-JOHN N.
R-JOHN S.
1
L
'S///A
5/%£/;/j
V////7/A
Y//
^
1
\7///A
\$y/
vt///\
J 0.5
VS/SA
'%/y\
///
y/
^
v/
V////A
•^^^
w,
V///A
vw^A
//
Y^fflfy
^2^1 '
//yf/////l
'. j, i
fy/////^//,
„
'////A '
'///A
TSS
COD
TOT. P
TKN
TOT. CU
TOT. Pb
1.0 1.5 2.0
Figure 6-16. Range of Normalized EMC Medians at IL1
6-30
-------
w
tl
H
o o
En O
w
w E-1
U <
s o
w
s p
H
a a
rH
I
H
J
«
«!
EH
eQ
c
j}
p
c
0
z
0)
8f
r-H
•rH
u
V-l
0)
o
u
•o
a;
X
•rH
2
rH
(0
•rH
c
CU
•H
tn
-------
Some caution in the interpretation of the information presented in Table 6-12
is in order since statistical confidence limits are not given. These are
indicated in Figure 6-17 (a through k), which illustrates land use differ-
ences graphically, with additional statistical detail derived from the basic
parameters listed in Table 6-11, to assist in interpretation and comparisons.
The box plots which compare characteristics of all sites within a land use
category identify the land use, median EMC, its 90 percent confidence limits,
and the 10, 25, 75 and 90 percent quantities for the sites. Careful perusal
of these box plots leads one to the conclusion that only the open/non-urban
land use category appears to be significantly different overall. Responses
of the other land use categories are varied and inconsistent among con-
stituents. This may be seen in a somewhat different way by observing the
plotting positions of the land use categories presented in Figures 6-4
through 6-13. Here also, there are no consistent tendencies. There are
undeniably some trends. For example, in Figure 6-7 commercial sites occupy
the lowest plotting position at each project for total phosphorus (Mil and
one WI1 site are exceptions), which certainly suggests that there might be a
land use category difference for this constituent.
Review of Figure 6-17(j), however, suggests that while a trend to lower total
phosphorus EMC values is apparent as one goes from residential, to mixed, to
commercial land uses, the statistical significance may not be great. The
actual site median total phosphorus EMC probability density functions for
each land use are presented in Figure 6-18. Here it can be seen that
although three different pdfs can be drawn for residential, mixed, and com-
mercial land use categories, their degree of overlap is so great that there
is little statistical significance to the apparent difference. Since this
was the strongest tendency towards land use effect, we must conclude that
using this approach there is again no truly discernible and consistent effect
of land use on the quality of urban runoff,.
The one exception is the open/non-urban category which, as its name suggests,
includes atypical sites. The data in Table 6-12 and the box plots of
Figure 6-12 suggest that the pdfs for this land use category are quite dif-
ferent from those of the other land use categories, and this is in fact the
case. Figure 6-18 shows it dramatically for total phosphorus.
Thus, regardless of the analytical approach taken, we are forced to conclude
that, if land use category effects are present, they are eclipsed by the
storm to storm variabilities and that, therefore, land use category is of
little general use to aid in predicting urban runoff quality at unmonitored
sites or in explaining site to site differences where monitoring data exist.
Correlation Between EMCs and Runoff Volume. To examine the possible rela-
tionship between the event mean concentration of a particular constituent and
the runoff volume, linear correlation coefficients (r) were calculated. The
null hypothesis that the two variables are linearly unrelated was tested at
both the 90 and 95 percent confidence levels. Since it is possible for
correlation to be either positive or negative, the two-tailed test was used.
Failure to reject the null hypothesis is interpreted as meaning that linear
dependency between the two variables in the population has not been shown.
6-32
-------
LEGEND
90%
VALUE
75%
VALUE
STATISTICAL
OF THE
MEDIAN
90% \
CONFIDENCE^
„
nftWut
GROUP A GROUP B
BOD
RESIDENTIAL
SITES
11
MIXED
SITES
11
COMMERCIAL
SITES
OPEN
SITE
(a)
(b)
500
400
300
200
100
TSS
33
RESIDENTIAL
(C) SITES
19
MIXED
SITES
14
COMMERCIAL
SITES
OPEN
SITES
160
140
120
100
f 80
60
40
20
0
COD
-r
-
T
' 0 i\ /\ V/
- 1 ^ ^r V/
1C
33 16 13 5
RESIDENTIAL MIXED COMMERCIAL OPEN
SITES SITES SITES SITES
(d)
Figure 6-17. Box Plots of Pollutant EMCs for
Different Land Uses
6-33
-------
100
90
80
70
2
zo 60
< i- =
n tf U
Q "*• '"*
LU OC _ 4 7^.
E t a » 50
^ ^ 3.
UJ U. 0
1- "H
"o 40
u
30
20
10
-
-
\
>
.
/
{
P
.
TOTAL
COPPER
^
t—
^
Q
t—
1 ^
LU
o
• - — • \ ] u.
\ / \ / ^
A A
A ^r
23 12 10 2
RESIDENTIAL MIXED COMMERCIAL OPEN
SITES SITES SITES SITES
500
400
300
200
100
TOTAL LEAD
30
RESIDENTIAL
SITES
16
MIXED
SITES
11
COMMERCIAL
SITES
^
OPEN
SITES
(e)
(f)
500 r
400 h
300
aff-1
*£~-
£u
«* 200
100
TOTAL
ZINC
5000
4000
3000
2000
1000
TKN
26
RESIDENTIAL
SITES
12
MIXED
SITES
13
COMMERCIAL
SITES
4
OPEN
SITES
32
RESIDENTIAL
SITES
18
MIXED
SITES
14
COMMERCIAL
SITES
8
OPEN
SITES
(g)
(h)
Figure 6-17. Box Plots of Pollutant EMCs for
Different Land Uses (Cont'd)
6-34
-------
2000
1800
1600
1400
!3J*s '200
:S=*~ 1000
>0
u
800
600
400
200
NITRITE
AND
NITRATE
24
RESIDENTIAL
SITES
17
MIXED
SITES
11
COMMERCIAL
SITES
6
OPEN
SITES
1000
900
800
SITE MEDIAN
CONCENTRATIONS
TOTAL PHOSPHORUS
H9/I
K3 Ui f* U> Ol -J
o o o o o o
a o 0^ ^ o o
100
0
LAND USE
NO SITES
TOTAL PHOSPHORUS
-
h W T
i A X \7
^ 4
34 19 14 8
RESIDENTIAL MIXED COMMERCIAL OPEN
& S
INDUSTRIAL NON URBAN
(i)
(j)
(k)
250
200
150
100
50
SOLUBLE
PHOSPHORUS
16 14 8 6
RESIDENTIAL MIXED COMMERCIAL OPEN
SITES SITES SITES SITES
Figure 6-17. Box Plots of Pollutant EMCs for
Different Land Uses (Cont'd)
6-35
-------
t-Z 190ZE8
o>
3.
O
u
CO
ll)
Q
4J
•H
rQ 01
03 W
O
\4 T3
flj £
03
U ^
W 4-J
CLi 0)
O -M
H C
C M
(0 O
-H M-l
TJ
01 t/1
s c
0)
4-)
•rH
3
CP
6-36
-------
The rejection of the null hypothesis means that there is evidence of a linear
dependency between the two variables in the population, but it does not mean
that a cause-and-effeet relationship has been established.
General guidelines for the use of this test suggest that it be used with
caution for values of n less than ten due to the high uncertainties asso-
ciated with estimates of population variance with small samples. Further-
more, when n is 2 a perfect correlation will result but is meaningless. To
include as many sites as possible in this examination, all constituents for
which n was 5 or greater were included. At the other extreme, when n is very
large, say over 100, correlation coefficients are almost always significant
but can be so weak that they are meaningless. For n = 100 the critical value
of r at the 90 percent confidence level is 0.164, meaning that the correla-
tion explains less than 3 percent of the concentration variability.
A total of 67 sites from 20 of the NURP projects were examined for possible
correlation for nine constituents. Of the 517 linear correlation coeffic-
ients calculated (not all constituents were measured at all sites),
116 (22 percent) were significant at the 95 percent confidence level and
154 (30 percent) were significant at the 90 percent confidence level. Of the
r values that were significant, 83 and 87 percent were negative at the 90 and
95 percent confidence levels respectively. When sites with fewer than
10 events were dropped, the foregoing was essentially unchanged. Greater
detail in terms of the number of significant linear correlation by constit-
uent is provided in Table 6-13. There it can be seen that the greatest
tendency for positive values of r occurs with TSS, followed by soluble
phosphorus. The correlation coefficients for the other 7 constituents all
strongly tend to be negative.
When the results are examined by sites, however, a clearer picture emerges.
Although it can be correctly argued that unless a correlation coefficient is
statistically significant the number is meaningless, it also follows that in
such a case they are as likely to be positive as negative. On the other
hand, if all the correlation coefficients (whether significant or not) have
the same sign, it suggests a tendency for that site. The sign of the corre-
lation coefficient (if greater than 0.1) for each site and constituent
examined is given in Table 6-14. Giving appropriate weight to significant
r values but considering others as well, some 37 of the sites tend to have
negative correlations, 13 tend to be positive, and the remaining 17 tend to
be mixed. Perusal of Table 6-14 reveals that this tendency for sites to have
either positive or negative correlation coefficients is quite strong,
especially for sites with a large number of significant correlations. Sites
where erosion, scour, system lag, and such are present could be expected to
exhibit a tendency towards positive correlations. Sites lacking such effects
could be expected to have negative correlation due to dilution associated
with larger runoff events.
The magnitude of the correlation coefficients is indicated in Table 6-15.
Two points stand out in particular. First, the r values are not very large,
averaging around 0.55. Phis means that the correlation is only able to
explain about 30 percent, of the concentration variability. The few high
values are always associated with very few observations (n<10) for which the
6-37
-------
TABLE 6-13. NUMBER OF SIGNIFICANT LINEAR
CORRELATIONS BY CONSTITUENT
(a) ALL SITES
TOTAL #
POLLUTANT OF SITES
TSS
COD
TOT. P
SOL P
TKN
N02+3-N
TOT. Cu
TOT. Pb
TOT. Zn
TOTAL
PERCENT
67
64
67
34
64
57
49
59
56
517
90% SIGNIFICANT CORRELATION
TOTAL ft
13 (19%)
24 (38%)
20 (30%)
10 (29%)
19 (30%)
17 (30%)
17 (35%)
15 (25%)
19 (34%)
154
30%
#NEG.
4
23
16
6
18
15
15
13
18
128
83%
#POS.
9
1
4
4
1
2
2
2
1
26
17%
95% SIGNIFICANT CORRELATION
TOTAL #
7 (10%)
19 (30%)
15 (22%)
7 (21%)
14 (22%)
13 (23%)
13 (27%)
12 (20%)
16 (29%)
116
22%
#NEG.
3
19
12
4
14
11
12
11
15
101
87%
#POS.
4
0
3
3
0
2
1
1
1
15
13%
(b) SITES WITH n > 10
TSS
COD
TOT. P
SOL. P
TKN
N02+3-N
TOT. Cu
TOT. Pb
TOT. Zn
TOTAL
PERCENT
56
52
53
23
50
41
31
45
37
388
9 (16%)
21 (40%)
17 (32%)
8 (35%)
17 (34%)
14 (34%)
13 (42%)
13 (29%)
14 (38%)
126
32%
4
20
15
5
16
12
12
12
13
109
87%
5
1
2
3
1
2
1
1
1
17
13%
7 (12%)
16 (31%)
12 (23%)
6 (26%)
12 (24%)
12 (29%)
12 (39%)
11 (24%)
11 (30%)
99
26%
3
16
11
4
12
10
11
10
10
87
88%
4
0
1
2
0
2
1
1
1
12
12%
6-38
-------
££ 190ZE8
to
W
EH
EH
5S
o
H
w
o
o
§
EH
^
W
O
u
b-i
O
O
H
rH
I
W
iJ
9
NZ '101
qd 101
113 '101
N/E+ZON
NMi
d 'IDS
d 101
003
SSI
NZ 101
qd 101
no loi
N-E+Z ON
NMl
d 10S
d 101
003
CC 1
obi
NZ 101
qd loi
f!3 101
N.E+ZON
NMl
d 'IDS
d 101
003
SSI
©
+
ee
g
u
z
©
1
1
©
©
+
z
UJ
_J
^
2
1
1
©
©
-t-
1
+
X
o
X
u
1
+ 1
©
1
+
+
+
+
CEDAR
CRANSTON
CM n
z z
1 1
1
1 +
1
' ©
1 1
' ©
5 0
S 5
E i
+ +
1 4
- ©
1
© -
+
©
4- 4-
©©
Sif =
» m
. CO
CO 4
u
© '
©
©
1
© +
© 1
1 4-
E. ROCH.
SOUTHGATE
© -
i i
;
i
+
+ 4-
+
z
1 z
=. z
o «t
*-
I
1
1 +
1 1
©
' ©
1 4-
1
u
OC K
O LU
. Z
m u
©©
©©
©©
1 1
© +
©©
© +
1 +
©©
a
to —
o oc
z
4. ©
+ 1
1 ©
1 1
+ +
+-
+
_• z
LU a
z oc
o o
o ->
©0
© '
© '
© i
© '
© '
© i
©0
1
UJ UJ
z g
. 0
z oe
1
+
1 © -
© • +
' ©
1 1
' ©
+ 4- 4-
H-
CC
CNJ CJ «S
OC CO X
X
' + I +
+ +
0 © I
+ 1 +
+ + 1
(+) 1
+
(+) + \+
X
CO CO
35
5 i- a
_j oe <
CM
i
0 - '
©
© -
© i '
© -
© '
©0 -
© ' ©
i + i
si ^ S
> — a
u
a
©
©
+
©0)
1
- ©
1 1
a _:
0 Z
1 2
5
i
+
©
i i
+ i
©
4- 4-
1
W -jv'
O z
2 =
LU O
z m
»—
o
S
- ©
1
©
+
+ ©
+ +
+
+
UJ
g 2
© - '
© ©
-•-
© ^
©0 '
© ' ©
i
SURREY D.
BURBANK
HASTINGS
i
1 4- 1
4- 1
0 i +
+ i
i i
' 1 ; 4- 1
1 + 1
-'
a i
z en
a. LU S
E S •
z i E
0 1 ©
1 I
+ I ©
© ©
4- 1
+
' ' ©
+
STEDWICK
STRATTON
WESTLEIGH
1
- ©
' ©
©
' ©
' ©
©
5 ='
i i
^ a.
1 1
1 1
© ©
1
© 1
1
© '
i u)
CO «
LU QC
OC C9
E
©
©
0 +
i i
©
© i
© i
i
CHARTERIH
YOUNG
s
1 4- 4-
1
©© 1
© ' '
1 ©
©©
©©
OC L^ ej
LU
_1 UJ _
i- i- H
2> < 0
K CO S
© © 1
1 1
1 1
1
1
+ +
1 4-
1 1
Z
LU O Q
11°
O C3 £
1
1
1 1
+
© 1
0) 0
4- © ©
4- 4- 4
O.
< Z to
cc z z
i S ?
=
I ©
©
I © +
1 © 1
I ©
1
1 © +
1 ©
C±)
_l
CC
UJ
> C*» C*!
s i i
0
z
© ©
© CD
(u© -
+
(0 0) i
CO 0; +
i i
X CO
1 t -g
CO
©
©
©
©
©
1
©
1
CP
_
z
z
o
a
= ==
= < = t
CO C3 Z Jj
O UJ C3
CO CO OT 2
LU LU r.' —
I— (— |— O
o u 2 ~
= => 5 55
iQi
6-39
-------
K190ZE8
W
ffl
w
w
3
H
M
u
t,
W
o
u
2
O
H
w
OS
K
O
U
in
rH
i
NZ '101
qd 'ioi
f)3 MOI
N-E + ZQN
N»l
d IDS
d "101
003
SSI
NZ '101
qd 'ioi
ns 101
NE + Z ON
NM1
d IDS
d 101
003
SSI
NZ 101
qd -101
no 'loi
N-E+zON
NXl
d 10S
d '101
003
SSI
© ©©
© ©©
=====©©
©©= = = ©
° © ©
= = = = S © ©
f*\ ^A
iZJ (S)
33 1 r- j
'©
UJ
0 «
cc |_ 3; cs
_, CC CO CJ Z
— i «t Z O t-
cc a < oc 3 a
4UJOC ,oca«— CNCJ
CJCJOLUCOCJCCOCCO
•— CM ro •—
(5~) (?)
S
a
00
= = = = © S £
3 =
©g ©
(15) g (?)
00
slfiiss
rfC->^cc«>a = m
9»-_iLuZZCCu
ujiizC3>«sooo^-
ZiSzocto-aScc
2 i
©©©
© ©
© ©
© © ©
S S © ©
V_^ x X
= ©@ ©
®y — N S — N /" — X
© ©@
\ / v_y v^/
©©©
©03
u?
. ^ U ^
9 CC >• & UJ UJ ~
X£ = O:Q:=>Z>-
tf o
!_ 0
©©
©© ©
= = = = = =
©00
3 333 = 33
© S
S ©0©
= = ° == i z
o z uj «t S ^
1— O LU CC CD 1— °
cc » a- ec cc co cj
< ?
©
S
©/KT\ /S\
(5)©
UJ u z ^ Z
LU " cj o 2 £
a a — L_ LU u
"- S DC S 1— — ' I—
a 1 LU S « [T cc
"- £ x LU cc co <
3 3 3 H™ H~ Z
O u. _J CO CO 3 CJ
CJ Ij
a u.
C7
3 = = =
©«©
© s
= = = =,
5 ©
©@©
©©©
CC uJ tj
° iS LU a
t— t— 1— 0
CO CO < 2
o = i- S
a. oc « S
©= © ©
© ©
S © ©
© ©
© ©
3 = 3
©
®/-s\
(5)
©
• • >•
LU LU O § DC
CC CC CC z S ° => 3=
i i i
= = = =
®@
©©
33333
s fS)(s)
=3 = 33333
i « ) ( 3 ) ( S ) ( S 1
•- CO /5\ / n ( to 1 ( to i
aJ z to
0 < z co g co
z S z z t t -B
3 0= Z Z t t =
O(3 ?^ m *• ** »5
*—* O O — — rsJ
>- Z ^ -» S S 01
^ CO
UJ
LU
en
LU
>—
H-
LJJ
CC
CC
CO
cc
LU
5 _ i
""13
^ £ =
LU *" O
Z o h-
C9 U Z
« <
" S y
a OC u.
ill
^' LU CO
—.So
LTl CO
CTJ Z LU
c/i ** t;
ri ,« *
~ CO JJ
^1 w S
5 « ~
z a »
A- <
o=«
6-40
-------
test is suspect since one or two events may dominate the correlation or
otherwise cause it to be overstated due to uncertainties in parameter esti-
mation. Second, only 25 percent of the sites account for over two-thirds of
the significant correlations. In fact, 33 of the 67 sites had at most one
significant correlation, 16 had 2 or 3, and 18 had 4 or more significant
r values.
Data for the sites with many significant correlations are presented in
Table 6-16. It can be noted that the r values for all constituents are
around 0.55. Thus, there is no overall tendency to have strong correlations
for some constituents and weak correlations for others. On a site by site
basis, the strength of the apparent correlation varies inversely with n as
does the significance requirement. Discounting the sites with very low or
high values of n, however, the r values for the remainder are again around
0.55, which is the average for all 19 of these sites. Turning to land use,
it is significant that half of the sites with many significant correlations
have a large commercial/industrial component. Discounting sites with a small
number of observations (n _^ 12), the sites in Table 6-16 are smaller (average
size is 41 acres vs 126 acres for all sites) , more impervious (average of
65 percent vs 40 percent for all sites) , and have higher runoff coef-
ficients (0.5 vs 0.3 for all sites). Thus, one could conjecture that their
responses might tend to be somewhat less random and more ameanable to deter-
ministic analysis (i.e., with conventional modeling approaches). Since they
represent only around 25 percent of the total number of sites, however, and
the correlations are rather weak, any effect of EMC correlation with runoff
volume can be ignored without serious overall error.
This finding of no significant linear correlation between EMCs and runoff
volumes is important for several reasons. First, in stormwater monitoring
programs there is a natural and appropriate bias that favors emphasizing
resource allocation to larger storm events. This was generally the case with
the NURP projects as well. However, because of differences in local meteor-
ological conditions, degree of site imperviousness, and other factors, there
are appreciable differences in the average sizes of storms monitored by site
in the NURP database. Since no significant linear correlation was found,
such biases and differences are not expected to influence EMC comparisons to
any appreciable extent.
Secondly, the probabilistic methodologies for examining receiving water
impacts identified in Chapter 5 assume, as they are now structured, that con-
centration and runoff volume are independent (i.e., that there is no signif-
icant correlation). Although the methods can be modified to account for such
correlations if they exist, the finding of no significant correlation indi-
cates that such refinement is not warranted at this time.
Other Factors. We have not exhaustively analyzed all potential effects of
other factors that might influence and hence modify our interpretations and
conclusions regarding site differences. Factors such as slope, population
density, soil type, seasonal bias in monitored events, and precipitation
characteristics (average rainfall intensity, peak rainfall intensity,
rainfall duration, time since last storm event, etc.) all have a potential
6-41
-------
9E l90cE8
1N3I3IJJ303
JdONriH
^.nniAH^iAii
ON Ul ' iQ J tJ VV 1
%
3sn
QNtfl
u
J DAtf
rj 9AV
UZ
qd
"3
N-E+ZON
mi
d IDS
d 101
003
SSI
oo
OS!
=p
CO
LO
CO
CD
00
CNI
oo
CNI
LO
CO
CNJ
CD
«=!•
1
CNI
1
1 —
1
CNI
LO
CNI
r-
1
CNI
1
f —
1
CO
LO
1
|
1 —
CC
0
•^
CD
CO
CNI
COO
-J=
, —
cn
C_J
o
CD
p^
CNI
cn
LO
LO
00
LO
CD
1
LO
LO
1
CD
I
^
^-
03
CD
!
r--
co
oo
LO
1
CO
I —
1
1—
f^
zi
>
CO
3=
, —
CM
CC
oo
cn
LO
oo
LO
CO
CNJ
^.
^3~
i
|
CO
)
m
CO
I
1
)
ON)
|
|
1 1 1
LLJ
^
CO
a
00
,0
CD
CC
^
CO
CO
'•SJ
'
CO
CD
r —
oo
CD
LO
1
^
CD
1
LO
!
oo
CD
1
_,
^
1
CO
1
1
LU
CC
^
CO
—
u-
O3
00
CD
,j=
00
LO
CO
CD
LO
CNI
LO
1
CNI
_,
CD
1
CD
1
_!
CO
1
ID
cn
LO
1
CD
•z.
OO
•^
,_
—
C3 CD
ro Ln
c? £?
r*~~- *zf
OO ^±
cc ~
1 s
OO CD
00 —
LO CO
CD CD
Osf <3-
=3 |
CO 1
•=}• '
1
•=t 0
CO CO
1 1
ZD ^3
s
1
=3 ^3
LO ' —
LO LO
f l'
— CO
CD r-^
1 1
| ]
CO
co -a:
i 5
CO "
cn co —
CO CO CO
CNI =>
•— 1 CD
oo ' ^r
1
1 cn co
r^ LO
1 1
CO | |
00 '
CD
— J CO
H- — CO
CO DT —
CO CO •—
O LLJ CO
— 1 CC <—
5 g E
2 "
oo
LO
CD
_j=
( — i
CO
CJ
o
o
oo
oo
s
CD
CNJ
CO
LO
J
1
o
LO
1
CO
1
CO
1
_,
cc
1
1
o
o_
£
"^
LO
^=
CO
00
cc
CD
o
CO
CO
r —
LO
Osl
1
1
Osl
1
=
_,
CO
1
=>
^
CO
1
cn
1 — ,
1
1
cc
1 —
CO
LiJ
0
cc
LU
"
"^
CO CNJ
CO 00
CM CD
^ i°
cn oo
CO 00
CD CC
CD S-
CD CO
LO ' —
LO r*-
LO r~~
0 r~
ro LO
LO Osl
CD CO
1
•— LO
LO CO
r— CNI
LO P^
1
1 1
CD 1
LO '
!
1 — - Osl
«f CO
1 1
CNI 1
CD
1
1 i
CO CNI
«* oo
1
cn
CO r-
co cc
•^
1 —
00
cn
J=
r^
CO
DC
cn
CJD
CNI
cn
CO
J
r^.
n
I
CD
CNJ
=
_,
oo
1
_,
!
00
oo
1
1
1 1 1
<•
— 1
cf
s -
^
.=
cn
CNI
cc
C3
o
CO
CD
CNI
^
CD
oo
Osl
1
oo
1
=
_J
CNI
1
— ,
CO
CD
1
CO
1
1
a
>-
LU
cc
cc
CO
coo
CO
CO
-^
LO
CO
0
CO
a
cn
^
00
^.
00
Osl
^
oo
LO
1
CD
1
— ,
^
CNI
r
CO
CNI
1
CO
00
1
CD
a_
*
^
CO CNI
CT) CNI
r-^ CD
^ &
LO r~-
CTj f--
C_J C_J
CD ^
CD in
CNJ £NV
CO LT)
«d- tn
CO CD
^— CO
1 1
! (
~) Z3
r-- Csj
1 1
as 1
(TJ
1
~3 :D
r^
1
LO 00
LIT <3-
[ 1
r- r^
OT «T
1 1
cc
1 1 ' 1
H- 1—
CO >
<=r <=c
LO
&
O
I— I
EH
<
^
H
ftj
&
O
U
EH
O
H
CO
LO
w
EH
H
LO
w
J
m
6-42
-------
influence on the median and variability of pollutant concentrations at a
site.
On the basis of limited screening, however, we have concluded that such
factors do not appear to have any real consistent significance in explaining
observed similarities or differences among individual sites. Therefore,
although more detailed and rigorous analysis and evaluation of the NURP data-
base may well provide additional useful insight and understanding of the
influence of such other factors, we do not believe that the basic findings
and conclusions presented in this report will be significantly altered by the
results of such efforts. Furthermore, the value of any such insights as may
be developed are likely to have limited influence on general decisions on
control of urban runoff. For example, the finding of a strong seasonal
effect on EMC values would have little influence on a decision to require
detention basins in all newly developing urban areas, nor would it be likely
to influence their design.
Urban Runoff Characteristics
Having determined, as discussed in the preceding section, that geographic
location, land use category, or other factors appear to be of little utility
in explaining overall site-to-site variability or predicting the character-
istics of unmonitored sites, the best general characterization of urban
runoff can be obtained by pooling the site data for all sites (other than the
open/non-urban ones). This approach is appropriate, given the need for a
nationwide assessment and the general planning thrust of this report.
Recognizing that there tend to be exceptions to any generalization, however
realistic and appropriate, in the absence of better information the data
given in Table 6-17 are recommended for planning level purposes as the best
description of the characteristics of urban runoff.
TABLE 6-17. WATER QUALITY CHARACTERISTICS OF URBAN RUNOFF
Constituent
TSS (mg/1)
BOD (mg/1)
COD (mg/1)
Tot. P (mg/1)
Sol. P (mg/1)
TKN (mg/1)
Tot. Cu (yg/1)
Tot. Pb (yg/1)
Tot. Zn (pg/1)
Event to Event
Variability
in EMC's
(Coef Var)
1-2
0.5-1.0
0.5-1.0
0.5-1.0
0.5-1.0
0.5-1.0
0.5-1.0
0.5-1.0
0.5-1.0
0.5-1.0
Site Median EMC
For
Median
Urban Site
100
9
65
0.33
0.12
1.50
0.68
34
144
160
For
90th Percentile
Urban Site
300
15
140
0.70
0.21
3.30
1.75
93
350
500
-------
Coliform Bacteria
Coliform bacteria counts in urban runoff were monitored for a significant
number of storm events by seven of the NURP projects at 17 different sites.
Data were collected at twelve of these sites for more than five and up to
20 storm events. Data on either Fecal CoLiform or both Fecal and Total
Coliform counts are available for a total of 156 separate storm events.
Although the data base for bacteria is thus considerably more restricted than
for other pollutants, useful results have been obtained.
Table 6-18 summarizes the results of an analysis of these data. Some vari-
ability exists from site to site, and data are too limited to identify any
land use distinctions. However, results from the different sites and proj-
ects are consistent in showing a very dramatic seasonal effect. Coliform
counts in urban runoff during the warmer periods of the year are approxi-
mately 20 times greater than those in urban runoff that occurs during colder
periods.
The substantial seasonal differences which are observed do not correspond
with comparable variations in urban activities. This suggests that seasonal
temperature effects and sources of coliform unrelated to those traditionally
associated with human health risk may be significant.
In addition to the summarized data presented here, special study reports pre-
pared by the Long Island and Baltimore projects address the issue of animal
and other sources of coliform bacteria using information derived from field
monitoring and the technical literature. The Baltimore NURP project also
conducted small scale site studies which simulated washoff by storms and
identified that quite substantial differences in coliform levels can result
from the general cleanliness of an area, which they associate with the
socio-economic strata of the neighborhood. A special study by the
Long Island NURP project examined salmonella counts in urban runoff and in an
adjacent shellfish area influenced by uroan runoff. The Kncxville, TN
project also conducced a special study on Salmonella. These project reports
may be obtained through NTIS.
Other issues related to bacteria as a health risk were raised and warrant
further investigation. A better understanding is needed of the contribution
of domestic animals or such wildlife as may be expected in urban areas to
observed coliform levels.
Though high levels of indicator microorganisms were found in urban runoff,
the analysis as well as current literature suggests that indicators such as
fecal coliform may not be useful in identifying health risks from urban
runoff pollutions.
PRIORITY POLLUTANTS
'ackground
he NURP priority pollutant monitoring project was conducted to evaluate the
resence, concentration, and potential water quality impacts of priority pol-
utants in urban runoff. A total of 121 urban runoff samples were collected
6-44
-------
TABLE 6-18. FECAL COLIFORM CONCENTRATIONS IN URBAN RUNOFF
Project
and
Site
DC1 Burke
Westleigh
Stedwick
MD1 Homeland
Mt Wash
Res Hill
NCI (CBD) 1013
Res 1023
NH1 Pkg Lot
NY1 Carll
Unqua
SD1 Meade
TNI CBD
Rl
R2
SC
All Sites*
Warm Weather
Site
No.
Obs
1
1
2
7
1
1
11
2
20
12
7
9
7
6
6
7
76
Events
11
Median
EMC
(1000/
100 ml)
4.6
46
10
11
130
281
15
23
0.3
24
11
57
54
56
19
12
21
C.V.
_
-
—
1.8
-
-
1.6
-
0.5
0.9
1.6
0.7
1.5
2.0
6.2
2.8
0.8
Cold Weather
Site
No.
Obs
1
2
1
_
1
1
8
4
-
15
4
-
7
4
4
4
52
Events
9
Median
EMC
(WOO/
100 ml)
0.02
0.35
0.2
_
3.3
330
1.0
2.6
-
1.4
0.9
-
1.0
1.6
0.5
0.9
1
C.V.
_
-
-
_
-
-
0.6
1.1
-
1.5
14
-
1.4
1.9
2.4
1.7
0.7
Notes:
* For general characterization of urban runoff, exclude the
following sites:
NH1 - A small (0.9A) Parking Lot; concentrations low and
atypical.
Four sites with only one observation for season;
variability is too high for any confidence in representa-
tiveness of a single value.
6-45
-------
at 61 sites (two storm events per site) in 20 of the NURP projects that par-
ticipated in this phase of the program. These sites were predominantly in
the residential, mixed, or commercial land use areas as defined earlier.
Thus, the results of this effort cannot be attributed to runoff from indus-
trial facilities or complexes. Furthermore, an especially exhaustive quality
control component, over and above the standard NURP QA/QC effort, was imposed
on the priority pollutant portion of the program, resulting in the rejection
of nearly 14 percent of the data. Therefore, there is a hLgh level of con-
fidence in the results of this project.
Since only two samples were collected at each site, no meaningful site sta-
tistic could be calculated. Therefore the data were pooled for analysis. In
view of the discussion in the preceding section, however, this approach seems
to be justified.
A detailed compilation of NURP priority pollutant analytical results in-
cluding city and site where the sample was collected, date of collection;
discrete or composite sample, pH, and pollutant concentration can be found in
the final report on the NURP Priority Pollutant Monitoring Program soon to be
issued by the Monitoring and Data Support Division of the agency. A summary
of the findings taken from the December 5, 1983 draft of that report follows.
Pollutants Not Included in NURP. Asbestos and dioxin were excluded from the
NURP program. However, standard laboratory methods will reveal the presence
of dioxin at concentrations of 1 to 10 yg/1, and most laboratories did scan
their chromatograms for the possible presence of this pollutant. All such
scans were negative, and on this basis dioxin is included as "not detected1".
Results Not Valid. The NURP results for seven priority pollutants cannot be
considered valid. Recent EPA investigation has revealed that standard
methods are not appropriate for the measurement of hexachlorocyclopentadiene,
dimethyl nitrosamine, diphenyl nitrosamine, benzidine, and 1,2-diphenylhy-
drazine. Two othe_- pollutants, acrolein and acrylonitrile, must be analyzed
within three days of sample collection. Such a time constraint was an
impractical one for the NURP program.
Pollutants Detected in Runoff
Seventy-seven priority pollutants were detected in the NURP urban runoff
samples. This group includes 14 inorganic and 63 organic pollutants
(Table 6-19).
Inorganic Pollutants. As a group, the toxic metals are by far the most prev-
alent priority pollutant constituents of urban runoff. All 14 inorganics
(13 metals, plus cyanides; asbestos excluded) were detected, and all but
three at frequencies of detection greater than 10 percent. Most often
detected among the metals were copper, lead, and zinc, all of which were
found in at least 91 percent of the samples. Their concentrations were also
among the highest for any pollutant, and reached a maximum of 100, 460, and
2,400 yg/1, respectively. Other frequently detected inorganics included
irsenic, chromium, cadmium, nickel, and cyanide (Table 6-20). Twelve of the
:hirteen toxic metals (antimony excluded) were also sampled in the special
6-46
-------
TABLE 6-19. SUMMARY OF ANALYTICAL CHEMISTRY FINDINGS FROM
NURP PRIORITY POLLUTANT SAMPLES
(Includes information received through September 30, 1983)
Pollutant
I. PESTICIDES
1. Acrolein
2. Aldnn
3. o-Hexachlorocyclohexane (a-BHC)
(Alpha)
4. B-Hexachlorocyclohexane (6-BHC)
(Beta)
5. Y-HexachlorocyClohexane (Y-BHC)
(Gamma) (Lindane)
6. 6-Hexachlorocvclohexane (6-BHC)
(Delta)
7. Chlordane
8. ODD
9. DDE
10. DDT
11. Dieldrin
12. o-Endosulfan (Alpha)
13. B-Endosulfan (Beta)
14. Endosulfan sulfate
15. Endrin
16. Endrin aldehyde
17. Heptachlor
18. Heptachlor epoxide
19. Isophorone
?0. TCDD (2,3,7,8-tetrachlorodibenzo-
p-dioxi n)
21. Toxaphene
11. METALS AND INORGANICS
22. Antimony
23. Arsenic
24. Asbestos
25. Beryllium
26. Cadmium
27 . Chromium
28. Copper
29. Cyanides
30. Lead
31. Mercury
32. Nickel
33. Selenium
34. Sliver
35. Thallium
36. Zinc
HI. PCBs AND RELATED COMPOUNDS
37. PCB-1016 (Aroclor 1016)
38. PCB-1221 (Aroclor 1221)
39. PCB-1232 (Aroclor 1232)
40. PCB-1242 (Aroclor 1242)
41. PCB-1248 (Aroclor 1248)
42. PCB-1254 (Aroclor 1254)
43. PCB-1260 (Aroclor 1260)
44. 2-Chloronaphthalene
Cities Where Detected2
Holding times exceeded
4,7,26
7,8,22,26
7,8
7,8,22,26
7,26
2,8,21,26
Not detected
26
7
26,27
7,26,27
Not detected
Not detected
Not detected
Not detected
7,8,27
7,26
7
Not included in NURP program
Not detected
7,24,26
2,3,7,12,19,20,21,22,26,27
Not included in NURP program
7,12,20,21
1,2,3,7,12,20,21,27
1,2,7,8,12,17,19,20,21,22,26,
27,28
1,2, 3, 4, 7 ,8, 12, 17, 19, 20, 21, 22,
23,26,27,28
4,8,19,22,26,27
1,2,3,4,7,8,12,17,19,20,21 ,22,
26,28
7,20,28
2,3,7,12,20,21,26,27
7,19,23
3,17,27
7
1,2,3,7,12,17,19,20,21,22,
23,27,28
Not detected
Not detected
Not detected
Not detected
Not detected
Not detected
2
Not detected
Frequency of
Detection3
6
20
5
15
6
17
6
1
6
19
6
2
3
13
52
12
48
58
91
23
94
9
43
11
7
6
94
1
Range of Detected
Concentrations (ug/z)1*
0.002T-0.1M
0. 0027-0. 1M
0. 018-0. 1M
0. 007-0. 1M
0.004-0.1M
0.01L-10
0.007-0.027
0.1M
0.007-0.1
0.008-0.2
0.01-0.1M
0.003T-0.1M
10M
2.6-23A
1-50.5
1-49
0.1M-14
1-190
1L-100
2-300
6-460
0.6-1.?
1-182
2-77
0.2M-0.8
1-14
10-2400
0.03
6-47
-------
TABLE 6-19. SUMMARY OF ANALYTICAL CHEMISTRY FINDINGS FROM
NURP PRIORITY POLLUTANT SAMPLES1 (Cont'd)
(Includes information received through September 30, 1983)
Pol lutdnt
IV. hlALOGE^,ATEn ALIPHATICS
45. Methane, bronio- (methyl bromide)
46. Methane, chloro- (methyl chloride)
47. Methane, dichloro- {methvlene
chloride)
48. Methane, chl orodibromo-
49. Methane, dichlorobromo-
50. Methane, tribromo - (bromoform)
51 Methane, trichloro- (chluroform)
52. Hcthane, tetrachloro- (rarbon
tetrachlonde)
53. Methane, truhlorof luoro-1'
54 Methane, di chl orodi f 1 uoro-
( F reon-12) "^
55. Ethatie, chloro-
56. Ethane, 1 , 1-dichloro-
57. Ethane, 1 ,2-dichloro-
58. Ethane, 1 ,1 , 1-trichloro-
59. Ethane, 1 , 1 ,2-tr ichloro-
60 Ethane, 1 ,1 ,2 ,2-tetrachloro-
61. Ethane, hexachloro-
62. Ethene, chloro- (vinyl chloride)
63 Ethene, 1 ,1-dichloro-
b4 . Ethene, 1 ,2-trans-dichloro-
65. Ethene, trichloro-
66 Ethene, tetrachloro-
67. Propane, 1 ,2-dichloro-
68. Propene, 1 ,3-dichloro-
69. butadiene, hexachloro-
70. Cycl opentadi ene , hexachloro-
Cities Where Detected-^
hot detected
Not detected
4,17,??
28
28
28
4,17,20,22,23,27,28
4,28
2,4,24,28
Not detected
Not detected
4,28
28
4,2,7,22,24
28
4
Not detected
Not detected
28
20,28
2,4,8,24,28
8,17,22,28
28
28
Not detected
Standard methods inappropriate
V. ETHERS
7i. Ether, bi s(cMoromethyl ) 'J hot detected
72. Ether, bi s(2-chloroethyl ) Not detected
73. Ether, bi s(2-chloroisopropyl ) Not detected
74. Ether, 2-chloroethy 1 vinyl [lot detected
75. Ether, 4-bron'nphenyl phenyl Mot detected
76. Ether, 4-chl orophenyl phenyl Not detected
77. B's(^-chloroethoxy) methdne Not detected
VI. HOtiOCYCLIC ARMOMATICS (EXCLUOIIIG PHENOIS, CRESOLS, PHThALATES)
IK. benzene
79 Benzene, chloro-
80 Benzene, 1 ,2-dichloro-
83. Benzene, 1 ,3-dichloro-
82. Benzene, 1 ,4-dichloro-
83. Benzene, 1 ,2 ,4-trichloro-
84. Benzene, hexdChloro-
85. Benzene, ethyl -
86. Benzene, mtro-
87. Toluene
88. Toluene, 2,4-diriitrc-
89. Toluene, 2,6-dinitro
4,17,27
7,20,26,28
Not detected
Not detected
Not detected
Not detected
Not detected
4,8,17,20,26,28
[Jot detected
4,17
Not detected
Not detected
Frequency of
OetectionJ
11
1
1
1
9
3
5
3
1
(,
2
z
7
4
6
5
1
2
5
5
f
3
Range of Detected
Concent rat ]ofis (i,fj/. }"
5-14. 5A
2
2
1
0.2T-12L
1-2
0.6T-27
1.5A-3
4
1.6-1011
2-3
^0-3
I 5-4
1-:;
0.3T-12
1M-43
3
1-2
1-13
1G-10M
1 -2
3-1
6-48
-------
TABLE 6-19. SUMMARY OF ANALYTICAL CHEMISTRY FINDINGS FROM
NURP PRIORITY POLLUTANT SAMPLES1 (Cont'd)
(Includes information received through September 30, 1983)
Pol lutant
VII PHENOLS AND CRESOLS
90. Pnenol
91 Phenol , 2-chloro-
9?. Phenol, 2 ,4-dl chloro-
93 Phenol, 2 ,4 ,6- tri chl oro-
94. Phenol, pentachloro-
95 Phenol , 2-m tro-
96. Phenol , 4-m tro-
9 /' Phenol , 2,4-dimtro-
98. Phenol , 2 ,4-dimethyl -
99. m-Ciesol, p-cMoro-
100 o-Cresol 4,6-dimtro-
1 11 I . PHTHALATE ESTERS
101. Phthalate, dimethyl
102. Phthalate, diothyl
103 Phthalate, di-n-butyl
104 Phthalate, di-n-octyl
105. Phthdlate, bi s(2-ethyl hexy! )
106. Phthdlate, butyl benzyl
l>. POKCYCLlC AROMATIC HYDROCARBONS
iO/. Ac enaphthene
108. Acenaphthylene
109. Anthracene
110. Bpnzo (a) anthracene
111. Benzo (b) fluoranthene
112. Benzo (k) tluoranthene
in Benzo (g.h.i) perylene
114. Benzo (a) pyrene
115. Chrysene
116. Dibenzo (a,h) anthracene
117. Fluoranthene
118. Fluorene
119. Inaeno (l,2,j-c,d) pyrene
120. Naphthalene
121. Phonanthrene
122. Pyrene
Cities Where Detected'
4,7,26
28
'lot detected
Not detected
4, 8, 19, 20, 26, 27, ?8
p
4,7,8,20,26,28
Not detected
4,7,8,26
4
Not detected
8
3,4,17,20,21
4,22,?4
8,20,26,27,28
4,12,19,22,21,26
2,8,26
Mot detected
Mot detected
2,17,20,21,26,28
2,21,27
26,27
2 ,21,27
2 1
2,21,26,27
2,7,17,21 ,26,27
21
2,8,12,17,21,26,27,28
28
21
4,24,26,28
2,8,17,20,21,26,27,28
2,3,8,12,17,21,26,27,28
Frequency of
Detection^
14
1
19
1
10
8
1
1
6
6
6
22
6
7
4
5
2
1
6
10
16
1
1
9
12
15
Range of Detected
Concentrations (pg/1. )""
1L-13T
2
1T-11B
1M
1T-37
1T-10M
1.5A
1L
1-10M
0.5T-11
0.4T-2C
4T-62
1-10N
1-10C
1-10M
1-S
4-14
5
i-ioc,
0 6~-10H
:1
G.3T-21
1
4
0 8T-2.3
(1.3T-10M
0 3T-16
-------
TABLE 6-19. SUMMARY OF ANALYTICAL CHEMISTRY FINDINGS FROM
NURP PRIORITY POLLUTANT SAMPLES1 (Cont'd)
(Includes information received through September 30, 1983)
Pollutant
X. NITROSAMINES AND OTHER NITROGEN-CONTAININC
123. NUrosamine, dimethyl (OMN)
124. Ni trosamine , diphenyl
125. Nitrosamine, di-n-propyl
126. Benzidine
127. Benzidine, 3,3'-dichloro-
128. Hydrazine, 1 ,2-diphenyl -
129. Acrylomtnle
Cities Where Detected2
COMPOUNDS
Standard methods inappropriate
Standard methods inappropriate
Not detected
Standard methods inappropriate
Not detected
Standard methods inapproonate
Holding times exceeded
Frequency of
Detection3
Range of Detected
Concentrations (ug/l)1*
1 Based on 121 sample results received as of 9/30/83, adjusted for qua'ity control review.
2 Cities from which data are available:
1. Durham, NH 20. Little Rock, AR
2. Lake Quinsigamond, MA 21. Kansas City, KS
3. Mystic River, MA 22. Denver, CO
4. Long Island, NY 23. Salt Lake City
7. Washington, DC 24. Rapid City, SO
8. Baltimore, MO 26. Fresno, CA
12. Knoxville, TN 27. Bellevue, WA
17. Glen Ellyn, IL 28. Eugene, OR
19. Austin, TX
Numbering of cities conforms to NURP convention.
UT
3 Percentages rounded to nearest whole number.
h Some reported concentrations are qualified by STORE! quality control remark codes, to wit: A = Value reported is the
mean of two or more determinations; S = Value reported is the maximum of two or more determinations; L = Actual value
is known to be greater than value given; M = Presence of material verified but not quantified; T = Value reported is
less than criteria of detection. One value in this column indicates one positive observation or that all observations
were equal.
5 No longer included as a priority pollutant.
6-50
-------
TABLE 6-20. MOST FREQUENTLY DETECTED PRIORITY POLLUTANTS
IN NURP URBAN RUNOFF SAMPLES1
Priority Pollutants Detected in 75 Percent or More of the NURP Samples
Inorganics Organics
30. Lead (94%) None
36. Zinc (94%)
28. Copper (91%)
Priority Pollutants Detected in 50 percent to 74 percent of the NURP Samples
Inorganics Organics
27. Chrominum (58%) None
23. Arsenic (52%)
Priority Pollutants Detected in 20 percent to 49 percent of the NURP Samples
Inorganics Organics
26. Cadmium (48%) 105. Bis (2-ethylhexyl) phthalate (22%)
32. Nickel (43%) 3. a-Hexachlorocyclohexane (20%)
29. Cyanides (23%)
Priority Pollutants Detected in 10 percent to 19 percent of the NURP Samples
Inorganics Organics
22. Antimony (13%) 12. ot-Endosulfan (19%)
25. Beryllium (12%) 94. Pentachlorophenol (19%)
33. Selenium (11%) 7. Chlordane (17%)
5. Y~Hexachlorocyclohexane (Lindane) (15%)
122. Pyrene (15%)
90. Phenol (14%)
121. Phenanthrene (12%)
47. Dichloromethane (methylene chloride) (31%)
96. 4-Nitrophenol (10%)
115. Chrysene (10%)
117. Fluoranthene (16%)
1 Based on 121 sample results received as of September 30, 1983, adjusted
for quality control review. Does not include special metals samples.
6-51
-------
metals project in order to determine the relationships among dissolved,
total, and total recoverable concentrations. The discussion and result of
this separate effort are in a subsequent section of this chapter.
A comparison of individual urban runoff sample concentrations undiluted by
stream flow (i.e., end of pipe concentrations) with EPA water quality cri-
teria and drinking water standards reveals numerous exceedances of these
levels, as shown in Table 6-21. Freshwater acute criteria were exceeded by
copper concentrations in 47 percent of the samples and by lead in 23 percent.
Freshwater chronic exceedances were common for lead (94 percent), copper
(82 percent), zinc (77 percent), and cadmium (48 percent). One organ oleptic
(taste and odor) criteria exceedance was observed. Regarding human toxicity,
the most significant, pollutant was lead. Lead concentrations violated
drinking water criteria in 73 percent of the observations.
Whenever an exceedance is noted above, it does not necessarily imply that an
actual violation of criteria did or will take place in receiving waters.
Rather, the enumeration of exceedances is used as a screening procedure to
make a preliminary identification of those pollutants for which their pres-
ence in urban runoff requires highest priority for further evaluation. Ex-
ceedances of freshwater chronic criteria levels may not persist for a full
24-hour period, for example. However, many small urban streams probably
carry only slightly diluted runoff following storms, and acute criteria or
other exceedances may in fact be real in such circumstances..
Among the inorganics, the most frequently detected pollutants are also those
which are found at the highest concentrations, which most frequently exceed
water quality criteria and which are the most geographically well-
distributed. One additional observation can be made concerning the samples
from Washington, D.C. These samples accounted for a preponderance of the
detections of many of the less frequently detected inorganics, including
antimony, beryllium, mercury, nickel, selenium, and thallium. No sampling or
analytical irregularities have been identified which explain this result.
Organic Pollutants. In general, the organic pollutants were detected less
frequently and at lower concentrations than the inorganic pollutants.
Sixty-three of a possible 106 organics were detected. The most commonly
found organic was the plasticizer bis (2-ethylhexyl) phthalate (22 percent)
followed by the pesticide a-hexachlorocyclohexane (ot-BHC) (20 percent). An
additional 11 organic pollutants were reported with detection frequencies
between 10 and 20 percent; 3 pesticides, 3 phenols, 4 polycyclic aromatics,
and a single haloginated aliphatic (Table 6-20).
Criteria exceedances were less frequently observed among the organics than
the inorganics. One unusually high pentachlorophenol concentration of
115 ng/1 resulted in the only exceedance of the organoleptic criteria (Ta-
ble 6-21) . This observation and one for the chlordane exceeded the fresh-
water acute criteria. Freshwater chronic criteria exceedances were observed
for pentochlorophenol, bis (2-ethylhexyl) phthalate, y-hexachlorocyclohexane
(Lindane), a-endosulfan, and chlordane. All other organic exceedances were
in the human carcinogen category and were most serious for a-hexachloro-
cyclohexane (a-BHC), •y-hexachlorocyclohexane (y-BHC or Lindane), chlordane,
phenanthrene, pyrene, and chrysene.
6-52
-------
TABLE 6-21. SUMMARY OF WATER QUALITY CRITERIA EXCEEDANCES FOR
POLLUTANTS DETECTED IN AT LEAST 10 PERCENT OF NURP SAMPLES:
PERCENTAGE OF SAMPLES IN WHICH POLLUTANT
CONCENTRATIONS EXCEED CRITERIA1
Pollutant
1. PESTICIDES
3. a-Hexachlorocyf lohexanp
5. Y-Hexarhlorocyclohexane (Lindane)
7. Chlordanp
1J. a-Endosulfan
II. METALS AND INORGANICS
?2. Antimony
23 . Arsenic
25. Beryllium
26. Cadmium-
27. Chromium-16
28. Copper5
29. Cyamdes
30. Lead'
32. Nirke!5
33. Selenium
36. Zinc5
IV. HALOGENATED ALIPHATICS
47. Methane, dichloro-
V!I. PHENOLS AND CRESOLS
90. Phenol
94, Phenol, pentachloro-
96. Phenol , 4-nitro-
VIII. PHTHALATE ESTERS
105. Phthalate, bi s(2-ethylhexyl )
IX. POLYCVCLIC AROMATIC HYDROCARBONS
115. Chrysenp
117. Fluoranthene
121 . Phenanthrene
1?2. Pvrene
Frequency of
Detection (° )
20
15
17
19
13
52
12
48
58
91
23
94
43
11
94
11
14
19
10
22
10
16
12
15
Detections/
Samples^
21/106
15/100
7/42
9/49
14/106
45/87
11/94
44/91
47/81
79/87
16/71
75/80
39/91
10/88
88/94
3/28
13/91
21/111
11/107
15/69
11/109
17/109
13/110
16/110
Criteria Exceedances (' \
None
X
X
X
X
FA
2
8
47
3
23
14
1*
FC
8
17
10
6*
48
1*
82
22
94
5
5
77
11*
22*
OL
1
HH
1
4
73
21
10
HC'
8,18,20
0,10,15
17,17,17
52,52,52
1?,12,12
0,0,11
10,10,10
12,12,12
15,15,15
DW
1
1
1
73
10
Indicates FTA or FTC value substituted where FA or FC criterion not available (see bellow).
Based on 121 sample results received as of September 30, 1983, adjusted for quality control review.
Number of times detected/number of acceptable samples.
FA - Freshwater ambient 24-hour i nstantaneous maximum cri ten on ("acute" en ten on).
FC = Freshwater ambient 24-hour average criterion ("chronic" criterion).
FTA = Lowest reported freshwater acute toxic concentration. (Used only when FA is not ava"" 1 able. ^
FTC = lowest reported freshwater chronic toxic concentration. (Used only when FC is not available.1
OL = Taste and odor forganoleptic) criterion.
HH = Non-Carcinogenic human health criterion for ingestion of contaminated water and organisms.
HC ~ Protection of human health from carcinogenic effects for ingestion of contaminated water and organisms
DW = Primary drinking water criterion.
Fntries in this column indicate exceedances of the human carcinogen value at the 10 ,10 , and 10 risk level, respectively. The
numbers are cumulative, i.e., all 10~ exceedances are included in 10 exceedances, and all 10 exceedances are included in 10
exceedances.
Where hardness dependent, hardness of 100 mg/1 CaCO, equivalent assumed.
Different criteria are written for the trivalent and hexavalpnt forms of chromium. For purposes of this analysis, all chromium is
assumed to be in the less toxic trivalenT form.
6-53
-------
An additional 50 organic pollutants were found in one to nine percent of the
samples. These frequencies of detection are low, and the pollutant is noted
in Table 6-22.
Among the PCB group, there was only a single detection of one PCB type among
all the samples. Approximately two-thirds of the halogenated aliphatic com-
pounds were detected. Among those cities reporting these compounds, the city
of Eugene, Oregon, figured prominently. For example, eight pollutants from
this group were found in Eugene only. None of the pollutants in the ethers
group were detected.
Monocyclic aromatics were rarely detected in the samples. However, many
reported detections of benzene and toluene, two commonly reported pollutants,
had to be withdrawn due to contamination problems.
Of the 11 phenolics, four have not been reported in urban runoff, while three
have been observed only once. The remaining four have been found fairLy
frequently but at low concentrations. Exceedances of criteria were noted
only for pentachlorophenol.
All the phthalate esters were detected at Least once in the NURP program,
with bis (2-ethylhexyl) found most frequently. Several times the reported
concentration exceeded the lowest observed freshwater acute toxic concentra-
tion for this pollutant. Given the significant blank contamination problems
with the phthalates, however, these findings must be interpreted with
caution.
Only two of the polycyclic aromatic hydrocarbons were not detected in at
least one sample. Crysene, phenanthrene, pyrene, and fluoranthene were each
found at least 10 percent of the time. All the observed concentrations for
the first three of these pollutants exceeded the criteria for the protection
of human health from carcinogenic effects (there are no such criteria for
fluoranthene). Results for the polycyclic aromatics were generally free from
quality control problems.
There were no detections of nitrosamines or other nitrogen-containing com-
pounds. Due to methodological and holding time problems, however, results
for only two compounds can be used. Moreover, for one of these compounds,
3,3-dichlorobenzidine, performance evaluation results were unacceptable in
several cases.
Pollutants Not Detected In Urban Runoff
Some 43 priority pollutants were riot detected in any acceptable runoff sam-
ples (Table 6-22). All of these pollutants are organics. This group of sub-
stances should be considered to pose a minimal threat to the quality of
surface waters from runoff contamination.
While the priority pollutants which were not detected are of less immediate
concern than those pollutants found often, they cannot safely be eliminated
from all future consideration. Many of these pollutants have associated
water quality criteria which are below the limits of detection of routine
6-54
-------
TABLE 6-22. INFREQUENTLY DETECTED ORGANIC PRIORITY
POLLUTANTS IN NURP URBAN RUNOFF SAMPLES1
Priority Pollutants Detected in 1 percent to 9 percent of the NURP Samples
51. Trichloromethane (9%)
120. Naphthalene (9%)
98. 2,4-Dimethyl phenol (8%)
109. Anthracene (7%)
2. Aldrin (6%)
6. 6-Hexachlorocyclohexane (6%)
9. DDE (6%)
11. Dieldrin (6%)
17. Heptachlor (6%)
58. 1,1,1-Trichloroethane (6%)
65. Trichloroethene (6%)
85. Ethylbenzene (6%)
102. Diethyl phthalate (6%)
103. Di-n-butyl phthalate (6%)
104. Di~n-octyl phthalate (6%)
106. Butyl benzyl phthalate (6%)*
114. Benzo(a)pyrene (6%)
4. B-Hexachlorocyclohexane (5%)
53. Trichlorofluoromethane (5%)2
66. Tetrachloroethene (5%)
78. Benzene (5%)
79. Chlorobenzene (5%)
111. Benzo(b)fluoranthene (5%)*
64. 1,2-trans-dichloroethene (4%)
110. Benzo(a)anthracene (4%)
19. Isophorone (3%)
52. Tetrachloromethane (carbon tetrachloride) (3%)
56. 1,1-Dichloroethane (3%)
87. Toluene (3%)
112. Benzo(k)fluoranthene (3%)
18. Heptachlor epoxide (2%)*
59. 1,1,2-Trichloroethane (2%)*
60. 1,1,2,2-Tetrachloroethane (2%)*
63. 1,1-Dichloroethene (2%)
68. 1,3-Dichloropropene (2%)*
113. Benzo(g,h,i)perylene (2%)
10. DDT (1%)*
43. PCB-1260 (1%)*
48. Chlorodibromomethane (1%)*
49. Dichlorobromomethane (1%)*
50. Tribromomethane (bromoform) (1%)*
57. 1,2-Dichloroethane (1%)*
67. 1,2-Dichloropropane (1%)*
91. 2-Chloropheno: (1%)*
95. 2-Nitrophenol (1%)*
99. p-Chloro-m-creosol (1%)*
101. Dimethyl phthalate (1%)*
116. Dibenzo(a,h)anthracene (1%)*
118. Fluorene (1%)*
119. Indeno(l,2,3-cd)pyrene (1%)*
6-55
-------
TABLE 6-22. INFREQUENTLY DETECTED ORGANIC PRIORITY
POLLUTANTS IN NURP URBAN RUNOFF SAMPLES1 (Cont'd)
Priority Pollutants Not Detected in NURP Samples
8. ODD
13. g-Endosulfan
14. Endosulfan sulfate
15. Endrin
16. Endrin aldehyde
21. Toxaphene
37. PCB-1016
38. PCB-1221
39. PCP-1232
40. PCB-1242
41. PCB-1248
42. PCB-1254
44. 2-Chloronaphthalene
45. Bromomethane (methyl bromide)
46. Chloromethane (methyl chloride)
54. Dichlorodifluoromethane (Freon-12)2
55. Chloroethane
61. Hexachloroethane
62. Chloroethene (vinyl chloride)
69. Hexachlorobutadiene
71. Bis(chloromethyl) ether2
72. Bis (chloroethyl) ether
73. Bis(chloroisopropyl) ether
74. 2-Chloroethyl vinyl ether
75. 4-Bromophenyl phenyl ether
76. 4-Chlorophenyl phenyl ether
77. Bis(2-chloroethoxy) methane
80. 1,2-Dichlorobenzene
81. 1,3-Dichlorobenzene
82. 1,4-Dichlorobenzene
83. 1,2,4-Trichlorobenzene
84. Hexachlorobenzene
86. Nitrobenzene
88. 2,4-Dinitrotoluene
89. 2,6-Dinitrotoluene
92. 2,4-Dichlorophenol
93. 2,4,6-Trichlorophenol
97. 2,4-Dinitrophenol
100. 4,6-Dinitro-o-cresol
107. Acenaphthene
108. Acenaphthylene
125. Di-n-propyl nitrosamine
127. 3,3'-Dichlorobenzidine
6-56
-------
TABLE 6-22. INFREQUENTLY DETECTED ORGANIC PRIORITY
POLLUTANTS IN NURP URBAN RUNOFF SAMPLES1 (Cont'd)
Priority Pollutants Not Analyzed for or Withdrawn for Methodological
Reasons or Holding Time Violations
1. Acrolein
20. TCDD (Dioxin)
24. Asbestos
70. Hexachlorocyclopentadiene
123. Dimethyl nitrosamine (DMN)
124. Diphenyl nitrosamine
126. Benzidine
128. 1,2-Diphenyl hydrazine
129. Acryloriitrile
* Detected in only one or two samples.
Based on 121 sample results received as of September 30, 1983, adjusted
for quality control review.
^ No longer on the priority pollutant list.
analytical methods. Some of these substances may in fact have been present
in the NURP samples. Four priority pollutants not detected in runoff were
found in street dust sweepings from Bellevue, Washington, suggesting that
further urban runoff samplings can be expected to detect more priority pol-
lutants. More sensitive analytical methodologies must be used and dilution
effects considered before it can be said with assurance that these pollutants
are not found in urban stormwater runoff at levels which, without dilution,
pose a threat to human health or aquatic life.
ODD, chloromethane, 1,2-dichlorobenzene, and 2,4-dichlorophenol were detected
in runoff samples at least once, but these observations had to be withdrawn
for quality control reasons. Therefore, among the not detected pollutants,
these four can be considered to have a slightly elevated possibility of ac-
tually being present in the runoff samples.
RUNOFF-RAINFALL RELATIONSHIPS
A runoff coefficient (Rv), defined as the ratio of runoff volume to rainfall
volume, has been determined for each of the monitored storm events. As with
the EMCs, the runoff coefficient values at a particular site are, with rela-
tively few exceptions, well characterized by a lognormal distribution.
Table 6-23 summarizes the statistical properties of Rv's at the loading sites
in the data base.
Figure 6-19 illustrates the relationship between percent impervious area and
the median runoff coefficient for the site. Sites which monitored fewer than
5 storms are excluded. The upper plot (a) groups the results from 16 of the
6-57
-------
w
H
CO
W
CO
Q
O
CO
E-i
CJ
M
tw
CM
W
O
U
O
a
CM
I
W
6-58
-------
o
E
LL.
O
u
I.U
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
n
0
!••
0
0 .
•
- •*
B
»
9
" n*
«
o
>
1
o
o
0
a
6
0
«
a
»
0 10 20 30 40 50 60 70 80 90 100
% IMPERVIOUS
(a) 16 Projects
I.U
0.9
0.8
<£ 0.7
i—
1 0-6
iz
g 0.5
o
fe 0.4
« 0.3
0.2
0.1
n
o
0
•
«
o1
e
o
e
o
a
• 6
0
0
0 10 20 30 40 50 60 70 80 90 100
% IMPERVIOUS
(b) 4 Projects (KS1, Mil, TNI, TX1)
Figure 6-19. Relationship Between Percent Impervious Area
and Median Runoff Coefficient
6-59
-------
20 projects investigated. The lower plot (b) groups results from the re-
maining four projects (KS1, Mil, TNI, TX1). The reason for the difference is
unexplained. However, the separate grouping is based on the fact that the
relationship for these sites is internally consistent and significantly dif-
ferent than the bulk of the project results.
Figure 6-20 illustrates the same impervious area/runoff coefficient rela-
tionship, but shows the 90 percent confidence limits for median Rv's.
POLLUTANT LOADS
Although the EMC median concentration values are appropriate for many appli-
cations (e.g., assessing water quality impacts in rivers and streams), when
cumulative effects such as water quality impacts in lakes and comparisons
with other sources on a long-term basis (e.g., annual or seasonal loads) are
to be examined, the EMC mean concentration values should be used. Taking the
EMC median and coefficient of variation values given in Table 6-37, we have
converted them into mean values using the relationship given in Chapter 5.
These EMC mean concentrations and the values used in the load comparison to
follow are listed in Table 6-24.
The range shown for site mean concentrations for both the median and 90th
percentile urban sites reflects the difference in means depending on whether
the higher or lower value of coefficient of variation listed in Table 6-17 is
used to describe event-to-event variability of EMC's at urban sites. The
range in values shown for use in the load comparisons below reflects the
median and 90th percentile site mean concentrations, using the average of the
range caused by coefficient of variation effects.
TABLE 6-24. EMC MEAN VALUES USED IN LOAD COMPARISON
Constituent
TSS (mg/1)
BOD (mg/1)
COD (mg/1)
Tot. P (mg/1)
Sol. P (mg/1)
TKN (mg/1)
NCL -N (mg/1)
2+3
Tot. Cu (ug/1)
Tot. Pb (ug/1)
Tot. Zn (ug/1)
Site Mean EMC
Median
Urban Site
141 - 224
10 - 13
73 - 92
0.37 - 0.47
0.13 - 0.17
1.68 - 2.12
0.76 - 0.96
38 - 48
161 - 204
179 - 226
90th Percentile
Urban Site
424 - 671
17 - 21
157 - 198
0.78 - 0.99
0.23 - 0.30
3.69 - 4.67
1.96 - 2.47
104 - 132
391 - 495
559 - 707
Values Used in
Load Comparison
180 - 548
12 - 19
82 - 178
0.42 - 0.88
0.15 - 0.28
1.90 - 4.18
0.86 - 2.21
43 - ]18
182 - 443
202 - 633
6-60
-------
0 10 20 30 40 50 60 70 80 90 100
% IMPERVIOUS
(a) 16 Proiects
1.0
0.9
0.8
0.7
0.6
0.5
it 0.4
0.3
0.2
0.1
0 10 20 30 40 50 60 70 80 90 100
% IMPERVIOUS
(b) 4 Projects (KS1, Mil, TNI, TK1)
Figure 6-20. 90 Percent Confidence Limits for Median
Runoff Coefficients
6-63
-------
It is a straightforward procedure to calculate mean annual load estimates for
urban runoff constituents on a Kg/Ha basis by assigning appropriate rainfall
and runoff coefficient values and selecting EMC mean concentration values
from Table 6-24. In and of themselves, however, such estimates see;m to be of
little utility. Therefore, it was decided to do a comparison of the mean
annual loads from urban runoff with those of a "well run" secondary treatment
plant. We chose to use TSS = 25 mg/1, BOD = 15 mg/1, and Tot. P = 8 mg/1 for
the effluents from such plants for the purposes of this order of magnitude
comparison. For a meaningful comparison for a specific situation, locally
appropriate values should be used. Based upon Table 6-24, the corresponding
urban runoff mean concentrations used were TSS = 180 mg/1, BOD = 12 mg/1, and
Total P = 0.4 mg/1 as typical and TSS = 548 ug/1, BOD = 19 mg/1, and
Tot. P = 0.88 mg/1 as a "worst case" for comparison purposes.
The value of 0.35 was selected as a typical mean runoff coefficient. It is
the median of the NURP mean runoff coefficient database for the twenty
projects discussed earlier; their average is 0.42, but we believe that this
number is overly weighted by the disproportionate number of highly impervious
sites in the database. Assuming an average population density of 10 persons
per acre (the average of the NURP sites) and a mean annual rainfall of
40 inches per year, urban runoff averages 104 gallons per day per capita.
This is also a reasonable estimate of sewage generation in an urban area.
Therefore, as a first cut, the ratio of mean pollutant concentrations of
urban runoff and POTW effluents will also be the ratio of their annual loads.
Thus, we have;
TSS = -. = 7 ; BOD = -if = 0.8 ; Tot. P = ^~ ~ 0.05
2b ID o
using typical urban runoff values, and;
TSS - 548 s 22 BQD = 19 ^ 1>3 Tot> p = 2,88 ^ ^
2.b ID o
using the "worst case" values. These numbers suggest that annual loads from
urban runoff are approximately one order of magnitude higher than those from
a well run secondary treatment plant for TSS, the same order of magnitude for
BOD, and an order of magnitude less for Tot.. P.
If the hypothetical urban area just described were to go to advanced waste
treatment and achieve an effluent quality of TSS = 10 mg/1, BOD = 5 mg/1, and
Total P = 1 mg/1 and no urban runoff controls were instituted, the mean
annual load reductions to the receiving water would be :
TSS = - * 7% ; BOD = = 37% ' Tot' p -- ^ 83%
for our typical case, and;
TSS = lf * 3% ? BOD " lfr-15 = 29% '• Tot' p = oTeir-a * 79%
6-62
-------
for our "worst case." On the other hand, if urban runoff controls that
reduced TSS by 90 percent, BOD by 60 percent, and Total P by 50 percent were
instituted, (typical results from a well-designed detention basin) , the mean
annual load reductions to the receiving water would be:
TSS . = W , BOD .
for our typical case, and;
Too _ 548 - 55 __ _ _ 19-8 ^ _ _ 0.88 - 0.44 ^
TSS ~ 548 + 25 ~ 86° ; B°D - 19 + 15 ~ 32* ' T°tal P ~ 0.58 + 8 ~ 5%
Thus, if these pollutants are causing receiving water quality problems, con-
sideration of urban runoff control appears warranted for TSS, both urban
runoff control and AWT might be considered for BOD, and only AWT would be
effective for Total P.
The foregoing should be viewed as illustrative of a preliminary screening for
trade-off studies that can be performed using appropriate values for a
specific urban area, rather than as description of any particular real-world
case. They are, however, believed useful in providing order of magnitude
comparisons. Local values for annual rainfall, runoff coefficient, or point
source characteristics that are different than those used in the illustration
will of course change the results shown; although in most cases the changes
would not be expected to cause a significant change in the general
relationship.
As a final perspective on urban runoff loads, Table 6-25 presents an estimate
of annual urban runoff loads, expressed as Kg/Ha/year, for comparison with
other data summaries of nonpoint source loads which state results in this
manner. Load computations are based on site mean pollutant concentrations
for the median urban site and on the specified values for annual rainfall and
runoff coefficient. Typical values for mean runoff coefficient (based on
NURP data) have been assigned for residential land use (Rv = 0.3), commercial
land use (Rv = 0.8), and for an aggregate urban area which is assumed to have
representative fractions of the total area in residential, commercial, and
open uses (Rv = 0.35).
Several useful observations can be made. The annual load estimates which
results are comparable to values and ranges reported in the literature.
Although the findings presented earlier in this chapter indicated that the
land use category does not have a significant influence on site concentra-
tions of pollutants, on a unit area basis total pollutant loads are sig-
nificantly higher for commercial areas because of the higher degree of
imperviousness typical of such areas. For broad urban areas, however, the
relatively small fraction of land with this use considerably mitigates such
an effect.
Finally, the annual loads shown by Table 6-25 have been computed on the basis
of a 40 inch annual rainfall volume. For urban areas in regions with higher
6-63
-------
TABLE 6-25. ANNUAL URBAN RUNOFF LOADS KG/HA/YEAR
Constituent
Assumed Rv
TSS
BOD
COD
Total P
Sol. P
TKN
N02+3-N
Tot. Cu
Tot. Pb
Tot. Zn
Site Mean
Con.mg/1
180
12
82
0.42
0.15
1.90
0.86
0.043
0.182
0.202
Residential
0.3
550
36
250
1.3
0.5
5.8
2.6
0.13
0.55
0.62
Commercial
0.8
1460
98
666
3.4
1.2
15.4
7.0
0.35
1.48
1.64
All Urban
0.35
640
43
292
1.5
0.5
6.6
3.6
0.15
0.65
0.72
NOTE. Assumes 40 inches/year rainfall as a long-term average.
or lower rainfall, these load estimates must be adjusted. The results
presented earlier suggest that pollutant concentrations are not sensitive to
runoff volume; however, total loads (the product of concentration and volume)
are strongly influenced by the volume of runoff. For estimates using equiv-
alent site conditions (Rv), loads for areas with other rainfall amounts are
obtained by factoring by the ratio of local rainfall volume to the 40 inch
volume used for the table. Planners who believe that the average annual
runoff coefficients in their local areas are substantially different from
those used in the table can make similar adjustments.
6-64
-------
CHAPTER 7
RECEIVING WATER QUALITY EFFECTS OF URBAN RUNOFF
INTRODUCTION
The effects of urban runoff on receiving water quality are very sdte speci-
fic. They depend on the type, size, and hydrology of the water body, the
designated beneficial use and the pollutants which affect that use, the urban
runoff (URO) quality characteristics, and the amounts of URO dictated by
local rainfall patterns and land use.
A number of the NURP projects examined receiving water impacts in some de-
.tail, others less rigorously. Because of the uniqueness of URO water quality
impacts, individual project results are considered best used for confirmation
and support, rather than as a basis for broad generalizations.
Accordingly, this chapter is structured to address each of the principal cat-
egories of receiving water bodies separately; streams and rivers, lakes,
estuaries and embayments, and groundwater aquifers. Some can be addressed
more thoroughly than others at this time. The approach taken to develop a
general, national scale screening assessment of the significance of URO pol-
lutant discharges is to compute anticipated effects using analysis methodolo-
gies identified in Chapter 5, where these are appropriate and to compare
anticipated effects indicated by such generalizations to specific experiences
and conclusions drawn by relevant individual NURP projects.
As with any generalization, there will be exceptions. Specific local situa-
tions can be expected which are either more or less favorable than the gen-
eral case. The results presented herein should therefore be interpreted as
representative estimates of a substantial percentage of urban runoff sites,
but not all of them.
Receiving waters have distinctive general characteristics which depend on the
water body type (e.g., stream, lake, estuary) and relatively unique individ-
ual characteristics which depend on geometry and hydrology. Given a minimum
acceptable amount of data on water bodies and their setting, it appears pos-
sible to make useful generalizations regarding the quantitative effects of
urban runoff on concentrations of various pollutants in the receiving waters
and to draw inferences concerning the influence urban runoff may have on the
beneficial uses of the water bodies. However extending the results of such
an analysis to an assessment of the prevalence of urban runoff induced "prob-
lems" on a national scale cannot be accomplished in a way would provide an
acceptable level of confidence in any conclusions drawn therefrom. In addi-
tion to the importance of local hydrology, meteorology, and urban character-
istics, the emphasis placed on each of the three elements that influence
problem definition;
(1) Denial or serious impairment of beneficial use;
7-1
-------
(2) Violation of ambient water quality standards; and
(3) Local perception;
will result in a high degree of site-specificity to the determination of the
existence of a problem.
RIVERS AND STREAMS
General
Flowing streams carry pollutant discharges downstream with the stream flow.
For intermittent stormwater discharges, a specific stream location and the
biota associated with it are exposed to a sequence of discrete pulses con-
taminated by the pollutants which enter with urban runoff. Because of the
inherent variability of urban runoff (URO), the average concentrations in
such pulses vary, as do their duration and the interval between successive
pulses. Table 7-1 summarizes average values for storm duration and intervals
between storm events for selected locations in the U.S., based on analysis of
long term rainfall records using a methodology (SYNOP) presented in an
earlier NURP document (the NURP Data Management Procedures Manual). The
information presented provides a sense of the temporal aspects of such inter-
mittent pulses and, by inference, the intermittent exposure patterns to which
stream biota are subjected. For many locations, storm pulses are produced
for about six hours every three days or more, on average.
A probabalistic methodology has been used to examine the concentration char-
acteristics of the storm pulses produced in streams, given the variability of
the relevant processes which are directly involved. Stream flow rates, run-
off flow rates, and concentrations vary and result in variable stream concen-
trations. For streams, it is not the runoff volume per se that is important.
The combination of stream and runoff flow rates (together with runoff concen-
tration) determine the pollutant concentration in the stream pulse. The
duration of the runoff event and the stream velocity dictate the spatial
extent of the storm pulse in the stream. The analysis presented in this
section addresses the frequency and magnitude of pollutant concentrations in
the instream storm pulses which are produced.
Runoff and Stream Flow Rates
The local combination of stream and runoff flow rates for an urban location
are, as indicated, important determinants of the stream concentrations which
will result. For long-range projections, the most appropriate data sources
for characterizing these parameters are long-term stream flow gauging records
(USGS) and long-term rainfall records (USWS).
Figure 7-1 (a) illustrates the regional variation of average daily stream
flows expressed as cfs/sq mile of drainage area, based on long-term (50 years
or more) gauging records at over 1000 stations. Figure 7-1(b) presents a
somewhat simplified regional pattern for average rainfall intensity. The
data base for this plot is considerably smaller, consisting of rainfall
records (usually 10 to 30 years of record) for approximately 40 cities.
Localized peturbations exist, but are smoothed out by contours presented.
7-2
-------
TABLE 7-1. AVERAGE STORM AND TIME BETWEEN STORMS FOR
SELECTED LOCATIONS IN THE UNITED STATES
Location
Atlanta, GA
Birmingham, AL
Boston, MA
Caribou, ME
Champaign-Urbana, IL
Chicago, IL
Columbia, SC
Davenport , IA
Detroit, MI
Gainesville, PL
Greensboro, SC
Kingston, NY
Louisville, KY
Memphis, TN
Mineola, NY
Minneapolis, MN
New Orleans, LA
New York City, NY
Steubenville, OH
Tampa, FL
Toledo, OH
Washington, DC
Zanesville, OH
Mean
Denver, CO
Oakland, CA
Phoenix, AZ
Rapid City, SD
Salt Lake City, UT
Mean
Portland, OR
Seattle, WA
Mean
Average Annual Values in Hours
Storm
Duration
8.0
7.2
6.1
5.8
6.1
5.7
4.5
6.6
4.4
7.6
5.0
7.0
6.7
6.9
5.8
6.0
6.9
6.7
7.0
3.6
5.0
5.9
6.1
6.1
9.1
4.3
3.2
8.0
7.8
6.5
15.5
21.5
18.5
Time Between
Storm Midpoints
94
85
68
55
80
72
68
98
57
106
70
80
76
89
89
87
89
77
79
93
62
80
77
81
144
320
286
127
133
202
83
101
92
-------
Figure 7-1(a). Regional Value of Average Annual Streamflow (cfs/sq mi)
.025
03
045
.055
.065
.105
125
.075
Figure 7-1(b). Regional Value of Average Storm Event Intensity (inch/hr)
7-4
-------
Variability of daily stream flows was determined for a smaller sample (about
150 sites) of the stream sites. Variability of storm event average intensi-
ties was determined for all of the rain gauge locations in the current data
base. These results are summarized in Table 7-2.
Total Hardness of Receiving Streams
Where the beneficial use of principal concern is the protection of aquatic
life, the URO pollutants of major concern appear to be heavy metals, partic-
ularly copper, lead and zinc. The potential toxicity of these pollutants are
strongly influenced by total hardness, as indicated by Table 5-1 in Chap-
ter 5. Other beneficial uses deal with pollutants and effects that are not
influenced by total hardness or (as with drinking water supplies) do not
modify the assigned significance of heavy metal concentrations on the basis
of total hardness.
As with stream flow and precipitation, distinct regional patterns also exist
for receiving water total hardness concentrations. Figure 7-2 delineates the
national pattern of regional differences. These patterns impose an addi-
tional regional influence on the potential of urban runoff to create problem
conditions in streams and rivers.
Technical Approach To Screening Analysis
The magnitude and frequency of occurrence of intermittent stream concentra-
tions of pollutants of interest, that result from urban runoff, has been
computed using the probabilistic methodology discussed in Chapter 5.
The input data required for application of the methodology includes repre-
sentative values for the mean and variability of stream flow, runoff flow,
and runoff pollutant concentrations. The material presented earlier in this
chapter provides the basis for assigning values for the flows; the results
summarized in Chapter 6 provide the basis for specifying pollutant concen-
tration inputs. In order to translate the probability distribution of stream
concentrations (which is the basic output of the analysis methodology) to an
average recurrence interval, which is considered to provide a more under-
standable basis for comparisons, the average number of storms per year is
also required. This is estimated directly from the average interval between
storm midpoints generated by the statistical analysis of hourly rainfall
records.
For a general screening on a national scale, an estimate of typical values
for a selected geographic location must be made. This has been done, and the
set of input values considered to be typical of geographical location are
described and summarized below. The values used should be considered rea-
sonably representative of the majority of sites in the area, but it should be
recognized that not all potential sites will have conditions either as favor-
able or unfavorable as those listed.
We have worked with a limited sample in assigning typical values. A greater
data base on rainfall and stream flow would permit greater spatial definition
7-5
-------
61 190ZE8
w
in
0)
•a
sc
0)
-p
Q)
U
«S
4-1
H
S-l
o
en
Q)
p
O
•H
CM
I
•H
fa
7-6
-------
than shown in the results. Specific regions or states could, with develop-
ment of a more detailed spatial definition of stream flows and rainfall, ex-
tend the analysis presented to provide a considerably more comprehensive
assessment of problem potential for local areas. This would involve the
development of input parameters (rainfall and streamflow) readily derived
from available long term USGS stream flow records and USWS rainfall records
and their use in the methodology with quality parameters based either on the
NURP analysis presented in Chapter 6, or on local monitoring activities.
The analysis methodology presently available permits computation of the pro-
bability distribution of instream concentrations, incorporating the effect of
upstream (background) concentrations of the pollutant of interest. The re-
sults presented here assume upstream concentrations of zero, principally be-
cause of our inability at present to make reliable estimates of typical
values for the magnitude and variability for pollutants of interest, espe-
cially on the broad national scale being examined. As a result, the summa-
ries will show the effects of urban runoff contributions only. In cases
where the background is small relative to the URO contribution, the summaries
will represent actual conditions quite closely. However, where background is
high and has appreciable variability, the implications of the URO contribu-
tion will be overstated, particularly the inferred improvement which could
result from control of URO.
In order to perform a national screening of regional influences on urban run-
off impacts, eight geographical regions illustrated by Figure 7-3 have been
delineated. Using the information summarized by Figures 7-1 and 7-2, typical
values for the pertinent rainfall/runoff and stream parameters have been
assigned for each of the regions. Table 7-2 summarizes the values for these
parameters which are used in the screening analysis.
TABLE 7-2. TYPICAL REGIONAL VALUES
I
L
3
4
5
6
7
8
Event Average
Rainfall Intensity
Mean
(in/hr)
0.04
0.10
0.08
0.055
0.04
0.03
0.045
0.025
C. V .
1.00
1.35
1.35
1.25
1.10
1.10
1.20
0.85
Average
Number
of
Events/year
110
100
90
110
63
70
30
80
Average
Runoff Flow Rate
Mean Event
( c f s / s q mi)
5
1?
10
7
5
4
5
3
c .v .
0.85
1.15
1.15
1.05
0.95
0.95
1.00
0.75
Stream Flow Rate
(Dai ly Avq Flows)
Mean
(cfs/sq mi )
1.75
1.25
1.00
0.75
0.35
0.05
0.05
4.50
c.v .
1.25
l.?5
1.25
1.25
1.25
1.25
1.25
1.25
Stream
Total
Hardness
(mg/1 )
50
50
50
200
200
300
200
50
Average stream flow and rainfall intensity were taken from the plots, which
are based on sources previously described. The estimate for variability of
daily stream flows (coefficient of variation) is based on computed values for
a sample of about 150 perennial streams. Results for a number of regional
7-7
-------
a-l90Z-£8
w
•H
w
c
•H
c
0)
0)
1-4
o
0)
4->
O
w
c
O
•H
CP
01
a;
o
-H
O
a;
a
•H
fc
7-8
-------
groupings indicated median values for coefficient of variation to fall be-
tween approximately 1 and 1.5. Since there were no clear regional patterns
apparent, a uniform value for coefficient of variation of stream flows of
1.25 was assigned.
The coefficient of variation of rainfall intensities was taken directly from
the statistical analysis of the rainfall records examined. This was reduced
by 15 percent to provide estimates of the coefficient of variation of runoff
flow rates, based on a recent published report, "Comparison of Basin Perform-
ance Modeling Techniques", Goforth, Heaney and Huber, ASCE JEED, Novem-
ber 1983, using the SWMM model on a long-term rainfall record.
The quality characteristics of urban runoff used in the screening analysis
are listed in Table 7-3, and are based on the results summarized in Chap-
ter 6. The analysis results have been rounded in the selection of repre-
sentative site median EMCs and are interpreted as being representative of an
array of urban sites discharging into the receiving stream being analyzed.
Average site conditions are based on the 50th percentile of all urban sites.
Since the data analysis indicated that sites at some locations tend to clus-
ter at either the higher or lower ends of the range for all sites, high range
and low range site conditions were also selected for use in the screening
analysis. High range site conditions are nominally based on the 90th percen-
tile of all site median concentrations; the low range on the 10th percentile
site. The variability of EMCs from storm to storm at any site is based on
the median of the coefficients of variation of EMCs at sites monitored by
NURP. This value was used for the low range and average site condition and
was increased nominally for the high range site condition.
TABLE 7-3. URBAN RUNOFF QUALITY CHARACTERISTICS
USED IN STREAM IMPACT ANALYSIS
(Concentrations in yg/1)
Low Range of
Site Conditions
Average
Site Conditions
High Range of
Site Conditions
COPPER
Site Median
EMC
15
35
90
Coef
Var
C.6
0.6
0.7
LEAD
Site Median
EMC
50
135
350
Coef
Var
0.75
0.75
0.85
ZINC
Site Median
EMC
75
165
450
Coef
Var
0.7
0.7
0.8
An illustrative example of a site-specific application of the probabilistic
analysis methodology employed is presented in order to:
1. Illustrate the nature of the computational results produced;
-------
2. Assist in the interpretation of the tabulations presented Later
which summarize results of the national scale screening
analysis;
3. Indicate how magnitude/frequency of instream concentrations may
be interpreted for inferences concerning the absence or
presence of a "problem" and where a problem is concluded to
exist, its degree of severity; and
4. Demonstrate how alternative URO control options may be eval-
uated in terms of their expected impact on water quality and
potential effect on problem severity.
From selected representative values for mean and variability of stream and
runoff conditions, the probability distribution of resulting instream concen-
trations during storm events can be computed. Figure 7-4 illustrates a plot
of such an output. Uncertainty in estimates for specific inputs can be ac-
commodated by sensitivity analyses which incorporate upper and lower bounds
for specific parameter values. Results are then presented as a band rather
than a specific projection. The probabilities which are the basic output of
the analysis may be converted to average recurrence intervals to provide what
is believed to be a more understandable basis for interpreting and evaluating
results.
Figure 7-5 presents results converted to the average recurrence interval at
which specific stream concentrations will be produced during storm runoff
periods.
The significance of a particular magnitude/frequency pattern of stream con-
centrations caused by urban runoff can be evaluated by comparing them with
concentrations which are significant for the beneficial use of the water
body. In the example presented, we have excluded comparisons with drinking
water criteria on the basis that urban streams are not generally used as
domestic water sources, and in any event, the criteria relate to finished
water, and surface water supplies almost invariably receive treatment.
Protection of aquatic life is selected for the screening analysis of the im-
pact of urban runoff because it is believed to be the predominant potential
beneficial use for urban streams on a national scale. The concentrations
which result from urban runoff are compared with stream target concentrations
associated with different degrees of adverse impact, as discussed and tabu-
lated in Chapter 5.
In the site specific situation illustrated, the stream concentrations of
copper caused by untreated urban runoff discharges exceed the "EPA Maximum"
criterion more than ten times per year on average. The concentration level
suggested by the NURP analysis to be the Threshold level of adverse biologi-
cal impacts is exceeded an average of five times per year (recurrence inter-
val 0.2 year), and significant mortality of more sensitive biological species
occurs about once every three years on average. Although this stress level
may not be great enough to result in a total denial of the use, there are
many who would argue that it represents an unacceptably severe degree of im-
pairment of this beneficial use.
7-10
-------
100
99
90
10
COPPER
STREAM TOTAL HARDNESS = 50 mgll
DRAINAGE AREA RATIO = 100
1 10 50 90
PERCENT OF STORM EVENTS EQUAL TO OR LESS THAN
Figure 7-4. Probability Distributions of Pollutant Concentrations
During Storm Runoff Periods
COPPER
STREAM TOTAL HARDNESS - 50 mgll
DRAINAGE AREA RATIO - 100
1 2 5
MEAN RECURRENCE INTERVAL YEARS
Figure 7-5. Recurrence Intervals for Pollutant Concentrations
7-1]
-------
The projection labeled "treated urban runoff" may be taken to represent the
in-stream result for either the originally considered discharge following the
application of controls which effect a 60 percent reduction, or of an uncon-
trolled urban runoff site with lower levels of copper in the runoff. In this
case, threshold levels are reached only once every 3 or 4 years on average,
and significant mortality levels are virtually never reached. Even though
the ambient "EPA MAX" criterion is exceeded once or twice a year on averacje,
one might conclude that the implied degree of stress is tolerable and is not
interpreted to represent a significant degree of impairment of the use.
The Threshold and Significant Mortality levels are estimates, which have been
explained earlier. In addition, the "acceptable" frequency at which specific
adverse effects can be tolerated is subjective at this time, since there are
no formal guidelines. However, an approach of this nature must be taken in
any evaluation of the significance of urban runoff and the importance of
applying controJ measures. There are two reasons why this is necessary.
First, because of the stochastic nature of the system we are dealing with,
virtually any target concentration we elect to specify will be exceeded at
some frequency, however rare. Secondly, from a practical point of vie;w,
there are limits to the capabilities of controls, however rigorously applied.
In the illustration presented, the untreated urban runoff site assigned urban
runoff copper concentrations equivalent to the average urban site. Since
NURP analysis data indicate that the copper in urban runoff has a soluble
fraction of about 40 percent, the level of removal used in the example re-
flects a control efficiency approaching the practical limit. Receiving water
impacts are significantly reduced, but not totally eliminated.
Results of Screening Analysis
A projection of stream water quality responses has been made for each of the
eight geographical areas shown by Figure 7-3. The rainfall, runoff, and
stream flow estimates used in the computations are those summarized in
Table 7-2. The urban runoff quality characteristics used are those presented
in Table 7-3.
To consolidate screening analysis results for easier comparison, results are
not presented as continuous concentration/frequency curves as used in the
illustrative example presented above. Instead, the comparison plots which
follow show only the recurrence interval at which specified biological
effects levels are exceeded. The concentrations which correspond with these
effects are strongly influenced by stream total hardness, and hence vary
regionally. Table 7-4, based on information presented in Chapter 5, summa-
rizes the stream target concentrations used in the screening analysis
summary.
Analysis results are presented for Copper (Figure 7-6), Lead (Figure 7-7) and
Zinc (Figure 7-8). Each individual bar represents a diffe;rent geographical
region, and the analysis is performed for two drainage area ratios. Since
regional stream flow differences are based on unit flows (cfs/sq mile of
drainage area), actual flow in a receiving stream at a particular location is
7-12
-------
TABLE 7-4. REGIONAL DIFFERENCES IN TOXIC CONCENTRATION LEVELS
(Concentrations in yg/1)
Pollutant
Copper
Lead
Zinc
Stream
Total Hardness
yg/i
50
200
300
50
200
300
50
200
300
Geo-
graphic
Regions
1,2,3,8
4,5,7
6
1,2,3,8
4,5,7
6
1,2,3,8
4,5,7
6
FPZX
MAX
12
42
62
74
400
660
180
570
800
Suggested Values For
Threshold
Effects1
20
80
115
150
850
1400
380
1200
1700
Significant Mortality2
(a) (b)
50
180
265
350
1950
3100
870
2750
3850
90
350
500
3200
17,850
29,000
3200
8000
11,000
Threshold Effects - mortality of the most sensitive individual
of the most sensitive species.
Significant Mortality
Level (a) - mortality of 50 percent of the most sensitive
species.
Level (b) - mortality of the most sensitive individual of
25th percentile sensitive species.
a function of
drainage area.
both the unit flow rate
The "drainage area ratio"
and the size of the contributing
(DAR) used in the analysis is
Urban Area Contributing Runoff
— ~ ~ ~~: - - - - - --- - -_-,™™.L-.,.— _.. —
Stream Drainage Area Upstream of Urban Input
It is a measure of the location of the urban area relative to the headwaters
of the receiving stream.
The shading scheme used on the bars duplicates that used earlier in the
illustrative example (Figure 7-5), and identifies the recurrence interval for
each of the target concentrations. For example, instream copper concentra-
tions during storm runoff periods in geographic region 1, with average site
conditions for copper concentrations in urban runoff, and a DAR = 10, are
projected to be as follows (middle plot, Figure 7-6) .
EPA MAX - ambient criterion is exceeded at a frequency of
0.02 year (= 50 times/year) or about every other storm event on
average.
7-13
-------
• S
smtu -
nH« IVIN
-g
H
§
•H
•P
(C
0)
U
O
U
4J
(U
s
O
<4-l K
H
>i CM
U ft
C O
Q) U
O
<«
•O
0)
0)
O
X
w
(U
n
3
Oi
•H
7-14
-------
SHVM - IVAHUm 33111111113311 WIN
I
HI
5
SHV3A - l¥AB3i«l BOmiltHlMH «V]H
H I
lint 1
r-
r-
0)
3
•rH
7-15
-------
•a
- nitnim xmrnnsm mm
m
SHV3H - IVIMmm 30H3UH3« ««»
J
Is*
6
i-l
4-1
C
0)
u
0
0
4-'
a>
m
E-i
Q)
S-l
4-1
t/1
o
M-l
u
>1 &
O H
C N
CD
O
-o
CD
(1)
U
I
r-
3
-!?
CM
SUV3* - 1VAH1MI 33»3»»n031) ««3H
7-16
-------
Threshold concentration levels at which adverse biological
stress for short duration exposures is projected to occur have a
recurrence interval of about 0.05 years (20 times/year).
Significant mortality levels are exceeded at intervals of about
0.5 year (twice/year) for the less severe effect, to about once
in 5.5 year for the more severe impact specified.
The plot is terminated at an upper level for recurrence interval of 50 years.
Although the analysis procedure computes specific recurrence intervals in
excess of this value, a realistic interpretation suggests that such condi-
tions are for practical purposes quite unlikely to ever be reached or ex-
ceeded. At computed recurrence intervals of about 10 years or more estimates
are not considered tc be reliable and are very probably conservative. There-
fore, indicated mean recurrence intervals in excess of 10 years probably (and
50 years certainly) should be interpreted as "unlikely" or "highly unlikely".
Discussion
An inspection of the screening analysis results (Figures 7-6 through 7-8)
indicates the reason why it is unrealistic to attempt a broad generalization
on whether urban runoff is, or is not a "problem" in rivers arid streams.
Water quality impacts can vary widely, depending on regional rainfall and
stream hydrology, urban site quality characteristics, drainage area ratio
(reflecting the size of the receiving stream relative to the urban area), and
the total hardness of the receiving stream. While the screening analysis
results provide an informative and useful perspective on the issue, it should
be recognized that any specific site may differ considerably from the typical
conditions used to characterize rainfall and stream flow for the area, and
further, that local variations in runoff quality characteristics within the
range defined by the NURP data can also have significant influence. The dom-
inant indication of the analysis is that the problem potential for urban
runoff is highly site-specific. Nevertheless some useful generalizations can
be made.
Perhaps the major factor which dictates whether urban runoff discharges of
copper, lead, or zinc will adversely impact aquatic life is the natural hard-
ness of the receiving streams. As a result, the southeast and gulf coast
areas are consistently indicated to be more sensitive than other areas of the
country. Of the remaining soft water areas, the northeast is somewhat less
sensitive; the Pacific northwest markedly less. This is attributed to sig-
nificantly lower storm intensities in these areas, coupled in the northwest
with appreciably higher stream flows.
Drainage area ratios have an important effect, reflecting as they do the
magnitude of stream flow at the urban location. The effect is much greater
for geographical regions with high unit flow (cfs/sq mile) than for lower
stream flow regions.
Finally, the quality characteristics of the urban sites have a significant
influence. Stream concentrations differ markedly depending on whether the
local urban sites tend to cluster toward the lower or higher end of the range
of site median concentrations indicated by the NURP data base.
7-17
-------
A comparison of the relative position of the bars on Figures 7-6, 7-7 and
7-8, is sufficient to indicate the comparative sensitivity to urban runoff
pollutant discharges. However, it is also desirable to decide whether a
given stream effect constitutes a serious degree of impairment of an aquatic
life beneficial use. There are no formal guidelines, and interpretations
that are either more liberal or more restrictive than those suggested below
may be preferred by others dealing with specific stream segments. For the
interpretation of the national scale screening analysis, the following deci-
sion basis has been used to identify the situations in which urban runoff is
likely to result in a water use "problem", (i.e., cause an unacceptable de-
gree of use impairment): '
Threshold effects - (mortality of the most sensitive individual
of the most sensitive species) occur more often thar about once
a year on average.
Significant mortality - using the lower of the two levels (i.e.,
50 percent mortality of the most sensitive species), occurs more
often than about once every 10 years on average.
Using these guidelines for assessing the occurrence of problem situations,
copper is shown to be the most significant of the three heavy metals con-
sistently found in urban runoff at elevated concentration levels. Where site
concentrations are at the high range of observed urban site conditions, prob-
lems are expected in all geographic regions at a DAR = 10, and in all geo-
graphic regions except region 8 at DARs as high as 100. When site
concentrations are in the average range of observed conditions, problem
situations are restricted to geographic regions 2 and 3 (plus region 1 at
DAR = 10) . When site copper concentrations are in the lower range of
observed site conditions, problem situations are restricted to geographic
regions 2 and 3 at low DARs. They are marginal (significant mortality once
every 5 years) but remain a problem according to the definition adopted. The
"marginal" attribution is used here, because the more severe degree of
significant mortality (most sensitive individual of 25th percentile sensitive
species) is indicated by the analysis virtually never to occur.
Thus, copper discharges in urban runoff are indicated to represent a signif-
icant threat to aquatic life use in regions 2 and 3 (southeast and Gulf
Coast) under almost all possibilities for urban site runoff quality. In re-
gion 1 (northeast), problems would be expected at all but the lower range of
site concentrations. In the hard water areas (regions 4, 5, 6, 7) problems
are expected only where site runoff quality is in the high end of the range
of observed site median concentrations.
It should be noted that the analysis has been based on total copper concen-
trations in urban runoff. Toxic effects are usually considered to be exerted
by the soluble form of the metal, and EPA defines an "active" fraction based
on a mild digestion which converts some of the inactive particulates to
soluble forms, to account for transformations which may occur in the natural
water systems. Copper in urban runoff has a typical soluble fraction of
about 50 percent, and the active fraction would therefore fall somewhere
between 50 and 100 percent of the total concentration used in the analysis.
The analysis has been performed using the total fraction, since adequate
7-18
-------
information is not available at present to reliably adjust these values.
However, although the problem assessment presented above may be somewhat con-
servative, further refinement along these lines would not change the infer-
ences drawn from the screening analysis results.
Zinc, like copper, has an indicated soluble fraction in the order of
50 percent, and the screening analysis indications will also be unaffected by
this consideration. It is indicated to be unlikely to pose a significant
threat to aquatic life in most urban runoff situations. Exceptions are
restricted to soft water areas in the east and south, lower DARs, and sites
with high zinc concentrations in urban runoff.
Lead results must be viewed with greater caution, because soluble fractions
in urban runoff are indicated to be quite low (less than 10 percent).
Problem indications are therefore likely to be reasonably conservative, i.e.,
overstate the problem potential. Problem situations may be expected to be
restricted to soft water areas in the east and Gulf areas when urban sites
have average site concentrations and DARs are low, and even at high DARs
when site concentrations are in the high range. Lead is not indicated to be
a threat to aquatic life in the hard water areas of the country or in the
Pacific northwest, except for the combination of low DAR and high site
concentration.
In performing the screening analysis, upstream concentrations were assumed to
be zero; that is, the receiving stream had only a diluting effect on the
urban runoff pollution. In actual cases background concentrations will be
greater than zero, and in some instances upstream contributions (e.g., agri-
cultural runoff, another city) could be significant and result in more severe
conditions than those identified in the screening analysis.
On the basis of the foregoing, it appears appropriate to identify copper as
the key toxic pollutant in urban runoff, for the following reasons:
Problem situations anticipated for lead and zinc do not occur
under any conditions for which copper does not show up as a
problem as well - and with more severe impacts. On the other
hand, copper is indicated to be a problem in situations where
lead or zinc are not.
Based on the ratios between concentrations producing increas-
ingly severe effects, copper is suggested to be a more generic
toxicant. It has an effect on a broad range of species. This
is in contrast to lead and zinc for which a substantially
greater degree of species selectivity is indicated. Some spe-
cies are sensitive, others relatively insensitive to lead and
zinc.
From the NURP data, locations which tend to have site median
concentrations in the low, average, or high end of the range
have generally consistent patterns for each of the three heavy
metals.
7-19
-------
Control measures which produce reductions in copper discharges
to receiving waters could be expected to result in equivalent
reductions in zinc, and greater reductions in lead, by virtue of
its significantly greater particulate fraction.
Copper is accordingly suggested to be an effective indicator for all heavy
metals in urban runoff relative to aquatic life. It might be used as the
focus for control evaluations, site specific bioassays, monitoring
activities, and the like.
It should be noted that while immediate water column impacts of lead are not
as significant as those for copper, the high particulate fraction of lead
would tend to result in greater accumulations in the stream bed. This aspect
has not been addressed by the NURP program in sufficient detail to warrant
any comment on its potential significance.
The results of the screening analysis summarized by Figures 7-6 through 7-8
are approximate, because they are influenced by the suitability of the
typical values for stream and runoff flows which were assigned. This however
can be refined by the use of appropriate values which can be developed from
readily available data bases, and thus adjusted for local variations which
are to be expected. A second issue relative to the reliability of the pro-
jections is the validity of the computations, given that the input parameters
are representative. This has been confirmed by a number of validation tests,
discussed in the NURP supporting document referenced earlier, which addresses
the stream analysis methodology.
The remaining issue for evaluating the reliability of the indications of
problem potential produced by the screening analysis is the reasonableness of
the intermittent exposure concentration levels, which have been associated
with various biological effects levels, and the guidelines adopted for this
discussion, which determine whether or not a problem is expected. While
rather tenuous at this time, the information available does provide support.
Two of the NURP projects examined aquatic life effects in streams receiving
runoff from monitored sites.
- Bellevue, WA concluded that whatever adverse effects were ob-
served were attributable to habitat impacts (stream bed scour
and deposition) as opposed to chemical toxicity. For this
project, heavy metal concentrations in the monitored urban
runoff sites were typical of the average for all urban sites.
The screening analysis results under these conditions do not
indicate the expectation of a problem.
Tampa, PL conducted extensive bioassay tests but failed to show
any adverse effect of water column concentrations of pollutants
in urban runoff. The screening analysis results presented in
Figure 7-6 indicate marginal problem conditions at low DAR for
this geographic region. At this project however, all monitored
sites show heavy metal concentrations significantly lower than
the low range conditions used in the screening analysis. When
7-20
-------
the screening analysis is repeated using site concentrations
representative of Tampa monitoring results, a problem situation
is not predicted, even at DAHs lower than is probably the case
for this location.
LAKES
Because lakes provide extended residence times for pollutants, the signifi-
cant time scale for evaluating urban runoff impacts is at least seasonal, and
usually annual or longer, rather than the storm event scale used for streams.
The screening methodology identified in Chapter 5, uses annual nutrient loads
to assess the tendency for development of undesirable eutrophication effects.
Figure 7-9 illustrates the effect of urban runoff on average lake phosphorus
concentration. The very significant influence of area ratio is evident. The
larger the urban area which drains into a lake of a given size, the greater
the annual loading, and the higher will be the lake phosphorus concentration
and the eutrophication effects produced.
The phosphorus concentrations characteristic of the urban sites surrounding a
particular lake are also seen to be significant. The three bands shown re-
flect the range of possibilities, based on the NURP data. The same basis is
used to estimate the phosphorus loads from average urban sites and those at
the higher and lower ends of site conditions, as was described for heavy
metals in the previous section. In this case, because it is annual mass
loads which are of interest, site median concentrations have been converted
to site mean values for use in the computations.
Lake phosphorus concentrations are also influenced by the annual runoff
volume (annual precipitation and runoff coefficient). The results illus-
trated are based on an annual rainfall of 30 inches and an overall average
runoff coefficient of 0.2. Plotted results may be scaled up or down in pro-
portion to the ratio between local values for these parameters and those used
in the illustration.
Finally, the lake morphology and hydrology influence the outcome; specific-
ally depth (H) and residence time (T) . This is reflected by the width of
each of the bands, which are based on a range of values for H/T (1 to 10)
estimated to be fairly typical for lakes in urban settings.
If an average lake phosphorus concentration of 20 yg/1 is used as a reference
concentration to assess the tendency for producing undesirable levels of bio-
stimulation, it is apparent that only lakes with rather small area ratios are
likely to be unaffected by urban runoff nutrient discharges. Since the three
bands represent different concentration levels of phosphorus in urban runoff,
qualitative inferences may be drawn concerning the beneficial use impacts of
control activities. More detailed estimates may of course be made by use of
the methodology with site specific parameters.
The salient feature of the situation, as generalized by the analysis sum-
marized by Figure 7-9, is that the problem potential of urban runoff for
lakes is quite site specific. The illustration considers only urban runoff
loads; in an actual situation, all nutrient sources (point and nonpoint)
7-21
-------
1000
3-
z
o
o
o
BE
O
o.
en
URBAN SITE QUALITY
CHARACTERISTICS
SITE MEAN TP CONCENTRATION )jg/l
HIGH RANGE
AVERAGE
CT1 LOW RANGE
ANNUAL RAINFALL = 30 inlyear
RUNOFF COEFFICIENT = 0.2
DEPTH/RESIDENCE
RATIO FOR LAKE
H|T= 1 to 10m(yr
SETTLING VELOCITY Vs = iflmlyr
(TOTAL P)
1000
RATIO
URBAN AREA
LAKE SURFACE AREA
Figure 7-9. Effect of Urban Runoff on Lake Phosphorus Concentrations
7-22
-------
would be considered, and this would tend to modify the relative significance
of urban runoff on lake conditions.
Several of the NURP projects addressed impacts on lake quality in some depth.
These projects include the following:
- Irondequoit Bay, NY - Lake has been highly eutrophic, due to
point and nonpoint discharges. Sewage treatment plant and com-
bined sewer overflow discharges have been removed, so that
residual sources are recycle from lake sediments and nonpoint
sources, including urban runoff, from the contributing drainage
area. Further reductions are considered necessary to meet tar-
gets. (Area ratio is high at this location.)
Lake George, NY - Lake is oligotrophic; the study addressed the
concern that urban runoff from present and potential future de-
velopment would unacceptably accelerate degradation of existing
water quality. (Area ratio is low at this location.)
- Lake Quinsigamond, MA - Urban runoff was determined to be one of
a number of sources preventing water quality objectives from
being met. Some control of urban runoff phosphorus loads was
recommended as one of the elements of an overall management
plan.
Each of the above situations is sufficiently unique, and the mix of urban
runoff and other load sources is sufficiently different to suggest that it is
inappropriate to attempt a broad generalization. The interested reader may
refer to the individual project documents which are available through NTIS
for more information.
ESTUARIES AND EMBAYMENTS
These water bodies are normally of sufficient size and complexity that simple
screening analyses have not been considered to be sufficiently useful or
effective to justify their use.
The Long Island, NY NURP project examined and confirmed that urban runoff
sources of coliform bacteria are the principal contributors to the water
column concentrations that result in closure of shellfish beds in a number of
embayments (principally the Great South Bay). Estimates of control activi-
ties that would allow the opening of presently closed areas were also made.
The reader is referred to the project documents for further information.
The significance of urban runoff and other nonpoint source loads on eutrophic
levels in the Potomac estuary is being investigated under a study which is
not associated with the NURP program. However, among other objectives of the
WASHCOG NURP project, estimates of urban nonpoint source loads have been de-
veloped to support this study.
Although specific situations where urban runoff is significant have been
identified, no general assessment for water bodies of this type can be made
at this time.
7-23
-------
GROUNDWATER AQUIFERS
Much of the precipitation which falls on an area either percolates directly
into the ground, or does so after relatively short overland flow distances.
This condition is essentially uncontrollable and distinctly different from
the case where urban runoff from impervious areas is deliberately collected
and routed to a recharge device which causes it to percolate to groundwaters.
This type of control approach is a practical and effective technique for re-
ducing pollutant loads which would otherwise reach surface waters as dis-
cussed in Chapter 8. The concern addressed here is with the extent to which
groundwater aquifers may be contaminated by this practice.
The Long Island, NY and Fresno, CA NURP projects examined this issue through
extensive tests utilizing recharge basins ranging from recent installations
to others which have been in service in excess of 20 years. A somewhat
simplified consolidation of the salient findings of these two projects is
presented below. The interested reader is referred to the individual project
report documents, available through NTIS, for the important details and
qualifications.
Most pollutants of importance in urban runoff are intercepted
during the process of infiltration and quite effectively
prevented from reaching the groundwater aquifers underlying
recharge basins. The pollutants tested and found to behave in
this manner include the heavy metals, an appreciable number of
the organic priority pollutants and pesticides, and coliform
bacteria.
Chlorides, which are sometimes present in urban runoff at
elevated concentrations due to road deicing practices, are not
attenuated during recharge.
Pollutants accumulate in the upper soil layers. The concen-
trations found are a function of the length of time a basin has
been in service. Effective retention of pollutants takes place
with all soil types tested, ranging from clays to sands. The
depth of pollutant penetration is affected by soil type; however
in no case did contaminant enrichment of soil exceed several
meters depth, and highest concentrations were found near the
surface.
The limit of the ability of the soil to retain the pollutants of
interest is unknown. Additional study of this aspect is appro-
priate. However given the long service periods of a number of
the recharge basins studied, this does not appear to represent
an imminent concern.
At both of these NURP locations, groundwater surfaces were at
least 20 feet, and often appreciably more, below the base of the
recharge device. The indicated findings may not be applicable
at locations with shallow depths to groundwater.
7-24
-------
No significant differences in interception/retention of
pollutants is apparent for basins with bare versus vegetated
recharge surfaces. However vegetation does apparently help to
maintain infiltration rates normal for the soil type.
Surface soil accumulations of priority pollutants in dual pur-
pose installations used for both recharge and recreational use
warrants further investigation to determine whether such prac-
tice creates unacceptable health risks or requires appropriately
designed and conducted maintenance procedures.
7-25
-------
CHAPTER 8
URBAN RUNOFF CONTROLS
INTRODUCTION
This chapter summarizes the information developed by the individual NURP
project studies relating to performance characteristics of selected tech-
niques for the control of urban runoff quality. The number of control
practices addressed here is considerably smaller than the array of best
management practices suggested in prior studies and publications. This is
not intended to exclude consideration of other approaches. However, the
techniques discussed in this chapter may be taken as an expression of con-
trols considered by the agencies involved to be potentially attractive and
practicable at localized planning levels. They represent the practices for
which performance data were obtained under the NURP program and which can be
analyzed and evaluated in this report.
Most of the NURP projects provide in their project reports a detailed
analysis and evaluation of the controls that were studied. These reports are
available through NTIS. In addition to this information source, an analysis
was performed by EPAs NURP headquarters team, using results available from
all project studies. The objective was to provide an overview and a generic
description of performance characteristics in a format considered to be
useful for planning activities. Thus, in addition to providing a consoli-
dated summary of project results, this chapter presents a summary of the
results of applying analysis methodologies developed under the NURP program.
Further detail on the former can be obtained by reference to relevant project
report documents; a more comprehensive development of the latter is provided
in separate NURP documents ("Detention and Recharge Basins for Control of
Urban Runoff Quality", and "Street Sweeping for Control or Urban Runoff
Quality").
The types of control techniques which received attention (to a greater or
lesser degree) in the NURP program can be grouped into four general
categories.
- Detention Devices - These include normally dry detention basins
typically designed for runoff quantity control, normally wet
detention basins, dual purpose basins, over-sized drain pipes,
and catchbasins.
Recharge Devices - These include infiltration pits, trenches,
and ponds; open-bottom galleries and catchbasins; and porous
pavements.
- Housekeeping Practices - These are principally street sweeping,
but also include sidewalk cleaning, litter containers, catch-
basin cleaning, etc.
8-1
-------
Other - These include the so-called "living filter" sipproaches,
grassed swales, wetlands, etc.
DETENTION DEVICES
General
Detention basins proved to be one of the most popular approaches to urban
runoff quality control selected at the local level, based on the number of
individual projects which elected to study them and the number of detention
devices tested in the study. It is perhaps instructive to note that nearly
all the detention facilities studied were either already in place, or re-
quired only modifications of outlet structures before initiation of the
NURP-supported studies. In general, detention devices proved to provide a
highly effective approach to control of urban runoff quality, although the
design concept has a significant bearing on performance characteristics.
Table 8-1 lists the NURP projects that included detention devices as elements
of their study program. Both the number of devices, and the number of storms
analyzed vary considerably, as indicated in Table 8-1, depending on project
priorities and other relevant activities. As a result, not all of the sites
are incorporated in the summary presented below. The Washington Area Council
of Governments (WASHCOG) conducted a particularly thorough and comprehensive
investigation of control techniques, particularly detention basins. They
have prepared several useful and informative analyses of performance results
on these devices.
Dry Basins
This is a type of detention basin which is currently in fairly extensive
service in various parts of the country. The performance objective of such
basins is commonly called "peak shaving", that is, to limit the maximum rate
of runoff to some preselected magnitude, usually a maximum pre-developmerit
rate. The purpose is to control flooding and erosion potential in areas
downstream of new development. Such basins employ a bottom outlet having a
hydraulic capacity restricted to the maximum allowable flow. Runoff from
smaller storms flows along the bottom of the basin and is discharged without
restriction. Flows in excess of design are backed up in the basin tempor-
arily and ponding occurs only during larger storms and for relatively short
periods of time. This class of retention basin is thus normally dry.
Performance of such basins, from a pollutant removal aspect, range from
insignificant to quite poor. Accordingly, the limited data available are not
discussed in this chapter.
Wet Basins
This designation covers detention basins which maintain a permanent pool of
water. They may vary considerably in appearance, ranging from natural
ponds or small lakes dedicated urban runoff control to enlarged sections in
3-2
-------
TABLE 8-1. DETENTION BASINS MONITORED BY NURP STUDIES
Project
C01 Denver
DC1 Washington, D.C.
IL2 N. Illinois
Mil Lansing
MI 3 Ann Arbor
NY1 Long Island
Site
North Ave
Burke
Lakeridge
Stedwick
Westleigh
Lake Ellyn
Dryer Farms
Grace St. N*
Grace St. S*
Waverly Hills
Pitt-AA
Traver
Swift Run
Unqua Pond
Design Type
Dry Basin
Wet Basin
Dry Basin
Dual-Purpose
Wet Basin
Wet Basin
Dry Basin
Wet Basin
Wet Basin
Wet Basin
Wet Basin
Wet Basin
Wet Basin
Wet Basin
No. Events
in/out
39/21
60/35
49/41
48/34
41/45
29/23
2/8
23/21
20/22
35/30
6/6
5/5
5/5
8/8
* These are oversized storm drains installed below street level. Inverts of
control sections are below the general grade line, so a permanent pool is
maintained.
constructed drainage systems. Runoff from an individual storm displaces all
or part of the prior volume, and the residual is retained until the next
storm event. This pattern may or may not be modified by natural base inflows
during dry weather depending on the local situation.
Detention basins utilizing this design concept have been shown by the NURP
studies to be capable of highly effective performance in urban runoff appli-
cations, as summarized below. Although performance characteristics of
individual basins ranged from poor to excellent, analysis shows these differ-
ences to be attributable to the size of the basin relative to the connected
urban area and local storm characteristics. Performance data also indicate
that in addition to removal of particulate forms or pollutants by sedimenta-
tion, some basins exhibit substantial reductions in soluble nutrients
(soluble phosphorus, nitrate + nitrite nitrogen). This is attributed to
biological processes which are permitted to proceed in the permanent water
pool.
8-3
-------
There are a number of ways to characterize detention basin performance. The
primary basis selected by NURP for doing so is to define performance effi-
ciency on the basis of the total pollutant mass removed over all storms.
This provides a meaningful general measure for comparison, is relevant for
water quality effects associated with extended time scales (e.g., nutrient
load impacts on lakes) , and conforms with the capabilities of the NURP
analysis methodology developed to provide a planning-level basis for esti-
mating cost/benefit differences in size or application density of this type
control.
Table 8-2 tabulates performance in terms of reduction in pollutant mass loads
over all monitored storm events. The analysis methodology developed under
the NURP program activities suggests that performance should be expected to
improve as the overflow rate (QR/A = mean runoff rate -5- basin surface area)
decreases and as the volume ratio (VB/VR = basin volume ~. mean runoff volume)
increases. The NURP basins used in the analysis are listed in increasing
order of expected performance capabilities.
The wide range of relative basin sizes provided by this data base is
apparent, and performance is seen to generally correspond with expectations.
The poorest performance occurs in a basin with an average overflow rate
during the mean storm of about six timess the median settling velocity
(1.5 ft/hr) of particles in urban runoff. In addition, less than 5 percent
of the mean storm runoff volume remains in this basin following the event, to
be susceptible to additional removal by quiescent settling during the
interval between storms. The basins which exhibit high removal efficiencies,
at the other end of the scale, have size relationships which result in the
mean storm displacing only about 10 percent of the available volume, and
producing overflow rates which are only a small fraction of the median
particle settling velocity.
This rationale is described more completely in the supporting NURP document
on detention basins identified earlier. The testing of the methodology
against the NURP monitoring data is presented, and the basis for the per-
formance projections illustrated below is documented.
Figure 8-1 presents a projection of removal efficiency of urban runoff de-
tention devices as a function of basin size relative to the contributing
catchment area and regional differences in typical rainfall patterns. The
removal rates apply for TSS, which are all settleable, and must be factored
by the particulate/soluble fraction of other pollutants which have signif-
icant soluble fractions in urban runoff. It applies for the specific basin
average depth and area runoff coefficient indicated (which are fairly typical
based on NURP data). However performance relationships could be different
than indicated based on relevant local values for the controlling parameters.
An alternate approach for characterizing performance of detention basins con-
centrates on the variable characteristics of individual storm events and how
these are modified by the detention device. A comparison of the mean and
coefficient of variation of basin inflow and discharge concentrations pro-
vides another measure of performance of an urban runoff detention device.
8-4
-------
c
tsi
-- EH
(J
C
0) ,Q
U PJ
M
0) EH
^ ,.
3
w u
1 e>
-P
CO ro
4-
T>
• H g
r^ ^
° K
1 — i P-l
1— 1 •
< rH
O
i co
rH
rfl
> CM
i *
0)
a
Q
W O
tfl U
(0
S
0) Q
tr> O
rO CQ
SH
>
<; co
CO
Ratios
VB/VR
cu <
N \
•rH «
CO OJ
tn
e
O H-I VH
ZOO
4J
CO
Project
and
Site
.~ -^ ro in
1 1 rH
CTl ^D C^ •
CN U3
"^ X
T CN
^— r^ ^ o
1 rH CM
-~ ro ^-~ ^D
1 CM i in
. — CM CO •*
1 rH rH ro
^- ~-~ ro in
1 1 (N rH
•31 rO rH *~~
rH (N 1
^— CN CN in
1 ro ro
in E^ ON] vD
O rH in rH
O O O rH
in r^ <^o o
r~ ro CO ro
CO fN rH O
CO CO vP in
rH rH
& CO
-P -P
CO CO >-| < ^
O < 0 V4
Cn 0) fji 0) X! 1 A
•H rrj -H to < 4-> rt (0
W M W in -H in
CO CO Cd, CEn
tO rd C C
J J ^~. r-
rH 1 CN
CTl • rH
CNJ r-
ro in TI*
<* in
CN ^^ in
t~- ro
II
0
O
EH
in O rH
CO k£> CO
rsl [-~ i-i
O O ro
rH ro in
o co in
CN O O
o o o
in co CN
ro
u
Q
C -O - X5
3 C C tn
VH PS ro O -H
O •-< 4-1 d)
X! 4-1 to ro tn rH
VH MH M 3 C 4-1
< -H tj1 -H in
S &> C £ 0)
C CO CD W g
C O TO
< J S
rH
in
r>
iri
S
o
o
r-
CTl
Oi
rH
r-
m
r--
o
o
en
CM
Lansing
Waverly Hills
rH
CO
rH
r-
•
•
•
ro
•
•
•sf
00
o
o
rH
o
rH
O
ro
CM
c
rH
rH
W
0)
CJ rO
PJ J
H
z
rC
0
§
•H
-M
m
Cn
OJ
q
c
0)
S-i
ft)
a
a,
w
0)
4-1
m
u
•H
•o
01
O
e
-P
ro
-P
3
w
Q)
rO
O
•rH
W
(U
4J
0
3-5
-------
OZ L90ZC8
4J
G
•H
U)
'D
O
5
M
QJ
M-l
C
o
Cr>
0)
i
00
•H
PM
oa
ssi dJONnu Nvaun do
30VH3AV WH31 9N01
3-6
-------
This approach provides more useful information for subsequently evaluating
the effect of controls on water quality impacts on rivers and streams. As
evident from the discussion in Chapter 6, reductions in the mean and vari-
ability of runoff concentrations (and the inferred reduction in mean and
variability of runoff rates) will have a significant beneficial effect on the
severity of impacts on flowing streams.
Table 8-3 summarizes detention basin performance when assessed in this
manner. It should be noted that in most cases more inlet storm events were
monitored than discharge events, and that some inlet events do not have a
matching discharge event and vice-versa. Further, for the larger basins
where storm inflow displaces only a fraction of the basin volume, it is
unlikely that influent and effluent for a specific event represent the same
volume of water. The tacit assumption in this analysis is that the inflow
events which were monitored provide a representative sample of the total
population of all influent event mean concentrations (EMCs), Similarly, the
monitored effluent events are assumed to be a representative sample of all
basin discharge EMCs. The appropriateness of this assumption is obviously
more uncertain where the number of individual storm events monitored is
small.
For each basin influent and effluent, the arithmetic mean and variance were
computed based on the relationships for lognormal distributions. The percent
reduction in the mean concentration and the coefficient of variation are
tabulated (Table 8-3). Note that where the number of monitored events shown
in this table differ from those listed in Table 8-2, it is because the mass
removal computations were restricted to synoptic storms (i.e., matching
influent and effluent results were available for an event).
Performance characteristics are generally consistent using either approach,
even though each displays a different type of information. Performance
improves with detention basin size relative to catchment size and hence the
magnitude of the runoff processed. Giving greater weight to the sites moni-
toring large numbers of storms, indications are that for most pollutants wet
ponds also generally result in a considerable reduction in the variability of
pollutant concentrations.
A significant exception to this tendency to reduce variability is shown for
the soluble nitrogen forms (NC>2 -t- NC^) . The positive removal efficiency
indicated by reduction of mean concentrations must be attributed to bio-
logical processes rather than sedimentation. A substantial increase in
variability is consistently indicated by the data. Among the heavy metals,
lead which is nearly all in particulate form shows significant reductions in
variability. Copper and zinc which have high (40 to 60 percent) soluble
fractions show an ambiguous pattern with regard to changes in variability.
In a few of the cases where atypical results are indicated, unique local
conditions suggest plausible explanations. For example, at the Ann Arbor
(Traver) site, erosion from an unstabilized bank at the outlet of this newly
constructed basin is attributed to the poor suspended solids removal ob-
served. The poor removal characteristics at the Unqua site for TKN and
nitrate may be associated with the significant wildfowl population at this
site.
-------
TABLE 8-3. OBSERVED PERFORMANCE OF WET DETENTION BASINS
(PERCENT REDUCTION IN POLLUTANT CONCENTRATIONS)
(a) Mean EMC
Project
and
Site
Lansing
Grace St. N.
Lansing
Grace St. S.
Ann Arbor
Pitt-AA
Ann Arbor
Traver
Ann Arbor
Swift Run
Long Island
Unqua
Washington, D.C.
Westleigh
Lansing
Waverly Hills
NIPC
Lake Ellyn
No.
of
Storms
(1)
23/20
18/17
6/6
5/5
5/5
8/8
40/40
35/30
25/20
Percent Reduction in Mean EMC
TSS
(6)
22
38
0
83
34
83
87
92
BOD
(26)
4
17
(66)
11
COD
15
(3)
23
12
(3)
(TOC=26)
52
33
52
64
TP
(10)
6
28
37
(38)
38
59
69
61
Sol.P
(26)
0
(2)
63
21
70
56
62
TKN
11
(5)
11
19
25
(31)
19
30
•
N02+3
(1)
(20)
8
28
77
(10)
28
54
82
T.Cu
(9)
25
•
•
•
10
53
88
T.Pb
39
14
59
•
86
78
•
93
91
T.Zn
(9)
7
22
19
10
58
87
(b) Coefficient of Variation of EMCs
Project
and
Site
Lansing
Grace St. N.
Lansing
Grace St. S.
Ann Arbor
Pitt-AA
Ann Arbor
Traver
Ann Arbor
Swift Run
Long Island
Unqua
Washington, D.C.
Westleigh
Lansing
Waverly Hills
NIPC
Lake Ellyn
No.
of
Storms
(1)
23/20
18/17
6/6
5/5
5/5
8/8
40/40
35/30
25/20
Percent Reduction in Coef of Variation of EKCs
TSS
14
(7)
17
14
(5)
(87)
46
38
44
BOD
49
(59)
(6)
(109)
39
(TOC=
•
5
•
COD
35
39
10
58
50
66)
(26)
69
41
TP
(7)
13
28
(3)
(150)
47
15
34
71
Sol.P
(13)
0
(84)
42
0
•
20
26
48
TKN
30
20
37
(150)
20
19
41
(8)
•
N02t3
0
21
0
(82)
(150)
(66)
(280)
(198)
(115)
T.Cu
0
17
•
•
•
•
0
(22)
60
T.Pb
45
18
53
•
26
65
•
34
19
T.Zn
(31)
15
(5)
0
•
•
(14)
(36)
41
Notes: (1) In/Out; numbers are approximate, and vary with pollutant. Removals in parentheses indicate
negative removal.
Dot (•) indicates pollutant either not monitored or number of observations is too small for
reliable estimate of percent reduction.
8-8
-------
The ability of detention basins to reduce coliform bacteria concentrations is
also of considerable interest because of the significant impact these urban
runoff contaminants exert on recreational or shellfish harvesting beneficial
uses. Other than at the Unqua site of the Long Island NURP project, the
number of observations made for indicator bacteria were too few to support a
reliable assessment of the ability of detention basins to effect quality
improvements. However, extensive data of this nature were secured on deten-
tion basin influent and effluent during all monitored storms at the Unqua
site.
Since coliform bacteria have a high rate of die-off in natural waters, per-
formance characteristics based on total mass reductions are not particularly
meaningful. The Unqua site data were analyzed to evaluate performance in
terms of reductions in concentration levels. Over eight monitored storms at
this site, covering a wide range in storm size, the mean EMC (MPN/100 ml) was
reduced by 94 percent for total coliform, 91 percent for fecal coliform, and
95 percent for fecal streptococcus bacteria. Variability of bacteria
concentrations in the pond outlet increased, with effluent coefficients of
variation ranging from about 10 to 100 percent greater than influents.
Accordingly, detention basins employing permanent pools (wet ponds) are
indicated to be capable of substantial reductions in indicator bacteria.
Dual Purpose Basins
In the absence of a well defined terminology, we have adopted this designa-
tion to define basins that are normally dry, and hence retain their full
potential for flood control, but which have outlet designs that result in a
slow release rate for detained storm flows. Detention time is extended
considerably compared with that provided by dry basins employing conventional
outlet designs.
One of the detention basins examined by the WASHCOG NURP project, was of this
type. This project designates such designs as "Extended Detention Dry
Ponds." The pond was converted from a conventional dry pond by replacing the
outlet pipe with a perforated riser enclosed in a gravel jacket. The modifi-
cation was designed to detain stormwater runoff for up to 24 hours, instead
of the 1 to 2 hours typically observed in conventional dry ponds.
For undetermined reasons, average detention periods during the study were in
the order of 4 to 8 hours, and hence considerably shorter than the design
objective. Nevertheless, based on monitoring of more than 30 storm events,
the removal of particulate forms of urban pollutants was typically high and
comparable to the performance efficiency of wet ponds.
Observed removals for this site (Stedwick) are summarized by Table 8-4,
showing percent reductions in both mass and concentration distributions. The
principal differences in performance of dual purpose basins compared with wet
basins are suggested by the available data to consist of the following:
- Soluble pollutants (e.g., soluble P and Nitrate/Nitrite) are not
effectively reduced because of the absence of a permanent pool
within which biological reactions have an opportunity to occur
in addition to sedimentation.
8-9
-------
- The variability of pollutant EMC's does not appear to be
modified to the extent that this occurs in wet ponds.
TABLE 8-4. PERFORMANCE CHARACTERISTICS OF A
DUAL-PURPOSE DETENTION DEVICE
(Stedwick Site - Washington Area NURP Project)
Pollutant
TSS
COD
Total P
Sol P
TKN
Organic N
N°2+3
T. Cu
T. Pb
T. Zn
Percent Reduction In
Pollutant Mass
Load Over All
Monitored Storms
64
30
< 15
1
•
30
10
•
84
57
Pol!
El
Mean
63
41
11
(4)
8
•
13
•
•
42
.utant
IC's
Coef Var
(31)
17
0
(13)
(11)
•
6
•
•
33
Although the performance characteristics of basins of this type are indicated
to be somewhat inferior to the potential offered by wet ponds, there are a
number of considerations which make dual purpose basins highly attractive
candidates for quality control of urban runoff. These include the fact that
flood control requirements are likely to be more economically obtained than
with wet basins and that many existing stormwater management basins may be
readily modified to significantly enhance their capability for improving the
quality of urban runoff. In areas where ordinances requiring conventional
stormwater management ponds are already in existence, the only changes
required would be an alternate specification of the outlet design.
Costs
The information presented here is intended to provide an order of magnitude
estimate of the cost of providing different levels of control of urban runoff
pollutant discharges, whan wet detention devices are used as the best manage-
ment practice (BMP). The summary is based on the size versus performance
relationship presented earlier in Figure 8-1 and on the size versus cost re-
lationships presented below.
8-10
-------
The analysis is based on cost information developed by the WASHCOG NURP
project and discussed in detail in one of their project reports produced for
the NURP effort. Construction cost estimates as a function of basin volume
are shown by Figure 8-2, adopted from this source. This estimate compares
quite favorably with a similar cost/size relationship developed previously by
the Soil Conservation Service (SCS).
The cost relationship shown by this figure applies to "dry pond" designs and
relates only to expected cost of construction activities. For specific cost
estimates, the results derived from Figure 8-~ should be modified as appro-
priate, in accordance with the following:
The highly variable capital cost of land acquisition is not
included in the construction costs.
Outlet modifications to provide a dual purpose basin design will
increase construction costs by about 10 to 12 percent.
Pond designs which meet the peak shaving requirements of con-
ventional (dry) pond designs, but also provide a permanent pool
of water may have costs up to 40 percent greater than indicated
by the cost relationship shown by Figure 8-2.
An additional allowance equal to 25 percent of construction
costs is suggested to allow for planning, design, administra-
tion, and construction related contingencies.
Operation and maintenance costs are estimated to involve an
annual expenditure of approximately 3 to 5 percent of base
construction cost, that is, before application of the 25 percent
factor for design, planning, and administration. The total is
composed of two elements: 2 to 3 percent of construction cost
estimates the annual cost of routine maintenance and upkeep; an
additional 1 to 2 percent of construction cost estimates the
annualized cost of sediment removal operations for a 10 year
clean-out cycle.
Planning agencies often distinquish between "on-site" controls, which are
applied to relatively small urban catchments, often installed by the
developer of an urban property, and "off-site" controls, which involve larger
basins and serve substantially larger urban drainage areas. Because of the
appreciable economy of scale inherent in the cost relationship defined by
Figure 8-2, this factor must be taken into account in developing cost/
performance summaries for urban runoff quality control using detention
basins. Accordingly, the control costs presented below for wet basin designs
indicate the differences based on the size of the urban catchment the basin
is designed to serve.
Figure 8-3 presents a planning level approximation of both present value and
annual cost of wet detention basins. Amoritization of costs is based on a
20 year basin life and an interest rate of 10 percent.
8-11
-------
IM90Z-EB
o
•P
u
V)
o
u
o
00
ID
i O
S-l 4->
O W
-P O
0) 0)
e e
0)
-P en
(0 (1)
O -H
•P -P
W CO
w
<0 tn
M O
0) U
CN
I
00
0)
^
P
&'
•H
(SUVHOQ 0861 '330) 1S03 NOIiOHUlSNOO
8-12
-------
9£ 190ZE8
V3uv Nvaun imasNNOo JO 3nav »3d mm «3d $
(WSO * NOIianuiSNOa a3ZlidOWV>
SNISV8 NOU33130 JO 1S03 1VHNNV
IS
LU }—
cc u
c
•H
W
c
o
O
W
C
-H
« CQ
C C
(0 O
5 3
i «
z££5;
SSig
D C
01
U-n1 -P
O 0)
Q
4-1
(0 4-1
O 0)
0 S
ro
I
cc
P
tjl
•H
fcj
Nvaun aaiasNNoa do 3tnv UBH $)
SiSOO 33NVN31NIVW S NOIlVUJdO
QN« HOI13n»iSN03 Q3Z11MOWV JO 3I11VA lN3S3Ud
3-13
-------
The performance levels associated with a particular basin size are shown at
the top of the plots as a range for long-term average removal efficiencies
for TSS. The range associated with a particular size reflects the regional
differences in performance which can be expected (Figure 8-1) as a result of
regional differences in storm characteristics. Approximate removal efficien-
cies for pollutants other than TSS can be estimated by factoring the indi-
cated TSS removal by the particulate fraction of the pollutant of interest.
The supplementary NURP document dealing with detention basins provides in-
formation to permit further refinement. A more concise local summary of
cost/performance relationships can be developed using the NURP data and
analysis methods, if local rainfall and land use characteristics, and design
and planning preferences are utilized.
The generalized relationships shown by Figure 8-3 can be summarized as
follows, if an urban catchment size of 20 to 40 acres is taken to represent a
typical "on-site" control application, and an "off-site" application is
reflected by detention basins serving 640 to 1000 acres.
Control
Application
On-site
Off-site
Approximate
Level of
Control
(% TSS Reduction)
50
90
50
90
Cost Per Acre of Urban Area
(Approximate)
Present
Value
$500 - $700
$1000 - $1500
$100
$250
Annua L
Cost
$60 - $80
$125 - $175
$10
$25
RECHARGE DEVICES
Control measures which enhance the infiltration of urban runoff are indicated
by the NURP studies to be techniques which are practical to apply and capable
of effective reductions in urban runoff quantity and quality. This finding
is based on project reports and on the results of a screening analysis using
a probabilistic methodology described in a supplementary NURP document on
detention basins.
The issue of the potential contamination of groundwater aquifers due to
enhanced infiltration of urban storm runoff has been discussed in the
previous chapter dealing with receiving water impacts. The favorable
findings support further consideration of this technique. At the same time,
it must be emphasized that specific local conditions may make recharge
inappropriate. Such conditions can include steep slopes, soil conditions,
depth to groundwater, and the proximity of water supply wells. Sound
planning and engineering judgement must be applied to determine the accept-
ability of this control approach in a local situation.
however, where local conditions premit, a wide variety of design concepts are
available for use. These range from off-site applications consisting of
8-14
-------
large retention basins, to small individual on-site units which include in-
filtration pits and trenches, percolating catch basins, and porous pavement.
The operating principle is the same regardless of size or design concept.
The important elements are the surface area provided for sub-surface perco-
lation and the storage volume of the device. Overall performance will be
related to the size of the recharge device relative to the urban catchment it
serves and the permeability (infiltration rate) of the soil.
The context in which the performance capabilities of recharge devices are
evaluated is the extent to which urban runoff is "captured" and prevented
from discharging directly to surface waters. Pollutant removals are reduced
in direct proportion to the runoff volume which is intercepted and recharged.
Load reductions will be further enhanced if quality improvements occur in the
portion of the runoff which is not captured. The combination of soil infil-
tration rate and percolating area provided determines the "treatment rate" of
a specific recharge device. When storm runoff is applied to the device at
rates of flow equal to or less than this rate, 100 percent of the runoff is
captured during that event. At higher applied rates, the fraction of the
runoff flow in excess of the treatment rate will escape and discharge to
surface waters.
Most recharge devices other than porous pavement also provide storage volume.
This improves performance capability because portions of the excess runoff
can be retained for subsequent percolation when applied rates subside. Over-
flow to surface water occurs only when the available storage is exceeded.
The Long Island and Metropolitan Washington, D.C. (WASHCOG) NURP projects
examined the performance of on-site recharge devices. An interconnected
system of percolating catch basins in Long Island was estimated to reduce
surface water discharges of storm runoff by more than 99 percent. The
WASHCOG project found that a porous pavement site produced pollutant load
reductions on the order of 85 to 95 percent depending on the specific
pollutant considered. An infiltration trench studied by this project
produced reductions in the order of 50 percent.
The NURP analysis methodology was employed in a screening analysis to assist
planning evaluations by establishing the relationship between performance
level and device size and soil percolation rates. Figure 8-4 presents a
planning level estimate of the influence of size, soil characteristics, and
regional rainfall differences on the performance of recharge devices.
The upper plot illustrates the significant effect regional differences in
rainfall characteristics can have on the performance of identical recharge
devices. Basin depth, soil percolation rate, and runoff coefficient for the
urban catchment are the same for each case. The performance differences
result from differences in the intensity and volume of the average storms in
each region. Basin size is represented on the horizontal axis by expressing
the percolation area that is provided as a percentage of the area of the
contributing urban catchment. For example, a recharge device with a perco-
lating surface area equal to 0.10 percent of an urban catchment represents a
design which provides (43,560 sq ft/acre x 0.10/100% =) 43.5 square feet of
8-15
-------
100
90
80
S 7D
ae
j 6°
3
g 5°
2
i «o
£
* 30
20
10
NORTHWEST
NORTHEAS
' /
OUTHEAST
AVERAGE DEPTH = S FT.
SOIL PERCOLATION RATE = 3 INCH/HOUR
RUNOFF COEF = 0 25
.01
0.05 .10 0.5 1.0
PERCOLATING AREA AS % OF CONTRIBUTING CATCHMENT AREA
5.0
GREAT LAKES PRECIP
MEAN C.V.
VOLUME 0.25 1.8
INTENS 0.05 1.4
DELTA 72 1.0
RV = 0.2
P - SOIL PERC RATE (WHRI
H = HASIN DEPTH (FEET)
O.OS .10 0.5 1.0
PERCOLATING AREA AS S OF CONTRIBUTING CATCHMENT AREA
Figure 8-4. Long Term Average Performance of Recharge Devices
8-16
-------
percolating surface area for each acre of urban catchment it serves. The
long-term average reductions in urban runoff volume and pollutant load which
can be expected will be approximately 35 percent in the southeast, 45 percent
in the northeast and 65 percent in the Pacific northwest.
The lower plot illustrates the much more significant influence of the amount
of storage volume provided (incidated by basin average depth), and the perme-
ability of the soil through which the storm runoff must percolate. The rain-
fall characteristics used in this -analysis are typical of the Great Lakes
region of the United States and are roughly comparable to those in the
northeastern part of the country. As might be expected, the permeability of
the soil in which the recharge device is constructed has a dominant influence
on performance capability. However significant compensation for low percola-
tion rates can be achieved by increases in percolation area and storage
volume.
When the screening analysis results are considered along with the favorable
results from the NURP studies, the NURP findings indicate that with a reason-
able degree of design flexibility to compensate for soils with lower percola-
tion rates, recharge devices provide a very effective method for control of
urban runoff.
STREET SWEEPING
End-of-pipe urban runoff pollutant concentrations have been commonly viewed
as being a function of two prime factors — accumulation of contaminants on
street surfaces and rainfall/runoff washoff. The postulated beneficial ef-
fect of street sweeping was to reduce contaminant accumulation. Prior to
NURP, emphasis of street sweeping investigations was placed on street surface
mechanisms (e.g., accumulation and washoff) and sweeper equipment performance
in removing street dirt. While these studies provided valuable insights into
the possible benefits of street sweeping, measurements of end-of-pipe concen-
trations are the only direct measures of street sweeping effectiveness in
water quality terms.
Recognizing this, NURP was designed to provide a large data base of urban
runoff water quality concentrations for both swept and unswept conditions.
In addition, the NURP street sweeping projects gathered and evaluated data on
atmospheric deposition (i.e., wetfall and dryfall), street surface accumula-
tion and washoff, and street sweeper removal rates and costs. The individual
project reports look at these other issues, and the results are not repeated
herein. Of prime interest and provided below is the effectiveness of street
sweeping in reducing end-of-pipe urban runoff pollutant concentrations (and
ultimately receiving water impacts). The findings presented below are based
upon the analyses performed by the individual projects, as well as other
statistical techniques, and are generally consistent with the projects'
conclusions.
3-17
-------
Five of the 28 NURP prototype projects had the evaluation of street sweeping
as a central element of their work plans. These projects were as follows:
Project Number of Sites
Castro Valley, CA 1
Milwaukee, WI 8
Champaign-Urbana, IL 4
Winston-Salem, NC 2
Bellevue, WA 2
Long Island, NY and Baltimore, MD also collected limited street sweeping
data. The experimental designs of the projects varied in detail, but essen-
tially followed either a paired basin or serial basin approach to gather test
and control data, with some projects using both approaches. The general
concept was that during a test period street sweeping would be more intensive
(up to daily) and thorough (e.g., with operator training, parking bans, etc.)
than during control periods when the streets were to be swept as usual or not
at all.
In the paired basin approach, two adjacent or close-by basins were operated
in a "control" or unswept mode for certain periods of time to establish a
baseline comparison, and then street sweeping was performed in a "test" basin
while the other remained as a control. The data provided an overall compari-
son between basins as well as a series of synoptic events for both basins.
In the serial approach, a basin was periodically operated in either a control
or test mode, with the periods adjusted so that all seasons of the year were
represented in each mode. Here, rather than synoptic data pairs, one has
data strings for both "swept" and "unswept" conditions.
There are no well established or prescribed procedures for evaluating the
possible reduction in runoff concentrations due to street sweeping. Issues
of concern include storm size and intensity effects, time since last rain,
ability to select truly paired basins, seasonal effects, etc. In an attempt
to sort out these issues, an exploratory data analysis was performed, and the
following findings were established:
Street sweeping has not been found to change the basic proba-
bility distribution of event mean concentrations. That is, the
fundamental assumption of random, lognormal behavior is valid
during sweeping operations.
The runoff quality characteristics of a basin during swept or
unswept conditions is best measured by the maximum likelihood
estimator of the median EMC, with the uncertainty indicated by
the 90 percent confidence interval of the median.
8-18
-------
There is in most cases no significant correlation (and in a few
cases a weak negative correlation) between EMCs and storm runoff
volume. EMCs and storm runoff intensities are also generally
uncorrelated (but in isolated cases exhibit a weak positive cor-
relation) . The implication of these findings is that differ-
ences in concentrations between swept and unswept conditions
will be largely unaffected by the size of the storms during the
monitoring periods. Because of this independence between con-
centration and volume, effects of sweeping on EMCs will also
indicate effects on mass pollutant loads.
EMCs for synoptic events on paired basins are, in general, not
significantly correlated or in some cases are weakly correlated;
however, over the longer term (e.g., mean, frequency distribu-
tion, etc.), there are no significant differences between the
distribution of EMCs of paired basins. These results show that
basins are independent from storm to storm, and thus, compari-
sons between basins should not be attempted using synoptic
events, but the basins do have similar statistical properties
and thus can be considered paired.
To evaluate the effectiveness of street sweeping, a series of bivariate plots
were constructed for projects using the serial basin approach. The site
median EMCs for swept and unswept conditions form the data pairs of the
plots. Bivariate plots are presented in Figure 8-5 for TSS, COD, TP, TKN,
and Pb concentrations, respectively. Each plot contains swept or unswept
conditions for multiple project sites. The assumption of the analysis is
that a large enough data base was collected to negate any temporal effects
such as seasonal, land use conditions, parking patterns, and other possible
factors (as rioted earlier, storm volume and intensity effects are not
believed to be significant). Examining the bivariate plots, it is observed
that, for the NURP data, the median concentrations are as likely to be
increased as decreased by street sweeping. Further, street sweeping never
produced a dramatic (e.g., >50 percent) reduction in concentrations (or
loads).
Street sweeping performance, as measured by the percent change in the site
median EMC, for selected NURP sites is graphically displayed in Figure 8-6.
The results are for five constituents (TSS, COD, TP, TKN, and Pb) at 10 sites
nationwide). For each site, the median EMC is based on data from between
10 and 60 events, with 30 events typical. Based on Figure 8-6 a number of
important observations are evident.
Performance as measured by change in site median EMC is highly
variable.
Where reductions occur, they generally occur for all
constituents.
Reductions never exceed 50 percent.
8-19
-------
(TSS Concentrations)
(TKN Concentrations)
4.0
30
20
100 200 300 400
UNSWEPT TSS Img/l)
1.0 2.0 3.0 4.0
UNSWEP1 TKN Img/l)
(COD Concentrations)
(Pb Concentrations)
50 100 150 200
UNSWEPT COD Img/l)
0.2 O.t 0.6 0.8
UNSWEPT Pb Img/l)
(TP Concentrations)
10
08
CL 06
a.
UJ
«> 04
02
0 02 04 OS 08 10
UNSWEPT TP Img/l)
Figure 8-5. Bivariate Plots of Median EMCs for
Swept and Unswept Conditions
8-20
-------
^«a^^«
^•^^•OBM
^^^•^^^
•I^H
•^^•••M
^••M
<••
^^H
•^^•^^^•1
•^•^^•^••1
••^^^^^^^
•^^^i^^^M
^^^l^^MM
•^•i^^H
^— —
^i^MB^I^^
^^H>
•^N^^^
•M^M^
^
STATE FAIR
WISCONSIN
RUSTLER
WISCONSIN
SURREY DOWNS
WASHINGTON
LAKE HILLS
WASHINGTON
RESIDENTIAL
NORTH CAROLINA
CBD
NORTH CAROLINA
MATTIS S
ILLINOIS
MATTIS N
ILLINOIS
JOHN ST. S
ILLINOIS
JOHN ST. N
ILLINOIS
Pb
TKN
TP
COD
TSS
NOI10nQ3d 3W3
Figure 8-6. Street Sweeping Performance
8-21
-------
In evaluating the results, it is critical that the uncertainty in the
estimate of median EMCs based on limited obs;erved data, and thus the uncer-
tainty in performance estimates, be assessed. This is especially true for
the cases of apparent increases in concentrations indicated by Figure 8-6.
For each of the 10 sites considered, the 90 percent confidence intervals of
the site median EMCs were computed as indicated in Figure 8-7. This analysis
indicates that there is generally no significant difference between median
EMCs for swept and unswept conditions. The implications of this analysis of
uncertainty are as follows:
- Based on statistical testing, no significant reductions in EMCs
are realized by street sweeping.
The indicated changes in site median EMCs (increases or
decreases) are much more likely due to random sampling than
actual effects of sweeping operations.
- Benefits of street sweeping (if any) are masked by the large
variability of the EMCs, therefore the benefit is certainly not
large (e.g., >50 percent), and an even larger site data base is
required to further identify the possible effect.
In the above context, the hypothesis that street sweeping
increases EMCs is generally not shown by the data, though it
could occur in isolated, site specific cases.
Urban runoff loads are the product of long term (e.g., annual) runoff volume
and event mean concentration. Under this definition, statements concerning
EMCs also hold for loads.
OTHER CONTROL APPROACHES
Several best management practices (BMPs) in addition to those discussed above
should be identified on the basis that local planning efforts determined them
to be practical to apply and to have the potential to provide significant
improvements in the quality characteristics of urban runoff. They are
grouped together in this section and discussed only briefly, principally
because, for one reason or another, sufficient data to characterize their
performance capabilities was not developed during the NURP program.
Grass Swales
Three grass swales were monitored by the Washington, D.C. area NURP project.
No significant improvement is urban runoff quality was indicated for pollut-
ants analyzed. Increases in zinc concentration which were observed were
attributed to mobilization of zinc from the galvanized culverts which carried
runoff under the driveways at the monitored residential sites. However the
project study report concluded that modifications which would increase
residence of runoff in the swales and enhance infiltration capability could
make this BMP effective for control of urban runoff.
5-22
-------
STREET SWEEPING PERFORMANCE-
SITE MEDIAN EMC
2 30D - =
COD. mg/l
=,*
E
-
03 % ffl L
.
^o
t S
< -I
LAK
WA
£8
V) t/3
ii
II
sg
o 06
i
TOT PB, ing/l
ri 2
<* OC
0°.
ccS
cnS
Figure 8-7. Effect of Street Sweeping on
Site Median EMC Values
8-23
-------
OM90Z-E8
0.8
S 0.6
i
UJ
U
0.2
Ss
J
ES
HJ£
coo
oz
LU o
TOT P, mgll
=5
SE z
0.8
06
0.4
0.2
*
E
LU
5
jj
E
JJ
XI
E
/>
Z
E
3
-J
«
to
LU
u
I ILLINOIS
c
y
i
i
c
1
(
1
i
*
-
n
u
u
c.
n
E
r
s
T
• LM*
•
u
Z CO
J2 co <2 <£
z t z i-
0 n \ Hr,
u hi 03 q3
-------
The Durham, New Hampshire NURP project monitored performance of a carefully
designed artificial swale which received runoff from a commercial parking
lot. Over 11 monitored storms, both soluble and particulate fractions of
heavy metals (Cu, Pb, Zn, and Cd) were reduced by approximately 50 percent.
Reductions in COD, nitrate, and ammonia were on the order of 25 percent. The
swale did not prove to be effective in reducing concentrations or organic
nitrogen, phosphorus, or bacterial species. It should be noted that the
performance capabilities indicated are based only on the concentration
changes produced in the stormwater which passes completely through the swale.
To the extent that infiltration of a portion of the runoff is effected by a
swale, load reductions would be increased in proportion.
The NURP results suggest that grass swales represent a practical and poten-
tially effective technique for control of urban runoff quality; that design
conditions are of major significance; and that additional study is necessary
to establish such parameters.
Wetlands
The potential of either natural or artificially created wetland areas to
effect favorable modification of urban runoff pollutant loads (particularly
sediment, nutrients, and heavy metals) has been widely suggested. The NURP
experience reinforces this expectation, but has not developed the detailed
performance data to permit either characterizing general performance capa-
bilities or identifying general design principles and parameters. Additional
study will be required to develop such information.
Miscellaneous
This category encompasses a variety of BMPs which were identified at the
local level as techniques of quality control which appeared to be relevant
for the circumstances which were operative. They are grouped under this
category because (a) their applicability tends to be site-specific rather
than general, and (b) while their effectiveness as a BMP may be substantial
on a relatively small spatial scale, the broad-scale effect on urban runoff
loads has not been possible to document.
BMPs in this category include erosion control practices and urban house-
keeping practices. As an example of the former, the Little Rock, Arkansas
NURP project widened and stabilized (with rip rap) a segment of an urban
stream to reduce erosion potential. The Baltimore NURP project data clearly
indicated the substantial difference in urban runoff quality that can result
from the general level of cleanliness maintained in an urban neighborhood.
3-25
-------
CHAPTER 9
CONCLUSIONS
INTRODUCTION
The Nationwide Urban Runoff Program has addressed such issues as quantifying
the characteristic of urban runoff, assessing the water quality effects on
receiving water bodies attributable to urban runoff discharges, and examining
the effectiveness of control practices in removing the pollutants found in
urban runoff. This chapter summarizes NURP's conclusion relating to these
issues and is based on the results presented in Chapters 6, 7, and 8 of this
report. Conclusions reached by the individual NURP projects are also pre-
sented to further support the results of the national level analysis.
URBAN RUNOFF CHARACTERISTICS
General
Field monitoring was conducted to characterize urban runoff flows and pol-
lutant concentrations. This was done for a variety of pollutants at a sub-
stantial number of sites distributed throughout the country. The resultant
data represent a cross-section of regional climatology, land use types,
slopes, and soil conditions and thereby provide a basis for identifying pat-
terns of similarities or differences and testing their significance.
Urban runoff flows and concentrations of contaminants are quite variable.
Experience shows that substantial variations occur within a particular event
and from one event to the next at a particular site. Due to the high vari-
ability of urban runoff, a large number of sites and storm events were moni-
tored, and a statistical approach was used to analyze the data. Procedures
are available for characterizing variable data without requiring knowledge of
or existence of any underlying probability distribution (nonparametric
statistical procedures). However, where a specific type of probability dis-
tribution is known to exist, the information content and efficiency of sta-
tistical analysis is enhanced. Standard statistical procedures allowed
probability distributions or frequency of occurrence to be examined and
tested. Since the underlying distributions were determined to be adequately
represented by the lognormal distribution, the log (base e) transforms of all
urban runoff data were used in developing the statistical characterizations.
The event mean concentration (EMC), defined as the total constituent mass
discharge divided by the total runoff volume, was chosen as the primary water
quality statistic. Event mean concentrations were based on flow weighted
composite samples for each event at each site in the accessible data base.
EMCs were chosen as the primary water quality characteristic subjected to
detailed analysis, even though it is recognized that mass loading character-
istics of urban runoff (e.g., pounds/acre for a specified time interval) is
9-1
-------
ultimately the relevant factor in many situations. The reason is that,
unlike EMCs, mass loadings are very strongly influenced by the amount of
precipitation and runoff, and estimates of typical annual mass loads will be
biased by the size of monitored storm events. The most reliable basis for
characterizing annual or seasonal mass loads is on the basis of EMC and
site-specific rainfall/runoff characteristics.
Establishing the fundamental distribution as lognormal and the availability
of a sufficiently large population of EMCs to provide reliability to the
statistics derived has yielded a number of benefits, including the ability to
provide:
Concise summaries of highly variable data
- Meaningful comparisons of results from different sites, events,
etc.
Statements concerning frequency of occurrence. One can express
how often values will be expected to exceed various magnitudes
of interest.
A more useful method of reporting data than the use of ranges;
one which is less subject to misinterpretation
- A framework for examining "transferability" of data in a quanti-
tative manner
Conclusions
1. Heavy metals (especially copper, lead and zinc) are by far the most pre-
valent priority pollutant constituents found in urban runoff. End-of-pipe
concentrations exceed EPA ambient water quality criteria and drinking
water standards in many instances. Some of the metals are present often
enough and in high enough concentrations to be potential threats to bene-
ficial uses.
All 13 metals on EPA's priority pollutant list were detected in urban
runoff samples, and all but three at frequencies of detection greater
than 10 percent. Most often detected among the metals were copper, lead,
and zinc, all of which were found in at least 91 percent of the samples-
Metal concentrations in end-of-pipe urban runoff samples (i.e., before
dilution by receiving water) exceeded EPA's water quality criteria and
drinking water standards numerous times. For example, freshwater acute
criteria were exceeded by copper concentrations in 47 percent of the
samples and by lead in 23 percent. Freshwater chronic exceedances were
common for lead (94 percent), copper (82 percent), zinc (77 percent), and
cadmium (48 percent). Regarding human toxicity, the most significant
pollutants were lead and nickel, and for human carcinogenesis, arsenic
and beryllium. Lead concentrations violated drinking water criteria in
73 percent of the samples.
9-2
-------
It should be stressed that the exceedances noted above do not necessarily
imply that an actual violation of standards will exist in the receiving
water body in question. Rather, the enumeration of exceedances serves a
screening function to identify those heavy metals whose presence in urban
runoff warrants high priority for further evaluation.
Based upon the much more extensive NURP data set for total copper, lead,
and zinc, the site median EMC values for the median urban site are: Cu =
34 pg/1, Pb = 144 ng/1, and Zn = 16° yg/1. For the 90th percentile urban
site the values are: Cu = 93 ug/lr Pb = 350 yg/1, and Zn = 500 yg/1.
These values are suggested to be appropriate for planning level screening
analyses where data are not available.
Some individual NURP project sites (e.g., at DC1, MD1, NH1) found unus-
ually high concentrations of certain heavy metals (especially copper and
zinc) in urban runoff. This was attributed by the projects to the effect
of acid rain on materials used for gutters, culverts, etc.
2. The organic priority pollutants were detected less frequently and at
lower concentrations than the heavy metals.
Sixty-three of a possible 106 organics were detected in urban runoff
samples. The most commonly found organic was the plasticizer bis
(2-ethylhexl) phthalate (22 percent), followed by the pesticide
a-hexachlorocyclohexane (a-BHC) (20 percent). An additional 11 organic
pollutants were reported at frequencies between 10 and 20 percent;
3 pesticides, 3 phenols, 4 polycyclic aromatics, and a single halogenated
aliphatic.
Criteria exceedances were less frequently observed among the organics
than the heavy metals. One unusually high pentachlorophenol concentra-
tion of 115 yg/1 resulted in exceedances of the freshwater acute and
organoleptic criteria. This observation and one for chlordane also ex-
ceeded the freshwater acute criteria. Freshwater chronic criteria
exceedances were observed for pentachlorophenol, bis (2-ethylhexyl)
phthalate, gamma-BHC, chlordane, and alpha-endosulfan. All other organic
exceedances were in the human carcinogen category and were most serious
for alpha-hexachlorocyclohexane (alpha-BHC), gamma-hexachlorocyclohexane
(gamma-BHC or Lindane), chlordane, phenanthrene, pyrene, and chrysene.
The fact that the NURP priority pollutant monitoring effort was limited
to two samples at each site leaves us unable to make many generalizations
about those organic pollutants which occurred only rarely. We can spec-
ulate that their occurrences tend to be very site specific as opposed to
being a generally widespread phenomena, but much more data would be re-
quired to conclusively prove this point.
3. Coliform bacteria are present at high levels in urban runoff and can be
expected to exceed EPA water quality criteria during and immediately
after storm events in many surface waters, even those providing high
degrees of dilution.
9-3
-------
Fecal coliform counts in urban runoff are typically in the tens to hun-
dreds of thousand per 100 ml during warm weather conditions, with the
median for all sites being around 21,000/100 ml. During cold weather,
fecal coliform counts are more typically in the 1,000/100 ml range, which
is the median for all sites. Thus, violations of fecal coliform stand-
ards were reported by a number of NURP projects. High fecal coliform
counts may not cause actual use impairments, in some instances, due to
the location of the urban runoff discharges relative to swimming areas or
shellfish beds and the degree of dilution/dispersal arid rate of die off.
The same is true of total coliform counts, which were found to exceed EPA
water quality criteria in undiluted urban runoff at virtually every site
every time it rained.
The substantial seasonal differences noted above do riot correspond with
comparable variations in urban activities. The NURP analyses as well as
current literature suggest that fecal coliform may not be the most
appropriate indicator organism for identifying potential health risks
when the source is stormwater runoff.
4. Nutrients are generally present in urban runoff, but with a few individ-
ual site exceptions, concentrations do not appear to be high in compari-
son with other possible discharges to receiving water bodies.
NURP data for total phosphorus, soluble phosphorus, total kjeldahl nitro-
gen, and nitrate plus nitrite as nitrogen were carefully examined. Me-
dian site EMC median concentrations in urban runoff were TP = 0.33 mg/1,
SP = 0.12 mg/1, TKN =1.5 mg/1, and N02+3 - N = 0.68 mg/1. On an annual
load basis, comparison with typical monitoring data, literature values,
and design objectives for discharges from a well run secondary treatment
plant suggests that mean annual nutrient loads from urban runoff are
around an order of magnitude less than those from a POTW.
5. Oxygen demanding substances are present in urban runoff at concentrations
approximating those in secondary treatment plant discharges. If dis-
solved oxygen problems are present in receiving waters of interest, con-
sideration of urban runoff controls as well as advanced waste treatment
appears to be warranted.
Urban runoff median site EMC median concentrations of 9 mg/1 BOD5 and
65 mg/1 COD are reflected in the NURP data, with 90th percentile site EMC
median values being 15 mg/1 BODS and 140 mg/1 COD. These concentrations
suggest that, on an annual load basis, urban runoff is comparable in mag-
nitude to secondary treatment plant discharges.
It can be argued that urban runoff is typically well oxygenated and
provides increased stream flow and, hence, in view of relatively long
travel times to the critical point, that dissolved oxygen problems
attributable solely to urban runoff should not be widespread occurrences.
No NURP project, specifically identified a low DO condition resulting from
9-4
-------
urban runoff. Nonetheless, there will be some situations where con-
sideration of urban runoff controls for oxygen demanding substances in an
overall water quality management strategy would seem appropriate.
6. Total suspended solids concentrations in urban runoff are fairly high in
comparison with treatment plant discharges. Urban runoff control is
strongly indicated where water quality problems associated with TSS, in-
cluding build-up of contaminated sediments, exist.
There are no formal water quality criteria for TSS relating to either
human health or aquatic life. The nature of the suspended solids in
urban runoff is different from those in treatment plant discharges, being
higher in mineral and man-made products (e.g., tire and street surface
wear particles) and somewhat lower in organic particulates. Also, the
solids in urban runoff are more likely to have other contaminants
adsorbed onto them. Thus, they cannot be simply considered as benign,
nor do they only pose an aesthetic issue. NURP did not examine the
problem of contaminated sediment build-up due to urban runoff, but it
undeniably exists, at least at some locations.
The suspended solids in urban runoff can also exert deleterious physical
effects by sedimenting over egg deposition sites, smothering juveniles,
and altering benthic communities.
On an annual load basis, suspended solids contributions from urban runoff
are around an order of magnitude or more greater than those from second-
ary treatment plants. Control of urban runoff, as opposed to advanced
waste treatment, should be considered where TSS-associated water quality
problems exist.
7. A summary characterization of urban runoff has been developed and is
believed to be appropriate for use in estimating urban runoff pollutant
discharges from sites where monitoring data are scant or lacking, at
least for planning level purposes.
As a result of extensive examination, it was concluded that geographic
location, land use category (residential, commercial, industrial park, or
mixed), or other factors (e.g., slope, population density, precipitation
characteristics) appear to be of little utility in consistently explain-
ing overall site-to-site variability in urban runoff EMCs or predicting
the characteristics of urban runoff discharges from unmonitored sites.
Uncertainty in site urban runoff characteristics caused by high event-
to-event variability at most sites eclipsed any site-to-site variability
that might have been present. The finding that EMC values are essen-
tially not correlated with storm runoff volumes facilitates the transfer
of urban runoff characteristics to unmonitored sites. Although there
tend to be exceptions to any generalization, the suggested summary urban
runoff characteristics given in Table 6-17 of the report are recommended
for planning level purposes as the best estimates, lacking local informa-
tion to the contrary.
9-5
-------
RECEIVING WATER EFFECTS
General
The effects of urban runoff on receiving water quality are highly site-
specific. They depend on the type, size, and hydrology of the water body;
the urban runoff quantity and quality characteristics; the designated bene-
ficial use; and the concentration levels of the specific pollutants that
affect that use.
The conclusions which follow are based on screening analyses performed by
NURP, observations and conclusions drawn by individual NURP projects that
examined receiving water effects in differing levels of detail and rigor, and
NURP's three levels of problem definition. Conclusions are organized on the
basis of water body type: rivers and streams, lakes, estuaries and embay-
ments, and groundwater aquifers. Site-specific exceptions should be
expected, but the statements presented are believed to provide an accurate
perspective on the general tendency of urban runoff to contribute signifi-
cantly to water quality problems.
Rivers and Streams
I. Frequent exceedances of heavy metals ambient water quality criteria for
freshwater aquatic life are produced by urban runoff.
The Denver NURP project found that in-stream concentrations of copper,
lead, zinc, and cadmium exceeded State ambient water quality standards
for the South Platte River during essentially all storm events.
NURP screening analyses suggest that frequent exceedances of both EPA
24-hour and maximum water quality criteria for heavy metals should be
expected on a relatively general basis.
2. Although a significant number of problem situations could result from
heavy metals in urban runoff, levels of freshwater aquatic life use
impairment suggested by the magnitude and frequency of ambient criteria
exceedances were not observed.
Based upon the magnitude and frequency of freshwater aquatic life ambient
criteria exceedances, one would expect to observe impairment of this
beneficial use in most streams that receive urban runoff discharges.
However, those NURP project studies which examined this issue did not
report significant use impairment problems associated with urban runoff.
The Bellevue, Washington NURP project concluded that toxic effects of
urban runoff pollutants did not appear to be a significant factor.
The Tampa, Florida NURP project conducted biological studies of the
impact of stormwater runoff upon the biological community of the
Hillsborough River. They conducted animal bioassay experiments on five
sensitive species in two samples of urban runoff from the Arctic Street
drainage basin. Thirty-two bioassay experiments were completed including
22 acute tests and 10 chronic tests. Neither sample of stormwater was
acutely toxic to test organisms. Long-term chronic experiments were
9-6
-------
undertaken with two species and resulted in no significant effects attri-
butable to stormwater exposure.
NURP screening analyses suggest that the potential of urban runoff to
seriously impair this beneficial use will be strongly influenced by local
conditions and the frequency of occurrence of concentration levels which
produce toxic effects under the intermittent, short duration exposures
typically produced by urban runoff.
While the application of the screening analysis to the Bellevue and Tampa
situations supports the absence of a problem situation in these cases, it
also suggests that a significant number of problem situations should be
expected. Therefore, although not the general, ubiquitous problem situa-
tion that criteria exceedances would suggest, there are site-specific
situations in which urban runoff could be expected to cause significant
impairment of freshwater aquatic life uses.
Because of the inconsistency between criteria exceedances and observed
use impairments due to urban runoff, adaptation of current ambient
quality criteria to better reflect use impacts where pollutant exposures
are intermittent and short duration appears to be a useful area for
further investigation.
3. Copper, lead and zinc appear to pose a significant threat to aquatic life
uses in some areas of the country. Copper is suggested to be the most
significant of the three.
Regional differences in surface water hardness, which has a strong influ-
ence on toxicity, in conjunction with regional variations in stream flow
and rainfall result in significant differences in susceptibility to ad-
verse impacts around the nation.
The southern and southeastern regions of the country are the most sus-
ceptible to aquatic life effects due to heavy metals, with the northeast
also a sensitive area, although somewhat less so.
Copper is the major toxic metal in urban runoff, with lead and zinc also
prevalent but a problem in more restricted cases. Copper discharges in
urban runoff are, in all but the most favorable cases, a significant
threat to aquatic life uses in the southeast and southern regions of the
country. In the northeast, problems would be expected only in rather
unfavorable conditions (large urban area contribution and high site con-
centrations) . In the remainder of the country (and for the other metals)
problems would only be expected under quite unfavorable site conditions.
These statements are based on total metal concentrations.
4. Organic priority pollutants in urban runoff do not appear to pose a gen-
eral threat to freshwater aquatic life.
This conclusion is based on limited data on the frequency with which or-
ganics are found in urban runoff discharges and measured end-of-pipe con-
centrations relative to published toxic criteria. One unusually
high pentachlorophenol concentration of 115 yg/1 resulted in the only
exceedance of the organoleptic criteria. This observation and one for
9-7
-------
chlordane exceeded the freshwater acute criteria. Freshwater
chronic criteria exceedances were observed for pentochlorophenol,
bis (2-ethylhexyl) phlhalate, Y~hexachlorocyclohexane (lindane),
a-endosulfan, and chlordane.
5. The physical aspects of urban runoff, e.g., erosion and scour, can be a
significant cause of habitat disruption and can affect the type of
fishery present. However, this area was studied only incidentally by
several of the projects under the NURP program and more concentrated
study is necessary.
The Metropolitan Washington Council of Governments (MWCOG) NURP project
did an analysis of fish diversity in the Seneca Creek Watershed, 20 miles
northwest of Washington, D.C. In this study, specific changes in fishery
diversity were identified due to urbanization in some of the sub-
watersheds. Specifically, the number of fish species present are reduced
and the types of species present changed dramatically, e.g., environ-
mentally sensitive species were replaced with more tolerant species. For
example, the Blacknose Dace replaced the Mottled Sculpin. MWCOG con-
cluded that the changes in fish diversity were due to habitat deteriora-
tion caused by the physical aspects of urban runoff.
The Bellevue, Washington NURP project concluded that habitat changes
(streambed scour and sedimentation) had a more significant effect than
pollutant concentrations, for the changes produced by urbanization.
6. Several projects identified possible problems in the sediments because of
the build-up of priority pollutants contributed wholly or in part by
urban runoff. However, the NURP studies in this area were few in number
and limited in scope, and the findings must be considered only indicative
of the need for further study, particularly as to long-term impacts.
The Denver NURP project found significant quantities of copper, lead,
zinc, and cadmium in river sediments. The Denver Regional Council of
Governments is concerned that during periods of continuous low flow, lead
may reach levels capable of adversely affecting fish.
The Milwaukee NURP project reported the observation of elevated levels of
heavy metals, particularly lead, in the sediments of a river receiving
urban runoff.
7. Coliform bacteria are present at high levels in urban runoff and can be
expected to exceed EPA water quality criteria during and immediately
after storm events in most rivers and streams.
Violations of the fecal coliform standard were reported by a number of
NURP projects. In some instances, high fecal coliform counts may not
cause actual use impairments due to the location of the urban runoff
discharge relative to swimming areas and the degree of dilution or dis-
persal and rate of die off.
Coliform bacteria are generally accepted to be a useful indicator of the
possible presence of human pathogens when the source of contamination is
sanitary sewage. However, no such relationship has been demonstrated for
9-8
-------
urban runoff. Therefore, the use of coliforms as an indicator of human
health risk when the sole source of contamination is urban runoff, war-
rants further investigation.
8. Domestic water supply systems with intakes located on streams in close
proximity to urban runoff discharges are encouraged to check for priority
pollutants which have been detected in urban runoff, particularly those
in the organic category.
Sixty-three of a possible 106 organics were detected in urban runoff sam-
ples. The most commonly found organic was the plasticizer bis
(2-ethylhexl) phthalate (22 percent), followed by the pesticide
a-hexachlorocyclohexane (a-BHC) (20 percent). An additional 11 organic
pollutants were reported at frequencies between 10 and 20 percent;
3 pesticides, 3 phenols, 4 polycyclic aromatics, and a single halogenated
aliphatic.
Lakes
1. Nutrients in urban runoff may accelerate eutrophication problems and
severely limit recreational uses, especially in lakes. However, NURP's
lake projects indicate that the degree of beneficial use impairment
varies widely, as does the significance of the urban runoff component.
The Lake Quinsigamond NURP project in Massachusetts identified eutrophi-
cation as a major problem in the lake, with urban runoff being a prime
contributor of the critical nutrient phosphorus. Point source discharges
to the lake have been eliminated almost entirely. However, in spite of
the abatement of point sources, survey data indicate that the lake has
shown little improvement over the abatement period. In particular, the
trophic status of the lake has shown no change, i.e., it is still clas-
sified as late mesotrophic-early eutrophic. Substantial growth is pro-
jected in the basin, and there is concern that Lake Quinsigamond will
become more eutrophic. A proposed water quality management plan for the
lake includes the objective of reducing urban runoff pollutant loads.
The Lake George NURP project in New York State also identified increasing
eutrophication as a potential problem if current development trends con-
tinue. Lake George is not classified as eutrophic, but from 1974 to 1978
algae production in the lake increased logarithmically. Lake George is a
very long lake, and the limnological differences between the north and
south basins provide evidence of human impact. The more developed,
southern portion of the lake exhibits lower transparencies, lower hypo-
limnetic dissolved oxygen concentrations, higher phosphorus and chlor-
ophyll ja concentrations, and a trend toward seasonal blooms of blue-green
algae. These differences in water quality indicators are associated with
higher levels of cultural activities (e.g., increased sources of phos-
phorus) in the southern portion of the lake's watershed, and continued
development will tend to accentuate the differences.
9-9
-------
The Lake George NURP project estimated that urban runoff from developed
areas currently accounts for only 13.6 percent of the annual phosphorus
loadings to Lake George as a whole. In contrast, developed areas con-
tribute 28.9 percent of the annual phosphorus load to the NURP study
areas at the south end of the Lake. Since there are no point source
discharges, this phosphorus loading is due solely to urban runoff. These
data illustrate the significant impact of urbanization on phosphorus
loads.
The NURP screening analysis suggests that lakes for which the contribu-
tions of urban runoff are significant in relation to other nonpoint
sources (even in the absence of point source discharges) are indicated to
be highly susceptible to eutrophication and that urban runoff control may
be warranted in such situations.
Coliform bacteria discharges in urban runoff have a significant negative
impact on the recreational uses of lakes.
As was the case with rivers and streams, coliform bacteria in urban run-
off can cause violations of criteria for the recreational use of lakes.
When unusually high fecal coliform counts are observed, they may be par-
tially attributable to sanitary sewage contamination, in which case
significant health risks may be involved.
The Lake Quinsigamond NURP project in Massachusetts found that bacterial
pollution was widespread throughout the drainage basin. In all cases
where samples were taken, fecal coliforms were in excess of 10,000 counts
per 100 ml, with conditions worse in the Belmont street storm drciins.
This project concluded that the very high fecal coliform counts in their
stormwater are at least partially due to sewage contamination apparently
entering the stormwater system throughout the local catchment.
The sources of sewage contamination are leaking septic tanks, infiltra-
tion from sanitary sewers into storm sewers, and leakage at manholes. In
the northern basin, the high fecal coliform counts are attributed to
known sewage contamination sources on Poor Farm Brook. The data from the
project suggest that it would be unwise to permit body contact recreation
in the northern basin of the lake during or immediately following signif-
icant storm events. The project concluded that disinfection at selected
storm drains should be considered in the future, especially if the sewage
contamination cannot be eliminated.
The Mystic River NURP project in Massachusetts found various areas where
fecal coliform counts were extremely high in urban stormwater. Fecal
coliform levels of up to one million with an average of 178,000/100 ml
were recorded in Sweetwater Brook, a tributary to Mystic River, during
wet weather. These high fecal coliform levels were specifically attrib-
uted to surcharging in their sanitary sewers, which caused sanitary
sewage to overflow into their storm drains via the combined manholes
present in this catchment. Fecal coliform levels above the class B fecal
coliform standard of 200 per 100 ml were found in approximately one-third
of the samples tested in the upper and lower forebays of the Upper Mystic
Lake and occasionally near the lake's outlet. In addition, Sandy Beach,
a public swimming area on Upper Mystic Lake, exceeded the State fecal
9-10
-------
coliform criteria in July of 1982, and warnings that swimming may be haz-
ardous to public health were posted for several days. It is important to
note that sewage contamination of surface waters is a major problem in
the watershed. The project concluded that urban runoff contributes to
the bacteria load during wet weather but, comparatively, is much less
significant than the sanitary sources.
Estuaries and Embayments
1. Adverse effects of urban runoff in marine waters will be a highly speci-
fic local situation. Though estuaries and embayments were studied to a
very limited extent in NURP, they are not believed to be generally
threatened by urban runoff, though specific instances where use is im-
paired or denied can be of significant local and even regional impor-
tance. Coliform bacteria present in urban runoff is the primary
pollutant of concern, causing direct impacts on shellfish harvesting and
beach closures.
The significant impact of urban runoff on shellfish harvesting has been
well documented by the Long Island, New York NURP project. In this proj-
ect, stormwater runoff was identified as the major source of bacterial
loading to marine waters and, thus, the indirect cause of the denial of
certification by the New York State Department of Conservation for about
one-fourth of the shellfishing area. Much of this area is along the
south shore, where the annual commercial shellfish harvest is valued at
approximately $17.5 million.
The Myrtle Beach, South Carolina NURP project found that stormwater dis-
charges from the City of Myrtle Beach directly onto the beach showed high
bacterial counts for short durations immediately after storm events. In
many instances these counts violated EPA water quality criteria for aqua-
tic life and contact recreation. The high bacteria counts, however, were
associated with standing pools formed at the end of collectors for brief
periods following the cessation of rainfall and before the runoff perco-
lated into the sand. Consequently, the threat to public health was not
considered great enough to warrant closure of the beach.
Groundwater Aquifers
1. Groundwater aquifers that receive deliberate recharge of urban runoff do
not appear to be imminently threatened by this practice at the two loca-
tions where it was investigated.
Two NURP projects (Long Island and Fresno) are situated over sole source
acquifers. They have been practicing recharge with urban runoff for two
decades or more at some sites, and extensively investigated the impact of
this practice on the quality of their groundwater. They both found that
soil processes are efficient in retaining urban runoff pollutants quite
close to the land surface, and concluded that no change in the use of
recharge basins is warranted.
Despite the fact that some of these basins have been in service for rela-
tively long periods of time and pollutant breakthrough of the upper soil
9-11
-------
layers has not occurred, the ability of the soil to continue to retain
pollutants is unknown. Further attention to this issue is recommended.
CONTROL EFFECTIVENESS
General
A limited number of techniques for the control of urban runoff quality were
evaluated by the NURP program. The set is considerably smaller than prev-
iously published lists of potential management practices. Since the control
approaches that were investigated were selected at the local level, the
choices may be taken as an initial indication of local perceptions regarding
practicality and feasibility from the standpoint of implementation.
Conclusions
1. There is a strong preference for detention devices, street sweeping, and
recharge devices as reflected by the control measures selected at the
local level for detailed investigation. Interest was also shown in grass
swales and wetlands.
Six NURP projects monitored the performance of a total of 14 detention
devices. Five separate projects conducted in-depth studies of the
effectiveness of street sweeping on the control of urban runoff quality.
A total of 17 separate study catchments were involved in this effort.
Three NURP projects examined either the potential of recharge devices to
reduce discharges of urban runoff to surface waters or the potential of
the practice to contaminate groundwaters. A total of 12 separate sites
were covered by this effort.
Grass swales were studied by two NURP projects. Two swales in existing
residential areas, and one experimental swale constructed to serve a com-
mercial parking lot were studied.
A number of NURP projects indicated interest in wetlands for improvir-g
urban runoff quality at early stages of the program. Only one allocated
monitoring activity to this control measure, however.
Various other management practices were identified as having local inter-
est by individual NURP projects, but none of them was allocated the
necessary resources to be pursued to a point which allowed an evaluation
of their ability to control pollution from urban runoff. Management
practices in this category included urban housekeeping (e.g.,, litter
programs, catch basin cleaning, pet ordinances) and public information
programs.
2. Detention basins are capable of providing very effective removal of pol-
lutants in urban runoff. Both the design concept and the size of the
basin in relation to the urban area served have a critical influence on
performance capability.
Wet basins (designs which maintain a permanent water pool) have the
greatest performance capabilities. Observed pollutant reductions varied
from excellent to very poor in the basins which were monitored. However,
9-12
-------
when basins are adequately sized, particulate removals in excess of
90 percent (TSS, lead) can be obtained. Pollutants with significant sol-
uble fractions in urban runoff show lower reductions; on the order of
65 percent for total P and approximately 50 percent for BOD, COD, TKN,
Copper, and Zinc. Results indicate that biological processes which are
operative in the permanent pool produce significant reductions (50 per-
cent or more) in soluble nutrients, nitrate and soluble phosphorus.
These performance characteristics are indicated by both the NURP analysis
results and conclusions reached by individual projects.
Dry basins, (conventional stormwater management basins), which are de-
signed to attenuate peak runoff rates and hence only very briefly detain
portions of flow from the larger storms, are indicated by NURP data to be
essentially ineffective for reducing pollutant loads.
Dual-purpose basins (conventional dry basins with modified outlet struc-
tures which significantly extend detention time) are suggested by limited
NURP data to provide effective reductions in urban runoff loads. Per-
formance may approach that of wet ponds; however, the additional proc-
esses which reduce soluble nutrient forms do not appear to be operative
in these basins. This design concept is particularly promising because
it represents a cost effective approach to combining flood control and
runoff quality control and because of the potential for converting
existing conventional stormwater management ponds.
Approximate costs of wet pond designs are estimated to be in the order of
$500 to $1500 per acre of urban area served, for on-site applications
serving relatively small urban areas, and about $100 to $250 per acre of
urban area for off-site applications serving relatively large urban
areas. The costs reflect present value amounts which include both capi-
tal and operating costs. The difference is due to an economy of scale
associated with large basin volumes. The range reflects differences in
size required to produce particulate removals in the order of 50 percent
or 90 percent. Annual costs per acre of urban area served are estimated
at $60 to $175, and $10 to $25 respectively.
3. Recharge Devices are capable of providing very effective control of urban
runoff pollutant discharges to surface waters. Although continued atten-
tion is warranted, present evidence does not indicate that significant
groundwater contamination will result from this practice.
Both individual project results and NURP screening analyses indicate that
adequately sized recharge devices are capable of providing high levels of
reduction in direct discharges of urban runoff to surface waters. The
level of performance will depend on both the size of the unit and the
soil permeability.
Application will be restricted to areas where conditions are favorable.
Soil type, depth to groundwater, land slopes, and proximity of water
supply wells will all influence the appropriateness of this control
technique.
9-13
-------
Surface accumulations which result from the high efficiency of soils to
retain pollutants, suggest further attention in applications where dual
purpose recharge areas also serve as recreational fields or playground
areas.
4. Street sweeping is generally ineffective as a technique for improving the
quality of urban runoff.
Five NURP projects evaluated street sweeping as a management practice to
control pollutants in urban runoff. Four of these projects concluded
that street sweeping was not effective for this purpose.. The fifth,
which had pronounced wet and dry seasons, believed that sweeping just
prior to the rainy season could produce some benefit in terms of reduced
pollution in urban runoff.
A large data base on the quality of urban runoff from street sweeping
test sites was obtained. At 10 study sites selected for detailed analy-
sis, a total of 381 storm events were monitored under control conditions,
and an additional 277 events during periods when street sweeping opera-
tions were in effect. Analysis of these data indicated that no signifi-
cant reductions in pollutant concentrations in urban runoff were produced
by street sweeping.
There may be special cases in which street cleaning applied at restricted
locations or times of year could provide improvements in urban runoff
quality. Some examples that have been suggested, though not demonstrated
by the NURP program, include periods following snow melt or leaf fall, or
urban neighborhoods where the general level of cleanliness could be sig-
nificantly improved.
5. Grass swales can provide moderate improvements in urban runoff quality.
Design conditions are important. Additional study could significantly
enhance the performance capabilities of swales.
Concentration reductions of about 50 percent for heavy metals, and
25 percent for COD, nitrate, and ammonia were observed in one of the
swales studied. However the swale was ineffective in reducing concen-
trations of organic nitrogen, phosphorus, or bacterial species. Two
other swales studied failed to demonstrate any quality improvements in
the urban runoff passing through them.
Evaluations by the NURP projects involved concluded, however, that this
was an attractive control technique whose performance could be improved
substantially by application of appropriate design considerations. Addi-
tional study to develop such information was recommended.
Design considerations cited included slope, vegetation type and mainte-
nance, control of flow velocity and residence time, and enhancement of
infiltration. The latter factor could produce load reductions greater
than those inferred from concentration changes and effect reductions in
those pollutant species which are not attenuated by flow through the
swale.
9-14
-------
6. Wetlands are considered to be a promising technique for control of urban
runoff quality. However, neither performance characteristics nor design
characteristics in relation to performance were developed by NURP.
Although a number of projects indicated interest, only one assigned NURP
monitoring activity to a wetland. This was a natural wetland, and flows
passing though it were uncontrolled. Results suggest its potential to
improve quality, but the investigation was not adequate to associate
necessary design factors to performance capability. Additional attention
to this control technique would be useful, and should include factors
such as the need for maintenance harvesting to prevent constituent
recycling.
ISSUES
A number of issues with respect to managing and controlling urban runoff
emerge from the conclusions summarized above. In some instances they repre-
sent the need for additional data/information or for further study. In
others they point to the need for follow-up activity by EPA, State, or local
officials to assemble and disseminate what is already known regarding water
quality problems caused by urban runoff and solutions.
Sediments
The nature and scope of the potential long-term threat posed by nutrient and
toxic pollutant accumulation in the sediments of urban lakes and streams re-
quires further study. A related issue is the safe and environmentally sound
disposal of sediments collected in detention basins used to control urban
runoff.
Priority Pollutants
NURP clearly demonstrated that many priority pollutants can be found in urban
runoff and noted that a serious human health risk could exist when water sup-
ply intakes are in close proximity to urban stormwater discharges. However,
questions related to the sources, fate, and transport mechanisms of priority
pollutants borne by urban runoff and their frequencies of occurrence will
require further study.
Rainfall pH Effects
The relationship between pH and heavy metal values in urban runoff has not
been established and needs further study. Several NURP projects (mostly in
the northeastern states) attributed high heavy metals concentrations in urban
runoff to the effects of acid rain. Although it is quite plausible that acid
rain increases the level of pollutants in urban runoff and may transform them
to more toxic and more easily assimilated forms, further study is required to
support this speculation.
Industrial Runoff
No truly industrial sites (as opposed to industrial parks) were included in
any of the NURP projects. A very limited body of data suggests, however,
that runoff from industrial sites may have significantly higher contaminant
9-15
-------
levels than runoff from other urban land use sites, and this issue should be
investigated further.
Central Business Districts
Data on the characteristics of urban runoff from central business districts
are quite limited as opposed to other land use categories investigated bv
NURP. The data do suggest, however, that, some sites may produce pollutant
concentrations in runoff that are significantly higher than those from other
sites in a given urban area. When combined with their typically high degrees
of imperviousness, the pollutant loads from central business districts can be
quite high indeed. The opportunities for control in central business dis-
tricts are quite limited, however.
Physical Effects
Several projects concluded that the physical impacts of urban runoff upon
receiving waters have received too little attention and, in some cases, are
more important determinants of beneficial use attainment than chemical pol-
lutants. This contention requires much more detailed documentation.
Synergy
NURP did not evaluate the synergistic effects that might result from pollut-
ant concentrations experienced in stormwater runoff, in association with pH
and temperature ranges that occur in the receiving waters. This type of in-
vestigation might reveal that control of a specific parameter, such as pH,
would adequately reduce an adverse synergistic effect caused by the presence
of other pollutants in combination and be the most cost effective solution.
Further investigations should include this issue.
Opportunities for Control
Based upon the results of NURP's evaluation of the performance of urban run-
off controls, opportunities for significant control of urban runoff quality
are much greater for newly developing areas. Institutional considerations
and availability of space are the key factors. Guidance on this issue in a
form useful to States and urban planning authorities should be prepared and
issued.
Wet Weather Water Quality Standards
The NURP experience suggests that EPA should evaluate the possible need to
develop "wet weather" standards, criteria, or modifications to ambient crite-
ria to reflect differences in impact due to the intermittent, short dura-
tion exposures characteristic of urban runoff and other nonpoint source
discharges.
Coliform Bacteria
The appropriateness of using coliform bacteria as indicator organisms for
human health risk where the source is exclusively urban runoff warrants fur-
ther investigation.
9-16
-------
Wetlands
The use of wetlands as a control measure is of great interest in many areas,
but the necessary information on design performance relationships required
before cost effective applications can be considered has not been adequately
documented. The environmental impacts of such use upon wetlands is a
critical issue which, at present, has been addressed marginally, if at all.
Swales
The use of grass swales was suggested by two NURP projects to represent a
very promising control opportunity. However, their performance is very
dependent upon design features about which information is lacking. Further
work to address this deficiency and appropriate maintenance practices appears
warranted.
Illicit Connections
A number of the NURP projects identified what appeared to be illicit connec-
tions of sanitary discharges to stormwater sewer systems, resulting in high
bacterial counts and dangers to public health. The costs and complications
of locating and eliminating such connections may pose a substantial problem
in urban areas, but the opportunities for dramatic improvement in the quality
of urban stormwater discharges certainly exist where this can be accom-
plished. Although not emphasized in the NURP effort, other than to assure
that the selected monitoring sites were free from sanitary sewage contamina-
tion, this BMP is clearly a desirable one to pursue.
Erosion Controls
NURP did not consider conventional erosion control measures because the
information base concerning them was considered to be adequate. They are
effective, and their use should be encouraged.
Combined Sewer Overflows
In order to address urban runoff from separate storm sewers, NURP avoided any
sites where combined sewers existed. However, in view of their relative
levels of contamination, priority should be given to control of combined
sewer overflows.
Implementation Guidance
The NURP studies have greatly increased our knowledge of the characteristics
of urban runoff, its effects upon designated uses, and of the performance
efficiencies of selected control measures. They have also confirmed earlier
impressions that some States and local communities have actually begun to
develop and implement stormwater management programs incorporating water
quality objectives. However, such management initiatives are, at present,
scattered and localized. The experience gained from such efforts is both
needed and sought after by many other States and localities. Documentation,
9-17
-------
evaluation, refinement and transfer of management and financing mechanisms/
arrangements, of simple and reliable problem assessment methodologies, and of
implementation guidance which can be used by planners and officials c\t the
State and local level are urgently needed as is a forum for the sharing of
experiences by those already involved, both among themselves and with those
who are about to address nonpoint source issues.
9-18 «U.S. OOVERHMENT PRISTWO 0??IOS : 1984 0-421-082/520
-------