D-1
APPENDIX D
SITES IDENTIFIED USING THE XBAR PLOT
D.1 Description
The following appendix describe those sites described as unusual in other land
uses besides residential.
D.2 Censored Data Descriptions
In this section is presented a brief description of the characteristics of each
constituent. It includes the range of detected values in the database, the maximum and
minimum values, the most frequent non-detected observation and the frequency at this
level of censoring.
Conductivity is the ability of the water to carry electrical current. It depends of the
available ions present in the sample. The database contains 685 samples collected in
single land uses most of them in residential and industrial areas. Only 2 samples were
collected in Open Space. All the samples collected were detected. Residential, industrial
and the pooled dataset have almost the same range. Low conductivity values are around
10 mS/cm, whereas high values are close to 2000 mS/cm. A single observation of 5955
mS/cm was observed in the complete dataset. The BMP database has a range between
0.04 and 600 mS/cm included in their database. In the NSQD around five percent of the
conductivity values are greater than 600 mS/cm.
D-2
Hardness indicates the presence of metallic cations (Ca, Mg, Sr, Fe and Mn) in
the water. Concentrations lower than 75 mg/L CaCO3 are considered low. Concentrations
between 150 and 300 mg/L CaCO3 are considered hard (Sawyer, et al, 2003). Hardness
is highly correlated with the characteristic of the soil in the watershed. If there is an
elevated presence of the metallic cations in the soil of the catchments it is expected a high
hardness concentration. The range of the residential commercial and industrial land uses
is very similar. It varies between 1.9 and 900 mg/L. Around 99% of the values found in
the database are smaller than 400 mg/L. The BMP database has a range between 0.5 and
543 mg/L. There are six values higher than the maximum reported in the BMP database.
There is only one detection limit (<10 mg/L) for this constituent in the database. Around
5% of the detected values in the NSQD were lower than this level.
Oil and Grease is related with discharges coming from parking lots, commercial
areas and residential areas with failing septic tanks. It is important in stormwater
discharges because oil and grease have low solubility and trend to float on water. Open
space has the lower variability among land uses. It is not expected elevated
concentrations in this land use. Freeways and commercial land uses have the higher
percentage of detected values. The range of detected values varies between 0.2 and
11,000 mg/L. Only 1% of the total dataset has values above 400 mg/L and around 95% of
the dataset has values smaller than 60 mg/L. The range in the BMP database is between
1.5 and 66.7 mg/L. The concentration at any percentile value was always higher in
freeways than in any other land use. Around 50% of the non-detected values are located
in <1 mg/L. This censoring level was also the most frequent in all the land uses except for
freeways where <3 mg/L was the most frequent. The second most frequent level was <5
D-3
mg/L. About 2% of the detected values in all land uses were lower than 1 mg/L while
52% of the detected values were smaller than 5 mg/L.
Total Dissolved Solids calculates the fraction of solids that passes through a filter.
It is important in stormwater because it can be associated with the concentration of
dissolved metals in runoff samples. More than 97% of the collected samples were
detected. The maximum percentage of non-detected values was observed in open space
with 2.22%. For all the land uses only 0.2% of detected values were smaller than 5 mg/L.
In the other hand, around 1.8% of the detected values were higher than 470 mg/L. The
range between 5 mg/L and 300 mg/L contains around 95% of the total detected values.
The BMP database contains a range of dissolved solids between 12.5 mg/L and 7100
mg/L. They assume a detection limit of 12.5 mg/L after corrections. The most frequent
detection limit was 5 mg/L. This value is the most appropriate because even though the
scale can detect variations around 1 mg/L, there are more factors associated with the
variability during the measurement.
Total Suspended Solids calculates the fraction of solids that is retained in a 0.45
mm filter or smaller. The range of TSS was between 3 mg/L and 2500 mg/L
approximately except for a single sample in a freeway land use. Around 95% of the
detected concentrations were below 450 mg/L. Only 1.5 % of the detected samples were
observed below 5 mg/L. The percentage of non-detected values in all land uses were
below 2% except for open space that was close to 4.5%. The BMP database has a range
of TDS between 38 and 11000 mg/L. In the NSQD 34% of the detected samples were
located below 38 mg/L. Almost 80% of the non-detected values were located at <5 mg/L.
D-4
The second most frequent detection limit was <1 mg/L, in samples collected in open
space and industrial land uses.
Biochemical Oxygen Demand (BOD) represents the total amount of oxygen that
bacteria require to decompose the organic material in a sample under aerobic conditions
(Sawyer et al. 2003). More than 3000 samples are included in the database. More than
84% of the samples were detected. The range of commercial, residential and industrial
land uses is very similar except for a single sample of 6920 mg/L in an industrial site.
The lower range was observed in open land use with a maximum observed value of 20
mg/L. The BMP database has a maximum value of 230 mg/L. There were observed 10
detection limits varying between one mg/L and 100 mg/L. More than 80% of the nondetected values are between one and five mg/L. Detection limits larger than 20 mg/L are
considered no relevant in the analysis. Around 7.5% of the non-detected samples were
right censored.
Chemical Oxygen Demand calculates the total oxygen required for oxidation of a
sample to carbon dioxide and water (Sawyer et al., 2003). The database contains 2,750
samples mostly collected in residential land use. The smallest range was observed in open
space, and the largest was observed in industrial. The probability plot of the freeways was
always the highest while the probability plot of the open space was always the smallest.
The percentages of non-detected were lower than 2% except for open space that has
23.26%. The BMP database has a range of COD between <12.14 mg/L and 2,030 mg/L.
All the samples collected in open space were lower than 100 mg/L except for an
observation of 476 mg/L. The COD range varies between 1 mg/L and 1260 mg/L. The
highest observation was observed in an industrial land use. Six detection limits were
D-5
observed in the data set. The most frequent detection limit was <10 mg/L with 41% of the
total non-detects. Around 80% of the non-detected samples in open space were lower
than 20 mg/L while all the non-detected samples in freeways were less than 10 mg/L.
Fecal Coliform is an indicator of bacteria in water. The probability plot of the
open space was always higher when compared with the other land uses. The BMP
database has a range between <1,000,000 and 1,827,000,000 colonies /100 mL. Open
space has smallest concentration of fecal coliform with 650 colonies per 100 mL. Open
space also has the smallest range with a maximum detected concentration of 63,000
colonies/100 mL. Fecal coliform was detected in all the freeways samples. In open space
around 8% of the samples were non-detects whereas in residential, commercial and
industrial the percentage was close to 12%. The data set contains left and right-censored
values. Around 39% of the non-detected values were left censored, the remaining were
right censored. The highest frequency was observed in <1 col/100mL with 22% of the
non-detected values. Close to 45% of the non-detected values in industrial land uses were
lower than 1 col./100 mL. The censored values found in open space were larger than
160,000 colonies/100 mL. The second most frequent detection limit was >60,000
colonies/100 mL with a 15% of the total non-detected values. This was the most frequent
censored value in commercial land use and the second most frequent in residential land
use.
Fecal Streptococcus is indicator of fecal contamination. The probability plot of
commercial and industrial land uses was always smaller when compared with the other
land uses. The BMP database has a range between 2 and 1,500,000 colonies/100 mL. In
this case the largest minimum detected was observed in freeways with 560 colonies/100
D-6
mL. The smallest range was observed in open space with a maximum of 101,000
colonies/100 mL. There was observed a single observation above 1,100,000 colonies/100
mL in an industrial site. All the freeways samples were detected. Less than 10 percent of
the samples in each land use were non-detected. The most frequent non detected value
was <1 colonies/100 mL. Around 50% of the industrial non-detected samples are in this
group. The second most frequent detection limit was >60,000 colonies/100 mL, more
than 45% of the non- detected values in commercial land used are in this group.
Ammonium ion NH4+ is produced by the interaction of ammonia gas (NH3) and
water. The probability plot of freeways was always higher and the probability plot of
open. The smallest ammonia concentration values were observed in residential land uses.
The highest percentage of non-detected values was observed in open space. For the
remaining land uses the percentage of non-detected was smaller than 20%. The range of
ammonia in the BMP database is between <0.5 mg/L and 9 mg/L. Around 54% of the
non-detected values were <0.2 mg/L. All the non-detected freeway observations were
observed in this group. The second highest group was when the detection limit was not
reported. The third group was <0.5 mg/L. About 70% of the non detected values in open
space have censored observations at 0.5 mg/L.
Nitrite and nitrate (NO2 and NO3) are produced after the reduction or oxidation of
NO2. NO2 can be formed during combustion process such as internal combustion of
gasoline engines. NO3 can be formed also by oxidation of nitrogen or ammonia in the
production of fertilizers. Nitrate can be reduced to nitrite inside a warm body. If nitrite
reaches the bloodstream it can reduce the oxygen transport (Sawyer et al., 2003). The
range was between 0.01 and 18 mg/L. The BMP database has a range between <0.01 and
D-7
9.09 mg/L. Only 1.5% of the dataset was greater than 3 mg/L. The probability plot
indicates that the median nitrite-nitrate in industrial land uses was always lower than in
other land uses. The percentage of non-detected values was smaller than 4% in any land
use except for open space that has approximately 16% of non-detected values. The most
frequent detection limit was <0.1 mg/L. All the non-detected values in open space and
50% of the commercial land use were found in this group. All the non-detected values in
freeways were <0.3 mg/L. The second highest group was observed when the detection
limit was not reported.
The total Kjeldahl nitrogen (TKN) method destroys all the organic nitrogen using
sulfuric acid. The destruction releases the nitrogen as ammonia gas (Sawyer et al, 2003).
TKN is the sum of organic nitrogen and ammonia. The range of TKN was between 0.05
mg/L and 36 mg/L. About 6.5% of the reported values were larger than 5 mg/L. The
BMP database has a range for TKN between <0.025 mg/L and 18.31 mg/L. Less than 5%
of the dataset was not-detected in any land use except for freeways where the percentage
of non detected observations was close to 30% was non-detected. About 30% of the nondetected values were at <0.2 mg/L followed by <0.5 mg/L with 22%. All the nondetected values in open space were located at <0.5 mg/L. Half of the non-detected values
in freeways were at <0.1 mg/L the other half was at <1 mg/L. Almost 40% of the nondetected at industrial land uses were <1 mg/L. In approximately 16% of the all nondetected values the detection limit was not reported.
Dissolved Phosphorus corresponds to the fraction of phosphorous able to pass a
0.45 mm filter. Phosphorous as nitrogen are associated with fertilizer and domestic
wastewater. The observed range of dissolved phosphorous was between 0.01 mg/L and 7
D-8
mg/L. Concentrations higher than 2 were observed in industrial, mixed industrial and
freeways land uses. The range in the BMP database is between 0.0022 mg/L and 8.42
mg/L. The probability plot of freeways indicated higher medians than the other single
land uses. Less than five percent of the freeway observations were non-detected, for the
remaining land uses the percentage of non-detected varies between 12 and 21%. The
highest level of non-detected was observed in open space. In this case it is clear that the
level of non-detected observed in freeways was lower than in the other land uses because
more accurate equipment was used. In 34% of the non-detected cases the detection limit
was not reported, when reported <0.02 mg/L, <0.05 mg/L, <0.1 mg/L and <0.5 mg/L
were the most common detection limits. About 80% of the detected observations were
smaller than 0.5 mg/L.
Total phosphorus is important for the growth of organisms. They are divided in
three groups: orthophosphate, polyphosphate and organic phosphorus (Sawyer et al.,
2003). Notice that in this case the highest value was observed in an open space. The
probability plot indicates that the median total phosphorus concentration for open space
was always smaller than in other land uses. Less than one percent of the observations in
freeways were non-detected, in the other land uses it was around four percent except for
open space where the percentage was 15%. The range in the BMP database was between
<0.1 mg/L and 80.2 mg/L. All the non-detected observations in freeways were at 0.03
mg/L. The largest frequency of non-detected values was <0.5 mg/L followed by <0.1
mg/L. In around 14% of the non-detected values the detection limit was not reported.
The abundance of copper on the earth’s crust is around 68 ppm, and in streams is
between 4 and 12 mg/L (APHA, 1995). Copper can be a track irritant not harmful to
D-9
humans in concentrations below 1 mg/L, however it can be toxic for some species at this
concentration. Copper is considered more dangerous to the environment than the humans.
The mayor source in drinking water is the corrosion of copper pipes (APHA, 1995). The
range of total copper varies between 0.6 mg/L to 1.4 mg/L. The highest observations
were found in industrial and residential land uses. The range in the BMP database was
between <25 mg/L and 130 mg/L. The lower level of non-detected was in freeways with
1% and the largest open space with 25% of the observations. Residential, commercial and
industrial land uses are between 7 and 16% of the observations. The probability plot
indicates that the distribution of freeways was always the highest, and the distribution of
open space was the smallest. The largest level of non-detected values was at <10 mg/L
followed by <20 mg/L. These two levels have close than 71% of the total non-detected
values in the dataset. The third group is <5 mg/L with 15%. There is a low percentage of
non-detected were the detection limit was not reported (2.62%) mostly from open space
land uses.
The abundance of lead ranges from 2.5 to 25 ppm in soil; in water the average
concentration is 3mg/L. It is used mainly in batteries, alloy, piping, insecticides and
ammunition. Lead is toxic and probably carcinogenic. It can cause brain and kidney
damage (APHA, 1995). Lead ranges from 0.5 to 1200 mg/L. The BMP database ranges
from <30 mg/L to 1200 mg/L. The smallest range was observed in open space, the largest
in industrial land use, the remaining single land uses varied from 0.5 to 450 – 700 mg/L.
In all freeways samples lead was detected. The percentage of non-detected in the
remaining land uses varied from 14 to 28%. The probability plot indicates that lead was
always higher in freeways than in any other group. In the other hand open space had the
D - 10
largest level of non-detected (58%) and the lowest concentrations. In almost 15% of the
non-detected values the detection limit was not reported. The detection limit with the
highest frequency was <10 mg/L, followed by <5 mg/L and <1 mg/L with 12% and 6%
respectively. Notice that 23% of the non detected observations in open space the
detection limit was 100 mg/L. Around 85% of the detected values are smaller than this
detection limit.
Nickel is found usually in electroplating activities. The abundance in water is 1
mg/L and in soils 2.5 ppm. Nickel is used in alloys, magnets, protective coating and
batteries. It seems not to be highly toxic to humans by oral exposition by may be
carcinogenic if inhaled (APHA, 1995). The range varies between 1 and 120 mg/L. In the
BMP database the range was between <15 mg/L and 300 mg/L. In the probability plot of
the detected values residential land use had always the smallest concentration. Open
space had the highest concentration for detected values smaller than 10 mg/L. For higher
values industrial land use was the highest. All the ranges were similar for the five single
land uses. The smallest percentage of non-detected was observed in Freeways were 90%
of the samples were detected. The highest level of non-detects was observed in open
space with 82%. For the remaining single land uses the percentage of non-detects varied
from 37 to 55%. Around 28% of the total non-detected values were at <20 mg/L. The
second and third group were <15 mg/L and 5 mg/L with 14% and 12% respectively. In
more than 15% of the non-detected values the detection limit was not reported. Detection
limits higher than 20 mg/L are not relevant, more than 60% of the detected values are
below this threshold.
D - 11
Zinc is found in soils at concentrations varying from 25 to 68 ppm. In water the
estimated concentration is 20 mg/L. Zinc is found in batteries, fungicides, and galvanized
pipes. Zinc is also found in industrial processes (APHA, 1995). The observed values
range from 2 to 22,500 mg/L. The BMP database ranges from <50 to 6300 mg/L. The
distribution with lowest median was observed in open space followed by residential land
use. Concentration larger than 1mg/L were found mainly in industrial and mixed
industrial land uses. The smallest concentrations and higher level of non-detects were
observed in open space. For the remaining land uses the level of non-detects was smaller
than four percent. More than half of the detection limits in freeways and commercial land
uses were observed at <100 mg/L. More than 80% of the detected values in open space
are smaller than this detection limit, however, only 20% of the detected values were
smaller than 100 mg/L in freeways. In 15% of the non-detected values the detection limit
was not reported. The most frequent detection limit was <20 mg/L with almost 18% of
the total non-detected values followed by <100 mg/L (15.5%), <10 mg/L (13.6%) and
<25 mg/L (10.7%).
D.3 Expected Percentages
Using the distribution of non-detected and detected observations it is possible to
estimate what percentage of the samples will be non detected at a specific detection limit.
The following discussion briefly describes the most common methods to calculate the
concentrations of stormwater constituents. It is also mention the recommended detection
limit required to obtain low percentage of non-detected observations.
D - 12
It was observed that the conductivity range of detected values in the NSQD
database was between 2 and 6000 S/cm, larger than the observed in the BMP database.
Most of the commercial conductivity meters used a Wheatstone bridge to calculate the
resistance required until there is not electricity flowing in the circuit. Potassium Chloride
is used to calibrate the conductivity meter; a solution 0.0001 M of KCl produces a
conductivity of 14.9 mS/cm at 25oC. Most sensors had a range between 0.1 – 20 S/cm
or 2 – 20 S/cm or 200 – 2000 S/cm. A conductivity meter with a range between 20 –
2000 S/cm will be able to detect 99% of the samples collected in the different land uses.
The range of detected values for hardness in the NSQD database was between 1.9
and 1,100 mg/L, similar to those found in the BMP database (between 0.5 and 543
mg/L). There are two ranges commercially available 0 - 20 mg/L and 10 – 4000 mg/L. If
the high range is used, between 2 and 7 percent of the samples will not be detected.
The range of detected oil and grease values in the database was between 0.2 and
11,000 mg/L. The BMP database has a lower range with a maximum of 66.7 mg/L.
About 7% of the total dataset were larger than 100 mg/L. Only 20 samples were collected
in open space, from these samples 74% were lower than 1 mg/L. Oil and grease is
associated with streets and parking lots, the highest values were observed in freeways. A
detection limit of 0.5 mg/L is recommended for this analysis. In average more than 17%
of the samples will be not detected if the detection limit is equal or higher than 1 mg/L.
The range of detected total dissolved solids values in the database was between 3
and 17,900 mg/L. The BMP database has values between 12.5 and 7,100 mg/L. A
detection limit of 5 mg/L is appropriate for this analysis. Using standard methods, more
than 98% of the stormwater samples were detected.
D - 13
The range of detected total suspended solids values in the database was between 3
and 4,800 mg/L. The BMP database has values between 38 and 11,000 mg/L. In this case
there are more samples in the range 1 – 5 mg/L. A higher precision scale will reduce the
level of non-detected values. The recommended detection limit is 1 mg/L. A detection
limit of 5 mg/L is appropriate for this analysis. Using the standard method more than
98% of the stormwater samples were detected
The range of BOD detected values in the database was between 1 and 6,920
mg/L, the BMP database has a maximum value of 230 mg/L. The recommended
detection limit is 1 mg/L. It has the highest frequency of detected values, more than 99%
of the observations were above this level. Detection limits higher than 5 mg/L are not
recommended because about 20% of the observations were below this limit.
The range of detected COD values in the database was between 1 and 1,260
mg/L, the BMP database has a maximum value of 2,030 mg/L. Three ranges are
commonly used to calculate COD. The ultra low range that ranges between 0.7 and 40
mg/L; the low range that ranges between 3 and 150 mg/L and the high range for
concentrations between 20 and 1,500 mg/L. Around 88% of the observations were
smaller than 150 mg/L. The recommended range is between 3 and 150 mg/L. For larger
concentrations, it is recommended to dilute the sample until reach the specified range.
The range of detected ammonia values in the database was between 0.01 and 12
mg/L. The BMP database reported a highest concentration of 9 mg/L. The most frequent
detection limit was <0.2 mg/L with 54% of the total non-detected observations followed
by <0.5 mg/L and <0.1 mg/L with 15% and 10% respectively. In close than 15% of the
non-detected values the detection limit was not reported. One of the EPA approved
D - 14
methods uses Nessler’s reaction. The precision of this method is between 0.02 to 2.5 m/L
using HACH instruments, and 0.03 to 2.5 mg/L using Chemetrics. This method is
appropriate for the detection of ammonia in stormwater. A detection limit of 0.2 mg/L is
not recommended. More than 20% of the observations will be non-detected using this
threshold.
The range of detected Nitrite - Nitrate values in the database was between 0.01
and 18 mg/L. The BMP database has a highest value of 9.09 mg/L. The most frequent
detection limit was 0.1 mg/L, more than 35% of the non-detected observations were
observed at this level. The cadmium reduction method can be used in the range 0.01 to 1
mg/L NO3. This method will cover most of the range of the samples collected in
stormwater. Cadmium is a heavy metal hazardous to humans.
The range of detected TKN values in the database was between 0.05 and 36 mg/L.
The BMP database has a highest value of 18.31 mg/L. The most frequent detection was
0.2 mg/L. More than 95% of the samples were detected in all land uses except for
freeways. There were not observations between 0.1 and 0.2 mg/L in freeways that means
that < 0.2 mg/L is an appropriate detection limit for TKN.
The range of detected dissolved phosphorus values was between 0.01 and 7 mg/L.
The BMP database has a highest value of 8.42 mg/L. The most frequent detection limit
was <0.1 mg/L. One of the most common methods is the ascorbic acid method. The range
for this method varies between 0.02 to 2.5 mg/L. This method covers 99.8% of the total
observations in the database.
The range of detected total phosphorus values was between 0.01 and 15 mg/L.
The BMP database has a highest value of 80.2 mg/L. The most common detection limit
D - 15
was <0.5 mg/L. The ascorbic acid method can be used also for calculate the total
phosphorus concentration. The range of this method varies between 0.02 and 2.5 mg/L.
The range of detected total copper values was between 0.6 and 1360 mg/L. The
BMP database has a highest value of 130 mg/L. The most common detection limit was
<10 mg/L. One of the most common used methods is the bathocuproine method. The
minimum detectable concentration using this method is 20 mg/L. Last table indicates that
around 63% of the samples will be not detected using this detection limit. Low
concentrations as 5 mg/L can be detected using electrothermal atomic absorption
spectrometry but the cost per sample can be moderate (Pitt, 1997).
The range of detected total lead values was between 0.5 and 1200 mg/L. The
BMP database has a highest value of 1200 mg/L. The most frequent non-detected value
was <10 mg/L. Electrothermal atomic absorption spectrometry can detect samples as low
as 5 mg/L. This will indicate that approximately 17% of the samples will be nondetected. The dithizone method has been approved by EPA for the reporting of total lead.
The lower detection limit for this method us 3 mg/L suitable for samples collected in
freeways, industrial and commercial land use.
The range of detected total nickel values was between 1 and 120 mg/L. The BMP
database has a highest value of 300 mg/L. The most frequent non-detected value was <20
mg/L. Around 84% of the observations in the database were smaller than 20 mg/L. The
heptoxime method is approved by EPA to detect nickel; the method is adequate for
concentrations larger than 20 mg/L but does not cover the whole range of observations.
ICP methods can detect concentrations as low as 17 mg/L while eletrothermal atomic
absorption spectrometry methods can detect low concentrations down to 5 mg/L.
D - 16
The range of detected total zinc values was between 2 and 22,500 mg/L. The
BMP database has a highest value of 6,300 mg/L. The most frequent non-detected value
was <20 mg/L. Around 4% of the observations in the database were smaller than 20
mg/L. The zincon method is approved by EPA to detect zinc. This method is also
included in standard methods for examination of water and wastewater. The method can
be used for concentrations of zinc larger than 20 mg/L. Atomic absorption is more
accurate detecting low concentrations close to 5 mg/L. ICP methods can be used in the
range 7 to 7076 mg/L.
D.4 Evaluation of The Methods Selected to Estimate Non-Detected Observations
Three methods were used to estimate the non-detected observations: delete them,
replace them by half of the detection limit or estimate them using the Cohen’s maximum
likelihood method. The following discussion shows the analysis for each constituent and
single land use
D.4.1 Hardness
It was observed for total hardness that all the samples were detected except in
industrial land uses where more than 96% of the samples were detected. Changes in the
average, median, standard deviation and coefficient of variation were not significant if
the non-detected values were ignored, estimated or replaced by half of the detection limit.
Table D.1 shows that there are no important differences in industrial land use using any
of the three methods.
D - 17
Table D.1 Summary Statistics for Estimated Observations for Total Hardness (mg/L)
RESIDENTIAL
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
250
100.00
3.00
401.00
43.32
32.00
44.87
1.04
COMMERCIAL
HD
Ignore Estimate
250
250
3.00
401.00
43.32
32.00
44.87
1.04
3.00
401.00
43.32
32.00
44.87
1.04
139
100.00
1.90
356.00
62.03
38.90
65.17
1.05
Land use
Ignore Estimate
8
100.00
11.00
270.00
145.25
150.00
85.12
0.59
HD
139
139
1.90
356.00
62.03
38.90
65.17
1.05
1.90
356.00
62.03
38.90
65.17
1.05
OPEN SPACE
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
INDUSTRIAL
Ignore Estimate
138
96.38
5.50
888.00
68.83
39.00
104.55
1.52
HD
138
138
5.00
888.00
66.52
38.50
103.32
1.55
5.00
888.00
66.52
38.50
103.32
1.55
FREEWAY
HD
Ignore Estimate
8
8
11.00
270.00
145.25
150.00
85.12
0.59
11.00
270.00
145.25
150.00
85.12
0.59
HD
127
127
127
100.00
5.00
5.00
5.00
1000.00 1000.00 1000.00
57.19
57.19
57.19
34.00
34.00
34.00
105.95
105.95
105.95
1.85
1.85
1.85
Figure D.1 shows the probability plot for industrial land use. The plot indicates
that the mean value is smaller when the non-detected are either estimated or replaced by
half of the detection limit. The lower 40% of the distribution is displaced to the left. All
the non-detected values were observed at 10 mg/L. The upper 60% of the distribution is
not affected by the non-detected values.
D - 18
Lognormal Probability Plot for Hardness in Industrial Land Use
ML Estimates
99
Detection Limits
Percent
95
90
IGNORE
ESTIMATE
80
70
60
50
40
30
20
HALF
DETECTION
10
5
Goodness of Fit
AD*
1.405
1.484
1.484
1
10
100
1000
Hardness mg/L
Figure D.1. Estimated hardness distributions in industrial land use.
In the oil and grease case the level of censoring varied between 37% and 72%. It
was observed that the highest change in the coefficient of variation was observed in
freeways, where in contrast will all the other land uses the standard deviation increases
when the non detected were estimated or replaced by half of the detection limit. Table
D.2 shows the differences in the descriptive statistics using the three methods. There is an
increase between 30% to 60% in the mean oil and grease estimators when the censored
observations are ignored. There was observed differences below 4% when the censored
observation is estimated using Cohen’s maximum likelihood method or replace by half of
the detection limit.
D - 19
Table D.2 Summary Statistics for Estimated Observations in Oil and Grease (mg/L)
RESIDENTIAL
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
533
57.79
0.20
2980
22.85
3.85
175.53
7.68
COMMERCIAL
HD
533
533
0.02
2980
13.87
2.50
133.76
9.65
0.20
2980
13.89
2.50
133.76
9.63
Ignore Estimate
308
70.78
0.80
359
12.63
4.70
39.75
3.15
Land use
Ignore Estimate
19
36.84
0.50
3.70
1.53
1.30
1.07
0.70
HD
Ignore Estimate
308
308
0.03
359
9.42
3.00
33.80
3.59
0.25
359
9.39
3.00
33.81
3.60
OPEN SPACE
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
INDUSTRIAL
327
65.14
0.50
11000
62.87
5.00
753.77
11.99
HD
327
327
0.00
11000
41.40
2.50
608.56
14.70
0.25
11000
41.39
2.60
608.56
14.70
FREEWAY
HD
Ignore Estimate
19
19
0.50
3.70
1.09
0.50
0.93
0.85
0.50
3.70
1.09
0.50
0.93
0.85
60
71.67
3.00
30.00
8.49
8.00
5.28
0.62
HD
60
60
0.50
30.00
6.57
4.65
5.42
0.83
0.25
30.00
6.45
4.65
5.52
0.86
The probability plot in residential land uses indicates that the lower tail is better
described with the estimated method (Figure D.2). The upper tail was the same for the
estimated and the half detection limit method. Notice that around 40% of the nondetected values were observed at < 1 mg/L and another 40% was observed at < 5 mg/L.
The estimated values describe better the lower tail however there was not a significant
difference in the mean, standard deviation and coefficient of variation.
The standard deviation in both cases was the same, the difference between the two
means was 0.02 mg/L and the difference in the coefficient of variation was 0.02. These
results indicate that replacing by half of the detection limit or estimate the non-detected
has not an important difference. This case is very important because the level of
censoring was elevated (42.2%). Ignore the non-detected increases the mean value more
D - 20
than 64% and the standard deviation in more than 30%, in the other hand reduces the
coefficient of variation in 20%.
Lognormal Probability Plot for Oil and Grease in Residential Land Use
ML Estimates
Percent
99
NON DETECTED
95
90
IGNORE
80
70
60
50
40
30
20
HALF
DETECTION
ESTIMATE
Goodness of Fit
10
5
AD*
6.728
7.348
8.654
1
0.01
0.10
1.00
10.00
100.00
1000.00
Oil and Grease mg/L
Figure D.2. Estimated oil and grease distributions in residential land use.
The commercial land use has a similar trend as the observed in residential land
uses (Figure D.3). There is a better description of the lower tail but the mean, standard
deviation and coefficient of variation is almost the same if the censored data is replaced
by half of the detection limit or if is estimated. In this case the largest level of non
detected was observed at < 5 mg/L followed by < 1 mg/L. The average was increased in
34%, and the standard deviation in 18% when the censored data was ignored. The
coefficient of variation was reduced about 12% when the non-detected values were
ignored.
D - 21
Lognormal Probability Plot for Oil and Grease in Commercial Land Use
ML Estimates
99
NON DETECTED
Percent
95
90
IGNORE
ESTIMATE
80
70
60
50
40
30
20
HALF
DETECTION
Goodness of Fit
AD*
2.592
2.622
2.620
10
5
1
0.1
1.0
10.0
100.0
Oil and Grease mg/L
Figure D.3 Estimated oil and grease distributions in commercial land use.
Figure D.4 shows the probability plot in industrial land uses indicates the case
when an unusual value was present in the dataset. The maximum observation was larger
by a factor of 2200 compared with the median value of the distribution. This generate a
coefficient of variation of 12 in the case that the censored data is ignored or 14.7 in the
case that they are estimated or replaced by half of the detection limit. To estimate the
lower tail of the distribution does not indicate a significant difference in any parameter
when compared with replace by half of the detection limit.
The percentage of detected values of oil and grease in open space was very low,
only 7 from 19 observations were detected (Figure D.5). Almost all the non-detected
values were observed at < 1 mg/L. It was not possible to use the Cohen’s maximum
likelihood method in this case because a maximum number of detected values at the same
site were two.
D - 22
Lognormal Probability Plot for Oil and Grease in Industrial Land Uses
ML Estimates
99
NON DETECTED
Percent
95
90
IGNORE
ESTIMATE
80
70
60
50
40
30
20
HALF
DETECTION
Goodness of
Fit
AD*
10
5
3.549
3.882
3.222
1
0.0
0.0
0.1
1.0
10.0
100.0
1000.0
10000.0
Oil and Grease mg/L
Figure D.4 Estimated oil and grease distributions in industrial land use.
Lognormal Probability Plot for Oil and Grease in Open Space
ML Estimates
99
Percent
95
NON DETECTED
90
IGNORE
80
ESTIMATE
70
60
50
40
30
HALF
DETECTION
Goodness of Fit
20
AD*
10
1.710
1.856
1.856
5
1
0.1
1.0
10.0
Oil and Grease mg/L
Figure D.5 Estimated oil and grease distributions in industrial land use.
D - 23
Ignoring the non-detected will increase the mean value in almost 40% compared
with the case when the non-detected were replaced with half of the detection limit.
The probability plot for freeways indicate that estimating or replacing the
censored observations for half of the detection limit has not a significant difference in the
coefficient of variation (Figure D.6). The coefficient of variation was 3% larger when
half of detection limit was used instead of Cohen’s method. A different situation occurs
when the non-detected were ignored. The coefficient of variation was reduced in 30%
compared with the estimated method.
Lognormal Probability Plot for Oil and Grease in Freeways Land Use
ML Estimates
99
95
NON DETECTED
90
IGNORE
Percent
80
ESTIMATE
70
60
50
40
30
HALF
DETECTION
Goodness of Fit
AD*
20
0.928
0.782
1.643
10
5
1
1
10
Oil and Grease mg/L
100
Figure D.6 Estimated oil and grease distributions in freeways land use.
D.4.2 Total Dissolved Solids
In all the land uses the percentage of non-detected was very low. The lowest
percentage was observed in open space with 2%. No important differences were observed
D - 24
in the mean, standard deviation and coefficient of variation when the non-detected were
ignored or estimated. Descriptive statistics for each of the three methods are shown in
Table D.3
Table D.3 Summary Statistics for Estimated Observations in TDS (mg/L)
RESIDENTIAL
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
861
99.19
3.00
1700
96.26
72.00
102.45
1.06
COMMERCIAL
HD
Ignore Estimate
861
861
3.00
1700
95.54
70.50
102.35
1.07
0.50
1700
95.50
70.50
102.38
1.07
399
99.50
4.00
3860
109.94
74.00
208.76
1.90
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
45
97.78
32.00
542
151.41
124.50
109.83
0.73
HD
Ignore Estimate
399
399
4.00
3860
109.44
74.00
208.36
1.90
4.00
3860
109.42
74.00
208.37
1.90
OPEN SPACE
Ignore Estimate
INDUSTRIAL
412
99.51
4.50
11200
161.99
91.00
582.40
3.60
HD
412
412
1.78
11200
161.23
89.50
581.09
3.60
2.50
11200
161.22
89.50
581.09
3.60
FREEWAY
HD
Ignore Estimate
45
45
10.79
542
148.28
119.00
110.58
0.75
2.50
542
148.10
119.00
110.82
0.75
97
98.97
12.00
470
95.31
77.50
76.38
0.80
HD
97
97
5.85
470
94.39
77.00
76.52
0.81
0.50
470
94.34
77.00
76.59
0.81
Figure D.7 shows the probability plot for residential land uses. The plot indicates
that the half of detection limit estimate lower values than Cohen’s maximum likelihood
method. The upper 95% of the distributions are identical for the three cases. The
remaining land uses the probability plots don’t indicate significant differences among the
three methods. For example Figure D.8 shows the probability plots in commercial areas.
The three lines overlap except for a small fraction in the lower tail of the distribution.
D - 25
Lognormal Probability Plot for TDS in Residential Land Use
ML Estimates
99
NON DETECTED
Percent
95
90
IGNORE
ESTIMATE
80
70
60
50
40
30
20
HALF
DETECTION
10
5
Goodness of Fit
AD*
1
3.177
4.603
7.845
1
10
100
1000
TDS mg/L
Figure D.7 Estimated TDS distributions in residential land use.
Lognormal Probability Plot for TDS in Commercial Land Use
ML Estimates
Percent
99
NON DETECTED
95
90
IGNORE
80
70
60
50
40
30
20
HALF
DETECTION
ESTIMATE
Goodness of Fit
AD*
10
5
1.207
1.342
1.596
1
10
100
1000
TDS mg/L
Figure D.8 Estimated TDS distributions in commercial land use.
D - 26
D.4.3 Total Suspended Solids
This case was similar to the total dissolved solids. The maximum level of nondetected was observed in open space, where about 5% of the observations were censored.
Table D.4 indicates that there are not relevant differences in mean, standard deviation or
coefficient of variation were observed in any of the three methods.
Table D.4 Summary Statistics for Estimated Observations in TSS (mg/L)
RESIDENTIAL
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
991
98.59
3.00
2462
99.84
49.00
179.12
1.79
COMMERCIAL
HD
Ignore Estimate
990
991
0.63
2462
98.53
48.00
178.29
1.81
0.25
2462
98.46
48.00
178.22
1.81
458
98.25
3.00
2385
110.06
42.00
218.51
1.99
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
44
95.45
3.00
980
176.88
48.50
263.04
1.49
HD
Ignore Estimate
457
458
1.56
2385
108.45
41.00
217.22
2.00
0.25
2385
108.18
41.00
217.05
2.01
OPEN SPACE
Land use
INDUSTRIAL
427
99.06
3.00
2490
142.44
78.00
218.76
1.54
HD
426
427
0.43
2490
141.36
76.36
218.35
1.54
0.50
2490
141.12
76.00
218.15
1.55
FREEWAY
HD
Ignore Estimate
44
44
1.22
980
168.98
39.00
259.44
1.54
0.50
980
168.91
39.00
259.49
1.54
134
99.25
3.00
4800
173.39
99.00
448.85
2.59
HD
134
134
3.00
4800
172.13
98.50
447.39
2.60
0.50
4800
172.10
98.50
447.41
2.60
The probability plots indicate that lower values were estimated using half of the
detection limit rather than the Cohen’s method. This indicate that with large number
observations and small percentage of non detected values replace by half of the detection
limit will produce smaller means that those obtained using the maximum likelihood
method. Figure D.9 shows the probability plot for TSS in residential land use. The three
curves overlap indicating than the three methods will produce practically the same result.
D - 27
Lognormal Probability Plot for TSS in Residential Land Use
ML Estimates
99
Percent
95
90
NON DETECTED
IGNORE
80
70
60
50
40
30
20
ESTIMATE
HALF
DETECTION
10
5
Goodness of Fit
AD*
1
0.812
1.029
1.299
0.1
1.0
10.0
100.0
1000.0
TSS mg/L
Figure D.9. Estimated TSS distributions in residential land use.
The probability plot for open space has the lower number of observations among
the 5 land uses. In this case also the pattern observed in the three methods was almost the
same. The coefficient of variation increases only 3% when the censored data is estimated
or replaced by half of the detection limit.
D.4.4 Biochemical Oxygen Demand
The percentage of non-detected values was higher in open space and freeways
compared with the other land uses (Table D.5). The lowest concentrations were observed
in open space with a value close to 6 mg/L. Freeways, commercial and residential land
uses have similar concentrations with 15 mg/L in average. The highest concentration was
observed in industrial land use, however a single unusual observation of 6,920 mg/L has
a significant effect in the mean, standard deviation and coefficient of variation.
D - 28
Table D.5. Summary Statistics for Estimated Observations in BOD (mg/L)
RESIDENTIAL
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
941
97.56
1.00
350
15.05
9.00
22.25
1.48
COMMERCIAL
HD
941
941
1.00
350
14.97
9.00
22.34
1.49
0.50
350
14.84
9.00
22.11
1.49
Ignore Estimate
432
97.45
2.00
150
18.16
11.00
20.25
1.12
Land use
Ignore Estimate
44
86.36
1.00
20
6.25
5.40
4.30
0.69
HD
Ignore Estimate
432
432
0.75
220
18.58
11.00
22.59
1.22
0.50
150
18.14
11.00
20.63
1.14
OPEN SPACE
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
INDUSTRIAL
406
95.32
1.00
6920
35.92
9.00
351.89
9.80
HD
406
406
0.55
6920
34.65
9.00
343.62
9.92
0.50
6920
34.47
9.00
343.61
9.97
FREEWAY
HD
Ignore Estimate
43
44
0.62
20
5.68
4.00
4.34
0.76
0.50
20
5.74
4.00
4.38
0.76
26
84.62
2.0
89
14.86
8.0
18.68
1.26
HD
26
26
1.5
89
13.06
6.5
17.67
1.35
1.5
89
12.88
6.5
17.76
1.38
The lognormal probability plot in industrial land uses shows the unusual
observation. This observation is 35 times larger than the second highest observation. This
unusual value increases the standard deviation almost 18 times compared with the other
land uses. Figure D.10 shows the probability plot for industrial land use.
Open space and freeways had the largest level of non-detected values. The mean
value in open space increases almost 10% when the censored data is ignored. It was not
observed a significant difference in the variance (Figure D.11). Estimated the nondetected or replace by half of the detection limit creates almost the same mean, standard
deviation and coefficient of variation values.
D - 29
Lognormal Probability Plot for BOD in Industrial Land Use
ML Estimates
99
Percent
95
90
IGNORE
ESTIMATE
80
70
60
50
40
30
20
HALF
DETECTION
Goodness of Fit
10
5
AD*
4.411
3.651
3.354
1
1
10
100
1000
10000
BOD mg/L
Figure D.10 Estimated BOD distributions in industrial land use.
Lognormal Probability Plot for BOD in Open Space
ML Estimates
99
95
NON DETECTED
90
IGNORE
Percent
80
ESTIMATE
70
60
50
40
30
HALF
DETECTION
20
Goodness of Fit
10
AD*
0.621
0.632
0.733
5
1
1
BOD mg/L
10
Figure D.11 Estimated BOD distributions in open space land use.
D.4.5 Chemical Oxygen Demand
Differences in the mean, average and coefficient of variation between ignore the
censored data or estimated are not important except for the open space land use where the
D - 30
level of non-detected was close to 25% (Table D.6). In the remaining land uses the level
of non-detected was smaller than 2%.
Table D.6 Summary Statistics for Estimated Observations in COD (mg/L)
RESIDENTIAL
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
796
98.87
5.00
620
74.34
55.00
69.12
0.93
COMMERCIAL
HD
Ignore Estimate
796
796
1.74
620
73.55
53.60
69.12
0.94
0.50
620
73.52
53.60
69.15
0.94
373
98.39
4.00
635
94.11
60.00
94.39
1.00
Land use
Ignore Estimate
44
75.00
8.00
476
51.47
42.10
79.11
1.54
HD
373
373
1.96
635
92.70
59.00
94.28
1.02
0.50
635
92.63
59.00
94.34
1.02
OPEN SPACE
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
INDUSTRIAL
Ignore Estimate
361
98.89
2.00
1260
103.23
60.00
127.35
1.23
HD
361
361
2.00
1260
102.26
59.00
126.97
1.24
2.00
1260
102.17
59.00
127.03
1.24
FREEWAY
HD
Ignore Estimate
44
44
3.70
476
40.93
24.85
70.73
1.73
5.00
476
40.76
24.85
70.78
1.74
HD
67
67
67
98.51
2.44
2.44
2.44
1012.82 1012.82 1012.82
140.99
139.10
138.96
100.00
100.00
100.00
148.89
148.56
148.69
1.06
1.07
1.07
One characteristic of the COD probability plot is that the lower tail does not
follow the trend showed by the rest of the distribution. Figure D.12 shows for example
the COD distribution in residential land uses. This effect is increased when the censored
data is estimated or replaced by half of the detection limit.
The mean value in open space land use can be increase by 25% when the
censored data is ignored (Figure D.13). In the other hand the coefficient of variation can
be reduced in almost 12 % when the non-detected values are ignored. No significant
differences can be observed when the censored data is estimated or replaced by half of
the detection limit.
D - 31
Lognormal Probability Plot for COD in Residential Land Use
ML Estimates
NON DETECTED
Percent
99
95
90
IGNORE
80
70
60
50
40
30
20
HALF
DETECTION
ESTIMATE
Goodness of Fit
10
5
AD*
1.460
3.039
5.344
1
1
10
100
1000
COD mg/L
Figure D.12 Estimated COD distributions in residential land use.
Lognormal Probability Plot for COD in Open Space Land Use
ML Estimates
99
95
NON DETECTED
90
IGNORE
Percent
80
ESTIMATE
70
60
50
40
30
HALF
DETECTION
20
Goodness of Fit
10
AD*
0.887
0.654
1.015
5
1
10
100
1000
COD mg/L
Figure D.13 Estimated COD distributions in open space land use.
D - 32
D.4.6 Ammonia
This constituent has one of the largest levels of censoring observations in all the
land uses (Table 3.16). The percentage of non-detected is close to 20% except for open
space where is more than 80%. The highest concentrations were observed in freeways.
To ignore the censored observations will increase in 15% the mean values. In the other
hand ignoring the non-detected will increase the coefficient of variation in almost 15%
Table D.7 Summary Statistics for Estimated Observations in Ammonia (mg/L)
RESIDENTIAL
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
595
81.51
0.01
5.60
0.47
0.32
0.51
1.09
COMMERCIAL
HD
595
595
0.00
5.60
0.40
0.27
0.48
1.20
0.01
5.60
0.39
0.25
0.48
1.22
Ignore Estimate
299
83.28
0.02
7.80
0.85
0.50
1.02
1.20
OPEN SPACE
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
32
18.75
0.07
1.80
0.64
0.18
0.79
1.24
INDUSTRIAL
HD
Ignore Estimate
299
299
0.00
7.80
0.73
0.41
0.97
1.32
0.01
7.80
0.73
0.40
0.97
1.33
253
83.40
0.03
9.84
0.78
0.47
0.96
1.23
HD
252
253
0.00
9.84
0.68
0.38
0.91
1.35
0.01
9.84
0.68
0.36
0.91
1.35
FREEWAY
HD
Ignore Estimate
32
32
0.02
1.80
0.27
0.25
0.38
1.43
0.01
1.80
0.26
0.25
0.38
1.44
79
87.34
0.08
11.87
1.73
1.07
2.24
1.30
HD
79
79
0.08
11.87
1.53
0.90
2.16
1.41
0.08
11.87
1.52
0.90
2.16
1.42
It was observed in the probability plots that replacing the non-detected by half of
the detection limit estimated lower values than the Cohen detection limit. It was observed
also that the Anderson Darling statistic for normality increases when the censored data is
estimated. Figure D.14 shows the probability plot for ammonia in commercial land uses.
In open space the estimated values seems to don’t follow a log normal distribution
D - 33
(Figure D.15). To estimate the censored observations with more than 80% of nondetected does not indicate that the predicted observations followed a straight line.
Lognormal Probability Plot for Ammonia in Commercial Land Use
ML Estimates
99
Percent
95
90
NON DETECTED
IGNORE
80
70
60
50
40
30
20
ESTIMATE
HALF
DETECTION
Goodness of Fit
10
5
AD*
0.698
1.567
1.211
1
0.001
0.010
0.100
1.000
10.000
Ammonia mg/L
Figure D.14. Estimated ammonia distributions in commercial land use.
Lognormal Probability Plot for Ammonia in Open Space
ML Estimates
99
95
Percent
90
NON DETECTED
80
IGNORE
70
60
50
40
30
ESTIMATED
HALF
DETECTION
20
Goodness of Fit
10
AD*
5
2.129
2.482
6.676
1
0.01
0.10
1.00
Ammonia mg/L
Figure D.15 Estimated Ammonia distributions in open space land use.
10.00
D - 34
D.4.7 NO2 + NO3
The percentage of non-detected values was smaller than 5% in all the land uses
except for open space where the level of censored values was higher than 15%. There are
not significant differences in the mean, standard deviation and coefficient of variation
except for the open space.
Table D.8 Summary Statistics for Estimated Observations in NO2 + NO3 (mg/L)
RESIDENTIAL
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
927
97.41
0.01
18.00
0.76
0.59
0.87
1.14
COMMERCIAL
HD
927
927
0.01
18.00
0.75
0.58
0.86
1.15
0.01
18.00
0.74
0.58
0.86
1.16
Ignore Estimate
425
98.12
0.03
8.21
0.86
0.61
0.91
1.06
OPEN SPACE
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
44
84.09
0.09
3.33
0.99
0.59
0.88
0.89
INDUSTRIAL
HD
Ignore Estimate
425
425
0.02
8.21
0.85
0.60
0.91
1.08
0.01
8.21
0.85
0.60
0.91
1.08
417
96.16
0.02
8.40
0.98
0.73
0.87
0.89
HD
417
417
0.02
8.40
0.95
0.72
0.86
0.91
0.01
8.40
0.94
0.70
0.86
0.91
FREEWAY
HD
Ignore Estimate
44
44
0.02
3.33
0.84
0.50
0.88
1.04
0.05
3.33
0.84
0.50
0.88
1.04
25
96.00
0.10
3.00
0.51
0.28
0.63
1.23
HD
25
25
0.10
3.00
0.50
0.28
0.62
1.23
0.10
3.00
0.50
0.26
0.62
1.25
The probability plots for residential, commercial and industrial land uses show a
different trend for the lower tail of the distribution up to the 10th percentile. The
departures from normality are more evident in the case that the censored observations are
replaced by half of the detection limit than when estimated by Cohen’s method (Figure
D.16). In open space it was observed that when the censored data is estimated or replaced
the coefficient of variation increases almost 17% due the elevated level of censoring
D - 35
(Figure D.17). There were observed not main differences in the mean, standard deviation
and coefficient of variation when the censored values were replaced by half of the
detection limit or estimated using Cohen’s method.
Lognormal Probability Plot for NO2 - NO3 in Commercial Land Use
ML Estimates
99
NON DETECTED
Percent
95
90
IGNORE
80
70
60
50
40
30
20
ESTIMATE
HALF
DETECTION
10
5
Goodness of Fit
AD*
1
0.942
1.778
2.035
0.01
0.10
1.00
10.00
NO2 - NO3 mg/L
Figure D.16 Estimated nitrate - nitrite distributions in commercial land use.
Lognormal Probability Plot for NO2 - NO3 in Open Space Land Use
ML Estimates
99
95
NON DETECTED
90
IGNORE
Percent
80
ESTIMATE
70
60
50
40
30
HALF
DETECTION
20
Goodness of Fit
AD*
10
0.935
0.938
1.051
5
1
0.01
0.10
1.00
10.00
NO2-NO3 mg/L
Figure D.17 Estimated nitrate - nitrite distributions in open space land use.
D - 36
D.4.8 TKN
The level of censoring was smaller than 4% in all land uses except for open space.
The highest concentrations were observed in freeways and the lowest in open space
(Table D.9). Main changes in the coefficient of variation were observed in open space
where using Cohen’s method will increase it in 15% and replacing by half of the
detection limit will increase it in 22%.
Table D.9 Summary Statistics for Estimated Observations in TKN (mg/L)
RESIDENTIAL
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
957
96.76
0.05
36.00
1.96
1.43
2.05
1.05
COMMERCIAL
HD
957
957
0.00
36.00
1.91
1.40
2.04
1.07
0.01
36.00
1.90
1.40
2.04
1.07
Ignore Estimate
449
97.33
0.05
15.00
2.23
1.59
2.08
0.93
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
45
71.11
0.20
4.70
1.35
0.74
1.20
0.89
Ignore Estimate
449
449
0.02
15.00
2.18
1.55
2.07
0.95
0.01
15.00
2.17
1.55
2.08
0.96
OPEN SPACE
Land use
INDUSTRIAL
HD
439
95.90
0.05
25.00
2.23
1.40
2.56
1.15
HD
439
439
0.01
25.00
2.17
1.37
2.53
1.17
0.01
25.00
2.16
1.37
2.54
1.18
FREEWAY
HD
Ignore Estimate
45
45
0.20
4.70
1.08
0.50
1.10
1.02
0.20
4.70
1.03
0.50
1.13
1.09
125
96.80
0.20
36.15
3.29
2.00
4.49
1.37
HD
125
125
0.19
36.15
3.20
1.93
4.44
1.39
0.05
36.15
3.19
1.93
4.45
1.39
Notice that the lognormal probability plot follows a straight line except for the
lower tail up to the 5th percentile (Figure D.18). The effect in the Anderson Darling
statistic is increased when the censored data is estimated. The effect is higher when the
non-detected are replaced by half of the detection limit instead of estimate by the Cohen’s
maximum likelihood estimator. In open space when the level of censoring is elevated and
the number of observations is low the Cohen’s estimated method did not follow a
D - 37
lognormal distribution. In Figure D.19 it seems to exist two groups, but is important to
mention that more than 44% of the total observations were lower than 0.5 mg/L. All the
censored values in this land use were located at 0.5 mg/L.
Lognormal Probability Plot for TKN in Residential Land Use
ML Estimates
99
NON DETECTED
Percent
95
90
IGNORE
80
70
60
50
40
30
20
ESTIMATE
HALF
DETECTION
10
5
Goodness of Fit
AD*
1
2.104
3.594
8.225
0.01
0.10
1.00
10.00
TKN mg/L
Figure D.18. Estimated TKN distributions in residential land use.
Lognormal Probability Plot for TKN in Open Space
ML Estimates
99
95
NON DETECTED
90
IGNORE
Percent
80
ESTIMATE
70
60
50
40
30
HALF
DETECTION
20
Goodness of Fit
10
AD*
1.167
2.210
1.989
5
1
0.1
1.0
TKN mg/L
Figure D.19. Estimated TKN distributions in open space land use.
10.0
D - 38
D.4.9 Dissolved Phosphorus
This constituent has an elevated level of non-detected in all the land uses except
for freeways were only 5% of the observations were censored. The remaining land uses
had levels of censoring varying from 13% to 21%. In general ignore the non-detected
values increased the mean and standard deviation and reduced the coefficient of
variation. Table D.10 shows the descriptive statistics for this constituent.
Table D.10 Summary Statistics for Estimated Observations in Dissolved Phosphorus
(mg/L)
RESIDENTIAL
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
738
84.15
0.009
1.69
0.23
0.17
0.21
0.94
COMMERCIAL
HD
Ignore Estimate
738
738
0.001
1.69
0.20
0.14
0.21
1.04
0.005
1.69
0.20
0.14
0.21
1.05
323
81.11
0.01
1.60
0.21
0.11
0.27
1.24
OPEN SPACE
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
44
79.55
0.010
0.52
0.18
0.13
0.16
0.89
INDUSTRIAL
HD
323
323
0.00
1.60
0.18
0.09
0.25
1.35
0.01
1.60
0.19
0.09
0.25
1.34
Ignore Estimate
325
87.38
0.003
1.60
0.17
0.11
0.20
1.18
HD
325
325
0.003
1.60
0.16
0.10
0.19
1.23
0.003
1.60
0.16
0.10
0.19
1.23
FREEWAY
HD
Ignore Estimate
44
44
0.003
0.52
0.16
0.09
0.15
0.95
0.005
0.52
0.17
0.14
0.15
0.87
22
95.45
0.06
6.97
0.78
0.20
1.66
2.13
HD
22
22
0.01
6.97
0.75
0.20
1.63
2.18
0.01
6.97
0.75
0.20
1.63
2.18
As in the previous cases, ignore the censored observations will produce larger
means. There were not observed practical differences between the maximum likelihood
method and replacing the non-detected values for half of the detection limit (Figure
D.20). In freeways the dissolved phosphorus had the lowest level of censoring. The
D - 39
probability plot indicates that the distribution is heavy in the tails; the slope between the
20th and 60th percentiles is higher than in the tails (Figure D.21).
Lognormal Probability Plot for Dissolved Phosphorus in Industrial Land Use
ML Estimates
99
NON DETECTED
Percent
95
90
IGNORE
ESTIMATE
80
70
60
50
40
30
20
HALF
DETECTION
Goodness of Fit
10
5
AD*
0.863
0.571
0.675
1
0.01
0.10
1.00
Dissolved Phosphorus mg/L
Figure D.20 Estimated dissolved phosphorus distributions in industrial land use.
Lognormal Probability Plot for Dissolved Phosphorus in Freeways
ML Estimates
99
95
NON DETECTED
Percent
90
80
IGNORE
70
60
50
40
30
ESTIMATE
HALF
DETECTION
Goodness of Fit
20
AD*
10
1.650
1.408
1.481
5
1
0.01
0.10
1.00
10.00
Dissolved Phosphorus mg/L
Figure D.21 Estimated dissolved phosphorus distributions in freeways land use.
D - 40
D.4.10 Total Phosphorus
This constituent has a low level of censored observations (smaller than 5%) in all
land uses, except for open space where close to 15% of the observations were nondetected (Table D.11). Variations in the coefficient of variation were not significant
except in open space where ignore the censored observations will reduce the coefficient
of variation on almost 7%.
Table D.11 Summary Statistics for Estimated Observations in Total Phosphorus (mg/L)
RESIDENTIAL
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
963
96.88
0.01
6.90
0.42
0.30
0.47
1.13
COMMERCIAL
HD
Ignore Estimate
963
963
0.00
6.90
0.41
0.30
0.47
1.14
0.01
6.90
0.41
0.30
0.47
1.14
446
95.74
0.02
3.35
0.35
0.22
0.40
1.16
OPEN SPACE
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
46
84.78
0.02
15.40
0.68
0.31
2.43
3.54
INDUSTRIAL
HD
446
446
0.01
3.35
0.34
0.22
0.39
1.16
0.01
3.35
0.34
0.22
0.39
1.16
Ignore Estimate
434
95.85
0.02
7.90
0.46
0.26
0.64
1.39
HD
434
434
0.01
7.90
0.45
0.25
0.63
1.41
0.01
7.90
0.45
0.25
0.63
1.40
FREEWAY
HD
Ignore Estimate
46
46
0.01
15.40
0.59
0.22
2.24
3.77
0.01
15.40
0.60
0.25
2.24
3.74
128
99.22
0.06
7.19
0.43
0.25
0.76
1.76
HD
128
128
0.05
7.19
0.43
0.25
0.76
1.77
0.02
7.19
0.43
0.25
0.76
1.77
When the censored data is ignored the observations followed a lognormal
distribution. However if the non-detected are replaced by half of the detection limit or
estimated, the lower tail has lower values than the expected.
In the open space case there is a combination of factors: there is an unusual
observation 20 times higher than the second highest observation (Figure D.22). The most
frequent non-detected observation was <0.5 mg/L, is from this concentration that the
D - 41
effect of deviation of the lower rail is increased. Notice that replacing the censored
observations by half of the detection limit will produce values smaller than the estimated
by the Cohen’s method. In the freeways plot it was observed that the higher observations
are higher than the lognormal trend. The upper 20th percentile has a different slope than
the remaining observations
Lognormal Probability Plot for Total Phosphorus in Open Space Land Use
ML Estimates
99
NON DETECTED
95
IGNORE
90
ESTIMATE
Percent
80
HALF
DETECTION
70
60
50
40
30
20
Goodness of Fit
AD*
10
1.141
0.910
1.217
5
1
0.01
0.10
1.00
10.00
Total Phosphorus mg/L
Figure D.22 Estimated total phosphorus distributions in open space land use.
D.4.11 Total Cooper
Levels of censoring in this case vary from 1 to 15% among the land uses. When
the non-detected values are estimated or replaced by half of the detection limit the
coefficient of variations increases between 1% and 6%, in addition there is a reduction in
the mean and standard deviation. Table D.12 shows the descriptive statistics for each
method by land use.
D - 42
Table D.12 Summary Statistics for Estimated Observations in Total Cooper (g/L)
RESIDENTIAL
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
799
83.60
1.00
590
21.06
12.00
38.51
1.83
COMMERCIAL
HD
Ignore Estimate
799
799
0.25
590
18.54
10.00
35.70
1.93
0.23
590
18.51
10.00
35.69
1.93
387
92.76
1.50
384
29.02
17.00
42.92
1.48
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
HD
Ignore Estimate
387
387
1.50
384
27.47
15.60
41.73
1.52
1.00
384
27.30
15.00
41.79
1.53
OPEN SPACE
Land use
INDUSTRIAL
415
89.64
1.97
1360
47.00
21.88
93.81
2.00
HD
415
415
1.77
1360
43.37
20.00
89.47
2.06
1.00
1360
42.98
20.00
89.60
2.08
FREEWAY
HD
Ignore Estimate
39
74.36
2.00
210
19.15
10.00
38.97
39
39
2.00
210
15.79
5.30
33.98
2.04
2.15
HD
97
97
2.00
210
15.65
5.00
34.00
97
98.97
5.00
244
48.29
34.70
45.91
5.00
244
47.86
33.40
45.87
5.00
244
47.85
33.40
45.89
2.17
0.95
0.96
0.96
The lognormal probability plot for residential and commercial land uses indicate
that the upper 5th percentile have higher values than the expected if the distribution is
lognormal. This observation is important because the upper tail of the distribution has an
important effect in the mean and standard deviation of the dataset.
In open space estimated or replace the non-detected values reduce the mean and
standard deviation of the distribution in 18% and 13% respectively. The probability plot
for freeways is almost a perfect log-normal trend (Figure D.23). In this case the level of
non-detected values was only 1% and the differences in the coefficients of variations was
also 1%
D - 43
Lognormal Probability Plot for Total Cooper in Open Space
ML Estimates
99
95
NON DETECTED
Percent
90
IGNORE
80
ESTIMATE
70
60
50
40
30
HALF
DETECTION
20
Goodness of Fit
10
AD*
5
1.434
1.064
1.649
1
1
10
Total Cooper g/L
100
Figure D.23 Estimated total cooper distributions in open space land use.
D.4.12 Total Lead
The level of non-detected values varied from 0 to 58%. All the observations in
freeways indicate presence of lead, in addition to the highest concentration among the
land uses. Open land uses had the highest level of non-detected values. There is around a
10% reduction in the coefficient of variation when the censored data is ignored. Table
D.13 shows the descriptive statistics for each method.
The probability plots indicate that to replace the censored data by half of the
detection limit will generate smaller observations than Cohen’s method (Figure D.24). To
estimate the censored values reduces the Anderson Darling statistic, providing a better fit
with a lognormal distribution. In open space, most of the censored values were observed
in percentiles where located in < 40 mg/L, < 50 mg/L and < 100 mg/L. In all land uses
almost 80% of the observations were smaller than 50 mg/L. In open space the estimated
D - 44
mean, standard deviation and coefficient of variation are dubious because most of the
censored observations were located in the upper part of the distribution.
Table D.13 Summary Statistics for Estimated Observations in Total Lead (g/L)
RESIDENTIAL
Ignore Estimate
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
788
71.32
0.50
585
26.00
12.00
48.98
1.88
COMMERCIAL
HD
Ignore Estimate
723
788
0.03
585
21.03
8.20
44.21
2.10
0.10
585
22.08
10.00
43.17
1.96
377
85.41
1.00
689.07
37.42
18.00
59.53
1.59
INDUSTRIAL
HD
355
377
0.21
689.07
34.27
17.00
57.56
1.68
0.35
689.07
33.84
17.00
56.07
1.66
OPEN SPACE
Land use
Ignore Estimate
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
411
76.40
1.00
1200
70.10
25.00
128.57
1.83
Ignore Estimate
45
42.22
0.20
150
28.39
10.00
47.36
29
45
0.08
150
19.21
3.16
40.10
1.67
2.09
107
0.10
150
23.98
10.00
33.70
1.60
450
48.77
25.00
70.74
1.60
450
48.77
25.00
70.74
1.41
1.45
1.45
1.45
NON DETECTED
Percent
95
90
IGNORE
80
70
60
50
40
30
20
ESTIMATE
HALF
DETECTION
10
5
Goodness of Fit
AD*
1.030
0.441
1.543
1
10.0
0.50
1200
57.49
20.00
115.57
2.01
107
99
Total Lead
0.21
1200
59.52
20.00
119.79
2.01
107
100.00
1.60
450
48.77
25.00
70.74
ML Estimates
1.0
411
HD
Lognormal Probability Plot for Total Lead in Industrial Land Use
0.1
377
FREEWAY
HD
100.0
g/L
1000.0
Figure D.24 Estimated total lead distributions in industrial land use.
HD
D - 45
D.4.13 Total Zinc
The percentage of non-detected values was smaller than 4% except for open space
where it was close to 30% (Table D.14). No important changes in the coefficient of
variation were observed except for open space where ignoring the censored values
reduced the coefficient of variation in 13%.
Table D.14 Summary Statistics for Estimated Observations in Total Zinc (g/L)
RESIDENTIAL
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
Ignore Estimate
810
96.42
3.00
1580
116.70
73.00
151.81
1.30
COMMERCIAL
HD
Ignore Estimate
810
810
0.48
1580
113.53
70.00
150.25
1.32
0.30
1580
113.23
70.00
150.24
1.33
392
98.98
5.00
3050
225.32
150.00
275.81
1.22
Land use
Observations
% Detected
Minimum
Maximum
Average
Median
Standard Dev.
Coeff. of Var.
HD
Ignore Estimate
392
392
5.00
3050
224.06
150.00
274.74
1.23
5.00
3050
223.55
150.00
274.96
1.23
OPEN SPACE
Ignore Estimate
INDUSTRIAL
432
98.61
5.77
8100
318.25
209.50
474.36
1.49
HD
432
432
3.05
8100
315.02
204.50
471.89
1.50
2.00
8100
314.34
201.00
472.21
1.50
FREEWAY
HD
Ignore Estimate
45
71.11
5.00
390
72.44
40.00
96.88
45
45
2.00
390
55.90
20.00
85.85
1.34
1.54
HD
93
93
2.50
390
55.62
20.00
85.99
93
96.77
6.00
1829
279.43
200.00
281.16
6.00
1829
271.63
194.49
279.87
2.50
1829
271.52
194.49
279.98
1.55
1.01
1.03
1.03
The probability plot indicates that in the lower tail replacing the non-detected
observations by half of the detection limit will create smaller values than to estimate them
using the Cohen’s method (Figure D.25). In open space if the censored data is estimated
using Cohen’s method there is a reduction in the mean and variance of the dataset of 23%
and 12% respectively, however the coefficient of variation increases by 15% (Figure
D.26).
D - 46
Lognormal Probability Plot for Total Zinc in Residential Land Use
ML Estimates
NON DETECTED
Percent
99
IGNORE
95
90
80
70
60
50
40
30
20
10
5
ESTIMATE
HALF
DETECTION
Goodness of Fit
AD*
2.524
5.012
5.698
1
1
10
100
Total Zinc
1000
 g/L
Figure D.25 Estimated total zinc distributions in residential land use.
Lognormal Probability Plot for Total Zinc in Open Space Land Use
ML Estimates
99
Percent
95
NON DETECTED
90
IGNORE
80
ESTIMATE
70
60
50
40
30
HALF
DETECTION
Goodness of Fit
AD*
20
0.673
0.819
0.840
10
5
1
1
10
Total Zinc
100
 g/L
1000
Figure D.26 Estimated total zinc distributions in open space land use.
D - 47
D.5 Sites With Unusual TSS Concentrations in Main Single Land Uses
In this section is presented the continuation of the example presented in chapter 3,
section 3.4.1. A similar procedure was performed for the commercial, industrial and
mixed areas.
D.5.1 Residential and Mixed Residential Locations
The box plot indicates that there is only one site that seems to have different
distribution that the remaining sites in this group. The site is located in a residentialcommercial area in Wooden Bridge Run, Philadelphia (PAPH1051). There were
collected two samples with concentrations below 15 mg/L at this site. Figure D.27 shows
the box plots for TSS concentration by rain zone and location.
Boxplot TSS in Mixed Residential Landuses
Total Suspended Solids mg/L
1.000
100
10
Figure D.27 Box plot for TSS in mixed residential land uses.
9COCSA004
9KAWISBWY
8IDADA003
7ORPOA005
6CAALAL04
7ORGRA002
5TXPLA001
6AZMCA004
5TXGAA003
5TXGAA001
5TXDAA006
EPA Rain Zone and Location ID
5TXFWA005
5TXDAA003
4TXHCA002
4TXHCA001
3GAFUCOS3
3GACOCOL2
3GAFUCOS1
3ALJCC010
3GACOC1A2
2VAVBTYV5
2VAPMTYP3
2VAFFCOF8
2VANNTNN2
2VAFFCOF7
2VAFFCOF2
2TNKXTYFC
2TNKXTYGV
2NCRASIT3
2PAPH1051
2NCRASIT2
2NCFVROSE
2NCCHROSE
2MDCLCOSD
2KYLOTSR6
2MDBCTYHA
2KYLOTSR5
1MNMISD05
1MABOA001
1MNMISD04
-
D - 48
The results from the Xbar S chart analyses in mixed residential land uses are
presented in Table D.15.
Table D.15 Sites failing Xbar and S chart in Mixed Residential Land Uses
Sites Failing Xbar chart
Rain zone
ALL
1
2
3
4
5
6
7
8
9
9COCSA004(H) 2NCFVROSE(L) 7ORPOA005(H)
2TNKXTYGV(H) 5TXFWA005(H) 2VAVBTYV5(L)
None
Sites Failing S Chart
GAFUCOS3(H)
None
TNKXTYGV(H) VAVBTYV5(L)
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
From 40 mixed residential sites only six sites were out of control. Two sites with
means (log) below the lower control limit were located in the rain zone 2 (North Carolina
and Virginia). From the sites with mean (log) above the upper control limit three were
located in rain zones 5, 7 and 9. Rain zone 9 appears again with a site above the control
limit and rain zone 2 with two sites below the control limit. The site located at long island
creek in Fulton, Georgia shows to have the largest standard deviation among the mixed
residential sites, however, when region 3 is analyzed independently the S chart indicate
that the site is in control. It was observed that three out of four sites in Georgia have
drainage areas larger than 3.000 acres.
There are 21 sites located in mixed residential land use and rain zone 2, 17 with
more than one observation. Two sites, one above the upper control limit and one below
the control limit were observed in the Xbar chart. The site with high median (log) is
located in Gallaher view, Knoxville, Tennessee (TNKXTYGV, 38 observations, median
TSS = 105 mg/L). This site shows construction activity in the north part of the watershed
D - 49
including a self-storage business, north and east of Cedar Hills apartments. The site
located in Holland road, Virginia Beach (VAVBTYV5, 26 observations, median TSS =
32 mg/L) does have wet ponds in the watershed that seem to control high concentrations
but in average is the same as the other mixed residential sites.
The ANOVA analyses indicate that there is at least one rain zone with TSS
concentration different than the other rain zones with a p-value smaller than 1%. The
Dunnett’s comparison test at a family error rate of 5% will indicate that rain zones 5, and
7 have higher concentrations than those observed in rain zone 2.
In summary it was observed at a family error rate of 5% higher concentrations in
rain zones 5 (six sites, median TSS = 108 mg/L) and 7 (two sites, TSS = 175 mg/L)
compared with rain zone 2 (21 sites, TSS = 59 mg/L). The Kruskall-Wallis test indicates
that there is a significant difference in the TSS median concentrations with a p-value
close to zero. The site TNKXTYGV has not the same characteristics than the remaining
residential mixed sites because there was construction activity close to the outfall
location.
D.5.2 Commercial and Mixed Commercial Locations
Box plots, Xbar and S charts and ANOVA were calculated for commercial land
uses. Figure D.28 identifies a site with high TSS concentration in rain zone 4
(KATOJACK, 15 observations, median TSS = 603 mg/L). In general it seems that sites in
rain zone 7 and 9 have higher concentrations than the other rain zones. No other trend or
variation among rain zones was identified from the box plot.
D - 50
Boxplot TSS in Commercial Land Uses
Total Suspended Solids mg/L
1.000
100
10
EPA Rain Zone and Location_ID
Figure D.28 Box plot for TSS in commercial land uses.
The second approach was to identify unusual sites by rain zone using Xbar S
charts. The results in commercial land uses are presented in Table D.16
Table D.16 Sites failing Xbar and S chart in Commercial Land Uses
Sites Failing Xbar chart
Rain zone
ALL
1
2
3
4
5
6
7
9
2MDPGCOS1(H) 2VACHCCC4(L) 3ALHUWERP(L)
4KATOJACK(H) 4TXHCA005(L) 4TXHOA004(L)
9KAWITOWN(H)
None
Sites Failing S Chart
None
None
2MDPGCOS1(H) 2VACHCCC4(L)
None
None
None
4KATOJACK(H)
None
None
None
None
None
None
None
None
None
9CODEA002
9CODEA008
9KAWITOWN
7ORSAA002
9CODEA001
7OREUA001
7ORGRA004
7ORPOA001
6AZTUA003
7ORCCA005
5TXARA001
5TXDAA004
6AZMCA005
4TXHCA003
4TXHCA005
4TXHOA004
3ALMOSITV
4KATOJACK
3ALHUMASM
3ALHUWERP
3ALMODAPH
2VANNTNN3
2VAPMTYP1
2VAHCCOC1
2VAHCCOC2
2VANFTYN4
2VACPTYC4
2VAHATYH1
2NCGRMERR
2VAARLRS3
2VACHCCC4
2NCFVELMS
2NCGRATHE
2MDMOCOWP
2MDPGCOS1
2NCCHSTAR
2MDHOCODC
2MDMOCOBC
2MDBCTYKO
2MDCLCOJS
2MDHACOCF
2KYLXWHL1
2MDAACOPP
1MABOA003
2KYLXNEL3
-
D - 51
The Xbar S plot did not indicate any trend by geographical region for the 45 sites.
Sites with low concentrations were observed in rain zone 2,3 and 4. There were observed
three sites with concentrations above the control limit, one in rain zone 2, another in rain
zone 9, and the site identified by the box plot in rain zone 4.
In rain zone 2 two sites were found outside the control limits. MDPGCOS1 is
located a shopping center in Arena Plaza. There were collected 26 samples at this
location. The area of the watershed is 19.7 acres. The median TSS concentration for this
site is 158 mg/L. No specific reasons were found for high concentrations. The second site
is located in Clover Leaf mall in Chesterfield County (VACHCCC4, 12 observations, 60
acres, median TSS = 14 mg/L). There is not clear evidence in the aerial photographs that
explain the low concentrations found at this location. No sites outside the control limits
were found in other rain zones except for rain zone 4. The outfall is located in Jackson
Street in Topeka, Kansas (KATOJACK). It might be affected by the presence of a sand
quarry close to the watershed. There were collected 16 samples between April 1998 and
Septembers 2002.
The ANOVA analysis indicates that there are significant differences among EPA
rain zones (P-value = 0). The Dunnett’s comparison test with a family error of 5%,
indicates that TSS concentrations compared with EPA rain zone 2 (median TSS = 48
mg/L) are larger in EPA rain zones 4 (median TSS= 82 mg/L) and 9 (median TSS = 128
mg/L). The median TSS in the remaining zones are not statistically different than the
observed in rain zone 2.
There are 24 sites located in mixed commercial land use with more than one
observation. Rain zone 2 has the largest number of sites with 10, followed by region 5
D - 52
with 5 locations. Figure D.29 shows the box plots for mixed commercial land uses by
rain zone.
Boxplot of TSS in Mixed Commercial Land Use
-
100
-
Rain Zone - Location ID
Figure D.29 Box plot for TSS in mixed commercial land uses.
The box plot indicates that there is a mixed commercial site located in Plano,
Texas and a site in Colorado with a higher concentration than the remaining sites in this
category. Because the low number of sites sampled by geographical region it is no
possible to identify if there is a trend by rain zone. Table D.17 indicates those sites
outside the control limits by rain zone and for the all the mixed commercial sites.
9COCSA001
8IDA DA001
7ORSA A001
7ORPOA002
7OREUA002
5TXPLA004
6CA ALAL07
5TXPLA003
5TXMEA001
5TXGA A004
4TNMET260
5TXFWA003
3ALJC004R
3ALJC004L
2VA VBTYV3
2VANFTYN1
2NCRA SIT7
2TNKXTYWE
2NCRA SIT4
2NCCHYARD
2MDPGCOS3
2MDCLCOBP
2MDBACOLC
10
2MDA ACOSC
Total Suspended Solids mg/L
1.000
D - 53
Table D.17 Sites failing Xbar and S chart in Mixed Commercial Land Uses
Sites Failing Xbar chart
Rain zone
ALL
1
2
3
4
5
6
7
9
2TNKXTYWE(L) 2VANFTYN1(H) 2VAVBTYV3(L)
5TXPLA004(H) 9COCSA001(H)
None
Sites Failing S Chart
None
None
2VANFTYN1(H) 2VAVBTYV3(L)
None
None
None
None
None
5TXPLA004(H)
None
None
None
None
None
None
None
The Xbar chart for all mixed commercial observations indicated that sites with
high TSS concentrations were observed in EPA rain zones 5 and 9. In EPA rain zone two
three sites were outside the control limits, two below the lower control limit and one
above the upper control limit. As in the commercial site analyses EPA rain zone 9 seems
to have higher TSS concentrations than the remaining rain zones.
The analysis by rain zone indicates that only rain zones 2 and 5 have sites outside
the control limits. In EPA rain zone 2. the site with high concentrations (VANFTYN1) is
located in Armisted Avenue in Norfolk, Virginia. A total of 28 observations were
collected at this site. The median TSS for this location was 117 mg/L. In the other hand a
site with unusual low median TSS concentration was found in Haygood, Virginia Beach,
Virginia (VAVBTYV3). A total of 33 observations were collected and stored in the
database. The median TSS concentration at this location was 26 mg/L. This site is 79%
commercial and 13% open space.
The site located in EPA rain zone 5 is located in Spring Creek, Plano, Texas
(TXPLA004). There were stored 7 events from this site in the database. The median TSS
concentration is 575 mg/L. No information was found to explain the elevated
D - 54
concentration. Another site that appear to be outside the control limits for all the sites but
not in its group is located in Sixteenth hole valley, Colorado Springs, Colorado. The
median concentration for this site was 251 mg/L. It is more evidence of a possible
geographical region difference. This median concentration will be considered high for
commercial sites located in EPA rain zone 2. The site it has two automobile dealerships
and a gas station, evidence of erosion was observed in the aerial photograph.
The ANOVA analysis indicates that there are significant differences among EPA
rain zones (P-value = 0) in mixed commercial land uses. The Dunnett’s comparison test
with a family error of 5%, indicates that TSS concentrations compared with EPA rain
zone 2 (median TSS = 46 mg/L) are larger in EPA rain zones 5 (median TSS = 72 mg/L)
and 9 (median TSS = 254 mg/L). The median TSS in the remaining zones are not
statistically different than the observed in rain zone 2.
D.5.3 Industrial and Mixed Industrial Locations
Box plots, Xbar and S charts and ANOVA were calculated for industrial land
uses. Figure D.30 shows the box plots by rain zone and location. It was observed that
sites located in EPA rain zones 6 and 9 seem to have higher concentrations than the
remaining industrial sites. An unusual site with two unusual low concentrations was
observed in Boston, Massachusetts.
1M A BO A 004
1M N M IS D03
2KYLO TS R2
2KYLO TS R4
2KYLXTBL2
2M DA A C O M W
2M DBA C O BC
2M DBA C O TC
2M DBC TYBO
2M DBC TYF M
2M DHO C O F M
2M DH O C O O C
2M DP G C O S 5
2M DP G C O S 6
2N C C H BRE V
2N C C H HO S K
2N C F V WIN S
2N C G RHU S T
2V A A RLTC 4
2V A C P TYC 5
2V A F F O F 10
2V A F F O F 11
2V A H A TYH 2
2V A HC C O N 1
2V A HC C O N 2
2V A V BTYV 4
3A LH U C HIP
3A LM O S IIV
3A LM O TH E O
3G A A TA T01
3G A C LC O S I
4KA TO S TF E
4TN M E T211
4TXH C A 004
4TXH O A 002
5TXDA A 001
5TXDA A 002
5TXF WA 004
6A ZM C A 001
6A ZM C A 003
6A ZTU A 004
6C A A LA L09
7O RP O A 003
7O RP O A 004
7O RS A A 003
9C O DE A 007
9KA WIM C LE
Total Suspended Solids (mg/L)
D - 55
Boxplot of TSS in Industrial Land Uses
-
1
2
3
4
5
6
7
9
1.000
ALL
-
100
-
10
-
EPA Rain Zone - Location ID
Figure D.30 Box plot for TSS in industrial land uses.
Table D.18 shows those industrial sites that are outside the control limits by the
pooled dataset and by each EPA rain zone.
Table D.18 Sites failing Xbar and S chart in Industrial Land Uses
Rain zone
1MABOA004(L) 2VACPTYC5(L) 2VAVBTYV4(L)
3GAATAT01(L) 5TXFWA004(H) 6AZMCA003(H) None
6AZTUA004(H)
None
None
MDPGCOS6(H)
VACPTYC5(L)
None
VAVBTYV4(L)
None
None
Sites Failing Xbar chart
Sites Failing S Chart
None
None
TXFWA004(H)
None
AZMCA003(H)
None
None
None
None
None
D - 56
As in the other land uses, sites with concentrations below the control limit were
observed in EPA Rain zones 1,2 and 3. Sites with median concentrations larger than the
upper control limit were located in rain zones 5 and 6. By rain zone three sites were
outside the control limits in rain zone 2, one in rain zone 5, and one in rain zone 6. The
two sites in EPA rain zone 2 with low concentrations were located in Virginia, and the
site with high concentrations was located in Maryland. One of the sites located in
Virginia is located in Cavalier Industrial Park in the city of Chesapeake (VACPTYC5).
This 16 acres site is 92% industrial and the remaining 8% open space. A total of 15
samples were stored from this site in the database during the period 1993 to 1999. The
median TSS concentration for this site is 13 mg/L. No additional information was
observed in the aerial photos that might explain the low concentrations.
The second site was located in Viking Drive, Virginia Beach (VAVBTYV5). This
29-acre site has 55 percent of impervious. There are stored 30 samples from this site in
the database. The samples were collected between 1992 and 1999. The median TSS
concentration is 29 mg/L. Tall trees surround the industrial facility.
The site with elevated concentration in rain zone is located in Pennsy Drive in
Riverdale, Prince George County, Maryland (MDPGCOS6). This 42.4-acre size site has a
grass swale drainage system. There are 30 samples stored in the database from this
location. The samples were collected between 1994 and 1997. The median TSS
concentration is 98 mg/L. The site is located next to Glenridge Elementary School. The
aerial photo shows construction activity by the industry in the northwest part of the
watershed.
D - 57
The site with high concentration in EPA region 5 is located in Dry branch, in Fort
Worth, Texas (TXFWA004). A total of 21 samples are stored in the database. The
median TSS for this location is 288 mg/L. Several open space areas without cover were
observed in the aerial photograph. The site located in EPA region 6 is located in 27th
Avenue at Salt River in Maricopa County Arizona (AZMCA003). There are 27 samples
from this location stored in the database. The median TSS concentration is 660 mg/L.
The scarce vegetation and the type of soils can be the reason of this elevated median
value.
The ANOVA analysis indicates that there are significant differences among EPA
rain zones (P-value = 0) in industrial land uses. The Dunnett’s comparison test with a
family error of 5%, indicates that TSS concentrations compared with EPA rain zone 2
(median TSS = 53 mg/L) are larger for EPA rain zones 4 (median TSS = 92 mg/L), 5
(median TSS = 147 mg/L), 6 (median TSS = 288 mg/L), 7 (median TSS = 120 mg/L),
and 9(median TSS = 170 mg/L). The median TSS in the zones 1 and 3 are not
statistically different than the observed in rain zone 2.
The box plots in mixed industrial land uses are shown in Figure D.31. It was
observed that most of the box plots have the same median except for those located in
EPA rain zone 9. The sites that fail the quality control charts are shown in Table D.19.
Three sites are outside the control limits for mixed industrial land uses. Two sites in
Colorado and one site in North Carolina are out of control. This result is similar to those
observed in the other land uses. When each rain zone was analyzed individually no sites
were found to be out of control.
D - 58
Boxplot TSS in Mixed Indsutrial Land Uses
1.000
100
10
Rain Zone - Location ID
Figure D.31 Box plot for TSS in industrial land uses.
Table D.19 Sites failing Xbar and S chart in Mixed Industrial Land Uses
Rain zone
ALL
2
3
5
6
7
9
Sites Failing Xbar chart
Sites Failing S Chart
9COCSA002(H) 9CODEA006(H) 2NCGRUNIO(L) None
None
None
None
None
None
None
None
None
None
None
None
None
The ANOVA analysis indicates that there are significant differences among EPA
rain zones (P-value = 0) in mixed industrial land uses. The Dunnett’s comparison test
with a family error of 5%, indicates that TSS concentrations compared with EPA rain
zone 2 (median TSS = 82 mg/L) are larger only for EPA rain zone 9 (median TSS = 341
9CODEA006
9CODEA004
9COCSA003
9COCSA002
7ORGRA001
6CA ALAL10
6CA ALAL03
5TXIRA004
5TXIRA003
5TXIRA002
5TXGA A002
5TXARA004
3GAFUCOS2
3ALJCC002
3ALJCC001
2TNKXTYA P
2NCRA SIT6
2NCRA SIT5
2NCGRUNIO
2MDMOCOSL
Total Suspended Solids (mg/L)
-
D - 59
mg/L). The median TSS in the zones 3, 5, 6, and 7 are not statistically different than the
median in rain zone 2.
D.6 Table With Unusual Sites By Land Use
The following tables summarize those sites with unusual concentrations by land
use, constituent and rain zone.
Table D.6 Sites failing Xbar in Residential Land Uses.
CONSTITUENT
COD mg/L
Where
EPA Rain
Zone and
Location_ID
All
Observations
2KYLXTBL1
100.41
12
HIGH
2VAARLCV2
2VACHCOF3
2VACHCOF5
2VANFTYN3
2VAVBTYV1
3ALHUHURI
3GAATAT02
3GACLCOTR
3GACOC1A3
6AZMCA006
6AZTUA001
6AZTUA002
7ORCCA004
7ORSAA004
2KYLXTBL1
2VAARLCV2
2VACHCOF3
2VACHCOF5
2VANFTYN3
2VAVBTYV1
5TXARA002
6AZTUA001
22.17
16.83
28.19
81.17
88.64
18.94
26.20
31.75
13.75
110.20
242.29
100.84
18.11
22.06
100.41
22.17
16.83
28.19
81.17
88.64
41.59
242.29
9
11
16
28
25
9
9
22
6
20
12
12
5
5
12
9
11
16
28
25
20
12
LOW
LOW
LOW
HIGH
HIGH
LOW
LOW
LOW
LOW
HIGH
HIGH
HIGH
LOW
LOW
HIGH
LOW
LOW
LOW
HIGH
HIGH
LOW
HIGH
By Group
By Group
By Group
Median
Sample Size Characteristic
D - 60
Table D.6xx Sites failing Xbar in Residential Land Uses
CONSTITUENT
NO2 + NO3
(mg/L)
Where
EPA Rain
Zone and
Location ID
All
Observations
1MNMISD01
0.19
10
LOW
1MNMISD02
2KYLXTBL1
2VAHATYH3
2VAVBTYV2
4TXHOA005
6AZMCA006
6AZTUA001
7ORPOA006
2KYLXTBL1
2MDBACOSC
2NCCHHIDD
2VAVBTYV2
0.19
1.15
0.30
0.05
1.04
1.13
1.34
0.22
1.15
0.79
1.38
0.05
9
12
17
29
15
20
12
10
12
26
5
29
LOW
HIGH
LOW
LOW
HIGH
HIGH
HIGH
LOW
HIGH
HIGH
HIGH
LOW
By Group
Median
Sample Size Characteristic
Table D.6XX Sites failing Xbar in Residential Land Uses
CONSTITUENT
Ammonia
(mg/L)
Where
EPA Rain
Zone and
Location ID
All
2NCCHSIMS
Observations
2VAPMTYP4
Sample Size Characteristic
1.68
4
HIGH
17
LOW
29
LOW
4
HIGH
20
HIGH
9
LOW
6
LOW
5
HIGH
4
HIGH
16
HIGH
2VAPMTYP4
0.10
0.05
1.08
0.89
0.07
0.07
0.83
1.68
0.47
0.10
17
LOW
2VAVBTYV2
0.05
29
LOW
2VAVBTYV2
4TNMET231
6AZMCA006
7OREUA003
7ORGRA003
2NCCHHIDD
By Group
Median
2NCCHSIMS
2NCGRWILL
D - 61
Table D.6 Sites failing Xbar in Residential Land Uses
CONSTITUENT
TDS (mg/L)
Where
EPA Rain
Zone and
Location ID
All
2KYLOTSR3
Observations
2KYLXTBL1
2NCFVCLEA
2VAARLLP1
2VACHCOF3
2VACHCOF5
2VACPTYC3
2VAFFCOF1
2VAVBTYV2
4KATOATWO
4KATOBROO
6AZTUA001
7ORPOA006
7ORSAA004
9KAWIHUNT
By Group
2KYLOTSR3
2KYLXTBL1
2NCFVCLEA
2VAARLLP1
2VACHCOF3
2VACHCOF5
2VAFFCOF1
2VAVBTYV2
By Group
4TXHOA003
By Group
6AZTUA001
Median
Sample Size Characteristic
294.6
3
HIGH
124.8
30.96
247.8
28.10
40.95
41.38
22.92
112.8
191.2
169.9
147.6
33.07
20.93
118.7
294.7
124.8
30.96
247.8
28.10
40.95
22.92
112.8
63.28
147.6
12
14
8
11
16
15
3
29
15
16
13
11
5
16
3
12
14
8
11
16
3
29
14
13
HIGH
LOW
HIGH
LOW
LOW
LOW
LOW
HIGH
HIGH
HIGH
HIGH
LOW
LOW
HIGH
HIGH
HIGH
LOW
HIGH
LOW
LOW
LOW
HIGH
LOW
HIGH
Table D.6 Sites failing Xbar in Residential Land Uses
CONSTITUENT
Where
EPA Rain
Zone and
Location ID
Oil and Grease
All
2VACHCOF3
(mg/L)
Observations
5TXIRA001
By Group
2VACHCOF3
By Group
7ORGRA003
Median
Sample Size Characteristic
0.34
11
LOW
7.38
0.34
0.57
21
11
7
HIGH
LOW
LOW
D - 62
Table D.6XX Sites failing Xbar in Residential Land Uses
CONSTITUENT
Hardness
(mg/L)
Where
EPA Rain
Zone and
Location ID
All
Observations
2KYLXTBL1
68.08
12
HIGH
2VACHCN1A
2VACHCOF3
2VACHCOF5
7ORPOA006
9KAWIHUNT
2KYLXTBL1
2VACHCOF3
5TXMEA002
7OREUA003
7ORPOA006
85.55
9.64
18.79
8.76
51.00
68.08
9.64
49.42
28.37
8.76
4
7
12
11
15
12
7
7
20
6
HIGH
LOW
LOW
LOW
HIGH
HIGH
LOW
HIGH
HIGH
LOW
By Group
By Group
By Group
Median
Sample Size Characteristic
Table D.6 Sites failing Xbar in Residential Land Uses
CONSTITUENT
BOD (mg/L)
Where
EPA Rain
Zone and
Location ID
All
Observations
2KYLXTBL1
21.09
12
HIGH
2MDHACOBP
2MDHOCOGM
2NCCHNANC
2VAARLCV2
2VAVBTYV2
3GACOC1A3
4KATOBROO
6AZTUA001
6AZTUA002
7ORGRA003
7ORSAA004
2KYLXTBL1
2MDHACOBP
2MDHOCOGM
2NCCHNANC
2VAARLCV2
2VAVBTYV2
3GACOC1A3
2.95
46.54
2.81
3.66
6.16
2.42
20.99
32.99
24.45
3.27
2.50
21.09
2.95
46.54
2.81
3.66
6.16
2.42
18
3
4
9
29
6
16
9
9
6
5
12
18
3
4
9
29
6
LOW
HIGH
LOW
LOW
LOW
LOW
HIGH
HIGH
HIGH
LOW
LOW
HIGH
LOW
HIGH
LOW
LOW
LOW
LOW
By Group
By Group
Median
Sample Size Characteristic
D - 63
Table D.6 Sites failing Xbar in Residential Land Uses
CONSTITUENT
TKN (mg/L)
Where
EPA Rain
Zone and
Location ID
All
1MNMISD01
Observations
2KYLXTBL1
2MDBCTYHR
2MDHACOBP
2VAHATYH3
6AZMCA006
6AZTUA001
7ORGRA003
7ORSAA004
By Group
2KYLXTBL1
2MDBCTYHR
2MDHACOBP
2VAHATYH3
By Group
7ORSAA004
Median
Sample Size Characteristic
3.07
10
HIGH
2.92
10.56
0.74
0.77
3.23
3.73
0.55
0.12
2.92
10.56
0.74
0.77
0.12
12
2
18
17
19
13
6
5
12
2
18
17
19
HIGH
HIGH
LOW
LOW
HIGH
HIGH
LOW
LOW
HIGH
HIGH
LOW
LOW
LOW
D - 64
Table D.6 Sites failing Xbar in Residential Land Uses
CONSTITUENT
Dissolved
Phosphorus
(mg/L)
Where
EPA Rain
Zone and
Location ID
Median
Sample Size Characteristic
All
2VACHCOF3
Observations
0.05
10
LOW
2VAHATYH3
2VAHATYH5
2VANFTYN2
2VAVBTYV1
2VAVBTYV2
3GACLCOTR
4TXHCA006
5TXIRA001
6AZMCA006
7ORGRA003
7ORSAA004
2KYLXTBL1
2NCGRWILL
2VACHCOF3
2VAHATYH3
2VAHATYH5
2VANFTYN2
2VAPMTYP2
2VAPMTYP4
2VAVBTYV1
0.07
0.24
0.27
0.22
0.04
0.03
0.38
0.24
0.25
0.03
0.01
0.25
0.24
0.05
0.07
0.24
0.27
0.07
0.07
0.22
14
17
30
26
29
22
6
21
20
6
2
12
15
10
14
17
30
17
17
26
LOW
HIGH
HIGH
HIGH
LOW
LOW
HIGH
HIGH
HIGH
LOW
LOW
HIGH
HIGH
LOW
LOW
HIGH
HIGH
LOW
LOW
HIGH
By Group
D - 65
Table D.6 Sites failing Xbar in Residential Land Uses
CONSTITUENT
Total
Phosphorus
(mg/L)
Where
EPA Rain
Zone and
Location ID
All
1MNMISD01
Observations
By Group
By Group
By Group
1MNMISD02
2KYLXTBL1
2MDHACOBP
2MDPGCOS2
2NCFVCLEA
2VACHCOF3
2VACHCOF5
2VANFTYN2
2VAVBTYV1
3ALHUHURI
3GACLCOTR
3GACOC1A3
4TXHCA006
5TXIRA001
6AZTUA001
7ORCCA004
7ORSAA004
1MABOA006
2KYLXTBL1
2MDHACOBP
2MDPGCOS2
2NCFVCLEA
2VAARLCV2
2VACHCOF3
2VACHCOF5
2VANFTYN2
2VAVBTYV1
5TXARA003
Median
Sample Size Characteristic
0.64
10
HIGH
0.62
0.66
0.12
0.45
0.14
0.06
0.17
0.44
0.49
0.13
0.13
0.09
0.78
0.53
0.60
0.09
0.08
0.19
0.66
0.12
0.45
0.14
0.15
0.06
0.17
0.44
0.49
0.27
9
12
18
63
13
9
14
30
26
9
22
6
6
21
13
5
5
3
12
18
63
13
9
9
14
30
26
7
HIGH
HIGH
LOW
HIGH
LOW
LOW
LOW
HIGH
HIGH
LOW
LOW
LOW
HIGH
HIGH
HIGH
LOW
LOW
LOW
HIGH
LOW
HIGH
LOW
LOW
LOW
LOW
HIGH
HIGH
LOW
D - 66
Table D.6 Sites failing Xbar in Residential Land Uses
CONSTITUENT
Copper (µg/L)
Where
EPA Rain
Zone and
Location ID
All
1MABOA006
Observations
2KYLXTBL1
2MDBCTYHR
2MDHACOBP
2MDPGCOS2
2MDSHDTPS
2NCFVCLEA
2NCFVTRYO
2VACHCN2A
2VACHCOF3
2VACPTSF2
2VAVBTYV2
3GAATAT02
3GACLCOTR
4KATOATWO
4KATOBROO
4TXHOA005
5TXMEA003
6AZMCA006
6AZTUA001
6AZTUA002
By Group
1MABOA006
By Group
2KYLXTBL1
2MDBCTYHR
2MDHACOBP
2MDPGCOS2
2MDSHDTPS
2NCFVCLEA
2NCFVTRYO
2VACHCN2A
2VACHCOF3
2VACPTSF2
2VAVBTYV2
By Group
3GAATAT02
By Group
4TXHOA003
By Group
6AZMCA006
Median
Sample Size Characteristic
152.32
3
HIGH
46.61
161.6
5.49
20.05
27.93
5.36
4.64
2.50
25.77
1.82
2.69
32.11
4.15
20.94
27.98
29.41
4.36
22.15
4.25
2.80
152.32
46.61
161.65
5.49
20.05
27.93
5.36
4.64
2.50
25.77
1.82
2.69
32.1
9.23
22.15
12
3
18
63
10
14
18
4
11
3
3
8
22
15
16
16
7
20
11
10
3
12
3
18
63
10
14
18
4
11
3
3
8
13
20
HIGH
HIGH
LOW
HIGH
HIGH
LOW
LOW
LOW
HIGH
LOW
LOW
HIGH
LOW
HIGH
HIGH
HIGH
LOW
HIGH
LOW
LOW
HIGH
HIGH
HIGH
LOW
HIGH
HIGH
LOW
LOW
LOW
HIGH
LOW
LOW
HIGH
HIGH
HIGH
D - 67
Table D.6 Sites failing Xbar in Residential Land Uses
CONSTITUENT
Lead (µg/L)
Where
EPA Rain
Zone and
Location ID
All
1MNMISD01
Observations
2KYLXTBL1
2MDAACORK
2MDBACOSC
2MDHACOBP
2MDPGCOS2
2NCFVCLEA
2NCFVTRYO
2NCGRWILL
2VAARLCV2
2VAARLLP1
2VAPMTYP2
3ALHUHURI
5TXDAA005
5TXIRA001
6AZMCA006
6AZTUA002
7ORCCA004
7OREUA003
7ORSAA004
By Group
2KYLXTBL1
2MDAACORK
2MDBACOSC
2MDHACOBP
2MDPGCOS2
2NCFVTRYO
2VAARLCV2
2VAARLLP1
2VANFTYN3
2VANFTYN5
2VAPMTYP2
By Group
5TXARA002
5TXDAA005
5TXIRA001
5TXMEA003
By Group
6AZTUA002
By Group
7ORCCA004
7OREUA003
7ORSAA004
Median
Sample Size Characteristic
34.32
10
HIGH
34.95
1.25
1.84
1.46
30.95
3.56
2.90
3.68
1.16
1.78
186.4
75.50
39.75
18.87
22.53
1.38
1.53
20.89
1.73
34.95
1.25
1.84
1.46
30.95
2.90
1.16
1.78
18.82
17.80
186.4
6.26
39.75
18.87
5.49
1.38
1.53
20.89
1.73
11
3
26
18
63
14
15
16
9
8
3
2
6
21
20
8
5
15
5
11
3
26
18
63
15
9
8
11
11
3
20
6
21
7
8
6
21
20
HIGH
LOW
LOW
LOW
HIGH
LOW
LOW
LOW
LOW
LOW
HIGH
HIGH
HIGH
HIGH
HIGH
LOW
LOW
HIGH
LOW
HIGH
LOW
LOW
LOW
HIGH
LOW
LOW
LOW
HIGH
HIGH
HIGH
LOW
HIGH
HIGH
LOW
LOW
LOW
HIGH
LOW
D - 68
Table D.6 Sites failing Xbar in Residential Land Uses
CONSTITUENT
Zinc (µg/L)
Where
EPA Rain
Zone and
Location_ID
All
Observations
2KYLXTBL1
160.3
12
HIGH
2MDHACOBP
2MDPGCOS2
2VAARLLP1
2VACHCN2A
2VACPTSF2
2VAHATYH5
2VAPMTYP5
2VAVBTYV2
6AZMCA006
7ORSAA004
9KAWIHUNT
2KYLXTBL1
2MDCLCOCE
2MDHACOBP
2MDPGCOS2
2VAARLLP1
2VACHCN2A
2VACPTSF2
2VAHATYH5
2VAPMTYP5
2VAVBTYV2
7ORSAA004
26.82
190.7
17.01
8.66
7.94
12.02
15.10
4.11
184.1
18.76
147.9
160.26
270.76
26.82
190.70
17.01
8.66
7.94
12.02
15.10
4.11
18.76
18
63
8
4
3
3
3
3
20
5
14
12
3
18
63
8
4
3
3
3
3
20
LOW
HIGH
LOW
LOW
LOW
LOW
LOW
LOW
HIGH
LOW
HIGH
HIGH
HIGH
LOW
HIGH
LOW
LOW
LOW
LOW
LOW
LOW
LOW
By Group
By Group
Median
Sample Size Characteristic
Table D.6 Sites failing Xbar in Commercial Land Uses
CONSTITUENT
COD (mg/L)
Where
EPA Rain
Zone and
Location ID
All
2VAARLRS3
Observations
2VACHCCC4
3ALHUMASM
3ALHUWERP
4TXHCA005
6AZMCA005
6AZTUA003
By Group
2VAARLRS3
2VACHCCC4
Median
Sample Size Characteristic
27.18
8
LOW
34.75
30.13
16.59
25.92
178.7
257.9
27.18
34.75
13
9
9
6
25
11
8
13
LOW
LOW
LOW
LOW
HIGH
HIGH
LOW
LOW
D - 69
Table D.6 Sites failing Xbar in Commercial Land Uses
CONSTITUENT
NO2 + NO3
(mg/L)
Where
EPA Rain
Zone and
Location ID
All
2MDAACOPP
Observations
6AZMCA005
7OREUA001
7ORPOA001
9CODEA001
By Group
2MDAACOPP
Median
Sample Size Characteristic
0.30
26
LOW
1.07
0.26
0.25
2.54
0.30
23
14
12
3
26
HIGH
LOW
LOW
HIGH
LOW
Table D.6 Sites failing Xbar in Commercial Land Uses
CONSTITUENT
Ammonia
(mg/L)
Where
EPA Rain
Zone and
Location ID
All
Observations
2KYLXNEL3
0.05
11
LOW
6AZMCA005
7ORGRA004
7ORPOA001
2KYLXNEL3
1.95
0.11
0.14
0.05
23
6
11
11
HIGH
LOW
LOW
LOW
By Group
Median
Sample Size Characteristic