Evaluation of Stormwater Notions Using the Nsqd

Evaluation of Stormwater Notions Using the Nsqd

CHAPTER V

EVALUATION OF STORMWATER NOTIONS USING THE NSQD

5.1 Introduction

A classical approach to classify urban sites is based on land uses. This approach is generally accepted because is related with the activity in the watershed. Two drainage areas with the same size, percentage of impervious, slope, sampling methods, and controls will produce different stormwater concentrations if the main activity in the watershed is a metallurgical company (industrial) or a shopping center (commercial) for example. It is expected a higher concentration of metals in the industrial site due manufacturing processes involved in the plant, while in the commercial site a higher concentration of oil an grease due the frequency in the customer entering and leaving the parking lots.

The results from the previous chapter indicate that there are significant differences in the stormwater constituents based on land use. This is supported for other databases like NURP, CDM and USGS. The question is if there is a better classification of stormwater constituents based on rain zone, season, percentage of impervious, watershed size, or the interactions between them.

This chapter presents several approaches trying to explain the variability in stormwater discharges by location using the previous factors.

5.2 Main Factors

Rain zone, percentage of impervious, watershed size, land use and type of conveyance, controls in the watershed, sample analysis and type of sampling were selected for the preliminary analyses. All the mixed land uses were combined according to the land use with the highest percentage in the watershed. Single land uses will be used in the analysis and the mixed land uses will be used during the verification. The first step was an inventory of the total number of events in each of the possible combinations using the variables included in the database. Rain zone, land use, type of conveyance, controls in the watershed, sample analysis and type analysis are discrete variables, while percentage of impervious and watershed size are continuous. The total counts for each discrete variable is shown in table 5.1.

Table 5.1 Total samples and sites by land use

Land use

/ Events / % / Rain Zone / Events / % / Controls / Events / Percent
Residential / 1042 / 27.68 / 1 / 69 / 1.83 / Channel (CW) / 30 / 0.80
Mixed Residential / 611 / 16.23 / 2 / 2000 / 53.12 / Dry Pond (DP) / 50 / 1.33
Commercial / 527 / 14.00 / 3 / 266 / 7.07 / Detention Storage (DS) / 17 / 0.45
Mixed Commercial / 324 / 8.61 / 4 / 212 / 5.63 / Wet Pond (WP) / 113 / 3.00
Industrial / 566 / 15.03 / 5 / 485 / 12.88 / WP Watershed (WP_W) / 182 / 4.83
Mixed Industrial / 249 / 6.61 / 6 / 356 / 9.46 / WP in Series (WP_S) / 42 / 1.12
Institutional / 18 / 0.48 / 7 / 229 / 6.08 / None / 3331 / 88.47
Open Space / 49 / 1.30 / 8 / 24 / 0.64
Mixed Open Space / 168 / 4.46 / 9 / 124 / 3.29
Freeways / 185 / 4.91
Mixed Freeways / 26 / 0.69

Sample Analysis

/ Events / % / Sampler / Events / % / Type of Conveyance / Events / Percent
Composite no specified / 718 / 19.07 / Automatic / 3055 / 81.14 / Curb and gutter / 2454 / 65.18
Flow Composite / 2752 / 73.09 / Manual / 393 / 10.44 / Grass swale / 344 / 9.14
Time Composite / 295 / 7.84 / Not specified / 317 / 8.42 / Not specified / 967 / 25.68

Table 5.1 indicates that most of the events stored in the database were located in residential, mixed residential, commercial and industrial areas. It was also observed that the rain zones 2 thru 7 and 9 have each more than three percent of the total events stored in the database. Rain zones 1 and 8 together have less than three percent of the total events. In the previous chapter it was observed that wet ponds could have an important reduction in the concentration of suspended solids. About 88% of the events have no controls, three percent have wet ponds before the outfall, about 5 percent included ponds in the watershed but not at the outfall, and only one percent includes ponds as series. The evaluation for ponds should be done in specific for each constituent.

About 80 percent of the samples were collected using automatic samplers. It was observed that in some cases manual sampling have lower TSS concentrations that in those cases when automatic samplers were used. This can be explained because in cases when the sampling team arrives late to the site they can miss the first flush (if existed) having smaller concentrations. About 73% of the events were collected using flow composite samples, 8% time composite and in about 19% was not possible to identify which method was used. Flow composite samples are more accurate than time composite samples. Respect to type of conveyance and almost 66% of the events were collected in sites with curb and gutters, 9% using grass swales and in about 25% was not possible to identify which system was used. The grass swales are useful to reduce the concentration of suspended solids, and metals. They also have the advantage to infiltrate the water reducing the peak of the discharge.

5.3 Effect of Stormwater Controls on Stormwater Quality

It is hoped that stormwater controls located in a watershed, or at the outfall, would result in significant reductions in stormwater pollutant concentrations. Figure 5.1 shows the effect on effluent TSS concentrations when using various controls in residential area watersheds in EPA Rain Zone 2 (Maryland, Virginia, North Carolina, Tennessee and Kentucky). The controls noted for these locations included:

 weir: a flow measurement weir in an open channel that forms a small pool, a very small wet pond.

 dry pond (DP): a dry detention pond that drains completely between each storm event.

 detention storage (DS): Oversize pipes with small outlet orifices, usually under parking lots.

 wet pond (WP): a wet detention pond that retains water between events, forming a small lake or pond. If the pond is in the watershed but not before the outfall will be considered wet pond inside the watershed (WPM).

Figure 5.1. TSS distribution by controls in residential areas and rain zone 2

The stormwater monitoring was conducted at the outfalls of the drainage areas, after the stormwater controls. Wet ponds are seen to reduce the TSS concentration in the stormwater more than the other controls. Detention storage units and dry ponds also reduced the TSS concentrations, but to a smaller extent. Only one site (located in Virginia Beach) had a channel and weir control, but that did not reduce the observed TSS concentration.

For each constituent it was analyzed if there is a significant difference in the controls, sample analysis, sampler instrument and type of conveyance. As example TSS will be used to identify those sites with similar concentrations.

The first step was to identify how many observations existed in the database for the different possibilities. The following four single land uses and rain zones have sites with controls: residential, commercial and industrial in rain zone 2 and industrial in rain zone 3. For each group one-way ANOVA analyses were used to identify if there is any difference in the concentration of 13 constituents for those sites that include controls. Dunnet’s method was also used to compare if sites with a certain control are significantly different to sites without control using a family error rate of 5%. Table 5.2 shows the results from the ANOVA analyses for each group.

1

Table 5.2.a. One-Way ANOVA Results by Control in Residential Land Use, Rain Zone 2*

Hardness mg/L / Oil and Grease mg/L / TDS mg/L / TSS mg/L
p-value / 0.024 / 0.999 / 0 / 0
n / median / Dunnet / n / median / Dunnet / n / median / Dunnet / n / median / Dunnet
Weir / 3 / 2.50 / = / 29 / 112.80 / 29 / 55.54 / =
DP / 3 / 2.68 / = / 3 / 58.88 / = / 21 / 26.67 / =
DS / 7 / 44.38 / = / 9 / 2.19 / = / 8 / 98.45 / = / 9 / 14.46
No Control / 61 / 30.77 / 202 / 2.38 / 424 / 62.42 / 559 / 40.10
WP / 10 / 66.45 / 13 / 2.50 / = / 12 / 120.39 / 12 / 9.25
BOD mg/L / COD mg/L / Ammonia mg/L / NO2 + NO3 mg/L
p-value / 0 / 0 / 0 / 0
n / median / Dunnet / n / median / Dunnet / n / median / Dunnet / n / median / Dunnet
Weir / 29 / 6.16 / 29 / 49.02 / = / 29 / 0.05 / 29 / 0.05
DP / 21 / 3.44 / 3 / 33.45 / = / 3 / 0.41 / = / 21 / 0.59 / =
DS / 9 / 3.66 / 9 / 22.17 / 9 / 0.40 / = / 9 / 0.98 / =
No Control / 533 / 11.07 / 418 / 56.91 / 409 / 0.24 / 546 / 0.54
WP / 12 / 3.10 / 12 / 24.58 / 12 / 0.07 / 12 / 0.28
TKN mg/L / Dissolved Phosphorus mg/L / Total Phosphorus mg/L / Total Copper g/L
p-value / 0.012 / 0 / 0 / 0
n / median / Dunnet / n / median / Dunnet / n / median / Dunnet / n / median / Dunnet
Weir / 29 / 1.49 / = / 29 / 0.04 / 29 / 0.23 / = / 3 / 2.69
DP / 21 / 0.79 / 3 / 0.15 / = / 21 / 0.12 / 21 / 6.16
DS / 9 / 1.38 / = / 8 / 0.11 / = / 9 / 0.15 / 9 / 20.75
No Control / 549 / 1.34 / 404 / 0.14 / 550 / 0.30 / 403 / 11.01
WP / 12 / 1.04 / = / 12 / 0.03 / 12 / 0.07 / 4 / 3.13
Total Lead g/L / Total Zinc g/L
p-value / 0 / 0
n / median / Dunnet / n / median / Dunnet
Weir / 3 / 6.41 / 3 / 4.11
DP / 21 / 1.50 / 21 / 29.63
DS / 9 / 1.16 / 9 / 103.25
No Control / 364 / 7.73 / 405 / 67.56
WP / 4 / 1.00 / 4 / 10.44

Table 5.2.b One-Way ANOVA Results by Control in Commercial Land Use, Rain Zone 2

Hardness / Oil and Grease mg/L / TDS mg/L / TSS mg/L
p-value / 0.717 / 0.082 / 0.477 / 0
n / median / Dunnet / n / median / Dunnet / n / median / Dunnet / n / median / Dunnet
DS / 8 / 58.17 / = / 8 / 1.84 / = / 8 / 100.69 / = / 8 / 19.54
No Control / 35 / 58.97 / 100 / 4.20 / 174 / 74.89 / 244 / 48.13
WP / 11 / 71.80 / = / 17 / 2.84 / = / 26 / 89.99 / = / 26 / 19.47
WPW / 9 / 47.11 / = / 13 / 3.36 / = / 13 / 71.12 / = / 13 / 16.85
BOD mg/L / COD mg/L / Ammonia mg/L / NO2 + NO3 mg/L
p-value / 0 / 0 / 0 / 0
n / median / Dunnet / n / median / Dunnet / n / median / Dunnet / n / median / Dunnet
DS / 8 / 4.44 / 8 / 27.18 / 8 / 0.30 / = / 8 / 1.18 / =
No Control / 241 / 14.66 / 174 / 73.62 / 174 / 0.39 / 242 / 0.60
WP / 26 / 7.06 / 26 / 35.99 / 26 / 0.13 / 26 / 0.48 / =
WPW / 12 / 5.41 / 13 / 23.88 / 13 / 0.16 / 13 / 0.22
TKN mg/L / Dissolved Phosphorus mg/L / Total Phosphorus mg/L / Total Copper g/L
p-value / 0.057 / 0 / 0 / 0
n / median / Dunnet / n / median / Dunnet / n / median / Dunnet / n / median / Dunnet
DS / 8 / 1.04 / = / 7 / 0.09 / = / 8 / 0.16 / = / 8 / 14.14 / =
No Control / 241 / 1.59 / 161 / 0.11 / 238 / 0.25 / 194 / 17.53
WP / 26 / 1.19 / = / 25 / 0.05 / = / 26 / 0.13 / 6 / 5.57
WPW / 13 / 1.03 / = / 13 / 0.03 / = / 13 / 0.08 / 4 / 6.00
Total Lead g/L / Total Zinc g/L
p-value / 0 / 0
n / median / Dunnet / n / median / Dunnet
DS / 8 / 1.61 / 8 / 82.57
No Control / 194 / 16.41 / 197 / 188.02
WP / 7 / 4.90 / 7 / 44.26
WPW / 4 / 2.49 / 4 / 39.68

Table 5.2.c One-Way ANOVA Results by Control in Industrial Land Use, Rain Zone 2

Hardness mg/L / Oil and Grease mg/L / TDS mg/L / TSS mg/L
p-value / none / 0 / none / 0.693
n / median / Dunnet / n / median / Dunnet / n / median / Dunnet / n / median / Dunnet
No Control / 81 / 3.85 / 205 / 51.96
WP / 37 / 1.43 / 29 / 48.05 / =
BOD mg/L / COD mg/L / Ammonia mg/L / NO2 + NO3 mg/L
p-value / 0.466 / none / none / 0
n / median / Dunnet / n / median / Dunnet / n / median / Dunnet / n / median / Dunnet
No Control / 200 / 10.63 / 197 / 0.61
WP / 29 / 9.30 / = / 29 / 0.30
TKN mg/L / Dissolved Phosphorus mg/L / Total Phosphorus mg/L / Total Copper g/L
p-value / 0.166 / none / 0 / 0
n / median / Dunnet / n / median / Dunnet / n / median / Dunnet / n / median / Dunnet
No Control / 198 / 1.22 / 200 / 0.23 / 150 / 16.00
WP / 29 / 0.98 / = / 29 / 0.09 / 29 / 7.38
Total Lead g/L / Total Zinc g/L
p-value / 0.353 / 0
n / median / Dunnet / n / median / Dunnet
No Control / 142 / 11.16 / 157 / 180.01
WP / 29 / 8.66 / = / 29 / 60.44

Table 5.2.d One-Way ANOVA Results by Control in Industrial Land Use, Rain Zone 3

Hardness mg/L / Oil and Grease mg/L / TDS mg/L / TSS mg/L
p-value / None / None / 0.112 / 0.281
n / median / Dunnet / n / median / Dunnet / n / median / Dunnet / n / median / Dunnet
No Control / 44 / 69.53 / 44 / 48.35
WP / 25 / 49.84 / = / 25 / 70.40 / =
BOD mg/L / COD mg/L / Ammonia mg/L / NO2 + NO3 mg/L
p-value / 0.221 / 0.395 / 0.165 / 0.193
n / median / Dunnet / n / median / Dunnet / n / median / Dunnet / n / median / Dunnet
No Control / 44 / 6.41 / 44 / 37.00 / 3 / 0.12 / 30 / 0.57
WP / 23 / 5.14 / = / 25 / 43.06 / = / 25 / 0.03 / = / 25 / 0.40 / =
TKN mg/L / Dissolved Phosphorus mg/L / Total Phosphorus mg/L / Total Copper g/L
p-value / 0.807 / 0.191 / 0.438 / 0.106
n / median / Dunnet / n / median / Dunnet / n / median / Dunnet / n / median / Dunnet
No Control / 43 / 1.18 / 39 / 0.07 / 43 / 0.16 / 38 / 16.66
WP / 25 / 1.12 / = / 25 / 0.06 / = / 25 / 0.19 / = / 25 / 12.58 / =
Total Lead g/L / Total Zinc g/L
p-value / 0.454 / 0.608
n / median / Dunnet / n / median / Dunnet
No Control / 31 / 8.49 / 38 / 143.28
WP / 25 / 6.73 / = / 25 / 156.93 / =

Note. Dunnet test compared if sites with control produces larger concentrations “>”, smaller concentrations “<” or not statistically diference “=” than sites without control at a family error of 5%.

None indicates no samples were collected for this constituent in the group.

1

Table 5.2 shows that for all the constituents with observations in industrial land uses and rain zone 3 indicate that there is no significant difference between sites with or without wet ponds. For the sites located in rain zone 2, it was observed that in nitrite-nitrate, total phosphorus, total copper and total zinc sites with wet ponds before the outfall have significantly lower concentrations than sites without control. Wet ponds did not reduce the TKN concentrations in any of the four groups. It was observed a significant reduction in the TSS concentration for those sites with ponds in residential and commercial land uses, but not in industrial.

Dry ponds were evaluated only in residential land use and EPA rain zone 2. No difference was found in TSS, nitrite-nitrate in sites that used dry ponds. However, it was observed a reduction of BOD, TKN, total phosphorus, total copper, total lead and total zinc.

Some communities are installed in their parking lots underground storage tanks to reduce the runoff volume and increase the amount of water infiltrated after a storm. More than 400 underground pipes are located in Arlington, Virginia for example. A significant reduction in the TSS, BOD, COD, total lead and total zinc concentrations were observed in those sites with underground devices. In the other hand, these controls did not indicate a significant difference in the concentration of nutrients. No difference was observed in ammonia, nitrite-nitrate, TKN, dissolved phosphorus and total phosphorus. An unusual situation was observed in EPA rain zone 2 for total zinc; in residential land uses it seems to increase the concentration while in commercial areas the test indicate that the storage device reduces the concentration. There is only one site with storage devices in rain zone 2. The elevated zinc concentration is specific from this location. (VAARLCV2)

5.4 Sampling Analysis

The use of manual or automatic sampling is a factor that is sometimes mentioned as having an affect on the quality of collected samples. Manual sampling is usually preferred when the number of samples is small and when there are not available resources for the purchase, installation, operation, and maintenance of automatic samplers. Manual sampling may also be required when the constituents being sampled require specific handling (such as for bacteria, oil and grease, and volatile organic compounds) (ASCE/EPA 2002). Automatic samplers are recommended for larger sampling programs, when better representations of the flows are needed, and when site access is not difficult or unsafe. In most cases, where a substantial number of samples are to be collected and when composite sampling is desired, automatic sampling can be much less expensive. Automatic samples also improve repeatability by reducing additional variability induced by the personnel from sample to sample (Bailey, 1993). Most importantly, automatic samplers can be much more reliable compared to manual sampling, especially when the goal of a monitoring project is to obtain data for as many of the events that occur as possible, and sampling must start near the beginning of the rainfall.

Only six sites in residential land use, EPA rain zone 2 have enough samples collected by both automatic samplers and manual sampling to allow a simple statistical comparison of the collected data.

Table 5.3 Sites in Residential Land Use EPA rain zone 2

Constituent / AUTOMATIC SAMPLING / MANUAL SAMPLING / P-VALUE
Total Samples / Median concentration / Total Samples / Median concentration
TSS mg/L / 362 / 44 / 48 / 23.5 / 0.0005
Ammonia mg/L / 299 / 0.23 / 45 / 0.27 / 0.3335
TDS mg/L / 304 / 67 / 45 / 60 / 0.1427
Hardness mg/L / 14 / 38.8 / 29 / 21 / 0.0381
Oil and Grease mg/L / 82 / 2.5 / 34 / 1.5 / 0.0726
BOD mg/L / 343 / 9 / 48 / 11.2 / 0.1041
COD mg/L / 302 / 55 / 45 / 60 / 0.1406
NO2 + NO3 mg/L / 355 / 0.46 / 46 / 0.67 / 0.0018
TKN mg/L / 355 / 1.31 / 48 / 1.38 / 0.6838
Dissolved P mg/L / 291 / 0.12 / 43 / 0.22 / 0.0247
Total P mg/L / 362 / 0.29 / 45 / 0.38 / 0.3247
Total Cu g/L / 149 / 11 / 44 / 7 / 0.0018
Total Pb g/L / 144 / 6 / 44 / 3.8 / 0.5549
Total Zn g/L / 149 / 55 / 44 / 4.1 / 0.0167

Table 5.3 indicates that TSS, hardness, copper and zinc have higher concentrations when automatic samples were used instead of manual sampling. In the other hand nitrite-nitrate and dissolved phosphorus show significant higher concentrations when the sample was collected manually. Figure 5.2 shows the box plots comparing automatic versus manual sampling for four of the six constituents.

Figure 5.2. comparison between automatic and manual sampling in residential land use and EPA rain zone 2.

5.5 Sampling Frequency

Another potential factor that may affect stormwater quality is the sampling procedure. Automatic samplers are not capable of sampling bed load material, and are less effective in sampling larger particles (>500 µm) than typically suspended solids. Manual sampling, if able to collect a sample from a cascading flow, can collect from the complete particle size distribution. However, automatic samplers can initiate sampling very close to the beginning of flow, while manual sampling usually requires travel time and other delays before sampling can be started. It is also possible for automatic samplers to represent the complete storm, if of very long duration, as long as proper sampler setup programming is performed (Burton and Pitt 2001).

The NPDES stormwater sampling protocols only required collecting composite samples over the first three hours of the event instead of the whole event. Truncating the sampling may have effects on the measured stormwater quality.

Selecting a small subset of the annual events could also bias the results. In most stormwater research projects, the goal is to sample and analyze as many events as possible during the monitoring period. As a minimum, about 30 samples are usually desired in order to adequately determine the stormwater characteristics (Burton and Pitt 2001). With only three events per year required per land use, the accuracy of the EMC is questionable until many years have passed. Also, the three storms need to be randomly selected from the complete set of rains in order to be most statistically representative, even it requires about ten years to collect an adequate number of samples.

Flagstaff Street in Prince George MD had the most events collected for any site in the NSQD. They collected 28 events during the two years of 1998 and 1999. A statistical test was made choosing 6 events (three for each year) from this set, creating 5,600 different possibilities. Figure 5.3 shows the probability density function of this 5600 possibilities. The median TSS of the 28 events was 170 mg/L, with a 95% confidence interval between 119 and 232 mg/L. About 60% of the 5,600 possibilities were inside the confidence interval. Almost half of the possibilities for the observed EMC would therefore be outside the 95% confidence interval for the true median concentration if only three events were available for two years. As the number of samples increase, there will be a reduction in the bias of the EMC estimates. In Southern California, Leecaster (2002) determined that ten years of collecting three samples per year was required in order to reduce the error to 10% (Leecaster, 2002).

Figure 5.3. Probability density distribution of possible TSS concentrations in Flagstaff Street using 3 samples per year.

5.6 Sampling Analysis

Time or flow weighted composite samples were also evaluated in residential, commercial, and industrial land uses. Mann-Whitney test were used to evaluate if there is a significant difference between time composite or flow composite sampling. Table 5.4 show the results from the test.

Table 5.4 Mann Whitney results comparing flow composite versus time composite sampling

/

Residential 2

Median values / Industrial 2
Median values / Commercial 2
Median values / Industrial 3
Median values
Constituent / Flow / Time / Pvalue / Flow / Time / Pvalue / Flow / Time / Pvalue / Flow / Time / Pvalue
Ammonia mg/L / 0.23 / 0.86 / 0.08 / 0.28 / 1.4 / 0 / 0.4 / 1.7 / 0.02 / N/A / N/A / N/A
TDS mg/L / 65 / 84 / 0.23 / 56.5 / 166 / 0 / 61 / 117 / 0.04 / 92.8 / 40.7 / 0.02
O&G mg/L / 2.5 / 4.47 / 0.50 / 2.5 / 8.5 / 0.15 / 4 / 18 / 0.01 / N/A / N/A / N/A
BOD mg/L / 9.79 / 8 / 0.40 / 8 / 10.95 / 0.15 / 12 / 13 / 0.25 / 7.6 / 7.5 / 0.63
COD mg/L / 57 / 51.5 / 0.59 / 54 / 68 / 0.37 / 63 / 91 / 0.25 / 38 / 41 / 0.38
NO2 mg/L / 0.49 / 0.7 / 0.18 / 0.59 / 1.49 / 0 / 0.57 / 0.60 / 0.77 / 0.24 / 0.90 / 0.024
TKN mg/L / 1.32 / 1.1 / 0.24 / 0.96 / 1.25 / 0.39 / 1.41 / 1.70 / 0.33 / 0.88 / 1.75 / 0.01
DP mg/L / 0.13 / 0.12 / 0.89 / 0.08 / 0.09 / 0.97 / 0.08 / 0.26 / 0.05 / 0.03 / 0.10 / 0
TP mg/L / 0.29 / 0.34 / 0.58 / 0.22 / 0.34 / 0 / 0.22 / 0.18 / 0.25 / 0.18 / 0.16 / 0.825
Cu g/L / 9.25 / 12 / 0.32 / 15.49 / 25.98 / 0.2 / 14.5 / 33.3 / 0 / 14 / 18 / 0.37
Pb g/L / 4.9 / 4.9 / 0.95 / 7.5 / 19.5 / 0.7 / 10 / 46 / 0 / 19 / 5.7 / 0.21
Zn g/L / 47 / 63 / 0.48 / 171 / 288 / 0.02 / 150 / 346 / 0 / 120 / 155 / 0.23
TSS mg/L / 43 / 24 / 0 / 42 / 86 / 0 / 34 / 134 / 0 / 53 / 39 / 0.78

There is a significant difference in TSS concentration between flow composite and time composite samples for samples collected in single land uses and EPA rain zone 2. It was observed that for residential land uses the time composite sampling has lower concentrations than the flow composite. For industrial and commercial was the opposite.