SDP120116f

Z. Rakib, et al., Int. J. Sus. Dev. Plann. Vol. 12, No. 1 (2017) 155–164
© 2017 WIT Press, www.witpress.com
ISSN: 1743-7601 (paper format), ISSN: 1743-761X (online), http://www.witpress.com/journals
DOI: 10.2495/SDP-V12-N1-155-164
This paper is part of the Proceedings of the 3rd International Conference on Design, Construction,
Maintenance, Monitoring and Control of Urban Water Systems (Urban Water 2016)
www.witconferences.com
CLIMATE CHANGE IMPACTS ON URBAN STORMWATER
BEST MANAGEMENT PRACTICES
ZUBAYED RAKIB1, MICHAEL BARBER1 & ROBERT MAHLER2
1Civil and Environmental Engineering, University of Utah, Salt Lake City, UT, USA.
2Soil Science Division, University of Idaho, Moscow, ID.
ABSTRACT
Total maximum daily load (TMDL) studies determine the amount of contaminant(s) that can be dis-
charged daily from point (waste load allocation – WLA) and nonpoint (load allocation – LA) sources
including a margin of safety (MOS) and then layout the path for achieving these levels by reductions
in loadings. This has caused environmental agencies to require best management practices (BMPs) for
control of urban stormwater contributions. Design storms for volume-based and peak discharge BMPs
are typically determined from historic precipitation and runoff records that do not adequately address
the impacts of climate change. We examine a 10-year period of predicted flows in the Spokane River
watershed under 2050 climate predictions to determine the amount of additional LA removal required
to meet water quality goals. While the current TMDL proposes a 50% reduction of nonpoint loading,
our results indicate this will not be adequate. The implication is that urban BMPs are currently inad-
equately designed to handle nonpoint pollution in areas projected to experience increased precipitation
events. The problem is particularly acute for rain on snow events where BMP performance is already
impaired.
Keywords: algal blooms, hydrodynamic simulation, nonpoint source pollution, nutrients, total maxi-
mum daily loads, waste load allocation, water quality modeling.
1 INTRODUCTION
In the United States, under section 303(d) of the Clean Water Act, states are required to cre-
ate pollution budgets and reduction plans for waters not meeting beneficial use standards [1].
These total maximum daily load (TMDL) studies determine the amount of contaminant(s)
that can be discharged daily from point (waste load allocation – WLA) and nonpoint (load
allocation – LA) sources including a margin of safety (MOS) and then layout the path for
achieving these levels by reductions in loadings. Stormwater runoff from impervious urban
and suburban development can contribute to significant degradation of streams and rivers
[2]. Since many TMDL studies have been performed using steady state models such as
QUAL2Kw [3–6], owning to the fact that point sources were historically identified as both
the main source and the main opportunity, properly accounting for stormwater pollution has
not been adequately addressed. Previous studies have shown that low flow analysis do not
necessarily capture the largest inputs of nutrients [7]. The likely result is that stormwater best
management practices (BMPs) may not be properly accounted for in the pollution budget.
This is especially true when considering the impacts of climate change on future precipita-
tion events.
Design storms for volume-based and peak discharge BMPs are typically determined from
historic precipitation and runoff records that do not adequately address the impacts of cli-
mate change. The stakes are immensely important as overdesigning BMPs wastes money but

156 Z. Rakib, et al., Int. J. Sus. Dev. Plann. Vol. 12, No. 1 (2017)
undersizing BMPs is not protective of the environment and potentially leads to more expen-
sive solutions in the future.
This study used the Spokane River Basin in northeastern Washington, USA as the basis
for evaluating the hypothesis that climate change should be factored into BMP designs when
planning LAs. We examine a 10-year period (1999–2009) of predicted flows in the Spokane
River watershed under 2050 climate predictions to determine if additional LA removal will
be required to meet water quality goals. This study modified an existing U.S. Army Corps of
Engineers CE-QUAL-W2 model [8] to simulate hydrology and water quality with particular
attention to flow, phosphorus, nitrogen, dissolved oxygen, and river temperature. The original
model was used to develop the Spokane River dissolved oxygen TMDL based on a 2001 low-
flow analysis [9]. By calibrating and applying the model for an extended period, we were able
to better predict the nutrient dynamics under varying a variety of flow conditions and facili-
tate investigation and decisions regarding permissible nutrient levels that include a variety of
hydrologic conditions. Implications to BMP design were discussed.
2 BACKGROUND
The Spokane River watershed is located in northeastern corner of the State of Washington,
USA (Fig. 1). As a result of low summer flow conditions, sections of the river have been
listed as impaired including Long Lake, which has dissolved oxygen (DO) concentrations
below beneficial use standards. In order to attain the DO requirement, significant reductions
in nutrient loadings much be achieved. In the U.S., this is done through WLA. WLA is the
pollutant load allocated to current and future point sources (including Municipal Separate
Storm Sewer System (MS4) and other permit-recognized nonpoint sources). It is part of the
TMDL calculation according to:
TMDL = WLA + LA + RC + MOS, (1)
where LA are additional nonpoint sources, RC is reserve capacity for future growth, and
MOS is margin of safety. Since TMDLs are typically triggered when water quality impair-
ment has been identified, WLAs often require some amount of reduction in current loading,
Figure 1: Spokane River Bain in Northeastern Washington State, USA [7].

Z. Rakib, et al., Int. J. Sus. Dev. Plann. Vol. 12, No. 1 (2017) 157
which may involve installation of costly new or improved treatment systems or nonpoint
source BMPs.
LA reduction is done primarily through implementation of stormwater BMPs such as
detention ponds, constructed wetlands, grassy swales, and other control devices [10]. BMPs
typically fall into one of two categories depending on whether sizing is controlled by flow
rate or flow volume. The Washington State Department of Ecology and the Washington
Department of Transportation recommend using a 3-hour short-duration storm distribution
for flow-based BMPs and a 30-hour long-duration storm hyetograph for volume-based BMPs
in the Spokane, Washington area [11]. These storms durations are the result of a large collab-
orative effort between various stormwater entities in eastern Washington. The return interval
(e.g. 0.5, 2, 5, 10, 25, 50, 100, PMF) varies depending on the nature of the structure and the
goal (quality vs quantity) of the BMP. The concerning feature of this is that Washington’s
isopluvial maps are based on NOAA Atlas 2, Volume 9 published in 1973 [12]. Thus, BMPs
are being designed ignoring more than 40 years of climate induced precipitation changes.
Accurately downscaling precipitation forecasts for future climate scenarios is an extremely
challenging pursuit given the high temporal and spatial variability of rainfall [13, 14]. As a
result, we did not attempt to redefine the isopluvial maps and conduct a microscale analysis
of individual BMPs for this project. Instead, we examined large-scale nonpoint source load-
ing under present and future scenarios and inferred this to BMP design requirements.
3 METHODOLOGY
This study modified the CE-QUAL-W2 model [12] to simulate hydrology and water quality
over a 10-year historic period (1999–2009), with particular attention to flow, phosphorus,
nitrogen, dissolved oxygen, and river temperature. The model consisted of 188 longitudi-
nal segments and an average of nearly 20 layers in vertical direction. The flow portion of
the model was calibrated/validated using existing USGS gaging stations and known reser-
voir levels. In addition to the main flow entering from Post Falls Dam, tributary inflows
from Hangman Creek, Little Spokane River, Coulee Creek, and wastewater discharges from
the area were added. As the majority of dams on the river represented so-called ‘run of
the river facilities’, few adjustments had to be made for changing reservoir storage levels.
Groundwater interactions, although significant sources/sinks of water, were assumed to be
constant through each simulation year.
Once the model successfully reproduced the flows of the 10-year period, water quality por-
tion of the model was calibrated and validated using available data. Concentration data from
State monitoring activities and reported Waste Water Treatment Plant discharges were used
to examine the predictive capabilities of the model.
The impacts of climate and population change were then added to the model. The 2050
model runoff scenarios were developed using flow projections developed by the Climate
Impacts Group at the University of Washington. Efforts were made to use the latest data
available for this study. So, while meteorological data were available from CMIP5 models for
RCP4.5 and RCP8.5 scenarios, flow projections were not yet completed. Therefore, CMIP3
flow data for B1 scenario were used in combination with CMIP5 meteorological data for the
RCP4.5 scenario, while CMIP3 flow data for A1B scenario were used with CMIP5 meteoro-
logical data for the RCP8.5 scenario. Such a combination provided a better job of bracketing
the range of plausible future greenhouse gas forcing in the Pacific Northwest.
Stream temperatures at the boundary conditions (WA-ID Stateline, Hangman Creek, Little
Spokane River, and Coulee Creek) for the projected time scale (2041–2050) were estimated

from available air temperature projections. Previous studies have demonstrated this rela-
tionship is well described by a continuous S-shaped function [15]. Spokane River and its
tributaries (Hangman Creek and Little Spokane River) generally do not experience freezing
temperatures. To take this into account, a parameter µ was added to represent the estimated
minimum stream temperature to produce:
Ts = µ + µ
α µ
γ β
+
−
+ −( )
1 e Ta
,
where Ts
is the estimated stream temperature, Ta
is the measured air temperature, α is the
estimated maximum stream temperature, γ is a function of the slope at the point of inflection,
and β represents the air temperature at the inflection point.
Changes were also made to account for population growth. According to the State Office
of Financial Management forecasts, the population in Spokane County will increase from
441,600 persons in 2005 to more than 593,000 by 2040. Discharge data at Spokane WWTP
was collected from the City of Spokane to understand the change in flow rate with chang-
ing population and operation practice. When compared, it was seen that there was a clear
increase in average flows from Spokane WWTP (1.4 m3/sec in 1980s, 1.7 m3/sec in 1990s,
2.0 m3/sec in 2000s; an increase of approximately 20% per decade). It was important to
look at the projected population that will be served by the WWTP, and not just the overall
population increase in the area. Subsequently, the population trend and discharge trends were
used to come up with the estimates for WWTP discharges for 2041–2050 – which was 6.8–
7.0 m3/sec. The discharge estimated for 2025 using the same trends was 2.6 m3/sec, which
was close to the estimate of 2.8 m3/sec given by the Spokane County. This provided a check
for the 2041–2050 estimates used in this study.
4 RESULTS AND DISCUSSION
Although most climate models predict a modest increase in precipitation over the entire
Pacific Northwest, model projections for the Spokane watershed show fairly large percent-
age increase particularly during the winter months. In examining these results we concluded
that rain (due to temperature increases) in winter months rather than snow disproportion-
ately skewed percentage increases during the December to March time frame (see Table 1).
Historically these were relatively low flow months with snow generating large May and June
runoff events. Over a 10-year simulation period, Fig. 2 illustrates the serious magnitude of
the flow increases.
Water temperature at surface in the Spokane River and Long Lake, as seen from climate
change scenario simulations, is expected to increase during 2040–2050 due to the projected
increase in air temperature. Figure 3 illustrates the projected stream temperatures at the Spo-
kane River at Spokane gage. Compared to the baseline scenario, water temperature at river
locations is expected to increase by about 0.6°C–0.8°C for the high and low emission sce-
narios during 2040–2050; while at Long Lake, the increase is about 1.8°C–2.2°C.
Figure 4 compares the projected average monthly water temperatures at the Spokane River
at Spokane gaging location. As illustrated, results indicate temperatures generally increased
in all months. These increases led to additional violations of the stream temperature stand-
ards. Violation of the temperature criteria under climate change impacts was also apparent
from vertical plots at a number of other reaches along the study corridor. As expected, the
vertical profiles indicated that surface temperatures were warmer than average temperature
exacerbating the problem.

Table 1: Projected % increases in streamflow by month.
Month
Spokane River @ Spokane Long Lake Reservoir
High emission
scenario
Low emission
scenario
High emission
scenario
Low emission
scenario
January 132.4 88.2 127.8 90.4
February 235.3 147.1 214.9 138.7
March 123.1 88.9 103.7 77.5
April 69.7 57.9 16.2 25.4
May 13.8 27.8 8.1 14.2
June 4.5 9.8 18.3 14.7
July 15.3 11.8 9.7 6.3
August 24.0 28.0 14.2 5.7
September 17.6 5.9 29.7 19.8
October 25.0 16.7 88.7 61.3
November 35.3 17.6 129.6 83.4
December 127.3 86.4 196.5 133.8
Figure 2: Future flow projections.
Similarly, Fig. 5 illustrates the changes in average monthly dissolved oxygen concentra-
tions at the same location as Fig. 4. While winter months indicate slightly improved DO
concentrations as a result of future climate change, the critical summer period concentrations
are nearly the same (in July) or slightly worse (in August). The daily fluctuations (shown
in Fig. 6) more clearly demonstrate the variability and potential problems associated with

Figure 3: Climate change model simulation results for stream temperature.
Figure 4: Climate scenario temperatures at monthly time scale.
low dissolved oxygen concentrations as several additional low concentration periods are
predicted. For the winter period, snowmelt impacts overcome the slight increase in air tem-
perature and DO is improved. However, during the summer, the increases in stream flow are
offset by the impacts of warmer air temperatures.
In fact, average concentrations during late summer are very close to the DO standard.
Statistical analysis showed the dissolved oxygen concentration differences between the
baseline and climate scenarios at different river locations were statistically significant
(0.05 significance level, p 0.005) but were not for concentrations in Long Lake (p 0.1).
This implies the issue with low DO at Long Lake during the summer will persist in the
future. Oxygen at depth seemed to suffer as well (see Fig. 7) presumably due to increased
temperature reducing oxygen flux from the atmosphere and higher algal activity shown
in Fig. 8.

Figure 5: Climate scenario dissolved oxygen concentrations at monthly time scale.
Figure 6: Daily dissolved oxygen concentrations at Long Lake.
The TMDL developed for the Spokane River calls for a 50% reduction in nutrient load-
ing from nonpoint sources. Simulation results from this investigation indicate that this will
not be sufficient to protect aquatic species dependent on dissolved oxygen for survival. The
number and size of stormwater BMPs in the watershed need to be increased to account for the
additional flow. While not all regions of the Pacific Northwest are expected to see increases
in climate-induced precipitation, it seems important to plan and design future BMPs with
additional capacity to meet water quality objectives.
5 CONCLUSIONS AND RECOMMENDATION
Design of stormwater BMPs taking climate change into consideration remains a challeng-
ing task. While our approach was different than other studies [14, 16], the wide variability
was similar. A 2011 study by Forsee and Ahmad found that 6-hour, 100-year precipita-
tion depth varied from a reduction of 30% to an increase of 300% depending on the model

Figure 7: Vertical profile of dissolved oxygen in Long Lake.
Figure 8: Climate impacts on total algae concentrations.
used. In our study, runoff amounts increased from 4.5% to 235%. While in some cases this
resulted in lower concentrations, total loads (concentration × flow) were still high and the
associated increase in algae production detrimentally impacted dissolved oxygen concen-
trations during die-off. However, despite the variability, it is readily apparent that BMPs

must be sized to treat additional volumes and increased peak discharges of flow. The precise
magnitudes of these increases still needs further study as the consequences are significant.
At the high end, BMPs would need to be more than double their size to treat 235% more
runoff.
One additional challenge that needs to be addressed is the wintertime shift in large rain-
on-snow runoff events. Many existing BMPs do not function well (or at all) during dormant
vegetation periods. Infiltration, stormwater inlets, and wetland vegetation all have significant
challenges during winter months. The cycling of nutrients in lakes, however, means that
stormwater controls must be effective for all stormwater design events.
More research is still needed to reduce the uncertainty of downscaled precipitation fore-
casts and frequency analysis. However, this study indicates that additional stormwater control
is not adequately accounted for in historic analysis of flow conditions.
ACKNOWLEDGEMENTS
This project was funded by the United State Department of Agriculture National Institute of
Food and Agriculture through their AFRI Climate Change program.
REFERENCES
[1] U.S. Environmental Protection Agency, Handbook for developing watershed TMDLs.
Office of Wetlands, Oceans, and Watersheds, Washington, DC, 2008.
[2] U.S. Environmental Protection Agency. National management measures to control
nonpoint source pollution from urban areas. EPA-8410B-05-004, Office of Water,
Washington, DC, 2005.
[3] Pelletier, G.J., Chapra, S.C. Hua, T., QUAL2Kw e A framework for modeling water
quality in streams and rivers using a genetic algorithm for calibration. Environmental
Modelling and Software, 21, pp. 419–425, 2006.
http://dx.doi.org/10.1016/j.envsoft.2005.07.002
[4] Turner, D., Pelletier, G. Kasper, B., Dissolved oxygen and pH modeling of a periphy-
ton dominated, nutrient enriched river. ASCE Journal of Environmental Engineering,
135(8), pp. 645–652, 2009.
http://dx.doi.org/10.1061/(ASCE)0733-9372(2009)135:8(645)
[5] Neilson, B.T., Hobson, A.J., Vonstackelberg, N., Shupryt, M. Ostenmiller, J., Using
Qual2K modeling to support nutrient criteria development and wasteload analyses in
Utah. Final Project Report. Utah Department of Environmental Quality. Division of
Water Quality. Salt Lake City, UT, 2012.
[6] von Stackelberg, N. Neilson, B., Collaborative approaches to calibration of a riverine
water quality model. Journal of Water Resources Planning and Management, 140(3),
pp. 393–405, 2014.
http://dx.doi.org/10.1061/(ASCE)WR.1943-5452.0000332
[7] Rakib, Z., Barber, M. Mahler, R., Modeling flow, nutrient and dissolved oxygen
concentrations in the Spokane River under multiple year conditions. 8th International
conference on Sustainable Water Resources Management, A Coruna, Spain, June 2015.
[8] Cole, T.M. Wells, S.A., CE-QUAL-W2: a two-dimensional, laterally averaged, hydro-
dynamic and water quality model, version 3.6. Department of Civil and Environmental
Engineering, Portland State University, Portland, OR, available at: http://www.cee.pdx.
edu/w2/

[9] Clark, D.L., Kasch, M. Brattebo, B., Spokane River watershed strategies for point
and nonpoint source management to meet the most restrictive TMDL in the nation.
Proceeding of the Water Environment Federation, pp. 632–640, 2011.
http://dx.doi.org/10.2175/193864711802864930
[10] International Stormwater BMP Database, available at: http://www.bmpdatabase.org/
[11] WSDOT Highway Runoff Manual M31-16.04, Washington State Department of
Transportation, Olympia, Washington, April 2014.
[12] NOAA National Weather Service, Hydrometeorological Design Studies Center, avail-
able at: http://www.nws.noaa.gov/oh/hdsc/currentpf.htm.
[13] Wong, G., Maraun, D., Vrac, M., Widmann, M., Eden, J. Kent, T., Stochastic model
output statistics for bias correcting and downscaling precipitation including extremes.
American Meteorological Society, 27, pp. 6940–6959, 2014.
http://dx.doi.org/10.1175/jcli-d-13-00604.1
[14] Forsee, W.J. Ahmad, S., Evaluating urban stormwater infrastructure design in
response to projected climate change. Journal of Hydrologic Engineering, 16(11),
pp. 865–873, 2011.
http://dx.doi.org/10.1061/(ASCE)HE.1943-5584.0000383
[15] Mohseni, O., Stefan, H.G. Erickson, T.R., A nonlinear regression model for weekly
stream temperatures. Water Resources Research, 34(10), pp. 2685–2692, 1998.
http://dx.doi.org/10.1029/98WR01877
[16] Mailhot, A., Duchesne, S., Caya, D. Talbot, G., Assessment of future change in inten-
sity-duration-frequency (IDF) curves for Southern Quebec using the Canadian Regional
Climate Model (CRCM). Journal of Hydrology, 347(1–2), pp. 197–210, 2007.
http://dx.doi.org/10.1016/j.jhydrol.2007.09.019

SDP120116f

More Related Content

What's hot (20)

Similar to SDP120116f (20)

SDP120116f