Abstract

To aid urban entities desiring to reduce runoff from precipitation while increasing aquifer recharge, we present an approach for simultaneously quantifying runoff and infiltration. Developing the approach involved using: (1) the Windows version of the Source Loading and Management Model (WINSLAMM) to estimate runoff from precipitation in areas with green infrastructure (GI); and (2) the SCS runoff curve method to estimate infiltration. Computed infiltration and runoff values enable the estimation of the runoff reduction and infiltration increase due to alternative GI construction modes. We relate infiltration ratios to land use for a range of event rainfall depths in southwestern USA. These ratios can aid estimation of aquifer recharge while improving storm water management. We apply the approach to a Salt Lake City residential area for current land use and three assumed runoff control practices. Although currently applicable for a wide range of precipitation and urban land use situations in southwestern USA, the approach is extensible to guide urban development elsewhere.

INTRODUCTION

Urban sustainability decreases where flooding, waterlogging, and water pollution increase and groundwater availability for public use decreases. Although waterlogging is a problem in some urban areas (Liu et al. 2016; Zhang et al. 2016), in other areas aquifer groundwater level declines are problematic. Concepts and techniques to reduce such problems include the ‘sponge city’, green infrastructure (GI), low impact development (LID), sustainable drainage systems (SDS) and water sensitive urban design (WSUD) (van Roon 2007; Benzerra et al. 2012; Qin et al. 2013). By increasing precipitation infiltration and storage, and reducing runoff, these technologies are important tools for future urban construction (Graham et al. 2004). The technologies function by increasing surficial perviousness, and infiltrating, retaining, and reusing storm water at or near the site of storm water generation.

Many scholars have evaluated the impacts of LID implementation. Dong & Han (2011) planned an ecological sponge system and prepared an index system for reducing runoff volume and water pollution. Via a rainfall–flood simulation of Changde City, PR China, Liu et al. (2016) showed that LID measures would greatly reduce runoff. From an urban hydrology perspective, Zhang et al. (2016) analyzed how sponge city concepts can address water problems of urbanizing China. Xiong et al. (2016) used the Storm Water Management Model (SWMM) to determine that LID would reduce water runoff and pollution in a typical Shanghai drainage area. Dietz (2007) reported that a green roof can reduce storm water volume by 60–70% compared with a conventional roof. Lee et al. (2012) calculated the flood reduction resulting from LID for a district of Asan Tangjung New Town, Korea. Qin et al. (2013) investigated the impact of LID techniques (swales, permeable pavements and green roofs) on Shenzhen urban flood control. After reviewing LID storm water literature, Eckart et al. (2017) summarized LID performance for various conditions and LID research and professional practice needs. Other researchers have investigated LID pattern design and the reduction of runoff volume and peak flow rate. The literature reviewed did not quantify the proportions of the reduced runoff volume that can become aquifer recharge: this paper addresses that lack.

The primary objectives are to develop and demonstrate a method for estimating the relative proportions of rainfall in GI areas that become infiltration after computing runoff. To quantify runoff from GI areas we employ the Windows version of the Source Loading and Management Model (WINSLAMM) software (Pitt 2003). To estimate infiltration we couple WINSLAMM with the SCS runoff curve method (SCS). We present a significant figure relating infiltration ratios to land use in southwestern USA. We then demonstrate application of the approach to a residential area of Salt Lake City, USA. We first estimate runoff and infiltration for the existing level of development; then we assume GI practices, and predict runoff infiltration for those practices. The transferable methodology enhances consideration of increasing root zone moisture and aquifer recharge while planning storm water management.

MATERIAL AND METHODS

Study area

The study catchment is a 6.180 ha (15.271-acre) residential zone within Salt Lake City, USA (Figure 1). The study area has silty soil and features no runoff control practices. Table 1 shows current land use in the study area.

Table 1

Land use in study area

Land use Area (hectare; acre) 
Roofs (pitched, directly connected) 0.906; 2.238 
Streets (medium texture) 1.109; 2.741 
Sidewalks (directly connected) 0.980; 2.421 
Small landscape (lawns, silty soils) 1.592; 3.935 
Undeveloped area (silty soils) 1.593; 3.936 
Total area 6.180; 15.271 
Land use Area (hectare; acre) 
Roofs (pitched, directly connected) 0.906; 2.238 
Streets (medium texture) 1.109; 2.741 
Sidewalks (directly connected) 0.980; 2.421 
Small landscape (lawns, silty soils) 1.592; 3.935 
Undeveloped area (silty soils) 1.593; 3.936 
Total area 6.180; 15.271 
Figure 1

The study area (modified from Google Earth).

Figure 1

The study area (modified from Google Earth).

Employed rainfall-runoff estimation methods

Developed in the 1970s and frequently improved, the WINSLAMM hydrologic simulation model calculates runoff volumes and urban pollutant loadings from individual rain events (Pitt & Clark 2008). WINSLAMM requires input of land use areas and rain depths, and uses internal runoff coefficients. WINSLAMM employs the general infiltration rate model shown in Figure 2. Equation (1) defines the line (Pitt 2003).  
formula
(1)
where F is cumulative infiltration after initial losses (in), g is an exponential coefficient, and P is cumulated rainfall (in). If b=0, then a = total initial losses, and no steady state infiltration losses occur (this is equivalent to the SCS model). When the slope of this line, b, does not equal 0, F is the infiltration occurring for a specific rain depth after runoff begins. WINSLAMM calculates CN values that are like SCS Runoff Curve Numbers.
Figure 2

Illustration of infiltration rate model used within WINSLAMM (modified from Pitt 2003).

Figure 2

Illustration of infiltration rate model used within WINSLAMM (modified from Pitt 2003).

The SCS Runoff Curve Number method developed by the United States Department of Agriculture (USDA) Soil Conservation Service (SCS) is widely used for predicting direct runoff or infiltration from rainfall after initial losses (USDA 1986). The basis of the curve number method is the empirical relationship between the retention (rainfall not converted into runoff) and runoff properties of the watershed and the rainfall (Equation (2), (USDA 1986)), and Figure 3 illustrates the components of the SCS runoff equation.  
formula
(2)
where Q is actual runoff (in); P is rainfall (in); Ia is initial abstraction (in), assumed to be 20 percent of S; S is potential maximum retention after runoff begins (in). In Figure 3, F is cumulative infiltration after initial losses, which is less than or equal to S. Equation (3) allows the estimation of S (USDA 1986). The maximum total rain loss equals Ia plus S.  
formula
(3)
where CN is the runoff curve number. To obtain S in millimetres, replace 1,000 by 25,400 and replace 10 by 254.

Presented method for computing infiltration coefficients

In order to estimate infiltration volume on different land use conditions, we use WINSLAMM and the SCS runoff curve method (SCS). WINSLAMM can estimate total rain loss and CN. After CN is known, one can use the SCS method to calculate S and Ia. Because Ia is initial abstraction, subtracting Ia from the actual total rain loss equals the infiltration (F) at a particular time after runoff begins. We define an infiltration coefficient (F/P) as the ratio between infiltration loss and total rain depth. For rainfall on three consecutive (assumed independent) days, Table 2 illustrates the stepwise process of computing infiltration coefficients.

Table 2

Illustration of process of computing infiltration coefficients

Event date Total rainfall, P (in) Runoff (in) Total loss (in) CN S (in) Initial loss, Ia (in) Infiltration F (in) Infiltration coefficient 
(1) (2) (3) (4) (5) (6) (7) (8) (9) 
04/01/84 0.20 0.07 0.13 97.90 0.21 0.04 0.09 0.44 
04/02/84 0.14 0.05 0.09 98.50 0.15 0.03 0.06 0.43 
04/03/84 0.07 0.01 0.06 99.00 0.10 0.02 0.04 0.57 
Event date Total rainfall, P (in) Runoff (in) Total loss (in) CN S (in) Initial loss, Ia (in) Infiltration F (in) Infiltration coefficient 
(1) (2) (3) (4) (5) (6) (7) (8) (9) 
04/01/84 0.20 0.07 0.13 97.90 0.21 0.04 0.09 0.44 
04/02/84 0.14 0.05 0.09 98.50 0.15 0.03 0.06 0.43 
04/03/84 0.07 0.01 0.06 99.00 0.10 0.02 0.04 0.57 

Note: depths are average study area values; column (2) is WINSLAMM input; columns (3), (4) and (5) are WINSLAMM outputs; column (6) results from using Equation (3); column (7) = column (6) × 0.2; column (8) = column (4) – column (7); and column (9) = column (8)/column (2).

In essence, for southwestern US conditions, we use the runoff coefficients (Pitt 2003) in WINSLAMM and the SCS loss-estimation approach to compute southwestern US infiltration coefficients for different land uses and rainfall depths (Figure 4). The Figure 4 infiltration coefficients are ratios (like runoff coefficients), useful for estimating infiltration volumes for specified area, land use, and rainfall.

Figure 4

Infiltration coefficients versus rainfall depth for land uses in southwestern USA (multiply inches by 25.4 to obtain millimetres).

Figure 4

Infiltration coefficients versus rainfall depth for land uses in southwestern USA (multiply inches by 25.4 to obtain millimetres).

Background scenario and three GI design scenarios

For a location the average proportions of precipitation that become initial abstraction (losses), runoff and infiltration differ, depending upon the land use. For the current area land use, after initial abstraction: (i) rainfall on roofs, streets, and sidewalks contributes to runoff that leaves the study area; and (ii) rainfall on small landscaping and undeveloped areas infiltrates or runs off and leaves the study area.

Here, we design three different GI scenarios that change the total proportions of infiltration and runoff from the study area. Scenario 1 GI connects the roofs and sidewalks to pervious areas such as the small landscaping and undeveloped areas of the study area, where the water will infiltrate or run off. Scenario 2 GI connects the roofs, sidewalks and streets to grass swales where some water will infiltrate and some will eventually run off the study area; scenario 2 sub-scenarios use different swale densities. Scenario 3 GI replaces current sidewalks with permeable pavements, and all rainfall upon the permeable pavements becomes initial losses or infiltration.

Below we compare the simulated runoff and infiltration volumes resulting from precipitation for the existing land use versus the three GI scenarios. The comparison employs daily April, May, and June rainfall values from 1973 to 1983 (data available from https://www.ncdc.noaa.gov). Those are the three months with the greatest rainfall.

RESULTS AND DISCUSSION

Background scenario and scenario 1 – drainage to pervious areas

For the existing development (background land use scenario), simulating the specified rainfall events produces 20.72 in (526.3 mm) of runoff and 12.78 in (324.6 mm) of infiltration. When daily precipitation increases, runoff and infiltration increase for both the existing development and scenario 1. However, by causing runoff from roofs and sidewalks to drain to a pervious area, scenario 1 reduces runoff by 57% (Figure 5(a)) and increases infiltration into the soil by 17.5% (Figure 5(b)) compared with existing development.

Figure 5

Comparison of current infrastructure versus scenario 1 volumes of: (a) runoff and (b) infiltration (multiply ft3 by 0.0283 to obtain m3, multiply inches by 25.4 to obtain millimetres).

Figure 5

Comparison of current infrastructure versus scenario 1 volumes of: (a) runoff and (b) infiltration (multiply ft3 by 0.0283 to obtain m3, multiply inches by 25.4 to obtain millimetres).

Scenario 2 – drainage to grass swales

To estimate impacts of installing grass swales, we assume five different total grass swale densities, 120 ft/ac (90.4 m/ha), 150 ft/ac (113.0 m/ha), 170 ft/ac (128.0 m/ha), 180 ft/ac (135.6 m/ha) and 200 ft/ac (150.6 m/ha), and simulate rainfall events for April–June of 1973–1983. Table 3 shows that by routing runoff from roofs, sidewalks, and streets to grass swales, even the lowest swale density significantly reduces runoff and increased infiltration. Increasing swale density reduces runoff only a little more and does not affect infiltration much. A 170 ft/ac (128.0 m/ha) swale density achieves the same 57.7% runoff reduction as scenario 1 (which routes runoff to pervious areas).

Table 3

Total runoff and infiltration results for different swale densities

Construction mode Total rainfall (in) Infiltration (in) Runoff (in) 
Existing construction 53.3 12.78 20.72 
Swale density (ft/ac) 120 53.3 22.09 10.81 
150 53.3 22.08 9.51 
170 53.3 22.05 8.73 
180 53.3 22.02 8.39 
200 53.3 21.61 7.73 
Construction mode Total rainfall (in) Infiltration (in) Runoff (in) 
Existing construction 53.3 12.78 20.72 
Swale density (ft/ac) 120 53.3 22.09 10.81 
150 53.3 22.08 9.51 
170 53.3 22.05 8.73 
180 53.3 22.02 8.39 
200 53.3 21.61 7.73 

Note: multiply inches by 25.4 to obtain millimetres, multiply ft/ac by 0.7532 to obtain m/ha.

Scenario 3 – apply permeable pavement

This scenario assumes replacing existing sidewalks with permeable pavement. The pavement thickness is 4 inches and there are no under-drains. Scenario simulations estimate total runoff and infiltration depths to be 15.22 in (386.6 mm) and 18.4 in (467.4 mm), respectively. Scenario 3 runoff volume is 26.5% less and infiltration is 44% greater than background scenario values. The porous pavement reduces the previous runoff from sidewalks by 100%.

CONCLUSIONS

In a novel approach, this paper first couples WINSLAMM simulation and the SCS Curve Number method to estimate infiltration coefficients for a range of rainfall values and different land uses. These infiltration ratios can aid integration of storm water management with aquifer water table and recharge management. Infiltration ratios can aid estimation of the impact of different land use structures on soil moisture and groundwater recharge. Urban construction and land use decision-makers can use infiltration ratios to address multiple water-management goals. To illustrate practical application, this paper estimates runoff reduction and infiltration increase for an existing southwestern US residential area and three GI scenarios.

Simulations predict that all three GI designs would substantially reduce the runoff volume and increase the infiltration volume at the existing residential area. If runoff from roofs and sidewalks drains to a pervious area, the runoff volume would decrease 57.5% and infiltration would increase 17.5%. A swale density of 170 ft/ac (128.0 m/ha) would be required to achieve the same runoff decrease, but that would increase infiltration by 72.6%. Routing roofs, sidewalks and streets runoff to swales could potentially provide much more aquifer recharge than routing roofs and sidewalks runoff to a pervious area. Replacing existing sidewalks with permeable pavements would reduce total runoff volume by 26.5% (the permeable pavement would produce no runoff and would increase infiltration by 44%). A combination of GI techniques could more effectively reduce runoff and increase infiltration than using only one of the GI practices.

ACKNOWLEDGEMENTS

This research was supported by the China Scholarship Council, U.S. EPA-STAR Grant#83582401, and the Utah Agricultural Experiment Station, Utah State University, and approved as journal paper number 9114. We appreciate Dr Ryan Dupont's leadership of the EPA-STAR grant project.

REFERENCES

REFERENCES
Benzerra
A.
,
Cherrared
M.
,
Chocat
B.
,
Cherqui
F.
&
Zekiouk
T.
2012
Decision support for sustainable urban drainage system management: a case study of Jijel, Algeria
.
Journal of Environmental Management
101
,
46
53
.
Dong
S.
&
Han
Z.
2011
Study on planning an ‘Eco-Sponge City’ for rainwater utilization
.
Urban Studies
18
(
12
),
37
41
.
Eckart
K.
,
McPhee
Z.
&
Bolisetti
T.
2017
Performance and implementation of low impact development – a review
.
Science of the Total Environment
607–608
,
413
432
.
Graham
P.
,
Maclean
L.
,
Medina
D.
,
Patwardhan
A.
&
Vasarhelyi
G.
2004
The role of water balance modelling in the transition to low impact development
.
Water Quality Research Journal of Canada
39
,
331
342
.
Lee
J.-m.
,
Hyun
K.-h.
,
Choi
J.-s.
,
Yoon
Y.-j.
&
Geronimo
F. K. F.
2012
Flood reduction analysis on watershed of LID design demonstration district using SWMM5
.
Desalination and Water Treatment
38
,
255
261
.
Liu
C.-m.
,
Yong-yong
Z.
,
Zhong-gen
W.
,
Yue-ling
W.
&
Peng
B.
2016
A preliminary study of planning and techniques for sponge city
.
Journal of Natural Resources
31
(
5
),
719
731
.
Pitt
R.
2003
WINSLAMM Users Guide. Available online: http://rpitt.eng.ua.edu/SLAMMDETPOND/WinSlamm/MainWINSLAMM_book.html (accessed 29 August 2018).
Pitt
R.
&
Clark
S. E.
2008
Integrated storm-water management for watershed sustainability
.
Journal of Irrigation and Drainage Engineering
134
,
548
555
.
USDA
1986
Urban Hydrology for Small Watersheds, Technical Release 55 (TR-55) USDA, Washington, DC, USA. Available online: https://www.nrcs.usda.gov/Internet/FSE_DOCUMENTS/stelprdb1044171.pdf (accessed 29 August 2018).
Xiong
L.-j.
,
Huang
F.
,
Xu
Z.-x.
,
Li
H.-z.
,
Gong
L.-l.
&
Dong
M.-k.
2016
Characteristics of rainfall and runoff in urban drainage based on the SWMM model
.
Chinese Journal of Applied Ecology
27
,
3659
3666
.
Zhang
J.
,
Wang
Y.
,
Hu
Q.
&
He
R.
2016
Discussion and views on some issues of the sponge city construction in China
.
Advances in Water Science
27
(
6
),
793
799
.