Urbanization has strongly changed the condition of the land surface and therefore rainfall runoff varies greatly. Peak flood flow and flood volumes increase with runoff volume. Low Impact Development (LID) is a sustainable concept that minimizes the effects of urbanization to maintain natural hydrological function in urban cities and has therefore gained increasing attention. This paper studies the effects of low impact development measures on the reduction of runoff generation and peak runoff at different locations in Longyan, China. The study was conducted using the SWMM model (5.1.006) with a newly developed LID module. In this study, the LID module, which includes rain gardens, green roofs, permeable pavements, and rain barrels, was used to simulate different layout scenarios and different rainfall patterns. The results show that the performance of a certain LID is similar at different locations but the reduction effect on runoff and peak flow varies. Rain gardens and permeable pavements perform a similar degree of reduction under different durations, but the peak flow reduction by rain barrels and green roofs varies greatly. Further research should focus on composite LID applications in other locations, combination with the local pipe network layout, which will ensure that the implemented system will be aesthetically pleasing, economically viable, and effective for reducing runoff and peak flow.

INTRODUCTION

Urbanization in many areas of the world has resulted in considerable changes in the natural geomorphology. The impervious area ratio has increased greatly and significantly influences rainfall runoff. In addition, changes to the water cycle may lead to an increase in the frequency and intensity of tropical storms, floods, and droughts (Huntington 2006), resulting in higher flood risk. Many cities in China are facing problems of flood control and drainage in areas with rapid urbanization. Studies have shown that increases in the impervious area result in increases in the mean, frequency, and duration of high flows (Fletcher et al. 2007; Ashbolt et al. 2012) by reducing surface infiltration and water storage capacity. People around the world continuously seek for solutions to reduce the negative impact of urbanization. As an innovative solution for storm water management, low impact development (LID) is designed in a sustainable way and treated as a no-regret strategy (Butler & Parkinson 1997; Pyke et al. 2011). LID practices first appeared in the USA in the 1990s, which resulted in significant effects on storm water management in urban areas. These practices have gained increasing attention in recent years as a new concept in urban development. The LID module controls surface runoff at its source, lessens urban runoff, reduces urban drainage pressure and pollution, and improves the natural landscape of urban areas. To better promote the LID practices, the reduction in flood runoff resulting from LID needs to be evaluated (Sin et al. 2014). LID components such as green roofs can delay, prolong, and reduce the peak rates of green roof discharge by 22–70% (Alfredo et al. 2009) compared with a standard roof surface. Data also show that various applications of barrels and porous pavements resulted in a 2–12% (Ahiablame et al. 2013) reduction in runoff and pollutants. Many factors influence the effect of LID and related studies are ongoing. Factors that affect runoff dynamics from LID modules, such as the depth of the substrate layer of green roofs, have been discussed in the literature (Mentens et al. 2006; Berndtsson 2010). There are few studies on LID module performance at different locations (Gilroy & McCuen 2009), and even fewer in an urban drainage system.

The aim of this work is to simulate the performance of low impact development in different locations with an urban drainage system. In previous SWMM versions, the commonly used rain garden and green roof measures for retention and infiltration of storm water were not included in the LID module. Here, a hydrological model was established using the new version of the SWMM model (5.1.006) with more LID types that allow engineers to accurately represent any combination of LID controls. Measured and design rainfall data were used for model calibration and simulation and the model was first applied to Longyan.

METHODS

Study area

Longyan city is located in China's southeastern province of Fujian. The city has a subtropical marine monsoon climate with an average annual rainfall of 1,450–2,200 mm, which is relatively high in China. The average rainfall from April to September accounts for 75% of the total annual rainfall. The study area is the main urban area of Longyan, which covers a total area of 134.56 km2. The main soil type is laterite and yellow loam, and the urban land use type consists mainly of residential (33.94%) and industrial land (21.74%), roads and squares (16.09%), green space (9.65%), public facilities (9.43%), and commercial and service facilities (4.74%). Based on the Longyan Urban Master Plan (2011) of the pipe network layout and considering the terrain factor, the study area was divided into 553 subareas, which contain 711 nodes, 698 pipes, and 11 weirs. The average slope of the study area is 2.9%, with an average impervious area of 39.3%. The Longyan location and pipe network layout are shown in Figure 1.
Figure 1

Longyan location and pipe network layout.

Figure 1

Longyan location and pipe network layout.

Hydraulic model

The Storm Water Management Model (SWMM) of the US Environmental Protection Agency (US EPA) has been widely used throughout the world since 1971 for planning, analysis and design related to stormwater runoff, combined and sanitary sewers, and other drainage systems in urban areas (Huber 2003; Rossman 2010).

A hydrological model of the Longyan drainage system was developed using the SWMM. LID practices have been developed since version 5.0.19 of the model, which reduce surface runoff and reduce flood risk. The new version of the SWMM (5.1.006) (http://www2.epa.gov/water-research/storm-water-management-model-swmm) was recently extended with the new LID modules of rain gardens and green roofs. This version allows planners to evaluate any combination of LID controls within a study area. Studies of LID analysis for BMP design aimed to achieve either maximum runoff control or total minimum system cost (Jia et al. 2012).

The parameters of the LID modules are divided into the following levels: surface, soil, drainage mat, pavement, storage and underdrain. The above parameters are set mainly reference to the model manual (Rossman 2010), and some selection considering the local soil characteristics. The parameters of the LID measures used in this study are listed in Table 1.

Table 1

Parameters of the LID measures used in this study

  Parameters Unit Rain Garden Green Roof Permeable Pavement Rain Barrel 
Surface Berm Height mm 3.8 2.5 – 
Vegetation Volume – 0.6 0.8 0.3 – 
Surface Roughness – 0.35 0.16 0.15 – 
Surface Slope 1.0 1.0 1.0 – 
Soil Thickness mm 500 100 – – 
Porosity – 0.5 0.5 – – 
Field Capacity – 0.33 0.33 – – 
Wilting Point – 0.1 0.1 – – 
Conductivity mm/h 60 60 – – 
Conductivity Slope – – – 
Suction Head mm 88.9 88.9 – – 
Drainage Mat Thickness mm – 70 – – 
Void Fraction – – 0.5 – – 
Roughness – – 0.03 – – 
Pavement Thickness mm – – 120 – 
Void Ratio – – – 0.15 – 
Impervious Surface Fraction – – – 0.2 – 
Permeability mm/h – – 2,500 – 
Clogging Factor – – – – 
Storage Thickness mm – – 300 – 
Void Ratio – – – 0.7 – 
Seepage Rate mm/h – – 500 – 
Clogging Factor – – – – 
Barrel Height mm – – – 750 
Underdrain Flow Coefficient – – – 
Flow Exponent – – – 0.5 0.5 
Offset Height mm – – 200 150 
Drain Delay hours – – – 
  Parameters Unit Rain Garden Green Roof Permeable Pavement Rain Barrel 
Surface Berm Height mm 3.8 2.5 – 
Vegetation Volume – 0.6 0.8 0.3 – 
Surface Roughness – 0.35 0.16 0.15 – 
Surface Slope 1.0 1.0 1.0 – 
Soil Thickness mm 500 100 – – 
Porosity – 0.5 0.5 – – 
Field Capacity – 0.33 0.33 – – 
Wilting Point – 0.1 0.1 – – 
Conductivity mm/h 60 60 – – 
Conductivity Slope – – – 
Suction Head mm 88.9 88.9 – – 
Drainage Mat Thickness mm – 70 – – 
Void Fraction – – 0.5 – – 
Roughness – – 0.03 – – 
Pavement Thickness mm – – 120 – 
Void Ratio – – – 0.15 – 
Impervious Surface Fraction – – – 0.2 – 
Permeability mm/h – – 2,500 – 
Clogging Factor – – – – 
Storage Thickness mm – – 300 – 
Void Ratio – – – 0.7 – 
Seepage Rate mm/h – – 500 – 
Clogging Factor – – – – 
Barrel Height mm – – – 750 
Underdrain Flow Coefficient – – – 
Flow Exponent – – – 0.5 0.5 
Offset Height mm – – 200 150 
Drain Delay hours – – – 

Detailed rainfall and water level data were obtained from the downstream Dongxing station and the upstream Longmen station established by the government. The study area boundary conditions were obtained from four upstream reservoirs of the Longmen River, Hongfang River, Dongxiao River, and Xiaoxi River. The model simulates the urban area upstream of the Dongxing station.

The model calculations are in metric units, and the Horton empirical equation was used for calculating the infiltration. The pipe network flow was calculated using the dynamic wave method to reduce error. Design rainfall of short duration lasted for 2 h; the simulation period was extended to 4 h, and the design rainfall of long duration lasted for 24 h; the simulation period was extended to 48 h. Evaporation during the rainfall period was not considered. Modeling parameters of subareas, nodes, conduits and weirs were set according to the actual situation, whereas other parameters such as the surface roughness coefficient and depression storage capacity were set based on the model manual (Rossman 2010).

Design scenarios and design storms

Three different districts were selected and divided into groups A, B, and C to test the performance of the LID measures. The areas of the three districts are 1.79 km2, 1.59 km2, and 1.85 km2, and the average impervious areas are 49.47%, 46.31%, and 44.45% for groups A, B, and C, respectively. In addition to the individual LID arrangement, a combination of a set of LIDs was included. The three regions have a high impervious area due to a large proportion of residential and commercial land use.

Four LID modules (i.e., rain garden, green roof, permeable pavement, and rain barrel) were selected for this study. According to the local urban planning and application manual, six types of schemes were set for each district independently: without LID measures; 10% green roofs; 20% rain gardens; 15% permeable pavement; 1% rain barrels; and a combination of 2.5% green roofs, 5% rain gardens, 4% permeable pavement, and 0.25% rain barrels.

Rainfall data spanning 41 yr (from 1971 to 1981 and 1983 to 2012) were used to calculate long duration frequency analysis. The annual maximum method was used to calculate the frequency of 24-h rainstorms. The short duration storm data of Longyan spans 25 yr (1966–1990), and the storm formula method uses multiple samples per year. The rainstorm intensity formula is as follows: 
formula
or: 
formula
where: I-calculated storm intensity value (mm/min); q-calculated storm intensity value (L/hm2•s); t-rainfall duration (min); and T-recurrence period (a).

The rainstorm intensity formula was combined with the Chicago rain type to calculate the short duration design storm. To simulate the effect of the selected LID module on the reduction of flood volume and flood peak (Tillinghast et al. 2012) at different locations, a total of six design storm events were used to simulate different frequencies. The 2-h short rainfall duration consisted of 1-, 2- and 5-year return periods ranging from 44.1 to 66.7 mm, and the 24-h long rainfall duration consisted of 10-, 20-, and 50-yr return periods ranging from 128.6 to 153.9 mm. Rainfall events were evaluated as the volume of runoff water relative to the total rainfall. Three districts of the flood abatement proportion were compared. The runoff of each district at different locations, and the peak value were also taken into account.

RESULTS AND DISCUSSION

Prior to conducting the simulation, a 24-h storm from June 2011 was used to verify the accuracy of the model (Figure 2). So that the model parameters more suitable for the study area. The measured rainfall and external boundaries were input the model to obtain the simulation water level process of a certain site. The result shows that the simulated water depth of Dongxing station agrees well with the measured value, and the water level trend is reasonable. The peak water level and time fit well, indicating a good model fit.
Figure 2

Comparison between simulated and measured water depth of Dongxing station.

Figure 2

Comparison between simulated and measured water depth of Dongxing station.

The runoff coefficient is calculated as the ratio between total runoff (mm) and precipitation (mm) within a certain catchment, and reflects the runoff generation capacity (Walsh et al. 2014). The ability of an LID to control runoff generation can be tested in various districts with different LID measures using the runoff coefficient of various rainfall scenarios.

Figures 35 Changes in the runoff coefficients under various rainfall scenarios are presented for districts A, B, and C. The left panels of the graphs represent design rainfall scenarios that last for 2 h with return periods of 1, 2, and 5 yr. The right panels of the graphs represent design rainfall scenarios that last for 24 h with return periods of 10, 20, and 50 yr.
Figure 3

Runoff coefficient of district A under various scenarios.

Figure 3

Runoff coefficient of district A under various scenarios.

Figure 4

Runoff coefficient of district B under various scenarios.

Figure 4

Runoff coefficient of district B under various scenarios.

Figure 5

Runoff coefficient of district C under various scenarios.

Figure 5

Runoff coefficient of district C under various scenarios.

The figures show that the runoff coefficient of a district increases with an increase in the length of the return period. It is clear that each LID measure reduces the runoff coefficient relative to where LID measures were not implemented. In general, the runoff coefficients of 2-h short duration rainfall with return periods of 1, 2, and 5 yr were lower than the 24-h long duration rainfall with return periods of 10, 20, and 50 yr. The same LID measure performed similarly in different districts in terms of the runoff coefficients under the 2-h and 24-h rainfall scenarios. The best LID performance was observed with the 20% rain gardens followed by 15% permeable pavement, combined measures, 1% rain barrels and 10% green roofs. The complexity of the LID module does not increase the model accuracy (Burszta-Adamiak & Mrowiec 2013), and the combined measures reduce the runoff coefficient. Compared with the condition where LID measures were not implemented, the reduction in runoff due to 1% rain barrels was more effective for a 2-h than a 24-h rainstorm due to the limited capacity of the rain barrels to store water from a 24-h rainfall period.

The runoff coefficients for districts A, B, and C differed under the various scenarios but the trends were similar. Compared with the condition where LID measures were not implemented, the greatest runoff reduction effect was observed in district A followed by districts C and B. The performance (in terms of runoff reduction) of a particular LID in different areas is not proportional to the area of the region or to the percent impervious area.

The peak flow of sub-catchments in urbanized regions will increase relative to the size of the natural watershed. The peak runoff reduction performances of LID measures implemented at different locations under various rainfall scenarios are useful for deciding on a particular LID. In this study, the average peak runoff values of the districts with different arrangements of LID measures are listed in Table 2.

Table 2

The average peak runoff values of the districts with different arrangements of LID measures (m3/s)

peak runoff rainfall duration return period without LID green roof rain garden Permeable rain barrel combined 
2 h 1 yr 1.227 0.899 0.955 1.005 0.901 0.977 
2 yr 1.436 1.062 1.121 1.190 1.064 1.152 
 5 yr 1.880 1.404 1.498 1.606 1.418 1.540 
24 h 10 yr 0.866 0.750 0.666 0.705 0.863 0.746 
20 yr 0.964 0.843 0.742 0.785 0.963 0.840 
  50 yr 1.089 0.961 0.861 0.886 1.088 0.963 
2 h 1 yr 1.512 1.297 1.214 1.266 1.141 1.204 
2 yr 1.794 1.544 1.432 1.521 1.374 1.436 
 5 yr 2.394 2.069 1.989 2.083 1.882 1.929 
24 h 10 yr 1.272 1.190 1.057 1.084 1.270 1.012 
20 yr 1.426 1.335 1.179 1.214 1.423 1.166 
  50 yr 1.615 1.517 1.356 1.374 1.614 1.415 
2 h 1 yr 1.505 1.138 1.178 1.235 1.106 1.132 
2 yr 1.767 1.339 1.388 1.462 1.312 1.352 
 5 yr 2.328 1.771 1.862 1.971 1.758 1.850 
24 h 10 yr 1.176 1.083 0.917 0.966 1.174 1.018 
20 yr 1.316 1.221 1.024 1.081 1.314 1.146 
50 yr 1.492 1.395 1.183 1.224 1.490 1.314 
peak runoff rainfall duration return period without LID green roof rain garden Permeable rain barrel combined 
2 h 1 yr 1.227 0.899 0.955 1.005 0.901 0.977 
2 yr 1.436 1.062 1.121 1.190 1.064 1.152 
 5 yr 1.880 1.404 1.498 1.606 1.418 1.540 
24 h 10 yr 0.866 0.750 0.666 0.705 0.863 0.746 
20 yr 0.964 0.843 0.742 0.785 0.963 0.840 
  50 yr 1.089 0.961 0.861 0.886 1.088 0.963 
2 h 1 yr 1.512 1.297 1.214 1.266 1.141 1.204 
2 yr 1.794 1.544 1.432 1.521 1.374 1.436 
 5 yr 2.394 2.069 1.989 2.083 1.882 1.929 
24 h 10 yr 1.272 1.190 1.057 1.084 1.270 1.012 
20 yr 1.426 1.335 1.179 1.214 1.423 1.166 
  50 yr 1.615 1.517 1.356 1.374 1.614 1.415 
2 h 1 yr 1.505 1.138 1.178 1.235 1.106 1.132 
2 yr 1.767 1.339 1.388 1.462 1.312 1.352 
 5 yr 2.328 1.771 1.862 1.971 1.758 1.850 
24 h 10 yr 1.176 1.083 0.917 0.966 1.174 1.018 
20 yr 1.316 1.221 1.024 1.081 1.314 1.146 
50 yr 1.492 1.395 1.183 1.224 1.490 1.314 

As can be seen from the table, the peak runoff flow varied with LID type, and LID performance varied with district and rainfall scenario. Rain gardens and permeable pavements performed a similar degree of reduction under different rainfall durations, but the peak flow reduction due to rain barrels and green roofs under different rainfall durations varied greatly. Within an individual LID, the best performance was observed in district A followed by districts C and B. The combined measures performed best in district B followed by districts C and A. The combined measures performance was similar across rainfall duration and district. Relative to a separate sub-catchment area, a district contains many subareas and interlinked, peak flow reduction ratio of some LID measures were affected by inflow from other subareas. Rain barrels were effective for rainwater harvesting, and peak flow reduction worked well under the 2-h rainfall return scenarios but were not effective under the 24-h scenarios. Green roofs performed similar to the rain barrels but were generally not as effective at reducing peak storm runoff during the 24-h rainstorms.

CONCLUSIONS

This study investigated the performance of LIDs at different locations and provided useful information for LID arrangement and guidelines for LID application. The performance of a certain LID measure was similar at different locations but the effect was different. This indicates that simulation calculations are necessary prior to LID implementation. The effect of different LID measures varied, e.g., the performance of 1% rain barrels and 10% green roofs varied greatly under different rainfall durations, whereas the difference between 20% rain gardens and 15% permeable pavements was small. A comprehensive consideration of total runoff and peak flow management is needed to reduce flood risk because the LID measures used for runoff and peak flow reduction behaved differently. Different rainfall return periods changed the effect of a particular LID, and a reasonable combination of various LID measures should be investigated to determine the optimum layout scheme at a specific location. The low impact development concept is suitable for future urban development, but requires more research and an expanded application. Further research should focus on LID composite applications in combination with the local pipe network layout, which will ensure that the implemented system is aesthetically pleasing, economically viable and effective for reducing runoff and peak flow.

ACKNOWLEDGEMENTS

This study was sponsored by the National Natural Science Foundation of China (No. 41471015).

REFERENCES

REFERENCES
Ashbolt
S.
Aryal
S.
Petrone
K.
McIntosh
B. S.
Maheepala
S.
Chowdhury
R.
Gardner
T.
2012
Can stormwater harvesting restore pre-development flows in urban catchments in South East Queensland?
Water Science and Technology
67
(
2
),
446
451
.
Burszta-Adamiak
E.
Mrowiec
M.
2013
Modelling of green roofs’ hydrologic performance using EPA's SWMM
.
Water Science and Technology
68
(
1
),
36
42
.
Butler
D.
Parkinson
J.
1997
Towards sustainable urban drainage
.
Water Science and Technology
35
(
9
),
53
63
.
Fletcher
T.
Mitchell
V.
Deletic
A.
Ladson
T.
Seven
A.
2007
Is stormwater harvesting beneficial to urban waterway environmental flows?
Water Science & Technology
55
(
4
),
265
272
.
Gilroy
K. L.
McCuen
R. H.
2009
Spatio-temporal effects of low impact development practices
.
Journal of Hydrology
367
(
3
),
228
236
.
Huber
W. C.
2003
Hydrologic modeling processes of the EPA Storm Water Management Model (SWMM)
. In
World Water & Environmental Resources Congress 2003
,
American Society of Civil Engineers (ASCE)
,
Reston, VA
, pp.
1
10
.
Jia
H.
Lu
Y.
Yu
S. L.
Chen
Y.
2012
Planning of LID–BMPs for urban runoff control: the case of Beijing Olympic Village
.
Separation and Purification Technology
84
,
112
119
.
Longyan Regional Plan
2011
Longyan Urban Master Plan 2011–2030
.
Longyan city government
,
China
.
Pyke
C.
Warren
M. P.
Johnson
T.
LaGro
J.
Jr
Scharfenberg
J.
Groth
P.
Main
E.
2011
Assessment of low impact development for managing stormwater with changing precipitation due to climate change
.
Landscape and Urban Planning
103
(
2
),
166
173
.
Rossman
L. A.
2010
Storm Water Management Model User's Manual, Version 5.0
.
EPA/600/R-05/040
,
US Environmental Protection Agency
,
Cincinnati, OH
,
USA
.
Tillinghast
E. D.
Hunt
W. F.
Jennings
G. D.
D'Arconte
P.
2012
Increasing stream geomorphic stability using storm water control measures in a densely urbanized watershed
.
Journal of Hydrologic Engineering
17
(
12
),
1381
1388
.