Hydrological modeling and field validation of a bioretention basin

Abstract

An emerging green infrastructure, the bioretention basin, has been deployed world-wide to reduce peak flows, encourage infiltration, and treat pollutants. However, inadequate design of a basin impairs its treatment potential and necessitates the development and validation of a suitable hydrological model for design and analysis of bioretention basins. In this study, an existing numerical model, RECHARGE, has been adopted to simulate hydrological performance of a basin in the tropical climate of Singapore over a half year that included 80 storm events. Comparison of the model predictions with field observations shows that RECHARGE successfully simulates the basin hydrology of 80 events of varying rainfall characteristics with mass balance error of 5.1 ± 7.5% per event and 0.3% overall. Using the verified model, we develop new design curves that predict bioretention basin performance as a function of three basin design parameters: detention depth; ratio of drainage basin area to bioretention area; and saturated hydraulic conductivity of the basin soil media. We evaluate basin performance in terms of the percentage of water that infiltrates and is treated in the subsurface portion of the basin and define an infiltration index to measure the change in infiltrated percentage caused by unit change in the basin design parameters. The marginal improvement in basin performance drops significantly when the basin depth (h_d) is increased above 40 cm, when the ratio of drainage area to bioretention area (R) is decreased below 20, or when the saturated hydraulic conductivity (K_s) is increased above 10 cm/h.

Introduction

Concrete-lined channels have often been used to manage stormwater in the tropics. However increasing urbanization and surface imperviousness can cause channel capacity to be exceeded, resulting in increased incidence of flooding in urban areas. Floods can be mitigated by encouraging local infiltration through Best Management Practices (BMPs) such as bioretention basins (Nehrke and Roesner, 2004). More importantly, such green infrastructure also reduces non-point source pollution and eutrophication by providing a surface basin and underlying soil treatment medium to remove pollutants such as nitrogen, phosphorus, and suspended solids (Payne et al., 2014). However, insufficient storage capacity in a basin impairs its treatment potential. An undersized basin will experience a high rate of overflow and allow stormwater to bypass storage and treatment, leading to pollutant removal rates below targets (Wang et al., 2017). How to size basins effectively to meet pollutant removal targets therefore is an important research and design question.

With the accumulation of experience in working with BMPs across the world over the past 20 years, BMPs have become part of a regulatory and planning requirement for land developers. Planners are required to demonstrate that with the installation of BMPs, the post-construction runoff will be able to replicate the pre-construction hydrology as closely as possible. It is therefore necessary to quantify accurately the effects of BMPs in encouraging infiltration and reducing stormwater discharge. However, current quantification of these effects, and hence design guidelines that arise from such quantification, has mostly been based on “rules of thumb,” such as the depth of rainfall to be captured (Wang et al., 2017), rather than predictive forecasting. The need to predict basin performance prompts the development and validation of the hydrological model for bioretention basins reported in this paper.

There are few hydrological models specifically for bioretention basins. Heasom et al. (2006) employed the Green-Ampt representation of infiltration to develop a simple model of bioretention basin infiltration and culvert bypass. Infiltration is represented as a two-stage process with two constant lumped rates: a rapid infiltration stage at the start of an event in the absence of surface ponding and a second steady infiltration stage when infiltration occurs under a surface ponding condition. As noted by the authors themselves, their method of modeling could be improved by relating infiltration to physical characteristics such as soil properties or by simulating saturated and unsaturated flow with a groundwater model. The HYDRUS software packages are finite-element variably-saturated models that numerically solve for the movement of water using the Richards' equation for unsaturated water flow (Šimůnek and Van Genuchten, 2006). Although this code models infiltration with an approach that is more physically based and with higher resolution than using lumped constants as by Heasom et al. (2006), it does not have a surface water balance to account for ponded infiltration nor is it specifically built for bioretention basins. A finite-element model of variably unsaturated flow was developed by He and Davis (2011) specifically for bioretention basins but this model considers theoretical rain storms rather than actual hydrographs and was not validated against field measurements. The model also lacks a complete accounting of surface-water flow into and out of the surface basin. Another complex model, RECHARGE, is an advanced hydrological model for bioretention basins that uses high-resolution modeling of unsaturated flow using the Richards' equation coupled with a surface water balance (Dussaillant et al., 2004). Since its development, RECHARGE has been verified against single events in a controlled experimental setting, but not at field scale or over a continuous time period that includes multiple storms and inter-storm periods.

The current literature lacks a field-verified, bioretention-basin-specific, high-resolution model to simulate ponded infiltration over a continuous time period. In this paper, we aim to bridge that gap by applying the RECHARGE model in a continuous simulation over a half year and validating the model against detailed field measurements of 80 events spanning a wide variety of rainfall depths and intensities in a tropical setting (Singapore). We present research on high-resolution single-event and continuous (half-year) modeling of hydrological behaviour that predicts time series of discharges from subsurface and surface outlets as well as the surface ponding depth and the soil moisture within the subsurface filter media. We then apply an infiltration index (ID) to describe predicted model performance for a range of engineering parameters. The objectives of this study were to (i) validate the applicability of the RECHARGE hydrological model for continuous simulation at field scale, (ii) predict bioretention basin performance in terms of infiltration capability as a function of engineering design parameters such as detention depth, ratio of drainage area to basin area, and saturated hydraulic conductivity of the soil media, and (iii) recommend basin configurations that improve performance efficiency.