1. INTRODUCTION
Waterlogging has become an increasingly common occurrence in recent years alongside climate-change-induced weather extremes and aggressive urbanization. Waterlogged structures and roads can severely negatively impact surrounding communities. Low impact development (LID) facilities are widely used in the United States, the United Kingdom, China, and other countries as an effective means of urban stormwater control and hydrological cycling (Dietz, 2007; Pyke et al., 2011; Ahiablame and Shakya, 2016; Jia et al., 2016; Hu et al., 2017; Wu et al., 2018a; Li et al., 2019b). Rainfall distribution is not uniform and evaporation exceeds rainfall in collapsible loess areas of China, where LID facilities should be designed not only to reduce the peak rainfall and minimize rainwater pollutants but also to promote regional rainwater circulation and effective utilization.
Collapsible loess, a water-sensitive type of soil with a sub-stable structure, is widely distributed throughout the world (Shroder et al., 2011; Chen et al., 2018; Lian et al., 2020). Loess soil readily disintegrates under the action of rainfall and irrigation, which can cause nearby structures (e.g., buildings, municipal roads) to crack and even to collapse. Slope instability, landslides, and other geological disasters are frequently reported in loess areas (Kozubal and Steshenko, 2015; Wu et al., 2017; Yates et al., 2018; Assadi-Langroudi, 2019). It is crucial to effectively understand the seepage law and moistening deformation of the infiltration rainwater of LID facilities in loess areas to ensure the safety of the facilities and surrounding structures.
LID facilities mainly include permeable pavement, green roofs, and bioretentions. Previous studies on LID facilities have mainly centered on peak reduction design, non-point source pollution risk reduction, and portfolio management (Ahiablame and Engel, 2012; Tang et al., 2016; Eckart et al., 2017; Ma et al., 2017; Winston et al., 2018; Yekkalar et al., 2018; Chen et al., 2019; Hou et al., 2019a; Sohn et al., 2019). There has been little research to date on risk prevention or rainwater detention control measures in loess area LID facilities. Academics and professionals agree that the LID facilities in collapsible loess areas must be specially equipped for safety; this may include shallow and small-scale designs and full paving with impermeable membranes at the bottom of underground facilities (Zhang, 2016; Si et al., 2018).
Wang et al. (2019) and Deng et al. (2020) studied the impact of rainwater on loess sites in cases of impervious membrane leakage to develop a construction optimization scheme. Their work has assisted in designing LID facilities in collapsible loess areas. However, the overall anti-seepage measures still have some drawbacks as the factors that further reduce the rainwater retention time and the total amount of infiltration and storage in the underlying surface of the facilities can cause the detention function ineffective. Rainwater infiltration and soil strength and deformation laws can be used to identify the controllable critical range of rainwater infiltration in the construction of LID facilities in loess areas. The numerical simulation Geo Studio has been used to analyze the seepage field and displacement field of different bioretention facilities in collapsible loess areas (Chai et al., 2019; Liang et al., 2020) and ultimately to optimize the length of underground bioretentions in various loess areas. The function of LID facilities in the loess area can be ensured using the methods described above. However, as a unique unsaturated soil, the infiltration mechanism of collapsible loess is not yet fully understood. The infiltration of loess is affected by initial water content, temperature, and pore characteristics (Wang et al., 2010; Haeri et al., 2012; Li and Li, 2017; Shao et al., 2018; Li et al., 2019a; Zhang et al., 2019). These characteristics are difficult to simulate numerically because the parameters and the calculation model of water movement law cannot effectively reflect water movement through loess; besides, the infiltration boundary conditions are unrealistic. Compared with the numerical simulation, the field test study can obtain the actual infiltration law of water in Loess more accurately.
Many previous researchers have investigated the law of water infiltration in loess. The depth of water infiltration is limited in collapsible loess sites if diversion measures are not taken. The influence of concentrated infiltration in loess areas on water content is more intense than that in natural rainfall infiltration areas (Tu et al., 2009; Huang et al., 2011; Min et al., 2017; Wu et al., 2018b; Hou et al., 2019a, 2020). However, previous studies have centered on the water infiltration law of single-layer collapsible loess. The structures of LID facilities vary but are mainly composed of replacement fill and gravel layers. LID facilities are also often arranged around municipal roads or buildings, which makes these sites vulnerable to the initial stress field of the subgrade. The law of water infiltration in a single loess site does not accurately represent changes in the seepage field and deformation field of the site after water infiltrates the facility.
A typical LID facility construction project in the loess area of Shaanxi Province, China, was taken as the engineering background for this study. Previous optimization results regarding the length of the anti-seepage membrane at the bottom of a typical bioretention, also a Q3eol collapsible loess site was used to selected a full-scale rainwater infiltration test of bioretention facilities near municipal roads under the worst-case rainfall intensity. The law of water infiltration was determined accordingly. Consolidation compression test data was used to calculate the wetting settlement of the site. The applicability of the modified Green-Ampt model in predicting the infiltration depth of LID facilities in the loess area was verified as discussed in detail below.