Field monitoring and numerical simulation study on the long-term response of rainfall and groundwater in collapsible loess areas

Abstract

Abstract

Long-term rainfall–groundwater response in deep collapsible loess remains insufficiently understood, particularly under extreme rainfall conditions. This study investigates rainfall infiltration, soil-water response, wetting-front migration, and groundwater dynamics in a deep loess area of Zhengning County, China. A 100 m observation well was used for continuous monitoring of meteorological conditions, soil moisture, and groundwater levels from 2015 to 2020, and targeted observations during an extreme rainfall event in 2021 were also analyzed. The physical properties of a 175 m thick loess profile were examined, and rainfall infiltration processes were simulated using HYDRUS-1D. The loess is classified as low-plasticity silty clay, with silt accounting for 60%–80%. Rainfall was concentrated from May to October, representing 81.9%–90.2% of annual precipitation. Soil moisture in the upper 0-2.0 m was strongly controlled by atmospheric conditions, especially above 1.0 m. A rainfall intensity threshold of 12.4 mm/d caused a >5% increase in water content at 0.2 m, while deeper layers showed increasing response lag. Wetting-front depth followed a power-law relationship with rainfall duration (R 2 = 0.95) and a logarithmic relationship with cumulative rainfall (R 2 = 0.97). Single short-term heavy rainfall produced limited infiltration, whereas multi-stage sustained rainfall increased infiltration depth to 7 m during the extreme event. Groundwater levels declined slowly by 0.5 m over five years, with no rapid rainfall recharge observed. Simulations showed that when the rainfall intensity/saturated hydraulic conductivity ratio was <1, infiltration depth and wetted-zone water content increased markedly; under equal cumulative rainfall, longer-duration, lower-intensity rainfall promoted deeper infiltration. These results indicate that rainfall duration, intensity, and temporal pattern strongly control infiltration and wetting-front migration in deep loess, whereas groundwater responds slowly because of the thick vadose zone and lack of preferential flow pathways. The findings provide useful data for water-cycle simulation, groundwater assessment, and geological hazard early warning in deep loess regions of the Loess Plateau.