Spatial-temporal behavior of precipitation driven karst spring discharge in a mountain terrain

Abstract

• Obtain the active time length of precipitation driven spring discharge (SD). • Rank the contribution of precipitation at each subarea (ESA) to the SD process. • Classify rainfall-SD in three patterns depending on landform and aquifer behaviors. • Quantify the contribution of precipitation at ESA to the maximum and medium SD. Behaviors of karst springs are manifestations of spatial and temporal dynamics involving multi-hydrogeological processes, including precipitation, surface water runoff, infiltration, groundwater flow, and anthropogenic activities. These dynamic processes are usually nonlinear and nonstationary. In this study, we couple the Shapley Additive exPlanation (SHAP) with a Long Short-Term Memory (LSTM) recurrent neural network to produce an interpretable deep learning model to explore the precipitation driven spring discharge mechanism and to predict spatial–temporal behaviors of karst springs. Applying the model to Niangziguan Springs catchment, China, we show that the precipitation infiltration volume of each catchment subregion is the primary factor driving the spring discharge, and the precipitation over the 12-month period has the most significant effect on the spring discharge. We categorize the precipitation-driven spring discharge at the catchment into three patterns according to each subregion's landform and karst aquifer characteristics based on the SHAP analysis. In the regions with deeply buried karst aquifers, moderate to light precipitation recharges the karst aquifer. On the other hand, heavy precipitation recharges the karst aquifer in the river valley regions more efficiently than others. In the regions where karst aquifers are exposed and groundwater discharges, the groundwater level is the primary factor dictating precipitation and spring discharge processes.