Quantifying climate factor contributions to groundwater level changes under different soil freezing-thawing states with the WT-PCMCI model

Abstract

With the intensification of global warming, the groundwater system has been increasingly affected, especially in the frozen soil area, where obvious changes were observed in the groundwater level and soil freezing-thawing states. However, the research remains limited concerning the soil freezing-thawing conditions under climate change, and the response of groundwater level to corresponding climate factors. This study developed a causal inference model for causal estimation, by coupling the wavelet transform (WT) with Peter and Clark Momentary Conditional Independence (PCMCI) algorithm, termed as the WT-PCMCI model. The Taoerhe alluvial-proluvial fan was selected as a typical area with seasonal frozen soils, and three soil freezing-thawing periods were partitioned based on different soil freezing-thawing states. After clarifying the soil freezing-thawing characteristics of each period and their spatial and temporal distribution variations of groundwater level and climate factors, the WT-PCMCI model was applied to determine key climate driving factors and quantitatively analyze the time lag relationships between them. The main findings are stated as follows: (1) the key climate driving factors for groundwater level and their time lag relationships with groundwater, vary significantly across different soil freezing-thawing states and spatial distributions; (2) during the thawing period, rainfall and snowmelt are the key climate driving factors of groundwater level, with direct impacts. Air temperature and surface soil temperature could indirectly affect the groundwater level through the snowmelt process. During the non-freezing period, rainfall remains the key climate driving factor affecting the groundwater level; and (3) within the thawing period, the area where rainfall was the key climate driving factor of groundwater level accounts for 63% of the total study area. These findings would provide a novel model for attributing groundwater level changes in the context of climate change, which is critical for the study of groundwater dynamics in seasonal frozen soil areas under climate change.