Causal Insights on the dynamics of Soil Moisture across Agro-climatic regions of India

Soil moisture stands as a pivotal element in land-atmospheric interplay, crucially affecting water, energy, and biogeochemical cycles. At a larger spatial scale, meteorological forces and land cover patterns significantly influence soil moisture dynamics. Prior studies primarily focused on investigating soil
moisture variability with soil texture and topography which have prominent influence factors at small spatial scales. Previous studies have employed various correlation techniques to explore these interactions however, correlation does not necessarily imply causation. Our study employs a causal discovery approach, specifically, Peter Clark Momentary Conditional Independence (PCMCI) using 8-day satellite-based gridded data from NASA's SMAP soil moisture, precipitation, leaf area index (LAI), evapotranspiration (ET), land surface temperature (LST), and vapor pressure deficit (VPD). The data was collected across diverse climate classes in India, and seasonal effects were removed to get valuable causal insights. Our results reveal that precipitation plays a predominant role in arid regions compared to humid ones, while LST and VPD significantly impact sub-arid regions. Notably, ET demonstrates substantial influence in the sub-humid regions. The maximum lag in causation is 2.5 to 3 months between the soil moisture and other variables. These findings significantly enhance our comprehension of soil moisture dynamics across various agro-climatic classes, thereby help in enhancing soil moisture predictions and improving agricultural drought forecasting.