Links between seasonal precipitation intermittency and soil moisture variability

Precipitation is an important control of soil moisture on land. Thus, many studies have focused on understanding the influences of mean or total precipitation variability on soil moisture. However, the relationship between precipitation intermittency (the temporal distribution of rainfall events) and soil moisture variability remains largely underexplored. This question requires more attention as climate models are known to be deficient in their representation of precipitation intermittency (PI), and PI is projected to increase in a future warmer climate, potentially affecting soil moisture variability. In this study, we examine the associations between seasonal PI and soil moisture (SM) across the globe in observation-based datasets (ERA5, MSWEP, and GLEAM) and model simulations (CESM2 Large Ensembles – LENS2) for the period 1981–2020. As a methodology to quantify the associations between PI and SM, we use a conditional regression analysis of 10cm soil moisture onto a metric of PI (reverted number of wet days in a season) after the removal of the influence of total seasonal precipitation from each variable. 

The result suggests that in many regions, higher PI leads to decreases in SM under the same amount of seasonal precipitation. These associations are explained by increased runoff under higher PI. Therefore, the spatial patterns of the magnitude and sign of the linkage between PI and SM align with the global patterns of PI-runoff interactions. Additionally, the regions where evapotranspiration (ET)–SM correlations are high (>0.5) present higher SM sensitivity to changes in PI. CESM2 exhibits spatial consistency in the PI–SM associations with ERA5, although noticeable differences exist in the magnitudes of the regression coefficients between the two datasets. In general, the PI–SM associations are weaker in CESM2. This disparity is attributed to the different runoff sensitivity to changes in precipitation and PI. CESM2 exhibits reduced runoff sensitivity to PI than ERA5 over the entire globe. This finding implies that how runoff is modeled and constrained in climate models will affect future projections of soil moisture.