Identifying multivariate controls of soil moisture variations using multiple wavelet coherence in the U.S. Midwest

Abstract

The spatiotemporal pattern of soil moisture (SM) is challenging to identify as it is defined by the complex interactions of the atmosphere, soil, and vegetation cover. We addressed this challenge by identifying the dominant controlling factors of soil moisture variations in i) a single factor, ii) pair factor, and iii) three-factor-combination scenarios using bivariate and multiple wavelet coherence. The analysis was done over the U.S. Midwest at four soil depths (0–10, 10–40, 40–100, and 100–200 cm) using grided datasets covering 1982–2020. The scale-dependent analysis in the time domain shows that vapor pressure deficit (VPD) is the primary controlling factor of the soil moisture dynamics in the topsoil layer, and the deeper layer's dynamic is primarily controlled by precipitation. Based on all scales, in single-factor analysis (i.e., explaining SM dynamics using only one factor), VPD, precipitation, and radiation have a higher coherency with SM than other factors. For two-factor analysis, the combination of precipitation-radiation, and for three-factor analysis, the combinations of precipitation-VPD and a vegetation index are the best predictors of SM variations. Overall, our analysis highlights that the dominant factors affecting the soil moisture variability could be different at various time scales, locations, and soil layers. The temporal scale dependency is mainly due to the seasonality of the factors, while space and location dependency is primarily due to the inherent characteristics of the climate, environment, and soil. Our findings provide useful insights into how time-scale dependency of soil moisture is linked with agro-meteorological factors, particularly for downscaling or upscaling issues in hydro-climatological modeling.