A spatio-temporal cross comparison framework for the accuracies of remotely sensed soil moisture products in a climate-sensitive grassland region

Soil moisture is a key variable in hydrological processes and is included in a wide range of global datasets; however, accurate indications of spatio-temporal characteristics of soil moisture remain limited for East Asia. This study focused on Hulunbeir, a typical climate-sensitive grassland area in East Asia. Four soil moisture datasets were retrieved from Global Land Data Assimilation System (GLDAS), CMA Land Data Assimilation System (CLDAS), Soil Moisture and Ocean Salinity (SMOS) and Soil Moisture Active Passive (SMAP), and the differences between these datasets and measured soil moisture data were analyzed by correlation analysis. The key factors affecting soil water, such as slope, precipitation, and normalized difference vegetation index (NDVI), were selected to identify the main causes of errors between the product dataset values and measured data by multiple stepwise regression analysis. The four products tended to underestimate the soil moisture but exaggerated the temporal variation in winter and spring. The CLDAS and SMAP products provided good representation of the amplitude and spatio-temporal variation in soil moisture. The GLDAS product performed well with acceptable error metrics but failed to capture the vertical variation in soil moisture. The SMOS product had a missing data problem, which reduced its utility in the study area. Notably, NDVI and precipitation were the main factors leading to the underestimation by the above four products. This study provides an important reference for accurately describing the hydrological characteristics of climate-sensitive grassland areas to achieve efficient management of water resources.

Automated summary

This study compared four soil moisture datasets in the climate-sensitive grassland area of Hulunbeir in East Asia, finding that the products tended to underestimate soil moisture but exaggerated temporal variations in winter and spring. Specifically, CLDAS and SMAP products performed well in representing the amplitude and spatio-temporal variation of soil moisture, while GLDAS struggled to capture vertical variation.