Bias detection of ISMN soil moisture measurement through soil water balance model and data assimilation

The international soil moisture network (ISMN) provides an important in-situ soil moisture dataset, which is widely utilized for hydrology, agriculture, environmental sciences, and remote sensing validation studies. ISMN soil moisture measurements are generally based on the relationship between soil moisture and other directly observable variables (e.g., dielectric constant) and therefore tend to be influenced by many factors at different installation sites, such as temperature, bulk density, texture, and mineralogy. Based on a previous study (Li, et al., 2020), it is found that coupling a linear bias-aware physical soil water model with data assimilation can effectively detect and calibrate the soil moisture measurement bias. The utilization of a sophisticated physical soil water model can accurately identify the bias but generally requires high costs, which makes extensive evaluation of the ISMN dataset on a large scale difficult. Therefore, a simplified model with less computational cost and satisfying simulation accuracy is needed. Herein, an efficient and bias-aware soil bucket balance model with a data assimilation scheme is developed. The newly developed model without significant accuracy loss has been used to evaluate ISMN data in the Conterminous United States (CONUS). Results show that the proposed model can effectively identify the bias direction based on soil water balance, and that there are many measurements with bias in the ISMN dataset over CONUS.