A Comparison of Soil Moisture Models Using Soil Climate Analysis Network Observations

Abstract

Abstract Because of the lack of field measurements, models are often used to monitor soil moisture conditions. Therefore, it is important to find a model that can accurately simulate soil moisture under a variety of land surface conditions. In this paper, three models of varying complexities [the Variable Infiltration Capacity (VIC), Decision Support System for Agrotechnology Transfer (DSSAT), and Climatic Water Budget (CWB) models] that are commonly used for simulating soil moisture were evaluated and compared using soil moisture data (1997–2005) from three Soil Climate Analysis Network (SCAN) sites (Bushland, Texas; Prairie View, Texas; Powder Mill, Maryland). Results demonstrated that DSSAT and VIC simulated soil moisture more accurately than CWB at the three SCAN sites. DSSAT and VIC both accurately simulated the annual cycle of soil moisture and the wetting and drying in response to weather conditions, as evidenced by the relatively strong correlations, but could not accurately simulate the actual soil water content in the upper soil layers (the mean coefficients of efficiency E for all DSSAT and VIC simulations were −0.8 and −2.6, respectively). CWB could not accurately simulate soil moisture at any of the SCAN sites. Model performance varied significantly not only from model to model but also from year to year and from location to location. Model sensitivity analysis using the factorial approach suggests that DSSAT is more sensitive than VIC and that model sensitivity varies by locations, indicating that parameter sensitivity is more strongly controlled by climatic gradients than by changes in soil properties.