Estimating soil water retention curves from co‐located in situ water content and matric potential sensors

Abstract The soil water retention curve (SWRC) governs water storage and redistribution in the vadose zone and is a key input for hydrologic and agroecosystem models. Laboratory evaporation methods provide high‐quality SWRCs but are destructive and represent static conditions that may not reflect in situ soil behavior under natural field conditions. Co‐located field sensors measuring volumetric water content ( θ ) and matric potential ( h ) offer a continuous, nondestructive alternative for estimating field‐based SWRCs. In this study, the extent to which SWRCs can be derived from field‐installed sensors by directly comparing in situ θ – h relationships with laboratory evaporation measurements was evaluated. Six parametric retention models (Brooks–Corey, Fredlund–Xing, Kosugi, van Genuchten‐mnvar, van Genuchten [ m = 1 − 1/ n ], and van Genuchten‐PDI) were fitted to laboratory and field datasets in silt‐loam soil. Model performance and parameter stability were assessed using root mean square error of water content (RMSE_ θ ), and corrected Akaike information criterion (AICc). Field capacity (FC), permanent wilting point (PWP), and plant‐available water (PAW) were also calculated. Field‐derived θ s was consistently lower than laboratory‐derived θ s , indicating effective saturation in field conditions than complete saturation in laboratory conditions. No single retention model consistently provided the best fit across all laboratory and field datasets. The FC was relatively consistent across models, but PWP and PAW were more sensitive to model structure and extrapolation for θ r . These results indicate that co‐located θ–h sensors can help field‐based SWRC estimation when model selection accounts for the limited suction domain. Additional high‐suction observations would improve constraint of dry‐end parameters and derived hydraulic indicators.