Soil-Water Characteristic Curve (SWCC) models define the relationship between soil moisture and matric suction, which is essential for geotechnical and hydrological analyses.
This study evaluated the performance of seven widely recognized SWCC models: van Genuchten, Fredlund and Xing, Brooks and Corey, Haverkamp, Kosugi, Gardner, and Campbell, against field-instrumented data collected from compacted clayey soil. The comparative analysis based on performance metrics (Coefficient of Determination, RMSE, and Index of Agreement) revealed that the van Genuchten model provided the most reliable fit to the field-observed SWCC, with the highest R² (0.948), the lowest RMSE (0.01299), and the highest IA (0.9686).
While the Fredlund, Xing, and Haverkamp models showed competitive performance, their lower explanatory power (R²) and slightly higher RMSE values support selecting the van Genuchten model as the most robust for compacted clay conditions. The study confirms that model performance is highly soil-type dependent, as evidenced by the lower reliability of models like Brookes and Corey, and Kosugi, which are typically better suited to sandy soils. The results also highlight that even small differences in statistical performance metrics can significantly influence model selection for precise hydrological predictions. Ultimately, the findings underscore the necessity of model calibration using field data to ensure reliable representation of soil-water interactions, especially in engineering and environmental applications involving unsaturated clayey soils.
Authors and Reources:
