Estimating the Soil Water Retention Inflection Point Using Pedotransfer Functions

Abstract

The inflection point of the soil water retention curve (SWRC) is increasingly recognized as a key indicator of soil physi- cal quality, as it reflects critical changes in pore structure and water availability. This study aims to develop and validate pedotransfer functions (PTFs) to estimate the water content (θi), matric suction head (hi), and slope (Si) at the SWRC inflexion point from basic soil physical properties. A dataset comprising 219 soil samples, including laboratory-measured and UNSODA database entries, was used. The inflection point parameters were computed analytically from van enuchten model fits. Linear, nonlinear, and polynomial regression techniques were applied to derive PTFs using soil organic matter, bulk density, geometric mean particle diameter, and geometric standard deviation as input variables. Model performance was evaluated using root mean square error (RMSE), normalized root mean square error (NRMSE), correlation coefficient (r), and Taylor diagrams. A dataset comprising 219 soil samples, including laboratory-measured and UNSODA database entries, was used. The inflection point parameters were computed analytically from van Genuchten model fits. Linear, nonlinear, and polynomial regression techniques were applied to derive PTFs using soil organic matter, bulk density, geometric mean particle diameter, and geometric standard deviation as input variables. Model performance was evalu- ated using root mean square error (RMSE), normalized root mean square error (NRMSE), correlation coefficient (r), and Taylor diagrams. Six PTFs were developed for θi (best model: r = 0.90; NRMSE = 8.6%), six for Si (best model: r = 0.71; NRMSE = 14.8%), and three for hi (best model: r = 0.46; NRMSE = 39.4%). θi and Si were estimated with good to excellent accuracy, while hi proved more difficult to predict due to its dependence on microstructural properties not captured by standard soil descriptors. The developed PTFs for θi and Si are reliable and practical tools for assessing soil hydraulic behavior and physical quality. In contrast, accurate estimation of hi remains challenging, suggesting the need for additional structural or imaging-based predictors in future models.