Abstract
The permanent wilting point (θPWP) is a crucial parameter in soil—plant—atmosphere research. The conventional θPWP, corresponding to a soil matric potential of –1500 kPa, is determined by the pressure-plate method, which necessitates specialized equipment that is not always readily accessible. Many pedotransfer functions have been developed to estimate θPWP, but they often face limitations due to soil type differences. This study aimed to construct a predictive model for θPWP using three hygroscopic water contents (θMRH, θRH80, and θRH50), which were measured at three relative humidity thresholds (98, 80, and 50%) and evaluate its performance. A new model for the dry—end section of the soil water retention curve (SWRC) was proposed as \(\theta = \left[ {1 - {{{\left( {\frac{{pF - p{{F}_{{{\text{MRH}}}}}}}{{p{{F}_{{{\text{RH50}}}}} - p{{F}_{{{\text{MRH}}}}}}}} \right)}}^{\beta }}} \right]\left( {{{\theta }_{{{\text{MRH}}}}} - {{\theta }_{{{\text{RH50}}}}}} \right) + {{\theta }_{{{\text{RH50}}}}}\) , and the hypothesis was that this model could be extended to estimate θPWP while keeping the parameter β unchanged. Data from 30 soils were analyzed, and the new model was compared with two existing models: θPWP = 1.5θMRH and θPWP = 3.0θRH50. The results showed that the new model had a slope of 1.019, a coefficient of determination (R2) of 0.989, a root mean squared error (RMSE) of 0.010 cm3 cm−3, and a mean absolute error (MAE) of 0.007 cm3 cm−3 in the validation, outperforming the other two models. Compared with five relevant studies in the literature, the standardized root mean square error (SRMSE) of the new model was 0.01, indicating higher prediction accuracy. The proposed model demonstrates robust applicability to soils exhibiting either linear or nonlinear dry-end behavior in their SWRCs, providing a more accurate and versatile way to estimate θPWP.
