A simplified approach to estimate water retention for Sicilian soils by the Arya-Paris model

Application of the Arya and Paris (AP) model to estimate the soil water retention curve requires a detailed description of the particle-size distribution (PSD) because the scale factor a, relating the pore length of an ideal soil to that of the natural one, depends on the particle size distribution parameters. For a dataset of 140 Sicilian soils that were grouped in five texture groups, the logistic and linear models were applied to evaluate a, and the water retention values predicted by the AP model were compared with the measured ones. Using the parameters proposed by Arya et al. (1999), the two models yielded similar unsystematic root mean error of estimate (RMSEu). Therefore, their potential accuracy was considered comparable. However, the water retention data predicted by the logistic model were more biased (higher systematic root mean error of estimate, RMSEs) than those predicted by the linear model. A calibration was conducted for the logistic model to obtain five sets of parameters specifically developed for Sicilian soils. The calibrated logistic model only minimally improved the prediction accuracy of the AP model. This result also supported the Arya et al.’s procedure (Arya et al., 1999) for soils that were not included in their original calibration dataset. With the aim to simplify AP model application, an alternative procedure was developed by optimizing a soil-specific a value in the range of measured water content values. For Sicilian soils, the optimised values of the scale parameter (OPT) were significantly correlated with clay content and bulk density. The empirical relationship that was obtained for the calibration dataset allowed to estimate the water retention data of the validation dataset (N = 70) with an estimation error (RMSE = 0.042 cm3cm-3) lower than that of the traditional approach based on the logistic model. Therefore, it can be considered as a reasonable alternative to the more complex logistic model for estimating the water retention curve of Sicilian soils