Development of class pedotransfer functions for integrating water retention properties into Portuguese soil maps

Hydrological modellers have recently been challenged to improve watershed models by better integrating soil information into model applications. Reliable soil hydraulic information is thus necessary for better describing the water balance components at the catchment scale. Frequently, that information does not exist. This study presents a set of class-pedotransfer functions (PTFs) for estimating the water retention properties of Portuguese soils. The class-PTFs were established from a dataset containing 697 soil horizons/layers, by averaging values of total porosity and volumetric water contents at -0.25, -1, -3.2, -6.3, -10, -33, -100, -250, and -1500 kPa matric potentials after grouping data by soil texture class, soil horizon, and bulk density. Fitted retention curves using the van Genuchten model were also obtained for every class-PTF. The root mean square error varied between 0.039 and 0.057 cm(3)/cm(3), with smaller values found when using the 12 texture classes of the International Soil Science Society (ISSS) system rather than the five texture classes of FAO, and when bulk density was also considered. The class-PTFs were then integrated into Portuguese soil maps and its usage was demonstrated by deriving maps of available water capacity to be used for modelling the water balance in a small catchment area with the SWAT model. The model successfully simulated the reservoir inflow when using the derived maps, but the results did not vary much whether using coarser or finer description of the catchment soils. Nonetheless, the class-PTFs contributed to a better soil characterisation than when using coarse-scaled information. The approach followed here was simple, inexpensive, and feasible for modellers with few resources but interested in considering the spatial variability of soil retention properties at large scales and in advancing hydrologic modelling in Portugal.