Pedotransfer functions for saturated hydraulic conductivity using a database with temperate and tropical climate soils

Abstract

Soil saturated hydraulic conductivity (Ks) is a key property in hydrological and environmental studies. As its measurement is unfortunately cumbersome, pedotransfer functions (PTFs) have been commonly used to estimate it. The literature brings a large number of Ks PTFs, most of which have been developed for temperate regions, small databases and/or limited territorial scales. The presence of soil structural variables in these models, such as those related to the water retention curve, is less common than the use of granulometric fractions and bulk density (BD). One objective of this work was to develop Ks PTFs for a database of tropical (Brazilian) and temperate (European) soils that include soil texture and/or BD, as well as effective porosity (EP) [total porosity minus θ (300-cm suction)], which is a soil structural property. This study evaluated the performances of the proposed PTFs and compared their predictions to those from seven models used in the literature, considering a testing database with characteristics similar to the calibration dataset. Comparison of the performances of the PTFs were made at three levels of representation of the study databases: all soils, temperate and tropical soils, and textural class groups. The results showed that the PTF proposed based only on EP was more accurate and reliable than the PTFs based only on texture and/or BD, also with greater reliability than the other models from the literature, reinforcing the importance of using structural variables in the estimation of Ks. We confirmed that the use of models developed for temperate soils based on texture and/or BD must be avoided in the prediction of Ks of tropical soils and vice versa. This study indicates the need for including new predictors in Ks PTFs, especially those that reflect macropore flows better. In addition, standard and more representative Ks measurement methods, based on an adequate sample size, are also required for achieving more accurate Ks predictions.