Coupling published pedotransfer functions for the estimation of bulk density and saturated hydraulic conductivity in stony soils

International audienceSaturated hydraulic conductivity (Ks) is one of the key parameters in the design of engineering and environmental structures built in the unsaturated zone for infiltration purpose. Unfortunately, its field determination is laborious and expensive and suffers non-uniqueness. Besides, pedotransfer functions (PTFs) have been established to translate some measured soil matrix properties such as bulk density, organic matter and soil texture into Ks. Especially when the soil is heterogeneous, these PTFs may exhibit large differences between the matrix-predicted and in situ measured Ks. Moreover, there is no published PTF to predict Ks in stony soils. In this study, the objective is to estimate Ks in heterogeneous stony soils by combining already published PTFs. A methodology has been developed to evaluate the performance of the combinations often published PTFs predicting Ks from bulk density in fine soil matrix (BDmo) with thirty two published PTFs predicting BDmo from the soil matrix texture properties. These 320 combined functions are likely to allow finding a correct estimation of Ks in most of fine soils but not in stony soils. To test this assumption, in situ Ks measurements with a Guelph permeameter and matrix texture and organic matter determinations in laboratory were carried out in a stony soil (colluvium from limestone). The magnitudes of measured Ks are much higher than the values predicted for the given texture of the matrix by 320 combined functions. This suggests that the rock fragment fraction in the soil plays an important role in accelerating the evacuation of water by increasing the cinematic porosity. Then, the empirical relation of Torn et al. (1994) is used to take into account this additional porosity within the real bulk density of the soil matrix. The retention capacity of the stony soil is controlled by the matrix content, which provides for a relation between rock fragment content and residual water content. Finally, six combined functions transformed by the Torri's relation are selected which predict Ks for a stony soil. (C) 2015 Elsevier B.V. All rights reserved