Development of pedotransfer functions for predicting hydraulic parameters of van Genuchten model by incorporating environmental variables on the Qinghai-Tibet Plateau

Abstract

The Qinghai-Tibet Plateau (QTP) region lacks large database of soil hydraulic parameters (SHPs) due to difficulty in soil sampling, and time- and labor-consuming measurement, which prevents accurate modeling of hydrological and biogeochemical processes. Pedotransfer functions (PTFs) using easily measurable soil properties may provide an alternative way to indirectly estimate SHPs. In this study, soil properties (bulk density (BD), clay, silt, sand, soil organic carbon, porosity, fitting parameter related to inverse of air entry pressure (α) and soil pore distribution (n), residual and saturated soil water content (θr and θs), and environmental parameters (altitude, slope gradient, annual average precipitation, air temperature, and evaporation) were determined from a north-south transect (ca. 500 km) in Qinghai Province and an east-west transect (ca. 1800 km) in Tibet Autonomous Region for establishing PTFs for SHPs on the QTP. The coefficient of variation of α was the largest, followed by θr, θs, and n, suggesting that α had the greatest spatial variability at the regional scale. There were significant differences in BD among 0–10, 10–20, and 20–30 cm soil layers (p < 0.05), whereas there were no significant differences in α, n, θr, and θs among these soil layers (p > 0.05). The five most important variables from the random forest analysis were selected to establish the PTFs of the SHPs and BD for the 0–30 cm soil layer on the QTP. The established PTFs for predicting SHPs and BD using multiple linear regression analysis were jointly affected by relief, climate, and soil properties. The accuracy of our newly established PTFs with the consideration of gravel content (< 10% or 10–30%) and mattic epipedon (existence or non-existence) was higher than that of the existing PTFs. We developed the first PTFs of SHPs on the QTP, providing a foundation to construct SHPs prediction models for other alpine regions in the world.