Estimation of soil thermal conductivity using hydraulic properties

Thermal properties of soils are very important in different areas of agriculture, engineering and soil science. Nevertheless their direct measurements are difficult, costly and time-consuming. Thus, in this study thermal conductivity of the soil was estimated using pedotransfer functions. In this regard, 156 data of Great Plains of South America were used to create pedotransfer functions in 12 steps by regression method for the estimation of thermal conductivity. At the first step, only the soil texture fractions were used as input variables. Later in other steps hydraulic and other physical properties were applied to estimate thermal conductivity. The model 2 using clay, silt/sand and bulk density as inputs and model 12 using the residual and saturated moisture contents, n and alpha parameters of van Genuchten model as inputs were the most accurate models. The model 2 showed RMSE and R2 values of 0.186 and 0.628, respectively, and the model 12 showed RMSE and R2 values of 0.235 and 0.319, respectively, in the evaluation step. The results showed that thermal conductivity was satisfactorily estimated using hydraulic and other physical properties as predictors.