Pedotransfer Functions of Soil Thermal Conductivity for the Textural Classes Sand, Silt, and Loam

Data of soil thermal conductivity (lambda) covering the whole range of soil water content (theta) are needed in many environmental disciplines. However, laboratory measurements of l(theta) are time consuming and costly. Therefore the use of pedotransfer functions is promising. In this study, we describe the procedure of deriving such lambda(theta) pedotransfer functions for the FAO texture groups (TG) of Sand, Silt and Loam. First, high-resolution lambda(theta) measurements from saturation to dryness of 43 soils were conducted by two laboratory methods: the evaporation approach and single-point measurements of lambda at distinct water contents. Measurements were obtained for a set of sandy, silty, loamy, and clayey soils, and for two urban soils and two peat soils. The slope and magnitude of the lambda(theta) curves show typical dependencies on bulk density (p(b)) and soil texture. In a second step, the lambda(theta) data were used to evaluate the lambda(theta) model as suggested by Lu et al. (2014). Fitting this model to the complete data set yielded a poor agreement with RMSE = 0.53 W m(-1) K-1. However, satisfying results were obtained by dividing the data into TGs Sand, Silt and Loam resulting in RMSEs of 0.18, 0.11 and 0.24 W m(-1) K-1, respectively. Estimated lambda values at low soil moisture (wilting point) were hardly different among texture classes, but lambda differences become highly relevant when soil moisture increased to field capacity. The derived pedotransfer functions are an easy-touse approach providing reliable lambda(theta) curves for sandy, silty and loamy soils, when only basic soil information (texture and bulk density) are available.