Thermal conductivity of undisturbed soil – Measurements and predictions

Abstract Soil thermal conductivity, λ, is decisive for a range of soil ecosystem services related to soil temperature. Soil functions affected by temperature comprise microbial as well as abiotic processes of importance to crop growth, leaching of nutrients, carbon sequestration, and emission of greenhouse gases. Models for crop production and climate change include λ for simulation of heat transfer in the natural environment. Nevertheless, most prediction equations for λ are based on measurements deriving from physically disturbed soil. We identified nine Danish soils with a gradient in clay from ~ 0.03 to 0.35 kg kg−1 and measured λ in situ at six soil depths from 5 to 85 cm with the heat pulse method. Soil cores were collected and used for measurements of λ at a range of moisture conditions in the laboratory. The degree of soil water saturation, S, ranged from 0.01 to 0.98, and measured λ was in the range 0.18–2.98 W m−1 K−1. A model for λ identified by multiple regression across all laboratory measurements explained 87% of the variation in data and pointed out a convex polynomial relation between S and λ. It included significant positive effects of bulk density (BD) and soil organic matter (SOM), while λ decreased with soil clay content for a given S. The stochastic model predicted λ for three independent data sets from the literature with little bias and root mean square errors in the range 0.15–0.23 W m−1 K−1. A polynomial regression of λ versus S was performed for combinations of site and soil depth. By this λ was estimated at completely dry conditions, λdry, and at the fully saturated state, λsat. The estimates of λdry as well as λsat had a wider range than reported in the literature. This study calls for re-evaluation of existing pedotransfer functions for λdry. Measurements of λ in soil of undisturbed structural conditions are encouraged.