Soil thermal properties from SoilGrids data: model tested in eastern Poland

Bogusław Usowicz (Białystok University of Technology and Institute of Agrophysics, Polish Academy of Sciences) and Jerzy Lipiec (Institute of Agrophysics, Polish Academy of Sciences) have published a paper in International Communications in Heat and Mass Transfer (DOI 10.1016/j.icheatmasstransfer.2026.112714) on predicting the thermal properties of soil. The article, made available on 29 September 2026, is assigned to volume 180 of the journal and is open access. The authors point out that a dependable estimate of these properties is needed to describe how energy is split at the ground surface and how heat moves through the soil.

Method

Their approach yields estimates of four quantities: thermal conductivity, thermal diffusivity, heat capacity and thermal inertia. It builds on a modified statistical-physical model that takes into account specific heat, conservation of energy and the arrangement of mineral, organic, water and air particles in the soil.

The input data come from the global SoilGrids database. They include the contents of the main grain-size fractions, of quartz and other minerals, of organic matter and of water, as well as air-filled porosity, bulk density and particle density, each given for three soil water potentials (minus 10, minus 33 and minus 1500 kPa). The method was tested on soils of different texture at nine sites in eastern Poland.

Results

Whatever the texture and depth (to 1.5 m), all four properties, from conductivity and diffusivity to heat capacity and inertia, were consistently lower at a potential of minus 1500 kPa than at the other two. At that potential, more sand and quartz went together with lower thermal conductivity and a wider spread of thermal inertia through the soil profile. At all three potentials, it also went together with a wider spread of thermal diffusivity.

The authors present the method as a proof of concept that bears out the assumptions behind it. In their view it allows quick regional-scale forecasts of soil thermal properties using input taken from the global SoilGrids database.