Experiments and COMSOL simulations: A comparative study of the heat flux plate method and the gradient method for soil heat flux measurements in barren sand

Soil heat flux (G) is important for studying the energy balance at the soil-atmosphere interface. In most cases, a reference flux (true value) is not available for soil heat flux. The heat flux plate method and the heat pulse sensor gradient method are two commonly used methods to measure soil heat flux. Both methods have inherent shortcomings that lead to unavoidable measurement errors. There is a need for a theoretical analysis to identify the origin of errors associated with the gradient method. In this study, by introducing a plane heater of known heating strength to generate a one-dimensional heat flux, we compared the measurements of soil heat flux by both methods at different soil moisture contents for laboratory (indoor) and field (outdoor) experiments, as well as computer simulations for a sandy soil. Differences between the COMSOL simulation results and the indoor measurements were less than 15%, indicating that the COMSOL simulated soil heat flux results were reliable. The indoor experiments provided close agreement between the heat flux plate method and the gradient method (Pearson's correlation coefficient r >= 0.993). The results of the field experiments showed that the correlation coefficient of the two methods decreased by at least 9.3%. Our field measurements indicated that the gradient method is not suitable for measuring G in the presence of alternating freezing-thawing soil conditions. Although heat flux plates have several short-comings, they may perform better than the heat pulse sensor gradient method in partially frozen soils. The finite difference approximation of nonlinear thermal gradients causes errors in the gradient method heat flux values. However, this gradient method error could be eliminated by altering the heat pulse probe positioning with depth to provide finer scale measurements of temperature with depth.