Soil Water Movement in Response to Temperature Gradients: Experimental Measurements and Model Evaluation

Abstract

AbstractTemperature gradients may have a significant effect on soil water movement under certain conditions, but inclusion of these effects adds complexity to the flow analysis. This study was conducted to help clarify the significance of nonisothermal water flow, and to examine theoretical and numerical descriptions of the transport processes. At initial water contents of 0.00, 0.049, 0.099, 0.151, and 0.282 m3 m‐3, isothermal and nonisothermal laboratory experiments were conducted to provide direct information on soil water movement in response to temperature gradients. These data were used to evaluate the numerical simulation model SPLaSHWaTr2, and to examine calculation of the thermal conductivity, λ(ϑ), the thermal vapor diffusivity, DTv(ϑ), and the temperature coefficient of the matric potential, Cψ, based on modifications to a theory proposed by Philip and de Vries in 1957. A statistically significant effect of the temperature gradient was found at an initial water content of 0.151 m3 m‐3, which corresponds to a pressure head of approximately ‐1.2 m of H2O. No effect of the temperature gradient was found at initial water contents of 0.00, 0.049, 0.099, or 0.282 m3 m‐3. Under isothermal conditions, the model provided simulated water‐content profiles that were in good agreement with measured profiles. Under nonisothermal conditions, profiles simulated by the model were in poor agreement with the measured data, using the original values of λ(ϑ), DTv(ϑ), and Cψ. Sensitivity analysis showed that λ(ϑ) and DTv(ϑ) had a negligible influence on nonisothermal water movement. On the other hand, Cψ had a significant influence on nonisothermal water movement. Adjusting Cψ from the original value of ‐0.0068 K‐1 to the value suggested by Philip and de Vries, ‐0.00209 K‐1, improved the agreement between simulated and measured water‐content profiles, particularly at an initial water content of 0.282 m3 m‐3. With the adjustment in Cψ, the model simulations were in good agreement with the measured data, indicating that the Philip and de Vries theory provides an adequate description of the nonisothermal transport processes.