Modeling coupled heat and moisture flow within a bare desert soil

Revegetation of semi-desert rangeland is dependent on rainfall, weather, and soil factors affecting seed germination and seedling establishment. To aid in predicting seed zone temperature and moisture following rainfall, a weather driven one-dimensional computer model was developed to simulate the simultaneous flow of heat and water within a bare semi-desert soil. The Newton-Raphson method was used to solve the surface energy budget equation for surface temperature. The coupled soil heat and water flow equations were then solved numerically using the weighted average finite-difference method to calculate the subsurface temperature (T{dollar}sb{lcub}rm s{rcub}{dollar}) and water content ({dollar}thetasb{lcub}rm v{rcub}{dollar}) profiles. Weather data and soil thermal and hydraulic properties were the only required inputs to the model.; The model was tested using two data sets collected in the Altar Valley of Arizona during the summer rainy season of 1988. Data set 1, collected from calendar day (CD) 198 to 205, was used to calibrate the model. Calibration tests revealed that the model markedly underestimated T{dollar}sb{lcub}rm s{rcub}{dollar} when measured values exceeded 37{dollar}spcirc{dollar}C. Underestimation of T{dollar}sb{lcub}rm s{rcub}{dollar} was found to be related to overestimation of latent heat flux. Therefore, the modelled latent heat flux was reduced as a linear function of air temperature (T{dollar}sb{lcub}rm air{rcub}{dollar}) when T{dollar}sb{lcub}rm air{rcub}>30spcirc{dollar}C. Also, soil thermal conductivity values predicted by the de Vries model had to be reduced 80% in order to achieve acceptable agreement between measured and modelled T{dollar}sb{lcub}rm s{rcub}{dollar}. Data set 2, from CD 191 to 195, was then used to validate the calibrated (modified) model.; Results obtained with data set 2 indicated that the modified model accurately simulated T{dollar}sb{lcub}rm s{rcub}{dollar} at 0.01 m depth even when the measured T{dollar}sb{lcub}rm s{rcub}{dollar} at that depth exceeded 50{dollar}spcirc{dollar}C. Simulated T{dollar}sb{lcub}rm s{rcub}{dollar} values for the soil profile were generally within {dollar}pm 3spcirc{dollar}C of the measured values. Results also showed good agreement between modelled and measured net radiation flux densities. In addition, the modified model predicted surface layer (0-0.03 m) moisture content remained wet enough for seed germination, i.e. {dollar}thetasb{lcub}rm v{rcub}>0.09{dollar} m{dollar}sp3{dollar} m{dollar}sp{lcub}-3{rcub}{dollar}, about 24 to 36 hours longer than indicated by measured (resistance block) {dollar}thetasb{lcub}rm v{rcub}{dollar} values.