Two source model of surface fluxes for millet fields in Niger

Fluxes from land surfaces with partial vegetative cover are being estimated with two source (soil and leaves) models. A network of aerodynamic resistances linking the two sources allows interaction of these fluxes after incident solar energy is partitioned between them using leaf area index. Many of these interactive two source models use a simple soil resistance approach to estimating fluxes from the soil which is rarely validated by field measurement. In this study a detailed model of soil water and heat fluxes is combined with a two source surface flux model to form the SWEAT model, which avoids use of the soil resistance approach. Output from the SWEAT model is compared with measurements of evaporation from soil, transpiration, soil temperature and soil water content made within a sparse millet field in Niger during the rainy season of 1993. Simulated and measured values of daily evaporation from soil are not significantly different. Simulated maximum soil temperatures at 0.1 m underestimate measurements by approximately 2.5 K. With a few exceptions, simulated soil water content is within one standard deviation of the mean measured value. The other field data provide further confirmation of the accuracy of simulation using SWEAT. Using the SWEAT model it was estimated that the greatest increase of transpiration (MJ) above the incident net radiation partitioned to the canopy, would occur at a leaf area index between 1.0 and 1.5. This increase in transpiration can be attributed to sensible heat fluxes from the soil underlying the canopy being absorbed by the canopy and dissipated as latent heat. To test the soil resistance approach, soil resistance was defined as a linear function of time since rain. It was shown that this linear resistance model for estimating fluxes from the soil may be adequate in a two source model when simulating fluxes from a land surface with a leaf area index of 2.0 or higher, but it is unlikely to be adequate for lower leaf area indices.