Single- and dual-source modeling of surface energy fluxes with radiometric surface temperature

Abstract

Single- and dual-source models of the surface energy transfer across the soil-vegetation-atmosphere interface were used in conjunction with remotely sensed surface temperature for computing the surface energy balance over heterogeneous surfaces. Both models are relatively simple so that only a few parameters are specified, making them potentially useful for computing surface fluxes with operational satellite observations. The models were tested with datasets collected from a semiarid rangeland environment with canopy cover generally less than 50% and a subhumid tallgrass prairie environment having canopy cover typically greater than 50%. For the semiarid site, differences between the single-source and dual-source model estimates of the sensible heat flux (H) and the observations averaged about 25%. For the tallgrass prairie, the disagreement between observations and single-source model estimates of H was significantly larger, averaging nearly 55%. The average difference between observations and the dual-source model predictions for the tallgrass prairie site increased slightly from the semiarid site to 30%. The latent heat flux (LE) was determined by residual from measurements of net radiation and model estimates of the soil heat flux. For the semiarid site, the single-source model estimates of LE differed on average with the observations by about 15%, whereas the LE values computed by the dual-source model differed by about 20%. For the tallgrass prairie site, the LE values from the single-source model differed from the observations by almost 35%, on average, whereas the dual-source model estimates produced an average difference of about 20%. Given the fact that energy flux observations by various techniques have been found to differ by at least 20%, the single-source model performed satisfactorily for the semiarid site but had difficulty reproducing the fluxes at the tallgrass prairie site. The dual-source model, however, performed reasonably well at both sites. To obtain results comparable to the dual-source model for the tallgrass prairie site, the single-source model required significant modifications to a parameter used in estimating the roughness length for heat.