We present an operational two-source (soil+vegetation) model for evaluating the surface energy balance given measurements of the time rate of change in radiometric surface temperature (T-RAD) during the morning hours. This model consists of a two-source surface component describing the relation between T-RAD and sensible heat flux, coupled with a time-integrated component connecting surface sensible heating with planetary boundary layer development. By tying together the time-dependent behavior of surface temperature and the temperature in the boundary layer with the flux of sensible heat from the surface to the atmosphere, the need for ancillary measurements of near-surface air temperature is eliminated. This is a significant benefit when T-RAD is acquired remotely. Air temperature can be strongly coupled to local biophysical surface conditions and, if the surface air and brightness temperature measurements used by a model are not collocated, energy flux estimates can be significantly corrupted. Furthermore, because this model uses only temporal changes in radiometric temperatures rather than absolute temperatures, time-independent biases in T-RAD, resulting from atmospheric effects or other sources, do not affect the estimated fluxes; only the time-varying component of corrections need be computed. The algorithm also decomposes the surface radiometric temperature into its soil and vegetation contributions; thus the angular dependence of T-RAD can be predicted from an observation of T-RAD at a single view angle. This capability is critical to an accurate interpretation of off-nadir measurements from polar orbiting and geosynchronous satellites. The performance of this model has been evaluated in comparison with data collected during two large-scale field experiments: the first International Satellite Land Surface Climatology Project field experiment, conducted in and around the Konza Prairie in Kansas, and the Monsoon '90 experiment, conducted in the semiarid rangelands of the Walnut Gulch Watershed in southern Arizona. Both comparisons yielded uncertainties comparable to those achieved by models that do require air temperature as an input and to measurement errors typical of standard micrometeorological methods for flux estimation. A strategy for applying the two-source time-integrated model on a regional or continental scale is briefly outlined. (C) Elsevier Science Inc., 1997.
