Incorporation of remotely sensed land surface temperature information into hydrological assimilation schemes can significantly improve modeling accuracy by constraining estimates of the surface energy balance. However, obtaining accurate surface temperatures at regional to global scales is difficult because of inadequate resolution, uncertain radiometric calibration, atmospheric variability and uncertain surface emissivities. To have practical value for assimilation models, instantaneous surface temperature estimates also need diurnal representation. A combination of spatial and temporal surface temperature estimates can be achieved by integrating remote sensing observations, which we demonstrate with three satellite sensors: MODIS, GOES and ASTER. MODIS provides frequent, global, 1 km temperature estimates at mid-morning and mid-afternoon times. GOES ½ hourly, 4 km observations allow diurnal extrapolation for clear sky regions. Using high resolution (90 m) multiband ASTER thermal observations, accuracy of surface temperature and emissivity estimates can be checked. An example of this three-way approach will be shown using remote sensing observations collected over two North American study sites--- Jornada, New Mexico and central Oklahoma--- during 2001.