Estimating evapotranspiration with ASTER thermal infrared imagery

The recent availability of multi-band thermal infrared imagery from the ASTER sensor has made feasible spatial estimation of evapotranspiration at 90 meter resolution. One critical variable in evapotranspiration models is surface temperature. With ASTER the temperature can be reliably determined over a wide range of vegetative conditions. The requirements for accurate temperature measurement include minimization of atmospheric effects, correction for surface emissivity variations and sufficient resolution for the type of vegetative cover. When ASTER imagery are combined with meteorological observations, these requirements are usually met and result in surface temperatures accurate within 1-2/spl deg/ C. Other sensors, such as Landsat TM and AVHRR, do not meet these requirements because they either lack the resolution of ASTER, or because their broadband radiometers cannot correct for surface emissivity. ASTER-based evapotranspiration estimates for a few days during 2000-2001 were made over the sub-humid grazing and wheat lands in central Oklahoma. Daily evapotranspiration was estimated by applying instantaneous ASTER surface temperatures, as well as ASTER-based vegetation indices from visible-near infrared bands, to a two-source energy flux model and combining the result with separately acquired hourly solar radiation data. The estimates of surface fluxes show reasonable agreement (50-100 W/m/sup 2/) with ground-based eddy-correlation measurements and illustrate how ASTER measurements can be applied to heterogeneous terrain. There are some discrepancies, however, and these may in part he due to difficulty quantifying fractional cover of senescent vegetation.