Estimation of air temperature from remotely sensed surface observations

Air temperature is an important descriptor of terrestrial environmental conditions across the earth. Standard meteorological observations generally provide reasonable descriptions of temporal variations in air temperature for the site sampled but may not describe the spatial heterogeneity typically encountered in this variable over larger land areas. If a reasonable estimate of spatial patterns of air temperature can be derived from satellite remote sensing, this pattern, in combination with the temporal precision of ground measurements, should significantly improve our knowledge of terrestrial environmental conditions. In this study, we explore a methodology for estimating air temperature directly from remotely sensed observations using the (observed) correlation between a spectral vegetation index and surface temperature (temperature-vegetation index). Inference of air temperature is based on the hypothesis that the bulk temperature of an infinitely thick vegetation canopy is close to ambient air temperature. Advanced very high resolution radiometer observations for five sites in northeastern Kansas were used to estimate air temperatures on 31 days during the 1987 growing season. These air temperature estimates were compared with coincident ground-measured air temperatures recorded at standard meteorological stations. A strong correlation (r=0.93) was found between the satellite estimates and measured air temperatures with a mean error of 2.92 C degrees. However, there was a consistent positive bias in the satellite estimates. It is not clear at this time whether the bias is due to an actual difference between air temperature and the temperature of an infinitely thick canopy or whether it is an artifact of the measurements themselves. Within the errors of the methods used, estimation of standard meteorological shelter height air temperatures recorded at the time of satellite overpass appears possible. Further refinements of the remote sensing methods used here are possible and can be expected in the era of the National Aeronautics and Space Administration's Earth Observing System. (C) Elsevier Science Inc., 1997.