Prince, S. D., S.J. Goetz, R.O. Dubayah, K.P. Czajkowski, M Thawley, 1998,
Surface temperature (Ts), air temperature (Ta), atmospheric precipitable water (U), and vapor pressure deficit (D) have been derived from algorithms that use satellite observations of visible, near-infrared and thermal infrared radiation in order to obtain higher spatial resolution data than is possible from meteorological station observations. The inferences of these four variables are compared with field instrument observations for sites with contrasted climates and vegetation including boreal forest (BOREAS), tropical bush savanna (HAPEX-Sahel), tall grass prairie (FIFE) and a region with mountain, prairie and intensive agriculture (Red and Arkansas river catchments). The results showed that Ts could be retrieved with RMS errors of 3.5oC for a range of 48oC; Ta with 3.9oC over a range of 36oC; U with 0.6 cm over a range of 3.6 cm; and D with 10.9 mb over a range of 58 mb. Although the results had low absolute accuracies the field data themselves are not without error - not least because, although the inferences were for a >1 km2 area made instantaneously, they were compared with point field values generally not measured at exactly the same times in the day. Maps of retrieved variables had good relative accuracy and possibly better absolute accuracy than the comparisons with point measurements suggest.