Recovery of surface environmental variables over large areas using AVHRR observations

The NOAA-AVHRR satellite record has provided a new view of terrestrial vegetation processes but has been exploited relatively little for the recovery of physical environment variables such as surface wetness, air temperature, or near-surface humidity. Recent advances in techniques to recover these variables over large areas has permitted the development of improved terrestrial primary production models. We report on the development of techniques to recover physical environment variables from 8-year time series of AVHRR observations. The recovery algorithms incorporate visible, near-infrared and thermal infrared radiation measurements in a contextual array. The accuracy of the recovered variables, when compared to surface meteorological stations over a broad range of environments, was shown to be within 95% confidence limits of +/-6.8 degreesC for a 36 degreesC air temperature range; +/-1.28 cm for a 3.6 cm atmospheric water vapor range; and +/- 11.2 mb for a 58 mb vapor pressure deficit range. Recovered values of surface soil moisture explained 77% of the observed variability at a temperate grassland site. Maps of retrieved variables for several study areas had good relative accuracy when compared to spatially interpolated surface observations. Multi-temporal global maps of these variables are presented and spatially analyzed relative to other information sources.