Study and modelling of temperature spatial variability by NOAA-AVHRR thermal imagery

Abstract

Several land management activities require the spatially continuous estimation of environmental parameters which are usually measured locally (temperature, rainfall, radiation, etc.). Numerous interpolation methods have been developed for this aim, among which the kriging approach is one of the most widely known and utilized. The approach is based on the experimental definition of spatial semi-variograms, which is often a difficult operation due to the insufficient coverage of the measurement points in the study areas. A typical example of this situation is the interpolation of maximum daily temperature in a topographically complex Italian region (Tuscany). Since NOAA-AVHRR measures surface temperature daily with a relatively high spatial resolution (1.1. km), it is here proposed that these data can be used for a better definition of the semi-variograms of maximum temperature. Following this approach, satellite thermal estimates from six scenes were first compared to ground measurements of maximum daily temperature. Experimental semi-variograms were then derived from both data sets as well as from the whole thermal infrared images. The results show that the estimated model parameters from the various methods are quite different and the fitting of the experimental points to the theoretical model is much better when using continuous satellite data. This finding has interesting implications for the more general problem of defining standard optimum interpolation methods.