THE ASSESSMENT OF REGIONAL CROP WATER CONDITIONS FROM METEOROLOGICAL SATELLITE THERMAL INFRARED DATA

Meteorological satellites (mainly NOAA AVHRR) have been extensively used these last years to monitor vegetation and crop conditions on a regional scale, using vegetation index NDVI data. Recent work with Nimbus-7 passive microwave measurements has also shown the complementary potential of that spectral domain. On the other hand, the thermal infrared channels, in spite of their well-known ability to detect water stress (as established by ground studies at a local scale), are less studied for the same purpose of long-term monitoring. This paper intends to demonstrate their capabilities in assessing crop water conditions on a regional scale and estimating the actual evapotranspiration (ET) to be used in agrometerological models. A brief analysis of past studies justifies the use of the cumulative SIGMA-(T(s) - T(a)) (difference between the midafternoon surface temperature by satellite and the maximum air temperature obtained from the meterological ground network), named stress-degree-day by Jackson et al. (1977) which may be related to ET by a simplified linear relationship. This criterion, already tested in Sahelian regions (Seguin et al., 1989), is applied to France on a large scale, corresponding to the entire country, using 5-day syntheses from Meteosat, calibrated by NOAA AVHRR on selected dates, for 3 years (1985-1987). Values of SIGMA-(T(s) - T(a)) may be considered as climatological data. They reveal both spatial differences in regional climates and the main features of each year. The use of the linear relationship, derived from ET values computed from a coupled energy budget-water balance model, allows one to estimate and map regional evaporation on a monthly to 6-month time basis. The variations obtained along a latitudinal transect display the relations between ET and potential evapotranspiration (PET), also leading to an indirect estimation of PET from remote sensing data. SIGMA-(T(s) - T(a)) / R(n) is proposed as an index of regional water stress, which may be derived from satellite data and appears as complementary to the intergrated NDVI, with the advantage of a more physically established relationship with ET.