It also appeared that to each site and each date corresponds a different atmospheric correction 
equation (Figure 3), depending on atmospheric conditions. It is evident that the wide difference between these 
equations (up to 25°C for DN = 190) does not allow using a single equation for all conditions, even on the same 
site. The presented method thus appears of great interest for operational uses of Landsat TM thermal IR data. 
Figure 3 : Intercomparison of the different atmospheric correction equations derived for each study site 
where correction equation could be derived and validated using both warm and cold targets.. 
4 - CONCLUSION 
This study aiming at designing a new atmospheric correction method of Landsat TM thermal 
infrared data has shown that: 
• surface temperatures of warm (bare soil, rangeland) and cold (irrigated fields, wet forest) sites 
can be successfully derived from surface energy balance equation on various study sites, using 
usual meteorological measurements; 
• this surface temperature estimation can be improved by at least one ground-measurement of 
the warm target surface temperature ; 
• when applied to a simple equation of radiative transfer (Wukelic et al., 1989), these warm and 
cold targets temperatures can be used to derive a local atmospheric correction equation for the 
day of Landsat TM overpass ; ground measurements of mid-temperature targets have shown 
the validity of such a derived equation. 
This method can be applied easily to a Landsat TM thermal IR image, even if it was acquired 
in the past, under the condition that meteorological data are available and sites where LE=LEp and LE=0 can 
be identified. It can therefore considerably increase the interest of using Landsat TM thermal IR data for local 
mapping of surface energy balance of various land surfaces, as it is only based on easily available 
meteorological and TM data. 5 
5 - ACKNOWLEDGEMENTS 
The authors would like to acknowledge the many people who were involved in the ground 
measurement acquisitions on the various sites : people of ORMVAG and IAV Hassan II, Morocco, for the 
Gharb measurement campaign of 1987, of USDA-ARS Water Conservation Lab., Phoenix Az., for the MAC 
campaigns of 1985-1986, and of USDA-ARS Tombstone Laboratory and Univ. of Arizona, Dept, of Hydrology 
and Water Resources and Dept of Soil and Water Science, for the Walnut Gulch '92 campaign. This study was 
conducted in the framework of the NASA EOS program (NASA Ref. Num. NAG-W2425, NSF grant BSC- 
8920851), of the CEMAGREF Natural Hazards program, and of the EOSAT Grant Program (EOS AT Grant