1217 
HIGH SPECTRAL RESOLUTION IN THE SOLAR SPECTRUM. 
F. Baret (1), R.O. Green (2) 
(1) ENRA Bioclimatologie, BP 91, 84 143 Montfavet cedex, (France) 
(2) NASA-JPL, 4800 Oak grave Drive, Pasadena, California 91 109 (USA) 
During this session dedicated to high spectral resolution in the solar spectrum, 23 papers were presented that 
correspond to some average in between the last two colloquiums (Courchevel: 30, Aussois: 20). The distribution 
between the several topics was approximately the same, with a majority dealing with vegetation: calibration of 
sensors (4); Atmospheric corrections (1), Geology and pedology’ (3), Inland water 3, and vegetation (12). 
Significant progresses were noted, with a development of the use of models or data bases. These will be 
highlighted for each topic. In the conclusion, we will also attempt to put in evidence the lacks and issues to 
address in priority. 
CALIBRATION 
The use of high spectral resolution data allows to monitor and compare in real time the spectral characteristics 
of various sensors. Examples were given for the evaluation of possible spectral shifts of the SPOT satellite 
sensors and the comparison between the NDVI computed from several satellites (SPOT, TM, NOAA AVHRR, 
ATSR-2). Progresses were also significant in the calibration procedures and the performances of the Airborne 
Visible Infrared Imaging Spectrometer (AVIRIS). It appears that the 10% absolute radiometric performances 
described during the last colloquium in Courchevelle was improved to about 5% now, with potentials for more 
improvements in the very near future. The geometric performances of such airborne sensors were also studied 
in detail, takin g into account all the plane possible movements and the local topography These significant 
progresses were necessary to achieve the transition towards a more quantitative use of this type of information. 
ATMOSPHERIC CORRECTION. 
A set of algorithms dedicated to correct from the atmospheric effects using only the spectral information 
gathered by imaging spectrophotometers was presented. They mostly use MODTRAN2a radiative transfer 
model that is inverted on certain portions on the spectrum to retrieve by non linear optimization techniques the 
aerosols characteristics (400-700nm), molecular and well mixed gases (with the oxygen absorption band at 
760nm) and water vapor (940nm). No a priori information is necessary for the soil background reflectance. 
These algorithms should be more widely used and tested on many sites to make them standard procedures 
applied during the many airborne experiments conducted these last years. They could also help designing the 
new generation of high spectral resolution sensors that aimed to deliver final products in ground level 
calibrated and atmospherically corrected reflectance values. A statistical procedure was also developed to 
retrieve the atmospheric effects from the pixel to pixel variation in the spectral response. 
GEOLOGY, PEDOLOGY. 
In geology/mineralogy, no very new approaches were presented but some applications making intensive use of 
laboratory spectral libraries of minerals and multiple spectral features mapping algorithms. This allowed to 
identify at least the abundance of 15 minerals. Special attention was paid to the alunite solid solution 
composition that allowed the mapping of hydrothermal alterations and the understanding of the geological 
processes. The comparison between laboratory and airborne data was possible because of the improved 
characteristics of the sensors and the algorithms used to calibrate and correct from the atmospheric effects. In 
pedology, high spectral resolution at ground level allowed to develop spectral indices to be used with broad 
band sensors to map the land surface degradation.