1218 
INLAND WATER 
A physical model that relates spectral reflectance to water characteristics was derived. The specific optical 
properties of several water materials were determined in the laboratory Algorithms to retrieve chlorophyll, 
cyanophycocyanin, seston dry weight, vertical attenuation of irradiance and Secchi depth transparency were 
developed, using only five 10-15nm width bands in the red domain. These algorithms were validated using 
airborne imaging spectrometers well calibrated and for which atmospheric corrections were applied. This very 
good work hi ghli ghts the potentials of the use of models to retrieve targets biophysical characteristics. Other 
studies were also presented with original approaches using chlorophyll fluorescence in the red domain to 
estimate chlorophyll concentration. The potentials of the detection of chlorophyll fluorescence via passive high 
spectral resolution remote sensing are certainly new approaches to be investigated more deeply. 
VEGETATION 
Empirical approaches were developed to relate leaf area index, crop color, or the efficiency with which canopies 
intercept photosynthetically active radiation, to high spectral resolution data, with very often comparison with 
broad band indices. In many cases, improvements were noticed when using high spectral resolution 
information, in particular for soil background effects correction. Similar empirical studies were devoted to 
relate high spectral information to leaf biochemistry. It appears that lignin and nitrogen could be directly 
estimated by multiple regression analysis. The concentrations in lignin and nitrogen were very dependent upon 
the specie composition of the canopy. These remote sensing estimates of lignin were assimilated into an 
ecosystem model to map the carbon balance. 
Models were also used for interpretation of high spectral resolution data. A model inversion 
study showed that leaf biochemical composition (chlorophyll, water) was attainable through model inversion 
with a reasonable accuracy. However, difficulties were encountered when retrieving canopy structural 
parameters without any a priori knowledge of the structure. Inversion performed when using only a selection of 
broad bands (the 6 TM bands) shows very similar results to what is obtained with the full spectral information. 
In all the cases, the spectra reconstructed with the retrieved values of canopy biophysical parameters were very 
close to the measured ones. These findings about the spectral information content are developed through a 
statistical approach performed on a wide collection of spectra. It shows that a limited number of bands, around 
25, are enough to represent most of the spectral information. The difficulty to get very accurate estimation of 
canopy structural variables from the spectral variation was stressed out. One way to encompass this limitation is 
to get bidirectional measurements that could complement the radiometric information. That was proposed in a 
paper, where high spectral resolution indices are used to retrieve leaf chlorophyll, and bidirectional broad band 
measurements to characterize both the leaf area index and the leaf inclination angle. 
CONCLUDING COMMENTS. 
This session showed a high degree of diversity in the topics and the approaches used. Significant improvements 
were noticed about the performances of the instruments, the radiometric and spectral calibration procedures as 
well as the geometric corrections applied. It was also highlighted that high spectral resolution data could 
provide by itself atmospherically corrected ground level calibrated reflectance values. These algorithms, based 
on the use of radiative transfer models should be tested in many experiments and then set up as standard 
routines. Very important advances were shown also in the use of radiative transfer models applied either on 
water bodies or vegetation. For water, the algorithms developed demonstrated the capacity to retrieve with great 
accuracy most of the water characteristics with few narrow bands in the red domain. For vegetation, results are 
not so clear, due to the complexity of canopies. However, model inversion demonstrated some potentials for the 
retrieval of leaf or canopy biochemical composition (at least chlorophyll and water). However, it seemed that 
canopy structural parameters could not be estimated without ambiguity unless using other sources of 
information and presumably bidirectional data. Several studies highlighted the high degree of redundancy 
contained in high spectral resolution data. However, further studies are required, with a particular attention to 
the soil background effects and demonstrate from the first principles the potentials of high spectral resolution to 
infer directly canopy biochemical composition such as lignin, cellulose, nitrogen ... These studies will be 
critical with the coming of the new generation of satellite borne imaging spectrometers such as MERIS, 
MODIS or PRISM.