485 
still very useful tools in the field of spaceborne earth-observations, when the very large sensor footprint 
provides data related to a complex mixture of numerous land-cover categories (McFarland et ai, 1990; 
Choudhury and Wang, 1990). When the footprint can be decomposed in homogenous areas, functional 
modeling of the remote sensing observations can be used. 
To reduce the number of input parameters of forward models dedicated to inversion, several 
approaches can be used. Sensitivity studies can be carried out for a variety of vegetation canopies, to 
weight the role of the different variables on the measured signal. Then the variables that have small or 
negligible effect on the measurements can be roughly estimated prior to the inversion procedure. More 
attention should also be paid to a priori information (statistical behavior, functional dependence of the 
parameters to be retrieved) that strongly increase the efficiency of the inversion process. Synthetic 
input parameters (above-mentioned parameters: T, to, b, (l z , l p ), etc.), that condense diffuse 
information, can also be used to reduce the number of input parameters. Prior to the inversion process, 
the synthetic parameters are calibrated and mapped (the spatial resolution is the sensor footprint for 
space applications) using ground measurements or they are retrieved simulteanously to the geophysical 
parameters of interest. For instance, instead of using a very accurate description of the whole 
vegetation canopy in terms of size, shape and distribution of orientation of leaves, branches or trunks 
(as done in the discrete approach), the correlation lengths are a very interesting tool to model the 
scattering properties of vegetation canopies. Based on similar concepts, the vegetation scattering phase 
function could also be constructed from a limited number of more representative physical parameters. 
As an illustration of this point, at the present time, most of retrieval studies based on actual passive 
microwave measurements use parametric approaches (Paloscia and Pampaloni, 1992; Calvet et ai, 
1993; Wigneron et al, 1993a; Van de Griend and Owe, 1993). 
Complex microwave models for homogeneous or discontinuous vegetation covers (Karam et ai, 1992; 
McDonald and Ulaby, 1993; Wang et ai, 1993) can be very useful to create models dedicated to the 
inverse problem. They are powerful tools to simulate sets of microwave signatures for a variety of 
vegetation canopies. These sets can be used to perform relevant sensitivity analysis or to train 
parametric approaches. Such composite approaches relying on both discrete and parametric modeling 
have been recently developed to perform retrievals both in the active and passive domains. 
REFERENCES 
Biard F. (1994), ‘Intérêt des mesures radiométriques en hyperfréquences passives sur couvert agricole 
de blé - Application à l’inversion des paramètres de surface à 1.4 et 10.65GHz’, Mémoire d’ingénieur 
ESITPA, promotion 1989, 76p. 
Brunfeldt D. R. and F. T. Ulaby (1984), ‘Measured microwave emission and scattering in vegetation 
canopies’, IEEE Trans. Geosci. Remote Sensing 22:520-524. 
Calvet, J.-C., J.-P. Wigneron, E. Mougin, Y. H. Kerr and J. L. Brito, ‘Plant water content and 
temperature of the Amazon forest from satellite microwave radiometry’, accepted for publication in 
IEEE Trans. Geosc. Remote Sens. 1993. 
Choudhury B. J. (1989), ‘Monitoring global land surface using Nimbus-7 37GHz data - Theory and 
examples’, Int. J. Remote Sens. 10(10): 1579-1605. 
Choudhury B. J. and J. R. Wang (1990), ‘Simulated and observed 37GHz emission over Africa’, Int. J. 
Remote Sens. 11:1837-1868.