Statistical framework of the inverse problem in the retrieval of vegetation parameters

The accuracy of the vegetation properties retrieved from radiometric data d epends mainly on the information about the target parameters contained with in each measurement. We postulate that a measurement provides the more info rmation the more it contributes to decrease the uncertainty in the retrieve d parameters. The mathematical definition of the information contents based on the statistical estimation theory, reveals clearly the role played by b oth the sensitivities of the measurements with respect to the desired param eters and the correlation between these sensitivities. In this paper, we discuss especially the application of this formal framewo rk to the retrieval of the vegetation properties, and investigate some fund amental problems such as the limitation of the accuracy of the target param eters in their retrieval. We also show how this limitation depends on the b iome properties and on the geometrical configuration of the illumination-ob servation. Furthermore, it should be emphasised that the present formalism provides a theoretical framework to treat different aspects of the retrieval of medium properties. In particular, it enables to define how to organise sampling o f most informative measurements.