Retrieval of biophysical parameters at global scale by inversion of a radiative transfer model with neural network

The retrieval of biophysical parameters through the fitting of a simple radiative transfer model against spectral and directional spaceborne POLDER optical data is feasible for the biosphere characterization. The LAI maps have shown the expected spatial variations for various types of vegetation as well as expected temporal evolution. The description of a new algorithm using a neural network is then given. The next step of this work is to apply it on POLDER 1 data.