Cost-effectiveness of Vegetation Biophysical Parameters Retrieval from Remote Sensing Data

In the context of vegetation studies Earth Observation (E.O.) data have been extensively used to retrieve biophysical parameters of land surface. In some cases, thanks to the availability of near-real-time data, tools and applications have been developed and implemented in the fields of precision agriculture, water resources monitoring and management. So far, empirical approaches based on vegetation indices (Vis) have been successfully applied. They may provide a satisfactory level of accuracy in the estimation of important vegetation biophysical parameters (e.g. LAI, fractional ground cover, biomass, etc). Such methods, however, require a reliable reference data-set to calibrate empirical formulas on different vegetation types; furthermore, they are generally based on a few spectral bands, with a consistent under-exploitation of the full spectral range available in new generation sensors. Alternative approaches based on inversion of radiative transfer models of vegetation represent a challenging opportunity for the estimation of vegetation parameters from data with high dimensionality. This work evaluates the effectiveness, in terms of accuracy and computational complexity, for retrieving the Leaf Area Index, on one hand, by means of empirical relationships, such as the simple CLAIR model proposed by Clevers (1989) and based on the Weighted Differences Vegetation Index (WDVI), and, on the other hand, by means of mathematical inversion of the combined radiative transfer model PROSPECT and SAILH (PSH). Both approaches, i.e. empirical relationship LAI (WDVI) and radiative transfer model inversion, have been tested by using super- spectral and multi-angular data in the solar domain from the Compact High Resolution Imaging Spectrometer on the PROBA experimental satellite.