A new inversion scheme for the RPV model

Abstract

Parametric bi-directional reflectance factor (BRF) models are often inverted against remote sensing data to describe the amplitude and shape of the anisotropy of geophysical media. For a given geophysical situation, this approach hinges on the availability of (1) a suitable inversion procedure delivering the most probable values of the coefficients entering these models, and (2) reliable and accurate observational data acquired under suitable geometries of illumination and observation. The Rahman, Pinty Verstraete (RPV) model one of the parametric BRF models, idealizes the BRF values through a simple non-linear equation requiring the values of three parameters to be estimated from the remote sensing data. One parameter describes the amplitude of the BRF and the other two jointly represent the angular shape of the anisotropy. The computational load and efficiency required to perform the inversion of the RPV model are critical for operational application. Most, if not all, of the suggested inversion procedures apply to linear versions of parametric models, including a linearized version of the RPV model. Although this approach enables fast inversion methods to be implemented, the linearization prevents the model from representing adequately the wide diversity of observed BRF fields. The present paper describes a computationally efficient inversion method that can be used with the original, i.e., non-linear, version of the RPV model. This method can be applied to retrieve the three model parameters of the original RPV model from an analysis of actual BRF data, while offering the opportunity of specifying the desired accuracy. The procedure delivers ranges of parameter values that depend on the user requirements and capitalizes on the performance of this non-linear model. Two applications are made with laboratory measurements and BRF data sets measured using the multi-angle imaging spectroradiometer (MISR) instrument of the Jet Propulsion Laboratory (JPL) on board the NASA EOS Terra platform.