A sensitivity criterion for BRDF model inversion analysis

The inversion of physical models in remote sensing is difficult due to its ill-posed essence. Though scientists have been realizing that the inversion result is much concerned with sensitivities of parameters, how to define the sensitivity of a parameter in inversion is still under discussion. In this paper, an "S Index" is proposed to derive S Ratio, a ratio of one input parameter\´s S Index to the sum of all other S Indices, as a useful sensitivity criterion. Moreover, we analyzed S index and S Ratio based on the information transfer theory. It is shown that S Ratio is related with information distribution ratio in inversion. The value of S Ratio may vary with different ground covers, soil types, moisture, geometries and bands. We took SAIL model as an example to illustrate the use of S Ratio under several typical scenes. Multi-angular datasets were generated for these scenes and further been used to retrieve 7 parameters of the model. The results suggest that the inversion accuracy is strongly correlated to S Ratio. Another two sensitivity indices are also demonstrated as a comparison. As a result, we could use it to estimate the sensitivity of parameters in a certain inversion step and which type of datasets is better for inversion under various cases. Such a priori information could be important before data selection