|
Retrieval of SIF and related vegetation biophysical variables from the top-of-atmosphere (TOA) radiance data ensures the consistency of physical meaning between different parameters and avoids the uncertainties and errors caused by atmospheric correction propagate to surface reflectance. For complex nonlinear models, global sensitivity analysis is the first step for retrieving biophysical parameters, which can quantitatively analyze the sensitive and non-sensitive input parameters for a model output, and gains insights into retrievable variables. In order to simulate the TOA radiance data, based on the four-stream radiative transfer theory, the coupling between the atmosphere and the non-Lambertian surface is described through a combination of radiative transfer models those were used to represent soil, vegetation and the atmosphere. A modified version of the Sobol`s method was used to analysis. The results show that:(i) From 650 nm to 850 nm, leaf area index(LAI), leaf inclination distribution function (LIDFa), leaf chlorophyll content(Cab) and leaf dry matter content(Cdm) were the most sensitive parameters affecting the output variance of the model among 12 input parameters. On the other hand, senescent material (Cs), leaf structure parameter (N), fluorescence quantum efficiency (fqe) and soil moisture percentage (SMp) have weak influences on the output radiance. (ii)When the atmospheric visibility is high (i.e.50km), the sensitivities of TOA radiance sensitivities are similar to that of TOC reflectance for the most sensitive parameters. (iii)The viewing and illumination geometry has a significant influence on the NIR wavelengths.
|
