6533b861fe1ef96bd12c42d5

RESEARCH PRODUCT

A Global Sensitivity Analysis Toolbox to Quantify Drivers of Vegetation Radiative Transfer Models

Juan Pablo RiveraJochem Verrelst

subject

010504 meteorology & atmospheric sciences0211 other engineering and technologiesProcess (computing)Sobol sequence02 engineering and technology01 natural sciencesToolboxOperator (computer programming)GeographyRadiative transferRadianceRange (statistics)Sensitivity (control systems)021101 geological & geomatics engineering0105 earth and related environmental sciencesRemote sensing

description

Abstract Global sensitivity analysis (GSA) enables to gain insight into the functioning of radiative transfer models (RTMs) by identifying the driving input variables of RTM spectral outputs such as reflectance, fluorescence, or radiance. This contribution introduces automated radiative transfer models operator's (ARTMO’s) new GSA toolbox. With the GSA toolbox the majority of ARTMO’s available RTMs can be decomposed into its driving variables. For a selected RTM output, a GSA identifies the most influential and noninfluential input variables according to Sobol' first-order and total-order indices. The toolbox can process RTM spectral outputs for any kind of optical sensor setting within the spectral range of 400–2400 nm. Multiple model outputs can be automatically analyzed within the same execution, which is of interest for more advanced, multioutput RTMs such as the soil–vegetation–atmosphere–transfer model SCOPE. To illustrate its functioning, total-order sensitivity indices results were calculated for reflectance and transmittance outputs of the leaf RTM PROSPECT-4/5, directional and hemispherical reflectance outputs of the canopy RTM PROSAIL, and fluorescence and photosynthesis outputs of SCOPE. Wavelength-dependent key driving and noninfluential input variables were quantified. In practice, the toolbox can be beneficial for the broader remote sensing community to gain insight into vegetation–light interactions and RTM input–output functioning.

https://doi.org/10.1016/b978-0-12-803011-0.00016-1