**Title of Presentation:** An Analysis of the Uncertainties in Radiative Transfer Models Used in Remote Sensed Data Product Generation.

**Primary (Corresponding) Author:** Robin D. Morris

**Organization of Primary Author:** Research Institute for Advanced Computer Science

**Co-Authors:** A. Kottas, R. Furfaro, M. Taddy, B. Ganapol

**Abstract**: The process of predicting a satellite observation of a vegetated region (e.g. a MODIS scene) involves running a Radiative Transfer Model (RTM). The RTM takes as input various biospheric and illumination parameters and computes the upwelling radiation at the top of the canopy that is ultimately observed by the satellite mounted sensor. The question we address is the following: which of the inputs to the RTM has the greatest impact on the computed observation?

We use the Leaf Canopy Model (LCM) RTM as a surrogate for the RTM used as the basis of the MODIS production algorithm. The LCM was designed to study the feasibility of observing leaf chemistry remotely. It takes as input leaf chemistry variables (chlorophyll, water, lignin, cellulose) and canopy structural parameters (leaf area index, leaf angle distribution, soil reflectance, sun angle). The influence of each input variable, or small subsets of the inputs, is captured through the determination of the ``main effects'' and ``sensitivity indices''. Computing these quantities numerically requires extensive runs of the RTM, which is computationally expensive. Using a Gaussian Process approximation to the LCM RTM we compute efficiently the main effects and sensitivity indices, and determine those input variables that are vital for accurate prediction.