Estimating spectral albedo and nadir reflectance through inversion of simple BRDF models with AVHRR/MODIS-like data

Abstract

In recent years, many computationally efficient bidirectional reflectance models have been developed to account for angular effects in land remote sensing data, particularly those from the NOAA advanced very high resolution radiometer (AVHRR), polarization and directionality of the Earth's reflectances (POLDER), and the planned EOS moderate-resolution imaging spectrometer (MODIS) and multi-angle imaging spectroradiometer (MISR) sensors. In this study, we assessed the relative ability of 10 such models to predict commonly used remote sensing products (nadir reflectance and albedo). Specifically, we inverted each model with ground-based data from the portable apparatus for rapid acquisition of bidirectional observations of the land and atmosphere (PARABOLA) arranged in subsets representative of satellite sampling geometries. We used data from nine land cover types, ranging from soil to grassland (First International Satellite Land Surface Climatology Project (ISLSCP) Field Experiment (FIFE)) to forest (Boreal Ecosystem-Atmosphere Study (BOREAS)). Retrieved parameters were used in forward model runs to estimate nadir reflectance and spectral albedo over a wide range of solar angles. We rank the models by the accuracy of the estimated products and find results to be strongly dependent on the view azimuth angle range of the inversion data, and less dependent on the spectral band and land cover type. Overall, the nonlinear model of Rahman et al. [993] and the linear kernel-driven RossThickLiSparse model [Wanner et al., 1995] were most accurate. The latter was at least 25 times faster to invert than the former. Interestingly, we found these two models were not able to match the various bidirectional reflectance distribution function (BRDF) shapes as well as other models, suggesting their superior performance lies in their ability to be more reliably inverted with sparse data sets. These results should be useful to those interested in the computationally fast normalization of bidirectional reflectance data and the estimation of radiation flux parameters (albedo, absorbed radiation) over diverse land covers.