Canopy Reflectance Model Benchmarking: RAMI and the ROMC

Summary

Canopy radiative transfer (RT) models simulate radiative transfer processes in the solar domain at or near the Earth’s terrestrial surface, i.e., within plant canopies and over bare soil surfaces. Such models are capable of simulating the transmitted, reflected and absorbed radiation fluxes, as well as, the angular distribution of the reflected light above vegetated surfaces in the optical domain of the solar spectrum. This directional aspect is important since the brightness with which a given canopy target appears in Earth Observation data (pixel) depends not only on the structural and spectral properties of the target area but also on the viewing and illumination geometry at the time of measurement. The quantitative interpretation of remote sensing data thus hinges more and more on the exploitation of RT models: In operational contexts they are used to pre-compute Look-Up-Tables of quantities required by the inversion algorithms, whereas, in off-line applications they may be directly inverted against the measurements to yield sets of state variables that describe the properties of the observed target. In both cases, the accuracy and reliability of the solutions to the inverse problems are determined by the performance of the RT model as well as the remote sensing instrument. Reducing the uncertainty of RT models will thus increase the quality of the information derived from remote sensing data (which is of interest to scientists and policy makers), and augment the reliability of RT model studies in forward mode (which is of interest to space agencies when testing new sensor concepts or mission strategies). This understanding has led to a series of model intercomparison projects (MIP) aiming either to document the spread of currently available simulation models, or, else to assess and benchmark the quality of their simulation results, e.g. (Bull. Am. Meteorol. Soc. 73:1962–1970, 1998; J. Geophys. Res. 106:11937–11956, 2001; J. Geophys. Res. 109:D06210, 2004; J. Geophys. Res. 110, 2005; Bull. Am. Meteorol. Soc. 86:1275–1293, 2005; J. Geophys. Res., 2007). Among these MIPs the RAdiation transfer Model Intercomparison (RAMI) activity focuses on the proper representation of the radiative processes occurring in vegetated environments in the optical domain of the solar spectrum.

Launched for the first time in 1999 the triennial RAMI community exercise encourages the systematic evaluation of canopy reflectance models on a voluntary basis. The first phase of RAMI focused on documenting the spread among radiative transfer simulations over a small set of primarily 1-D canopies (J Geophys Res 106:11937-11956, 2001). The positive response of the various RAMI-1 participants and the subsequent improvements made to a series of RT models promoted the launching of the second phase of RAMI (RAMI-2) in 2002. Here the number of test cases was expanded to focus further on the performance of RT models dealing with structurally complex 3-D plant environments. The main outcomes of RAMI-2 included (1) an increase in the number of participating models, (2) a better agreement between the model simulations in the case of the structurally simple scenes inherited from RAMI-1, and (3) the need to reduce the sometimes substantial differences between some of the 3-D RT models over complex heterogeneous scenes (J. Geophys. Res. 109:D06210, 2004). The latter issue was noted as one of the challenges that future intercomparison activities would have to face if some sort of reliable “surrogate truth” was to be derived for other RT models to be compared against. The third phase of RAMI (RAMI-3) was launched in 2005 and investigated the self-consistency, the relative and — to a limited extend — also the absolute performance of canopy RT models. RAMI-3 showed significant progress in the mutual agreement between RT models when compared to RAMI-2. In particular for 3-D Monte Carlo (MC) models the dispersion between simulated bidirectional reflectance factor (BRF) quantities was less than 1%, which supported the usage of these models in the generation of a “surrogate truth” data set covering all of the RAMI test cases (J. Geophys. Res., 2007). The availability of such a reference data set, in turn, lead to the development of the RAMI on-line model checker (ROMC), an open-access web-based interface allowing model developers and users to evaluate canopy RT models independently via the internet.