MC-FUME: A new method for compositing individual reflective channels

MC-FUME stands for median composite of fuzzy multispectral estimate. It is the name of a newly developed method for compositing individual reflective channels of the VEGETATION sensor onboard the SPOT-4 platform. MC-FUME is a two-step compositing methodology that uses combined angular and atmospheric corrections of reflectance measurements. The first step is an approximate BRDF correction. Considering the atmospheric influence to be stochastic, the top-of-canopy (TOC) reflectance at a reference geometry is estimated for each pixel by means of an extensive database of model-simulated top-of-atmosphere (TOA) reflectance values. The second step is compositing over a time period. This is done by taking the median of the estimated TOC reflectance values. The method is tested on simulated time series at different latitudes as well as on a time series of NOAA-AVHRR images. Tests performed on the simulated data set prove the ability of the MC-FUME algorithm to correctly reproduce TOC_nadir values. Moreover, it outperforms classic compositing strategies such as the maximum-value composite of the NDVI (MVC-NDVI) in this respect. Tests performed on AVHRR images show that the angular dependence of the MC-FUME algorithm is strongly reduced with respect to the classic MVC-NDVI method, as is the presence of speckle. This is especially remarkable for the individual reflective channels (RED and NIR). Thus, for individual reflective channels, MC-FUME produces speckle-free composites with reflectance values that are corrected for atmospheric and angular effects, and which therefore are independent of the observation/illumination geometry at the time of measurement