A 10-days compositing method accounting for bidirectional effects

Data provided by large field of view optical sensors present a strong dependency on the source-target-sensor geometry. Despite substantial reduction of these bidirectional effects through the calculation of vegetation indices (VIs), residual impacts still remain on 10-days products derived from the maximum value compositing (MVC). The association of different orbital tracks on a same image creates patchwork artifacts while the time behavior of reflectances and VIs exhibits erratic changes. The reduction of these noise-like fluctuations can be done by retrieving the bidirectional reflectance distribution function (BRDF). To fit a BRDF model requires the collecting of a data set, commonly achieved using a sliding time window of constant length from daily acquisitions. Once the BRDF retrieved, the data can be composed, by selecting a distinctive subset of data, correcting it from directional effects and choosing a way to combine it. The objective of this study is to adapt such a process, referred hereinafter as 'directional compositing', to the VEGETATION (SPOT satellite) system characteristics and production line requirements.