Super typhoon Haiyan makes landfall
Super typhoon Haiyan
Subscribe here to receive special images

VEGETATION - 2000

Lake Maggiore - Italy, 3-6 April 2000


NEW ALGORITHMIC CONCEPT FOR ATMOSPHERIC AND DIRECTIONAL CORRECTION OF THE SURFACE REFLECTANCES

P. MAISONGRANDE
B. DUCHEMIN, C. DUBEGNY,.G. DEDIEU , M. LEROY.
CESBIO (Centre d’Etudes Spatiales de la Biosphère) - CNES (Centre National D’Etudes Spatiales) 18, Avenue E. Belin,
31401 TOULOUSE, cedex 4
tel : 33 5 61 55 85 16 ---- fax : 33 5 61 55 85 00

Paper on Atmospheric correction (pdf file, 509 k)

Paper on Bidirectionnal Effects (pdf file, 1.58 Mo)

In the frame of the ‘Project for Improvement and Continuity of the VEGETATION Mission’, our goal was twofold. We have investigated new algorithmic schemes in order to improve the evaluation of Aerosol Optical Thickness (AOT to be used in the atmospheric correction) and for introduction of directional models in compositing methods. These new approaches are planned to replace respectively AOT climatology and the Maximum Value Composite (MVC) technique that are used in the production line of VEGETATION data set at Centre de Traitement des Images Vegetation.

The current algorithm already accounts for the SMAC method, which corrects the top of atmosphere reflectances for absorption by gases and scattering by molecules and aerosols. While ozone and water vapor are documented by remote sensing and meteorological models, AOT still requires real time estimations.

Taking advantage of the VEGETATION spectral range (Blue[0.415-0.455mm], Red[0.580-0.680mm], NIR[0.730-0.840mm] and SWIR[1.520-1.660mm]), we present a new approach for AOT retrieval. This procedure is based on the use of a time and target dependent ratio between blue and SWIR reflectances. After presenting the algorithm, we compare the newly processed reflectances time profiles and images with results of the current method.

Data provided by wide field of view optical sensors also present a strong dependency on the three dimensional geometry (source-target-sensor). Despite substantial reduction of these effects through calculation of Vegetation Indices (VIs), residual directional impacts still remain on reflectances time series. In the 10 days MVC products, the association of different orbital tracks on a same image creates patchwork artifacts while the time behavior of reflectances and VIs show erratic changes. This noise-like fluctuation is important since the currently applied MVC tends to select large angle configurations, where directional and atmospheric effects are pronounced.

Kernels driven semi-empirical Bidirectionnal Reflectance Distribution Functions (BRDF) offer a good efficiency/complexity compromise to quantify and model surface reflectance anisotropy. Taking such a model for granted, our purpose becomes its operational use in the real-time processing of global data set at full spatial resolution. The first originality of the method we developed, lies in the data selection scheme applied during the fit of the BRDF. Furthermore, we combined the smoothing effect of the average with noise reduction due to the normalization of the data to a standard viewing geometry. After presentation of the algorithmic outlines, new results will be compared with those obtained with the MVC technique.