The Mean Compositing method of temporal synthesis consists of
• a quality control of data (elimination of clouds, haze and erroneous values);
• calculating the average of valid data for each wavelength band. This method, which has the advantage of taking into account all available and useful data during a given period, was extended to the global level. The quality control of the pixel values was based on a sample of 110,000 reference pixels, permitting the identification of the appropriate thresholds and a statistical approach based on a discriminant analysis. The change detection methods developed by the project are based on the detection of temporal variations in the signal behaviour. The originality of the approach lies in the analysis of the time signal at all wavelength bands and not only the NDVI vegetation index that uses red and near infrared only. This rigorous approach is made possible thanks to the spatial and temporal coherence of the data produced by the Mean Compositing synthesis. The richer and more detailed multispectral analysis makes it possible to refine the quality of the change detection.

The research activities led to the production of several algorithms ready to be used by remote sensing specialists or by automatic environmental monitoring systems, in real time and on a global scale. A processing chain to produce composites based on the Mean Compositing method was developed. This highly flexible and rapid chain can be adapted by users in line with the applications and regional conditions of the observed environment. Scientific cooperation with international institutions led to the development of a number of tools including some for early warning and others for ecological monitoring: