Processing satellite images
Eyes watching from Space
The Guiana Space Centre sends rockets into Space in order to place satellites in orbit. Some provide telecommunications services, while others explore the cosmos or observe our planet, sending back optical and radar satellite images. The SEAS (*) technology platform (named for the French acronym for satellite monitoring of the Amazon environment) offers a whole range of satellite image applications and processing techniques.
Images from Spot 2, 4, 5 and Envisat satellites arrive in real time on the terminals following the SEAS project in Montabo. In response to the various incoming requests, SEAS supplies only the most suitable raw images - sometimes there is an awful lot of cloud cover in
Users must either have the technology needed for processing the images themselves, or call upon specialised companies. The IRD delivers raw images in a very rudimentary state, which have only undergone the most basic calibration pre-processing needed to make them useable. There are then an infinite number of real processing solutions, which are intimately linked with a project's theme. And regardless of what the satellite images will ultimately be used for, remote sensing must be used alongside information from the field. Image processing often involves emphasising certain natural or man-made phenomena in the region under study; the set of images has to be correlated with knowledge or indicators on the ground that confirm or modify the interpretation and therefore the processing method. This field knowledge is typically in the hands of the end users.
The SEAS project, which aims to further research, innovation and sustainable development in French Guiana, supplies satellite images to various public bodies free of charge after reviewing their proposals. CNES, which funded 75% of the platform (in cooperation with the French Guiana Regional Council and Prefecture, with ERDF assistance, and the IRD), and which - together with ESA - is also providing the telemetry received by the station, has already been able to benefit from these images. The range of applications is huge and can be associated with outside data; detecting such things as mobile phones at sea, illegal construction (by overlaying images with land registry plans), alluvial gold mining, specific ecosystems, areas of deforestation, etc. For instance, by combining botanical records with a field study of a landscape and analysis of satellite images, certain characteristics of a site's biodiversity can be determined. When applied to images predating the botanical study, this method can lead to relatively reliable hypotheses about changes to the site's biodiversity, thus supporting the longer-term monitoring. We are currently monitoring changes to the Amazonian coastal system, the results of which are to be given to local coastal management projects.
Although optical satellite images have much in common with what the eye actually sees, interpreting a radar image, which is the synthesis of a signal's echo, is much more difficult. The intensity of the return radar signal depends on the ruggedness of the land surfaces (their asperity) and their water content. Unlike the Spot satellites, radar is not hindered by clouds or rain, hence the advantage of coupling images of both types. By doing this, the IRD improved mapping of urban land in
(*)SEAS in 2006
Since SEAS was launched in 2005, 35 image requests corresponding to a similar number of local studies and projects have been dealt with. Received at a rate of 250 to 550 images a day from Spot satellites alone (each covering an area of 3600 square kilometres), the images acquired in real time are automatically archived in a database. In 2006, the three Spot satellites, which are permanently active, sent back some 150,000 images, including 6,000 of French Guiana, whereas Envisat, which 'only takes what it is told to take', transmitted 400 radar images.
For a full description of the SEAS project, see Latitude 5 No.68 (only available in French)
| Previous: Computer graphics |
Next: File photos |









