2. The service (2013)

Dataset and pre-processing:

Fig. 1: input datasets. RGB and Infra-Red composite color images in June and July 2005.

Pre-processing:

  • Precise geo-referencing by matching to Landsat imagery
  • Atmospheric correction

Method:

Fig. 2: Root Mean Square Deviation image (or sum of the quadratic difference bands between bands). The changes are given by the highest values in RMSD.

The changes due to the deforestation must be distinguished from the one due to other origins, such as:

  • clouds, snow, haze, fog...,
  • cloud or orography shadows,
  • land use/cover changes
  • changes of types of cultures

... in order to obtain a map where red pixels depict the area where a doubt has been raised on the ground exploitation, i.e. where forest might have been destroyed to be replaced by crops, as depicted in the following picture:

Fig. 3: Remaining suspect areas after image processing: These selected areas are worse sending a ground station mission.

Status of the development: the service is going to be tested by Primer users by the end of the first trimester 2012.

 

Notes: The precision of the ortho-rectification is 90 m, which corresponds approximately to one third of the spatial resolution of MERIS => this is the most part of the 4 hours of time required by CATENA to compute the result.