ABSTRACT
This chapter documents how the combination of the CAMP and GRFM radar mosaics gave support to the generation of thematic products covering different application areas, with underlying unique and novel characteristics:
High spatial resolution (100 m) combined with coverage of several million square kilometers, permitting the mapping of ecosystem-wide processes and, at the same time, local features.
Classification of forest types (e.g. lowland, swamp forest) by detecting structural properties through scattering mechanisms enabled by wavelength diversity, and by exploitation of spatial statistics (textures).
Mapping of flood conditions and time through the dual acquisitions.
Geolocation accuracy, allowing for the accurate mapping of features such as logging patterns and river courses.
In particular, the chapter focuses on a specific thematic product, a vegetation map of Central Congo compiled using the CAMP and GRFM radar mosaics (C-band and L-band), and data from the imaging spectrometer VEGETATION onboard SPOT 4. This last instrument was specifically designed for global-scale vegetation monitoring and acquires images in four spectral bands (VisBlue, VisRed, Near Infra-Red and Short-Wave Infra-Red) at 1 km spatial resolution. Thematic classes comprise: lowland rain forest, secondary forest, swamp forest, flooded swamp forest, savanna, water, urban and plantations.The map compilation methods are outlined, emphasizing the important complementarity of the thematic class visibility afforded by the observations in different parts of the electromagnetic spectrum.
Information fusion is achieved at the level of the classification maps derived independently from the microwave and the optical instruments. Each derived classification product is represented by a data structure composed of elementary variable size cells holding class labels. In the final product, each class will inherit the original spatial resolution.
Validation of regional-scale vegetation maps poses challenging problems, because the resolution and the extent of such maps preclude any classical ground data collection at an acceptable cost. The strategy followed for the GRFM–CAMP products, therefore, entailed a comparison with a compilation of national vegetation maps including the same thematic classes.
The chapter concludes by illustrating a number of thematically relevant cases that are provided by the classification map.
