ABSTRACT
Soon after the United Nations Framework Convention on Climate Change (UNFCCC) Kyoto protocol (2005) came into force, the Japanese Aerospace Exploration Agency (JAXA), in accord with the radar remote sensing community, responded to the observational needs raised by climate change studies by launching the Kyoto and Carbon (K&C) project in 2007. This initiative was the generator for a new effort in compiling systematic – in time and space – remote sensing observations, which could best support terrestrial carbon science and related international treaties.
Within the K&C project, the European Commission Joint Research Centre (JRC) was tasked with the development of a high-resolution wide-area radar mosaic over the whole African continent using data form the JAXA Phased Array L-band Synthetic Aperture Radar (ALOS PALSAR). The mosaic development, code-named KC Africa, is the subject of this chapter.
The KC Africa project marks another step forward in the evolution of the radar mosaics technique. It provided a snapshot of a whole continent (previously, only ecosystems were mapped), and the shortest time compatible with the sensor’s orbital cycle (two months in the case of Africa).
It afforded polarization diversity, thus enabling the better decoupling of volume and surface scattering. Indeed, data holding the cross-polarized backscattering coefficient (HV), on top of the co-polarized one (HH), were used.
The availability of the high-quality digital elevation model SRTM with comparable spatial resolution over the whole mosaic’s extent allowed the performance of precision range-Doppler geocoding and terrain correction, thus achieving better radiometric corrections for effects induced by topography, and terrain morphology measures as additional training sets in a classification procedure.
A bespoke processing engine was developed for the compilation of the K&C mosaic. The key functionalities are:
Calibration revision of the slant range strip-data for range-dependent trend and anomalies.
Geocoding (range-Doppler equations solution) into a latitude–longitude reference system, with provision of per-pixel local incidence, and further backscatter corrections for slope effects.
Assemblage of the geo-coded strips within a geographic bounding box. The module used an inter-strips amplitude-blending algorithm to avoid edge effects.
Radiometric revision for correcting seasonality effects and residual calibration errors.
Each processing step is discussed in detail in this chapter.An overview of the mosaic thematic information content is given, with a set of cases including a comparison with the GRFM dataset in connection with delineating the interface between forest and savanna; the capability of distinguishing between grass and woody savanna; mapping of swamp forests with qualification of soil condition (dry, flooded); mapping of plantations and secondary forests.
