Mapping deforestation in the Congo Basin forest using multi-temporal SPOT-VGT imagery from 2000 to 2004|
of SPOT-VEGETATION S10 imagery for the African continent for the period January
2000 to December 2004 have been analysed. A region of interest has been
extracted using the CROP-VGT software package for a Central African region
(15°S - 15°N; 5°E - 35°E). Clouds and cloud shadows were masked using
information from the associated Status Maps distributed with the S10 images. Hence,
a set of cloud free annual composites was produced using the ENVI band math
function, extracting the maximal Normalised Difference Vegetation Index (NDVI)
value for 36 decadal syntheses for
each year from 2000 to 2004. Subsequently, an unsupervised ISODATA
classification with 11 classes was performed using annual NDVI composites.
Finally, new vegetation maps for Central Africa were produced.
An RGB-NDVI change detection strategy to
detect and quantify major decreases or increases in green biomass - associated
with forest harvest or regeneration - was developed as well and applied for the
years 2000 till 2004. The NDVI analysis
will - in the near future - be applied to study reflectance behaviour in single
channels to obtain a better accuracy for our classification results. The use of
the Red (B2) and Near-Infrared (B3) channels, for example, provides complementary
information to make a distinction between bright soil with high reflectances in
both the B3 and B2 bands and vegetation,
where reflectance is low in the B2 band.