Basic HTML Version
3-73
Acknowledgements
We would like to thank the UN-REDD Programme for supporting these analyses.
Annex I – Generation of the preliminary biomass carbon map for the DRC
Several data sources were brought together to generate a preliminary biomass carbon map for
the DRC, comprising carbon stored in above- and below-ground live biomass. The above-ground
biomass was derived from a dataset for tropical Africa, based on remotely-sensed MODIS NBAR
data from 2000-2003 at a resolution of 1 km (Baccini
et al.
, 2008). Each pixel in this map contains
a value for biomass density in tonnes per hectare (t/ha). Ecosystem-specific root-to-shoot ratios
(FAO, 2006) were applied to these values to add below-ground biomass, using FAO ecological
zones to distinguish between ecosystems (FAO, 2001). The carbon mass of the resulting total
was estimated as half the biomass (Gibbs and Brown, 2007). The dataset provided by Baccini
et
al.
(2008) did not cover areas with less than 9 tonnes of above-ground biomass per hectare. A
National Land Cover map (Vancutsem
et al.
, 2009) was overlaid with these areas and the
following categories were assigned a value of 4 tonnes of carbon per ha, based on the values
from a global biomass carbon map (Ruesch and Gibbs, 2008):
•
Agriculture : Permanently cropped area with rainfed broadleaved tree crops (plantations)
or rainfed herbaceous crops or bare soils
•
Broadleaved deciduous woodland: Savanna woodland
•
Broadleaved deciduous woodland: Tree savanna
•
Broadleaved deciduous woodland: Woodland (Miombo)
A few remaining scattered pixels were assigned the value of their nearest neighbour while
ensuring that areas of water remained 0. This resulted in the preliminary map of biomass carbon
(Figure 15).
The shading of the map was produced using the GIS “quantile classification” method. This
method allocates the same number of pixels to each class. The variable deciding the class breaks
was carbon density. Depending on the number of classes to be generated, in this case seven, the
method allocates the pixels with the lowest carbon density to the lowest class until one seventh of
all pixels are in that class. The next pixel with a slightly higher carbon density than the previous is
then allocated to the next class, and so on. Collaboration partners assessed different ways to
classify and shade the map, for example, using a different number of classes, and agreed to the
shading that is shown in the report.