The density, canopy cover and biomass of woody plants regulate fundamental ecosystem properties, including the provision of forage and habitat critical for livestock and wildlife, biogeochemical cycles, fluxes of water, energy and carbon, and regulation of runoff and erosion. Humans also depend on woody plants for timber, fuel, fruits and medicinal products. The importance of accurate and spatially resolved information on woody vegetation structure has increased in recent decades in relation to climate change and international agreements mandating assessment of carbon sequestration in forests and woodlands. We will present improved assessments of woody vegetation canopy cover and above ground biomass across Sub-Saharan Africa with an emphasis on drier, more open and low-stature savannas where earlier analyses underestimate woody cover. We used woody cover data from over 1000 field sites to fit generalized linear mixed effects models with different combinations of phenological metrics from optical data (MODIS) and ku-band microwave data (Quick-Scatter) as predictors. The final, optimally predictive model was chosen based on out-of-sample predictive scores. A novel approach was used to estimate biomass by utilizing alternative allometric relationships.
Results/Conclusions
The woody cover and biomass maps exhibit the expected continental-scale spatial trends in mean and variability of tree cover across mean annual rainfall gradients. The dynamic range in tree cover estimates for the arid and mesic savannas is improved using the new product relative to earlier products. Across the savanna and forest regions of Sub-Saharan Africa, within-calibration-set root mean square errors (RMSE) of 14.0% and out-of-calibration-set RMSE of 14.9% (evaluated using fully independent field data) suggest reduced uncertainty relative to earlier products. Allometric relationships between canopy cover and tree height provide an indirect approach to above-ground biomass estimation that are less subject to saturation in densely populated forests, a weakness of optical and radar data when calibrated directly on field measured biomass. The new products provide improved woody cover and biomass estimates in a range of savanna and dry deciduous woodlands. The synergistic approach using radar and optical observations as correlates of woody cover variability, and allometric relationships to estimate biomass, provides an alternative framework that may overcome the limitations associated with earlier methods for canopy cover and woody biomass mapping.