Very high resolution derived biomass estimates in the Corridor Ankeniheny-Zahamena (CAZ), Madagascar

The products represent biomass estimates for four areas of interest in the Corridor Ankeniheny Zahamena (CAZ), Madagascar, generated using very high resolution imagery and based on field-collected plot information. The study extent aims to capture the drivers of deforestation in the Corridor Ankeniheny-Zahamena (CAZ). Multiple variables were incorporated into the sampling design, including elevation, slope, bioclimate zone, length of dry season, soil type, deforestation history (by epoch and elevation), and access. Based on these criteria, four Zones of Interest (ZOIs) were identified and modified in an iterative approach involving preliminary reconnaissance work, information on access provided by partners, and initial analyses of spatial data layers. The products were generated as part of a project, 'Can Paying 4 Global Ecosystem Services reduce poverty?' (P4GES), funded by the Ecosystem Services for Poverty Alleviation (ESPA) programme Random forest regression models were used to estimate biomass values for each of the ZOIs. Training data were a number of different biomass plots distributed roughly equal over the four ZOIs, so that most of the variation within the image was captured. These training data are then used to calculate the relationships between segment features and biomass in a random forest regression model. That model is then used to predict biomass estimates for each segment. The output is limited to a single predicted biomass image since per-segment prediction confidence metrics are not available.