Trial biomass map production in Riau Province, Indonesia using L-band SAR data

Biomass maps were produced from PALSAR data using two methods. Method 1 used land-cover classification results derived from PALSAR multi-temporal mosaic images. The average biomass value for a class was regarded as a representative biomass value for the class, and these were assigned to each respective class. Method 2 used the correlation between radar backscattering (gamma(0)(HV)) and forest aboveground biomass (AG-biomass). The RMSEs of the biomass estimation for method 2 were between 59.1 and 79.4, depending on the LiDAR-biomass model. These values are smaller than 99.1, which is the estimate from a simple average (method 1). This indicates that the variety of biomass within a natural forest class is correctly evaluated with method 2 by using the gamma(0)(HV)-(AG-biomass) relation. The gamma(0)(HV)-(AG-biomass) relation is useful for accurately estimating biomass, especially for a region in which biomass is less than 100 t/ha.