Potential of high-resolution ALOS-PALSAR mosaic texture for aboveground forest carbon tracking in tropical region

Estimating accurate aboveground forest carbon stocks (AFCS) is always challenging in tropical regions due to the complex mosaic of forest structure and species diversity. This study evaluates the potential of high-resolution ALOS/PALSAR mosaics data in the tropical forests of central Sumatra to improve AFCS estimates. The study region has an average AFCS (47% of the aboveground biomass) of 66 Mg C ha(-1) with a range of 1 to 334 Mg C ha-1 and consists of natural forests including peat swamp, dry moist, regrowth, and mangrove, and plantation forests including rubber, acacia, oil palm, and coconut. Field measurements of AFCS were carried out in 87 (ha(-1)) plots, where half of them were from plantation forests. Various possibilities including direct gamma naught back-scatters and their ratios and various types of textures of dual polarized mosaics from the years 2009 and 2010 were examined applying regression modeling in a five step framework. R-2, variable inflation factor (VIF), p-value, and root mean square errors (RMSE) were the major indicators considered for selection of best model in the calibration process. The potential models selected were cross validated by the leave-one-out (LOO) method where R-2, RMSE, mean deviation (MD), and Nash-Sutcliffe Efficiency (NSE, model performance indicator) were examined. The results indicate that a simple combination of backscatters and their ratios provides an AFCS estimate with a RMSE of 45 Mg C ha(-1), more efficient than the average of field measured AFCS (NSE of 0.54), and R-2 of 0.63. Inclusion of appropriate texture parameters derived from the high-resolution PALSAR mosaics further increases the potential for AFCS estimation by increasing R-2 and model performance (NSE) to 0.84 and 0.83, respectively and decreasing the uncertainty to 28 Mg C ha(-1). This SAR based method offers the low cost wall-to-wall forest carbon mapping with a high level of accuracy in the dense tropical forest regions of Southeast Asia where other methods are still rare. (C) 2015 Elsevier Inc All rights reserved.