Peat Swamp Forest Change Analysis During two Recent El-Nino Events in Sabah, Malaysia: Principal component analysis of multitemporal Normalized Difference Water Index from multisensor data

摘要

El-Nino associated fire is an important driving force to the recent disturbances on the tropical peat swamp forest. Klias Peninsular of Sabah contains largest peat swamp forest in the northern end of Borneo Island. The forest was severely damaged by fires occurred during the 1998 and 2003 El-Nino events. We emphasized the multitemporal disturbance of the peat swamp forest using multisensor data. Pre and postfire images were acquired for monitoring the disturbance during the 1998 (Landsat5-TM and Landsat7-ETM+) and 2003 El-Nino (Landsat7-ETM+ and SPOT4-HRVIR) events. The Normalized Difference Water Index (NDWI) was computed from the radiometrically corrected multisensor data. Applying principal component analysis (PCA) on the multitemporal NDWI allows the separation of unchanged and changed land covers into different principal components (PCs). The PC1 contained largely the unchanged peat swamp forests, mangrove and other land covers. The PC3 and PC2 allowed the identification of peat swamp forest disturbed during the 1998 and 2003 El-Nino events, respectively. The resulting PCs were analyzed with the unsupervised fuzzy c-mean clustering. The multitemporal change pattern of NDWI of each vegetation change class was used for extracting the disturbance classes. The disturbed areas detected were compared to that of the burned areas detected using a Normalized Burn Ratio (NBR) differencing with post-processing operations.