Combining ALOS/PALSAR derived vegetation structure and inundation patterns to characterize major vegetation types in the Mamiraua Sustainable Development Reserve, Central Amazon floodplain, Brazil

摘要: Remote sensing studies of vegetation cover and hydrologic dynamics in Amazonian wetlands have been mostly limited temporally or spatially, the distribution spatial configuration varzea habitats remains poorly known. This study uses multitemporal PALSAR L-band radar imagery combined with object-based image analysis, data mining techniques field to derive structure inundation patterns characterize major types forests Mamiraua Sustainable Development Reserve. Our results show that combination extent information can be a good indicator complex gradient along floodplain. The intersection between flood duration classes showed wider range combinations than suggested from based studies. Chavascal areas—chacaracterized as dense species-poor shrub/tree community developing old depressions, abandoned channels, shallow lakes—had shorter periods usually recognized hydroperiod 180–240 days flooding, while low varzea—a diverse fewest smallest species, highest individual density tolerate 120–180 flooding every year—was distributed ranges were higher reported by literature. Forest communities growing at sites never mapped flooded could indicate areas only during extreme hydrological events, for short time. emphasize potential contribution SAR remote monitoring management wetland environments, providing not accurate on landscape distribution, but also important insights ecohydrological processes ultimately determine floodplain habitat mosaics.