Abstract ID: 524

Mapping Varzea Ecosystems with Combined Radar and Optical Datasets: A Framework for Sampling and Modeling Aquatic Biodiversity

Freshwater ecosystems of the Amazon basin, and their associated wetlands, support one of Earth’s great biodiversity resources. However, current knowledge is severely lacking for nearly all taxonomic groups regarding the number of species, the spatial patterns of their distributions, and the life history information needed to formulate sound management plans to ensure their preservation. High degrees of spatial and temporal variability create particular challenges for surveying populations of aquatic and wetland species in the Amazon. Published studies of biodiversity for Amazonian wetlands often have analyzed species distributions in relation to habitat typologies that are not well defined and that may vary considerably between regions, e.g. using terms such as chavascal or high varzea. Fine-scale floodplain units based on vegetation structure, inundation periodicity, and water chemistry are key biodiversity surrogates for Amazonian freshwater systems, providing a framework for understanding the spatial patterns present in existing datasets, evaluating gaps in current inventories, designing future sampling strategies, modeling species distributions based on incomplete data, and carrying out systematic conservation planning. Systematically acquired SAR datasets from Japan’s ALOS satellite, combined with optical datasets, now make it possible to map floodplain habitats at high resolution both spatially and temporally. Here we present initial results of habitat mapping of the Mamiraua and Piagacu-Purus Sustainable Development Reserves on the Amazon varzea in Brazil. The object-based classification algorithm employs three modes of radar data from the ALOS PALSAR: Fine Beam Dual-pol (FBD; HH/HV polarization, 6.25 m pixels), Fine Beam Single-pol (FBS; HH polarization, 12.5 m pixels), and ScanSAR (HH, 100 m pixels). For both sites, 2-3 dates of fine-beam data and 8-10 dates of ScanSAR data acquired during 2007-2008 allowed characterization of vegetation structure and flooding state over a complete seasonal flood cycle. Historic Landsat TM imagery was used to map suspended sediment levels in open water of lakes and channels. Map accuracy is being assessed during field and aerial campaigns during October-November 2008. The habitat maps are being used by researchers for applications including studies of manatee, caiman, and pirarucu habitat, varzea forest biodiversity, and varzea forest structure and biomass.

Session:  Biodiversity - Modeling biodiversity, present and future.

Presentation Type:  Poster