Using MODIS to assess seasonal flooding in an Amazonian white sand ecosystem
J.
Marion
Adeney, Duke University, marion.adeney@duke.edu
(Presenting)
Distinct white sand ecosystems occur in patches throughout the Amazon and likely contribute to regional scale diversity patterns. Characterized by savanna-like vegetation and extreme conditions of nutrient scarcity, seasonal flooding and fire, they host a unique suite of species, many of them endemic to these ecosystems. The largest white sand area, in the northern Brazilian Amazon, drains its distinctive black water into the Rio Negro basin. It is characterized by extreme seasonal flooding, which likely drives the vegetation patterns. Little research and no remote sensing analyses have been done in this area. I used Landsat and MODIS imagery to characterize these ecosystems over space and time, addressing their unique biodiversity, disturbance regimes and conservation status. Here, I show the use of MODIS imagery to assess seasonal flooding events. I used a decision tree analysis framework to classify images and detect seasonal change. This framework allows for incorporation of diverse data, including vegetation and water indices, tree cover, rainfall, and spectral characteristics. Model results were then validated using Landsat imagery and ground truth data. These analyses contribute to our understanding of how white sand ecosystems function, at a time when they remain mostly intact but may be threatened by future human activities and climate change.
NASA Carbon Cycle & Ecosystems Active Awards Represented by this Poster: