This paper describes a remote sensing methodology for assessing the changes in water composition and volume of Amazon floodplain aquatic system in response to the annual flood pulse. The spatiotemporal dynamics of water composition was assessed through an integrated analyze of in situ (limnological and high resolution water reflectance spectra) and orbital data (multispectral Landsat/TM images). The spatiotemporal patterns of different water types were identified applying spatial analysis (ordinary kriging) to the limnological data. The effect of changes in water composition, along the hydrological cycle, on the water spectral response was assessed using spectral angle mapper and derivative analysis algorithms. Image segmentation and unsupervised classification of Landsat/TM images were applied to map and quantify the water types which were also characterized by the limnological parameters measured at four water stages. Bathymetric data, obtained from a high-resolution bottom topography survey was used to study the dynamic of flooded area and stored water volume in the floodplain. Regression models to predict flooded area and water volume from water level were constructed. As a result, a detailed conceptual model of the water circulation in the Curuai floodplain was proposed.
Science Theme: ND (Nutrient Dynamics)
Session: 2A: Hydrological and Meteorological Processes