Multiangular Contributions for Discriminate Seasonal Structural Changes in the Amazon Rainforest Using MODIS MAIAC Data

The sensitivity of the Amazon rainforests to climate change has received great attention by the scientific community due to the important role that this vegetation plays in the global carbon, water and energy cycle. The spatial and temporal variability of tropical forests across Amazonia, and their phenological, ecological and edaphic cycles are still poorly understood. The objective of this work was to infer seasonal and spatial variability of forest structure in the Brazilian Amazon based on anisotropy of multi-angle satellite observations. We used observations from the Moderate Resolution Imaging Spectroradiometer (MODIS/Terra and Aqua) processed by a new Multi-Angle Implementation Atmospheric Correction Algorithm (MAIAC) to investigate how multi-angular spectral response from satellite imagery can be used to analyze structural variability of Amazon rainforests. We calculated differences acquired from forward and backscatter reflectance by modeling the bi-directional reflectance distribution function to infer seasonal and spatial changes in vegetation structure. Changes in anisotropy were larger during the dry season than during the wet season, suggesting intra-annual changes in vegetation structure and density. However, there were marked differences in timing and amplitude depending on forest type. For instance differences between reflectance hotspot and darkspot showed more anisotropy in the open Ombrophilous forest than in the dense Ombrophilous forest. Our results show that multi-angle data can be useful for analyzing structural differences in various forest types and for discriminating different seasonal effects within the Amazon basin. Also, multi-angle data could help solve uncertainties about sensitivity of different tropical forest types to light versus rainfall. In conclusion, multi-angular information, as expressed by the anisotropy of spectral reflectance, may complement conventional studies and provide significant improvements over approaches that are based on vegetation indices alone.