Interpretation of variations in MODIS-measured greenness levels of Amazon forests during 2000 to 2009

Monitoring the Amazonian forests with satellite data is a topical research priority given their key roles in the global carbon and hydrological cycles. This is a challenging task as demonstrated by opposing reports of forest response to the once-in-a-century 2005 Amazon drought and divergent interpretations of observed seasonal variations in forest greenness. The objective of this research is to tackle these challenges using theory and satellite-based vegetation greenness measurements from the Moderate Resolution Imaging Spectroradiometer (MODIS) spanning a decade (2000-2009).;The previously reported patterns of large-scale greening of Amazon forests during the 2005 drought, from an earlier version of MODIS greenness data---Collection 4 (C4) Enhanced Vegetation Index (EVI), are found to be irreproducible, with both this earlier version and the improved, current version (C5), owing to inclusion of atmosphere-corrupted data in those results. There is no evidence of large-scale greening of the drought-stricken forests, which display greenness levels similar to non-drought years. Changes in surface solar irradiance are also contrary to earlier speculations on drought-induced greening.;Atmosphere-corruption of greenness data is pervasive in the Amazon---even in the dry season 60-66% of EVI data are corrupted with cloud and aerosol contamination. These influences create artifacts of enhanced greening from biomass-burning aerosols, and their removal biases estimates of greenness change. Small random patches of anomalous greening and browning appear in all ten years, irrespective of contemporaneous variations in precipitation, but with no persistence over time. These changes are insignificantly small---over 90% of EVI anomalies are within the error envelope of EVI, which warrants cautious interpretation.;Large seasonal variations in near-infrared reflectance of Amazon forests, observed in satellite measurements, could be due to either similar, but small, changes in leaf optical properties alone, or changes in both leaf area and leaf optical properties. The latter explanation is consistent with known phenological behavior of tropical forests, ground-based reports of seasonal changes in leaf area, litterfall, leaf optical properties and fluxes of evapotranspiration, and thus, reconciles various seemingly divergent interpretations.;The results presented in this dissertation provide a better understanding of seasonal and interannual variations of satellite-sensed greenness of Amazon forests.