The vegetation phenology detection in Amazon tropical evergreen forests using SPOT-VEGETATION 11-y time series

In tropical regions, the seasonal phenology and the interannual variability of carbon fluxes remain poorly understood, and its representation in global vegetation models highly simplified. However, previous field studies have explored the temporal dynamics of Amazonian vegetation and have shown unexpected and significant seasonal pattern in this ecosystem. Moreover, as a major component of the global terrestrial carbon cycle, the phenological behaviour of this tropical rainforest can significantly influence global dynamics of carbon fluxes and climate. In this context, it is crucial to detect the vegetation phenology in this region. However, field studies are rare and provide local information. By contrast, satellite data from medium resolution sensor offer the advantage of spatial and temporal resolution well adapted to phenological studies. Here, we explore the seasonal characterization of leaf phenology in the Amazon basin through vegetation indices, using a long time series of SPOT-VEGETATION data (2000-2010) at a spatial resolution of 1 km. The analysis is performed locally at a fluxtower site (Santarém) and at the basin scale. Temporal profiles are preliminary analyzed concerning potential artifacts on the cycle observed such as aerosols contamination or BRDF effects. The results indicate that the EVI is better suited than the NDVI to follow the vegetation dynamics in this region. An increase of this index is clearly observed during the dry season, suggesting a higher photosynthetic capacity as it matches higher gross primary production values measured from the fluxtower. Comparing with climate conditions, it is suggested that trees tend to produce new leaves in the dry season to optimize access to light and maximize the carbon uptake.