Understanding phenology of diverse tropical vegetation using high spatio-temporal resolution remote sensing

Abstract:

Vegetation phenology is an integrated and sensitive indicator of ecosystem function that responds to disturbance, seasonality, variability and extremes in weather and climate. Phenology modulates surface energy balance and hydrological processes at the landscape scale. Knowledge of tropical forest phenology, however, is limited because highly diverse communities of tree species exhibit a variety of often-subtle phenological patterns. Heterogeneous tropical forests also exhibit highly variable and heterogeneous responses to biotic and abiotic stressors. While satellite remote sensing is commonly used to study vegetation phenology, frequent cloud cover, smoke from fires, and sensor artifacts complicate the study of tropical phenology. Widely used satellite platforms like MODIS and Landsat suffer from limitations of coarse spatial resolution or low temporal repeat frequency. We used satellite remote sensing data from two platforms: a) Sentinel-2 MultiSpectral Instrument operated by European Space Agency; and b) VENμS, a cooperative Earth observation program of Israel and France using a minisatellite that combines high spatial resolution and frequent repeats for selected study areas.

We focused our analysis at a series of sites across a gradient of wet and dry tropical forests and varying land use, from evergreen tropical forests, to savannas, to grasslands and croplands, and where both Sentinel-2 and VENμS imagery were available. We developed Normalized Difference Red Edge Index (NDRE) time series from both platforms, which showed greater dynamic range than the frequently used Normalized Difference vegetation Index (NDVI) from MODIS. Analyses of high spatio-temporal resolution data help reveal the dominant phenological patterns in heterogeneous tropical vegetation, despite frequent occultation from clouds and smoke. Preliminary analysis shows varying phenological responses during wet vs dry seasons across broadleaf evergreen forest, savannas and grasslands.