Assessing the impact of green vegetation fraction on model’s latent heat fluxes in the Amazon basin

Abstract Two new Green Vegetation Fraction (GVF) sets were built from the Normalized Difference Vegetation Index (NDVI) and the Enhanced Vegetation Index (EVI), both based on the Moderate Resolution Imaging Spectroradiometer (MODIS) datasets, to improve greenness in land surface simulations. The Noah Land Surface Model (LSM) simulations used the newly created GVF sets at three station sites of the Large-Scale Biosphere-Atmosphere (LBA) Experiment in the Brazilian Amazon. A third GVF set, based on a vegetation index from the Advanced Very High-Resolution Radiometer (AVHRR), already implemented in Noah LSM, was added to the simulations. The three GVF sets, time-averaged over the 2000–2006 period, drove the offline simulations. A model-based product that assimilates satellite-based precipitation estimates provided the atmospheric forcing for all simulations. The intraseasonal variability of the GVF-EVI set had the best agreement with the predominant evergreen broadleaf (EB)-type phenology in the Amazon. The largest differences between the MODIS-based simulations of the latent heat fluxes occurred in June. A decrease in canopy evaporation in the EVI-driven simulations might result from a reduction in the vegetation fraction that intercepts precipitation. That might be responsible for an increase in direct soil evaporation compared to the NDVI-driven simulations. The AVHRR-based simulations exhibited the lowest (highest) transpiration (direct soil evaporation) annual means, likely due to a reduction in the EB pixels in comparison to the MODIS-based ones. Ultimately, this approach to enhancing greenness and seasonality aspects might improve the surface flux simulations in densely vegetated areas, such as the Amazon rainforest.