American Geophysical Union, Journal of Geophysical Research, D24(113), 2008
DOI: 10.1029/2007jd009370
Full text: Download
A new version of the land surface model of the European Centre for Medium-Range Weather Forecasts (Carbon-TESSEL, or CTESSEL) includes a vegetation growth model. This study describes a leaf area index (LAI) data assimilation system (LDAS) based on CTESSEL and satellite LAI for operational Net Ecosystem Exchange (NEE) predictions. The LDAS is evaluated over West Africa. A preliminary experiment shows a significant impact of the LAI on the CTESSEL NEE. The LAI is compared to two satellite products: the predicted annual cycle is delayed over the Sahel and savannah, and the LAI values differ from the satellite products. Preliminary to their use in the LDAS, the LAI products are rescaled to the CTESSEL predictions. The LDAS simulations are confronted to measurements of biomass and LAI for a site in Mali. The LAI analysis is shown to improve the predicted biomass and the annual cycles of the water (latent heat flux, or LE) and carbon (NEE) fluxes. Afterward, the LDAS is run over West Africa with the Moderate-Resolution Imaging Spectroradiometer products (2001–2005). The analysis of LAI shows a limited impact on LE, but it impacts strongly on NEE. Finally, the CTESSEL NEE are compared to two other models' outputs (simple biosphere (SIB) and Carnegie-Ames-Stanford (CASA)). The order of magnitude of the three data sets agrees well, and the shift in annual cycle of CTESSEL is reduced by the LDAS. It is concluded that a LAI data assimilation system is essential for NEE prediction at seasonal and interannual timescales, while a LAI satellite-based climatology may be sufficient for accurate LE predictions.