Diagnostic and prognostic modeling of the terrestrial biosphere with remotely sensed measurements. / Modélisation paramétrique et prédictive de la biosphère terrestre par des observations satellitaires

Abstract

This paper presents a recapitulative of the modeling studies we have developed to estimate carbon fluxes of the terrestrial biosphere, with various integrations of satellite observations to drive, test, or improve models. The diagnostic model aimed at estimating Net Primary Productivity (NPP) benefits from remotely sensed vegetation index (NDVJ) to monitor the fraction of absorbed photosynthetically active radiation (f(PAR)) with high space and time sampling. Applied to a pluriannual period, this method showed its ability to witness the response of vegetation to major interannual climatic events such as El Nino. Satellite f(PAR) estimates were also used to assess the outputs of a NPP process model that predicts Leaf Area Index (LAI). More complex approaches, based on the assimilation of satellite measurements within process models, are being developed in order to adjust some key model parameters. The balance between heterotrophic respiration (R(h)) and NPP determines the carbon Net Ecosystem Productivity (NEP ), i.e. the net carbon budget of ecosystems, which is the main variable of interest for global carbon cycle study. We describe an approach that combines satellite estimates of NPP with atmospheric CO₂ concentration and isotopic composition time series to adjust simple R(h) models. The method allowed to derive zonal distribution of the Q₁₀ parameter, that presents similar trends with values found in the literature. Besides the obvious advantage of remote sensing for monitoring the seasonal evolution of vegetation, this result proves the ability of satellite data to diagnose a realistic zonal trend of seasonal and annual NPP.