HIGH RESOLUTION SIMULATION OF LAI, CARBON AND WATER FLUXES OVER FRANCE

摘要

The vegetation/land component of GMES is called "Land Monitoring Core service" (LMCS). The geoland2 European project (FP7, 2008-2012) is a demonstrator of the evolution of the LMCS, including the consolidation of prototype services and the test of their operational capacity. The main mission of the land carbon core information service (LC-CIS) of geoland2 is to assess the impact of weather and climate variability on terrestrial biospheric carbon fluxes, in the context of international conventions. The LC-CIS aims at monitoring the global terrestrial carbon fluxes (e.g. to support reporting obligations in the course of the Kyoto Protocol) and setting-up pre-operational infrastructures for providing global products, both in near real time and off-line mode. A multi-model carbon accounting system is developed, coupled with EO data assimilation schemes. Emphasis is put on validation (in-situ data), with downscaling on reference European countries (F, NL, HU). The C-TESSEL and SURFEX modelling platforms (of ECMWF and Météo-France, respectively) are used for production. The ORCHIDEE modelling platform (LSCE) is used for benchmarking and validation purposes. We will present first test of the Geoland2 SURFEX model without data assimilation (Open-loop) over the French domain. We will focus on the simulation of the Leaf Area Index (LAI) and of carbon and water fluxes. The LAI is an important characteristics of the vegetation. It is a good measure of the amount of active vegetation and is linked to the canopy resistance, the water and carbon fluxes between the atmosphere and the vegetation. In this study, we compare 4 different estimations of LAI over France area : two LAI products derived from visible remote sensing and simulated LAI from 2 land surface models. The simulated LAI originates from two land surface models, The ISBA-A-gs model (developed by CNRM) and the ORCHIDEE model (developed by IPSL). The atmospheric forcing for the models is the SAFRAN dataset, a high resolution (8-km) operational product over France and the vegetation map is the ECOCLIMAP2 dataset. The LAI remote sensing product are derived from the MODIS and SPOT/VGT sensors : The Modis product (Collection 5) and the CYCLOPES LAI product. The study period ranges from 2000 to 2007 and allows to study the average seasonal cycle as well as the interannual variation. We analyse monthly values of LAI over the period. The average seasonal cycle over the seven years starts earlier in the ORCHIDEE simulation than in the ISBA-A-gs ones. ORCHIDEE simulate a strong maximum of interannual variability at springtime (and a very small variability at summertime), whereas in the ISBA-A-gs simulation the maximum is in summer (with small values in spring). These differences can be explained by the differences in model formulations. The interannual variability of both satellite products is more complex, with a weaker seasonal cycle and contrasted spatial patterns. Finally, we compare the water and carbon fluxes simulated by the two models. The two models present significative differences in the magnitude of the fluxes. We will show that this differences are linked to the parameterisation of the individual Plant Functional Type.