HR: 16:20h
AN: B42D-02    [PDF]
TI: Parameter Estimation in Biogeochimical Surface Model Using Nonlinear Optimization With Eddy-Covariance Measurements Over Pine Forests.
AU: * Santaren, D
EM: santaren@lsce.saclay.cea.fr
AF: LSCE/CEA, CEA, Orme des Merisiers, Bat 701, Gif/Yvette, 91191 France
AU: Peylin, P
EM: peylin@lsce.saclay.cea.fr
AF: LSCE/CEA, CEA, Orme des Merisiers, Bat 701, Gif/Yvette, 91191 France
AU: Viovy, N
EM: viovy@lsce.saclay.cea.fr
AF: LSCE/CEA, CEA, Orme des Merisiers, Bat 701, Gif/Yvette, 91191 France
AU: Ciais, P
EM: ciais@lsce.saclay.cea.fr
AF: LSCE/CEA, CEA, Orme des Merisiers, Bat 701, Gif/Yvette, 91191 France
AB: Global models of carbone, water, and energy exchanges between the biosphere and the atmosphere are usually validated and calibrated with intensive measurements made over specific ecosystems like those of the fluxnet networks. However the nonlinear dependance between fluxes and model parameters generally complicates the optimization of the major parameters and the assessment of their uncertainty. In this study, we estimate key parameters of the Land-surface ORCHIDEE model, using the seasonal and diurnal information from latent heat flux, sensible heat flux, net CO2 flux and net radiation measurements over pine forest (Landes, France). We first present the inverse methodology, the complications linked to the non linearity, and an approach to derive parameter uncertainties in this context. We optimize the major parameters controlling the rate of carbon assimilation, the carbon autotrophic and heterotrophic respiration, the aerodynamic resistance, and the surface energy balance. The results will be discussed in term of model to data fit and parameter estimation. Most parameters appear to be significantly changed from their a priori estimate. A detail analysis of the posterior covariance matrix will be performed including the correlation terms (which are high between certain parameters). This optimisation approach helped us to highlight discrepancies of the ORCHIDEE model and to improve some parametrizations. The results of the inverse procedure using measurements at other sites with similar ecosystem will be compared.
DE: 0315 Biosphere/atmosphere interactions
DE: 1615 Biogeochemical processes (4805)
DE: 3210 Modeling
DE: 3260 Inverse theory
SC: Biogeosciences [B]
MN: 2003 Fall Meeting