Articles | Volume 4, issue 4
Geosci. Model Dev., 4, 1103–1114, 2011
Geosci. Model Dev., 4, 1103–1114, 2011

Methods for assessment of models 05 Dec 2011

Methods for assessment of models | 05 Dec 2011

Evaluation of a Global Vegetation Model using time series of satellite vegetation indices

F. Maignan1, F.-M. Bréon1, F. Chevallier1, N. Viovy1, P. Ciais1, C. Garrec1, J. Trules1, and M. Mancip2 F. Maignan et al.
  • 1Laboratoire des Sciences du Climat et de l'Environnement, IPSL, UMR8212 (CEA/CNRS/UVSQ), CEA Saclay, Orme des Merisiers, 91191 Gif sur Yvette CEDEX, France
  • 2Institut Pierre Simon Laplace, Université Pierre et Marie Curie, 4 place Jussieu, 75252 Paris CEDEX 05, France

Abstract. Atmospheric CO2 drives most of the greenhouse effect increase. One major uncertainty on the future rate of increase of CO2 in the atmosphere is the impact of the anticipated climate change on the vegetation. Dynamic Global Vegetation Models (DGVM) are used to address this question. ORCHIDEE is such a DGVM that has proven useful for climate change studies. However, there is no objective and methodological way to accurately assess each new available version on the global scale. In this paper, we submit a methodological evaluation of ORCHIDEE by correlating satellite-derived Vegetation Index time series against those of the modeled Fraction of absorbed Photosynthetically Active Radiation (FPAR). A perfect correlation between the two is not expected, however an improvement of the model should lead to an increase of the overall performance.

We detail two case studies in which model improvements are demonstrated, using our methodology. In the first one, a new phenology version in ORCHIDEE is shown to bring a significant impact on the simulated annual cycles, in particular for C3 Grasses and C3 Crops. In the second case study, we compare the simulations when using two different weather fields to drive ORCHIDEE. The ERA-Interim forcing leads to a better description of the FPAR interannual anomalies than the simulation forced by a mixed CRU-NCEP dataset. This work shows that long time series of satellite observations, despite their uncertainties, can identify weaknesses in global vegetation models, a necessary first step to improving them.