Biogeosciences, 7, 2061-2080, 2010
© Author(s) 2010. This work is distributed under
the Creative Commons Attribution 3.0 License.
02 Jul 2010
Multi-model analysis of terrestrial carbon cycles in Japan: limitations and implications of model calibration using eddy flux observations
K. Ichii1, T. Suzuki1, T. Kato2,3, A. Ito2,4, T. Hajima2, M. Ueyama5, T. Sasai6, R. Hirata7, N. Saigusa4, Y. Ohtani8, and K. Takagi9 1Faculty of Symbiotic Systems Science, Fukushima University, 1 Kanayagawa, Fukushima, 960-1296, Japan
2Japan Agency for Marine-Earth Science and Technology, 3173-25, Showamachi, Kanazawa, Yokohama, 236-0001, Japan
3QUEST, Department of Earth Science, University of Bristol, Wills Memorial Building, Queen's Road, Bristol, BS8 1RJ, UK
4National Institute for Environmental Studies, 16-2, Onogawa, Tsukuba, 305-8506, Japan
5Graduate School of Life and Environmental Sciences, Osaka Prefecture University, 1-1 Gakuen-cho, Naka-ku, Sakai, Osaka, 599-8531, Japan
6Graduate School of Environmental Studies, Nagoya University, Furo-cho, Chikusa, Nagoya, 464-8601, Japan
7Graduate School of Agriculture, Hokkaido University, Kita 9 Nishi 9, Kita-ku, Sapporo, 060-8589, Japan
8Forestry and Forest Products Research Institute, 1 Matsunosato, Tsukuba, 305-8687, Japan
9Field Science Center for Northern Biosphere, Hokkaido University, Toikanbetsu, Horonobe, 098-2943, Japan
Abstract. Terrestrial biosphere models show large differences when simulating carbon and water cycles, and reducing these differences is a priority for developing more accurate estimates of the condition of terrestrial ecosystems and future climate change. To reduce uncertainties and improve the understanding of their carbon budgets, we investigated the utility of the eddy flux datasets to improve model simulations and reduce variabilities among multi-model outputs of terrestrial biosphere models in Japan. Using 9 terrestrial biosphere models (Support Vector Machine – based regressions, TOPS, CASA, VISIT, Biome-BGC, DAYCENT, SEIB, LPJ, and TRIFFID), we conducted two simulations: (1) point simulations at four eddy flux sites in Japan and (2) spatial simulations for Japan with a default model (based on original settings) and a modified model (based on model parameter tuning using eddy flux data). Generally, models using default model settings showed large deviations in model outputs from observation with large model-by-model variability. However, after we calibrated the model parameters using eddy flux data (GPP, RE and NEP), most models successfully simulated seasonal variations in the carbon cycle, with less variability among models. We also found that interannual variations in the carbon cycle are mostly consistent among models and observations. Spatial analysis also showed a large reduction in the variability among model outputs. This study demonstrated that careful validation and calibration of models with available eddy flux data reduced model-by-model differences. Yet, site history, analysis of model structure changes, and more objective procedure of model calibration should be included in the further analysis.

Citation: Ichii, K., Suzuki, T., Kato, T., Ito, A., Hajima, T., Ueyama, M., Sasai, T., Hirata, R., Saigusa, N., Ohtani, Y., and Takagi, K.: Multi-model analysis of terrestrial carbon cycles in Japan: limitations and implications of model calibration using eddy flux observations, Biogeosciences, 7, 2061-2080,, 2010.
Publications Copernicus
Special issue