Figure 2	Conceptual diagram of the hierarchical simulation of carbon cycle in terrestrial ecosystems, from a single point to broad regions (e.g. Monsoon Asia).

Our research group is working on modeling and simulations of carbon cycle at the global scale, including evaluation of forest productivity, biomass, and carbon budget. Because forest ecosystems are very complicated and time-scale of forest dynamics is very long (decades to centuries), models should contain a wide variety of ecological processes. Moreover, forest ecosystems have a remarkable diversity from a tropical rain forest to a boreal deciduous forest, demanding flexibility of models. Figure 1 shows the estimated net photosynthetic productivity of terrestrial ecosystems in Monsoon Asia with our model (Sim-CYCLE), based on an actual land-cover map using satellite data. Apparently, those areas covered by tropical humid forests (e.g. Southeast Asia) exhibited higher productivities, compared to inner continental areas covered by grasslands or deserts. Similarly, the model can be applicable to estimate plant biomass and soil carbon storage. At present, we are trying to apply the model to simulate carbon budget under global environmental change and human impact. In addition, we should incorporate new observational data into the model, so that these data can improve our model accuracy. For example, micrometeorological measurement and satellite observation of forest processes can be useful to validate our model.

Until now, models have regarded climatic conditions (e.g. temperature and precipitation) as major factors influencing carbon budget, so that climatic change may affect photosynthesis, respiration, and consequently net carbon budget. Recently, direct disturbances, such as wildfire and human deforestation, are expected to have even larger impacts on carbon budget. Since variations in carbon budget induced by disturbances are closely related to the issue of carbon accounting for the Kyoto Protocol, modeling of wildfire and deforestation is strongly required. Although these processes were difficult to merge into largescale models (e.g. Fig.1), we are attempting to construct a hierarchy model system, allowing us to parameterize local disturbance effect. For example, as shown in Figure 2, the hierarchy model system revealed that frequent wildfire in boreal forest of Siberia may reduce carbon storage considerably, and that climate change may result in further reduction of carbon storage through increased fire frequency. It is evident that through these simulation studies, we must take diverse factors into account, in order to make a reliable projection of carbon budget under changing global environment. Thus, further modeling studies are required to accomplish our purpose.