Development of the dedicated numerical platform feasible to very long-term climatic simulations based on the time-space method

Great progress of the performance of recent super computers has been further done. For example, the peak running speed of the "Earth simulator" developed in Japan and completed in 2002 reaches around 36 TFLOPS. Using these supercomputers, a lot of useful numerical studies for estimation of the future terrestrial environmental change and validations of many consumption scenarios of energy, water, food resources are conducted by using the numerical results of climate models. However, in such studies, the details of the terrestrial ecosystem, the most unknown factor on the earth, has not been included because of its complexity and uncertainty. In order to accurately investigate the future change in terrestrial environment, the role of the terrestrial ecosystem must be considered and carefully analyzed. From a numerical standpoint, such inclusion of the terrestrial ecosystem into the numerical modeling of the earth's climate is a hard work because it must be treated in a longer timescale (e.g. 1000 to 10000 years) than in the conventional modeling. Therefore, it is still necessary to make continuous efforts to improve performance of both numerical algorithms and computer architectures if we perform computer simulations with the higher degree of resolution compared with existing ones and handle very long time-scale problems. We have consistently used the time-space method (TSM) as an efficient dynamical-core solver for climate studies with very long time-scales like in ocean general circulation modeling, that is, the development of the dedicated numerical platform feasible to very long-term climatic simulations. The TSM has been proposed by Saitoh et al.(see ibid., 1992) as one of the fastest solvers for general heat transfer and fluid flow problems aiming at a great reduction of CPU time from the former standpoint. This paper presents numerical results of numerical climatic simulations using this efficient platform. The results show the realistic climate features. In addition, we will evaluate the roles of the land ecosystem in further longer-time scale than that treated in previous studies.