Simulation studies on terrestrial ecosystems are very important part of science. By making appropriate predictions, computer simulations give us valuable implications on environmental problems, issues concerning climate change, and social and economic applications of natural resources. However, the precision and accuracy of simulation studies largely depend on ways of representation of ecophysiological characteristics of vegetation and microbial communities. In this talk, we describe current conditions and problems of ecosystem models and show how ecophysiological aspects affect the results of the simulation.

Theoretical studies and simulation models cannot be realistic without the knowledge from empirical studies. However, it is usually problematic to simply extrapolate an empirical relationship to adjust the behavior of a model due to the difference in space and time. We discuss integration of theoretical and empirical studies: data assimilation. Data assimilation is an objective, statistical method to adjust a model, but has not been widely used in terrestrial ecosystem models partly due to abrupt behaviors that are characteristic to vegetation dynamics. Now we have an innovative measure that appropriately represents plant dynamics. In this talk we briefly introduce the benefits of our approach and preliminary results.