Improving Environmental Simulation Models to Assess Climate Change Impacts

Increased levels of atmospheric carbon dioxide and other by-products of human civilization are expected to disrupt the global energy budget and induce climatic change, which will aIter the distribution of plant species and the composition of plant communities. In spite of the importance of these projected effects, our understanding of the processes leading to such changes needs to be strengthened. This paper addresses the need to improve models that are used to assess spatial aspects of the response of vegetation to anthropogenic climate change. Several modifications are suggested, each of which relates to changes in forest community composition that occur during periods of climate change: spatial structure is an important determinant of community composition, and therefore, forest fragmentation, and spatial dispersal functions should be explicitly modeled. Other factors to be examined are the interactions of environmental change and competition and the effect of modeling scale on results. The described methodology uses a geographic information system to link interdependent forest plots in a computer simulation model of forest dynamics. Dispersal functions of species, forest fragmentation, and rates of climate change would be varied among runs of the simulation model, and projections would be analyzed for sensitivity using detrended correspondence analysis, variance analysis, and semivariogram techniques. The combination of physically-based dynamic models in a spatially-distributed multi-cell framework requires considerably more computational power than have prior studies. Approaches to computational intensity and efficiency are discussed.