2006 年 10 巻 2 号 p. 143-149
Recent advancements in the spatial resolution of climate models enable us to provide detailed future climate change scenarios including changes in regional climate and extreme events for impact assessment research. However, impact researchers should be aware that climate models are still burdened with sizable and inherent uncertainty caused by the incompleteness in the representation of physical processes, or “parameterizations.” The prediction of the quantitative amplitude of climate change, such as “climate sensitivity,” remains uncertain because it depends more on physical parameterizations than on resolution. However, some qualitative characteristics of projected climate change, such as geographical patterns, can be regarded as relatively reliable, though they remain at the level of “expert judgment” by climate researchers rather than fact. Criteria for determining the reliable aspects of projections include: a. statistical significance, b. inter-model consistency, c. interpretability, and d. the validity of key processes. It is difficult for impact researchers alone to judge whether these criteria are met by the projected climate change in question for their impact assessments. Therefore, active mutual communication between climate researchers and impact researchers is highly desirable to improve the overall reliability of impact assessment based on the results of high-resolution climate modeling.
