Scaling Methods in Regional Integrated Assessments: from Points Upward and from Global Models Downwards

摘要

No easy conclusions can be drawn from this comparison of up and down scaling methodologies. Data, expertise, time and financial resources are limited, and may drive the choice of method more strongly than the technical merits of different schemes. Based on the model testing in the CLIVARA project, however, it is possible to make some general statements. The simplest, site-driven approach may be justified in some homogeneous environments (Denmark) for some parameters (mean and variability of yield). For spatial grids a subset of the complete grid is adequate for many purposes. A baseline can be estimated on the whole grid and a series of sensitivity tests run on a stratified sample. Conversely, in complex terrain, a regional approach can expose key non-linearities that may not be apparent from site-based analyses. Including remote sensing techniques in impact studies allows full representation of landscape dynamics while not necessarily making the analysis overly complex. Similarly, sophisticated methods of downscaling climate change scenarios may be warranted where local variability in the terrain or in impacts are of great concern. However, where the coarse grain scenario varies significantly (e.g., different GCMs report significant increases or decreases in precipitation), simple downscaling methods may be sufficient to capture a sense of the risks.