ment and development and government versus market solutions. These narratives were then quantified with a focus on demographic, economic, and energy-related indicators, as well as sectoral indicators for water, agriculture, and health. These variables were quantified at the state level, taking into account decadal variations in the past and the qualitative story lines for the future.

Given the results of this study, Mathur concluded by asserting the need for further integrated assessment models and impact-adaptation-vulnerability analyses to generate more realistic and robust predictions on climate-related risks. This would require greater involvement of institutions at the regional and local levels to ensure that the assessed reduction potentials being considered in global studies allow for a better encapsulation of regional changes that are likely in the future.

QUANTITATIVE DOWNSCALING APPROACHES4

Tom Kram


Tom Kram made a presentation on behalf of Detlef Van Vuuren (who was unable to attend), based on an article in preparation by Van Vuuren and his colleagues on quantitative downscaling approaches. Kram noted that many methods of downscaling have been tried and that the chosen preferred method depends on purpose, coverage, resolution, and the availability of information. Downscaled data need to be consistent with both the larger and smaller scales, as well as internally. The article distinguishes four approaches: (1) algorithmic downscaling, which can be done (a) proportionally (assuming every unit at the smaller scale is equal), (b) by applying the change assumptions for the larger unit to the smaller units and assuming that they will converge toward the central estimate, or (c) by applying exogenous scenarios; (2) methods of intermediate complexity using simplified formulas that are calibrated differently for different subunits; (3) complex models that can be applied at a small scale; and (4) fully coupled physical-social models that use changes in the models for one year as inputs to the next year’s estimates. Some methods can lead to problems, such as when growth rate data for Asia are applied to Singapore. Kram concluded by saying that although there have been bad experiences with socioeconomic downscaling the past, better data and more advanced algorithms are now available. He said that although many methods are available, for global applications, simple methods might be adequate.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement