Uncertainty in Integrated Assessment Scenarios

Webster, Mort

University of North Carolina at Chapel Hill

关键词: Probability;    Productivity;    Economic Development;    Greenhouse Gases;    Production;   

DOI  :  10.2172/883668
RP-ID  :  DOE/ER/63468-1
RP-ID  :  FG02-02ER63468
RP-ID  :  883668

美国|英语

来源: UNT Digital Library

【 摘 要 】

The determination of climate policy is a decision under uncertainty. The uncertainty in future climate change impacts is large, as is the uncertainty in the costs of potential policies. Rational and economically efficient policy choices will therefore seek to balance the expected marginal costs with the expected marginal benefits. This approach requires that the risks of future climate change be assessed. The decision process need not be formal or quantitative for descriptions of the risks to be useful. Whatever the decision procedure, a useful starting point is to have as accurate a description of climate risks as possible. Given the goal of describing uncertainty in future climate change, we need to characterize the uncertainty in the main causes of uncertainty in climate impacts. One of the major drivers of uncertainty in future climate change is the uncertainty in future emissions, both of greenhouse gases and other radiatively important species such as sulfur dioxide. In turn, the drivers of uncertainty in emissions are uncertainties in the determinants of the rate of economic growth and in the technologies of production and how those technologies will change over time. This project uses historical experience and observations from a large number of countries to construct statistical descriptions of variability and correlation in labor productivity growth and in AEEI. The observed variability then provides a basis for constructing probability distributions for these drivers. The variance of uncertainty in growth rates can be further modified by expert judgment if it is believed that future variability will differ from the past. But often, expert judgment is more readily applied to projected median or expected paths through time. Analysis of past variance and covariance provides initial assumptions about future uncertainty for quantities that are less intuitive and difficult for experts to estimate, and these variances can be normalized and then applied to mean trends from a model for uncertainty projections. The probability distributions of these critical model drivers, and the resulting uncertainty in projections from a range of models, can provide the basis of future emission scenario set designs.