Emissions Scenarios - IPCC

Background and Overview 63
Box 1-1: Uncertainties and Scenario Analysis
In general, there are three types of uncertainty: uncertainty in quantities, uncertainty about model structure and uncertainties
that arise from disagreements among experts about the value of quantities or the functional form of the model (Morgan and
Henrion, 1990). Sources of uncertainty could be statistical variation, subjective judgment (systematic error), imperfect
definition (linguistic imprecision), natural variability, disagreement among experts and approximation (Morgan and Henrion,
1990). Others (Funtowicz and Ravetz, 1990J distinguish three main sources of uncertainty: "data uncertainties," "modeling
uncertainties" and "completeness uncertamties." Data uncertainties arise from the quality or appropriateness of the data used as
inputs to models. Modeling uncertainties arise from an mcomplete understanding of the modeled phenomena, or from
approximations that are used in formal representation of the processes. Completeness uncertainties refer to all omissions due to
lack of knowledge. They are, in principle, non-quantifiable and irreducible.
Scenarios help in the assessment of future developments in complex systems that are either inherently unpredictable, or that
have high scientific uncertainties. In all stages of the scenario-building process, uncertainties of different nature are
encountered. A large uncertainty surrounds future emissions and the possible evolution of their underlying driving forces, as
reflected in a wide range of future emissions paths in the literature. The uncertainty is fiirther compounded in going from
emissions paths to climate change, from climate change to possible impacts and finally from these driving forces to formulating
adaptation and mitigation measures and policies. The uncertainties range from inadequate scientific understanding of the
problems, data gaps and general lack of data to inherent uncertamties of future events in general. Hence the use of alternative
scenarios to describe the range of possible future emissions.
For the current SRES scenarios, the following sources of uncertainties are identified:
Choice of Storylines. Freedom in choice of qualitative scenario parameter combinations, such as low population combined with
high gross domestic product (GDP), contributes to scenario uncertainty.
Authors Interpretation of Stoi-ylines. Uncertainty in the individual modeler's translation of narrative scenario storyline text in
quantitative scenario drivers. Two kinds of parameters can be distinguished:
• Harmonized drivers such as population, GDP, and final energy (see Section 4.1. in Chapter 4). Inter-scenario uncertainty
is reduced in the hamionized runs as the modeling teams decided to keep population and GDP within certain agreed
boundaries.
• Other assumed parameters were chosen freely by the modelers, consistent with the storylines.
Translation of the Understanding of Linkages between Driving Forces into Quantitative Inputs for Scenario Analysis. Often the
understanding of the linkages is incomplete or qualitative only. This makes it difficult for modelers to implement these linkages
in a consistent manner.
Methodological Differences.
• Uncertainty induced by conceptual and structural differences in the way models work (model approaches) and in the ways
models are parameterized.
• Uncertamty in the assumptions that underlie the relationships between scenario drivers and output, such as the
relationship between average income and diet change.
Different Sources of Data. Data differ from a variety of well-acknowledged scientific studies, since "measurements" always
provide ranges and not exact values. Therefore, modelers can only choose from ranges of input parameters for. For example:
• Base year data.
• Historical development trajectories.
• Current investment requirements.
Inherent Uncertainties. These uncertainties stem from the fact that unexpected "rare" events or events that a majority of
researchers currently consider to be "rare future events" might nevertheless occur and produce outcomes that are fundamentally
different from those produced by SRES model runs.
such as population growth, socio-economic development, and
technological progress; thus to predict emissions accurately is
virtually impossible. However, near-tenn policies may have
profound long-term climate impacts. Consequently, policymakers
need a summary of what is understood about possible
future GHG emissions, and given the uncertainties in both
emissions models and our understanding of key driving forces,
scenarios are an appropriate tool for summarizing both current
understanding and current uncertainties. For such scenarios to
be useful for climate models, impact assessments and the
design of mitigation and adaptation policies, both the main
outputs of the SRES scenarios (emissions) and the main inputs
or driving forces (population growth, economic growth,
technological, e.g., as it affects energy and land-use) are
equally important.
GHG emissions scenarios are usually based on an internally
consistent and reproducible set of assumptions about the key