Criteria for a new scenario framework
One key objective of community scenarios is to integrate across different research disciplines. To do so, scenarios need to be relevant to questions regarding climate change research and assessment, including the interactions and trade-offs between climate impacts and adaptation and mitigation responses. Research on mitigation scenarios focuses on the economic and technological implications of different stabilization targets (in the future, research is expected to increasingly focus on governance and social aspects of mitigation policies). Research on adaptation includes the magnitude and extent of climate change and the vulnerability of the exposed system, which depends on the level of climate change but also on socio-economic conditions. Taken together, these research areas suggest two key factors around which community scenarios could be built:
-
The magnitude and extent of climate change and associated environmental changes;
-
The trends in human development in relation to the drivers of climate change, the capacity to mitigate greenhouse gas emissions, and the vulnerability and capacity to adapt to climate change.
Van Vuuren et al. (2012b) and Kriegler et al. (2012) proposed that a scenario framework should be based on these factors. Other criteria that are important for a useful scenario framework include that it should be:
-
1.
Incorporate the RCPs. The RCPs were developed to explore climate change under different levels of forcing in the first step of the scenario development process. Therefore, a new framework should incorporate these pathways.
-
2.
Streamlined: The number of scenarios should be as small as possible.
-
3.
Comprehensive. The framework needs to cover an adequate set of key variables required in IAM and IAV analyses, and to span a broad range of possible future climates and development pathways
-
4.
Comparable. The scenario set should facilitate comparison of different studies by providing common assumptions about climate outcomes and socio-economic developments. This will support the synthesis and assessment of results regarding climate change, impacts, adaptation and mitigation at multiple scales.
-
5.
Support uncertainty analysis. The framework should be capable of characterizing the range of uncertainty in the costs and other implications of mitigation, adaptation and impacts among alternative potential climate futures.
-
6.
Scalable. Scenarios should provide enough information at the scale of large world regions to support development of assumptions for studies at finer scales. Similarly, scenarios should include near- and long-term future conditions, the former providing links to ongoing trends and planning horizons and the latter including plausible large-scale divergences in key driving factors.
-
7.
Versatile. The scenario set should provide enough structure to facilitate consistency, and offer context and calibration points for IAV and mitigation analyses, but also offer flexibility for defining details relevant for sectors and regions in particular studies.
-
8.
Appropriate for social, institutional, governance and policy analysis. The framework should not only emphasize quantitative model-based analysis, but also be relevant for social science research.
The scenario matrix approach as an overall organizing principle
The two factors introduced in the previous section as organizing principles for the new framework (the level of climate change and socio-economic developments) determine to a large degree the ability to mitigate and/or adapt and the associated costs. Figure 1 shows how these sets of information are combined in the scenario matrix architecture. For the level of climate change, we propose to use related indicator i.e. the forcing level (see Section 2.3). This forcing axis is represented by the RCPs. The second axis of socio-economic assumptions are described in alternative future development pathways, called Shared Socio-economic Pathways (SSPs) (see Section 2.4). Next. scenarios can be defined for each cell in Fig. 1, where one level of anthropogenic forcing intersects with one set of socio-economic assumptions.
It is worth clarifying the distinction between the terms scenario and pathway in the matrix architecture. We use the term scenario to describe a plausible, comprehensive, integrated and consistent description of how the future might unfold (Nakicenovic et al. 2000) while refraining from a concrete statement on probability. The term scenario specifically refers to integration of socio-economic, climate change, and climate change policy assumptions within the cells of the matrix. In contrast, the term pathway is used for the conditions describing the rows and columns of the matrix (e.g., the RCPs and SSPs). In other words, the term pathway emphasizes that these conditions are not comprehensive scenarios, but are focused on a specific component of the future (climate change or socio-economic circumstances). Only when combined can they provide the basis of an integrated scenario. The word pathway also emphasizes the time-dependent nature of the conditions.
Four critical components of the new framework are discussed in Sections 2.3 to 2.6:
-
Radiative forcing axis (RCPs)
-
Socio-economic pathway axis (SSPs)
-
Shared climate Policy Assumptions (see further) (SPAs)
-
Projected climate change
The radiative forcing axis
Radiative forcing forms a natural choice for one axis of the scenario matrix because it constitutes a major interface between the IAM (translating emissions drivers to forcing) and the ESM communities (translating forcing to climate change). In fact, the four RCPs were defined in terms of their radiative forcing in the year 2100 and trajectory of change (Van Vuuren et al. 2011a) (Table 1). The RCPs provide information that is an essential input to ESMs, including emissions of greenhouse gases and short-lived species specified on a 0.5° latitude × 0.5° longitude grid, as well as land use and land cover information. The ESM community is conducting multiple modeling experiments to investigate the climate response to the RCPs (Taylor et al. 2012). Due to uncertainty about climate change, there will be a range of climate outcomes (in terms of temperature, precipitation, extreme events, etc.) that can be related to individual RCPs (see also Section 2.6).
Table 1 Representative concentration pathways in the year 2100
It should be noted that in the RCPs, forcing refers to the global average forcing on the basis of greenhouse gases and air pollutants. At the local scale, forcing (and therefore climate change) can be very different due to spatial patterns of land-use change and air pollutant concentrations. The impacts of land-use change and air pollutants can, certainly at the local scale, be so substantial that they could potentially complicate the combination of climate model output based on a specific RCP and other scenarios with a similar global forcing level. It is still an open research question how important the possible inconsistencies introduced by different land-use and climate patterns are compared to other uncertainties such as regional climate change, the actual forcing of land-use change and air pollution or those resulting from the current implementation of land-use in earth-system models. The answer to this question determines how easily different IAM scenarios for a specific forcing level can be combined with existing RCPs. Potentially, scaling methods could be developed to somewhat adjust for differences in assumptions.
The socio-economic pathways axis
The second axis of the framework is a set of socio-economic pathways consisting of quantitative and qualitative elements that describe the drivers of how the future might unfold in terms of population growth, governance efficiency, inequality across and within countries, socio-economic developments, institutional factors, technology change, and environmental conditions (see also O’Neill et al. 2013, who elaborate further on the SSPs). The SSPs are a small number of alternative characterizations of possible societal futures for use by different research communities, including narrative descriptions of future trends and quantitative information for some key elements. This information can be used as boundary conditions for the analysis of mitigation and adaptation policies.
A key assumption in the proposed framework is that it is possible to combine the socio-economic pathways with different levels of radiative forcing. This feature of the approach is achieved by defining SSPs as “reference” pathways that would occur in a hypothetical case without new climate policy interventions (mitigation and adaptation) and without being influenced by future climate change (O’Neill et al. 2013). These reference pathways would then be combined with climate policy assumptions (including mitigation) in order to achieve forcing outcomes, and with climate change outcomes to create scenarios. In these scenarios, socio-economic development will be affected by both climate change and policy interventions and so will differ to some extent from the assumptions in the SSPs. The separation of climate policies and impacts from the SSPs is a methodological choice that facilitates the analytical use of the SSPs (see also Section 2.5 on the formulation of policy). Still, it should be noted that the SSPs are influenced by many other policy assumptions and environmental factors. The distinction between climate policies (not included in SSPs) and closely related policies such as energy policies (included in SSPs) is discussed in detail in Kriegler et al. (Submitted for publication in this special issue), but in general these are differentiated according to the primary intent of the policy. Scenarios associated with lower radiative forcing than in the reference case can be created by assuming mitigation policies sufficient to achieve the forcing outcomes, whose extent and consequences may depend on the nature of related policies assumed in the SSP. The degree of global climate mitigation stringency is inversely related to the level of radiative forcing in the year 2100: all else equal, the wider the gap between reference forcing and an RCP level, the more effort will be required to close it. Thus, by definition, for any given SSP, lower radiative forcing in 2100 implies greater mitigation stringency. Such considerations bring into sharp focus the importance of specifying policy scenarios in relation to reference scenarios based on the SSPs.
A second key assumption in the framework is that different development pathways can lead to similar radiative forcing outcomes. In fact, the Special Report on Emission Scenarios (SRES) concluded that for each forcing level, multiple socio-economic scenarios could be identified (Nakicenovic et al. 2000). More recently, Van Vuuren et al. (2012b) observed that very little correlation exists between individual driver assumptions (such as population and economic assumptions) on the one hand and forcing levels on the other for climate policy scenarios reported in the literature (reconfirming the SRES finding).
In order to fulfill the objective of the framework, the SSPs as a set should describe a wide range of different futures with different mitigation and adaptation challenges. The conceptual approach to ensure this is discussed in detail in O’Neill et al. (2013). This approach is directly organized around the level of future challenges with respect to mitigation and adaptation, forming a second matrix. Four key distinct storylines can be identified describing low and high ‘challenge’ combinations: SSP1 (low mitigation and adaptation challenges), SSP3 (high mitigation and adaptation challenges), SSP4 (low mitigation and high adaptation challenges) and SSP5 (high mitigation and low adaptation challenges). In addition, a fifth storyline is defined that has medium assumptions for both types of challenges. O’Neill et al. discuss how these positions in the ‘challenges’ matrix can be translated into basic assumptions for elements of the SSPs such as population, governance and technology development, but the content of the SSPs still needs to be elaborated further in subsequent work.
Clearly, not every cell of the scenario matrix (Fig. 1) needs to be populated. For example, an SSP that is defined such that energy efficiency and the adoption of low carbon energy sources reduces the carbon intensity of economic production in the absence of climate policy, e.g. motivated by following a broader sustainable development agenda, may be inconsistent with radiative forcing reaching 8.5 W/m2 in 2100.
Shared climate policy assumptions
As discussed in Sections 2.2 and 2.4, additional assumptions are needed about adaptation and mitigation policies to derive a scenario consistent with a given combination of a RCP and SSP (a cell in Fig. 1). First of all, this concerns the mitigation target. However, assumptions also need to be made on the policies: is mitigation achieved via an universal carbon tax or through various technology standards, when is it introduced and who participates in mitigation policies? Similarly, adaptation policies also need to be described (e.g., how much international support is available to help the poorest country adapt?) as well as the way in which these policies are implemented, including imperfections in policy design and enforcement (e.g., is mitigation implemented with full or partial participation of all countries? Is adaptation finance accessible to all countries?).
As the effectiveness and costs of mitigation and adaptation will be very sensitive to the assumptions about climate policy, it is important to specify these clearly. Those policy factors can be seen as another dimension of the matrix architecture that characterizes the nature of the policy response (see Fig. 2) (for instance, in terms of participation, timing and international cooperation). While research teams will often make their own assumptions about climate policies, here, as for the SSPs, it is also useful to formulate a small number of shared (climate) policy assumptions (SPAs) that are common to different studies, hence improving the ability to compare scenarios across models and analyses. The concept of SPAs is discussed in detail in Kriegler et al. (Submitted for publication in this special issue). Because GDP and other variables could be affected by climate policies and by climate change impacts, the elaboration of scenarios that include one or both of these factors may well modify some of the SSP assumptions. Moreover, some SPAs are less likely for specific SSPs: for instance, it is not likely that all parties participate in international climate policy in a world that is characterized by fragmentation in other policy areas.
The climate dimension
The vertical axis in the scenario framework is defined in terms of RCPs, i.e. the level of radiative forcing. There are large uncertainties surrounding model projections of future climate for a given level of radiative forcing, due to factors such as the inherent unpredictability of natural climatic variations, global climate sensitivity in response to anthropogenic forcing and regional patterns of climate response (Christensen et al. 2007; Meehl et al. 2007; Tebaldi and Arblaster, Submitted for publication in this special issue; Van Vuuren et al. 2008). Regional projections of some climatic variables (such as precipitation and wind speed), which can be crucially important for impacts in certain sectors and systems, are even more uncertain than projections of others (such as air temperature). This is also true for the timing, pattern, frequency, duration, and intensity of weather events, which provides critical information for impacts assessments. Together, this implies that a specific climate model projection for a given RCP level might differ greatly from the projection from another climate model for the same forcing. This “climate change” uncertainty can be regarded as another axis of the framework (Fig. 3). It is important to address this uncertainty as much as possible by using a large range of ESM outputs (or pattern scaling results, see Tebaldi and Arblaster, Submitted for publication in this special issue). While analysts are increasingly applying multi-model ensemble climate projections in impact studies (e.g. Araújo and New, 2007; Diffenbaugh and Field, 2013), this is not always feasible. One way to handle the climate uncertainty while limiting the number of IAV calculations might be by identifying expected “best case” and “worst case” scenarios for specific purposes. For instance, if changes in precipitation are known to be a key determinant of agricultural impacts in a region, it is possible to identify, for a specific forcing level, climate models projecting the highest and lowest level of precipitation change in that region. Such methods are commonly applied in IAV analysis, although multiple criteria are normally applied in determining the final selection of representative climate scenarios to be used (IPCC-TGICA 2007; Wilby et al. 2009). Further guidelines on this would clearly be useful in application of the overall framework.