Climate Scenarios and Projections: the Known, the Unknown, and the Unknowable as applied to California

The purpose of this meeting was to discuss the likelihoods of different global climate scenarios, and how they could be effectively linked with regional projections of climate and utilized in better understanding impacts. Using California as a case study, participants explored how information such as the IPCC report (AR4) could help to inform and assess adaptation and mitigation strategies with a focus on particular sectors such as water and its management.

Keywords: climate variability and change; solutions

Introduction
What can be known about future climate for Earth? This is perhaps one of the most hotly contested scientific questions of the last 25 years, and an even more challenging question when scaled to a region such as the state of California. To estimate future climate changes, it is first necessary to make projections of population growth, economic product, and the technologies and organizations that will provide the estimated productivity levels and resultant energy and material flows in the biosphere. Although no “frequentist” data is possible for the future (until it actually happens, and thus is unknowable in principle), there is already considerable understanding of how physical, biological and social systems function. Combining such information into integrated assessment models allows scenario generation for alternative futures.

It is already known how to do this, but uncertainties are inherent in every step, and categorizing the uncertainties and trying to assess subjective probabilities for various elements of the scenario generation exercise is a frontier in climate assessment (e.g., Moss and Schneider, 1997 & 2000; Giles, 2002). Determining which of the major uncertainties can be narrowed by normal scientific investigation (e.g., see Schneider, Turner and Morehouse Gariga, 1998) in order to determine what is knowable, is an ongoing task of integrated assessors (e.g., Morgan and Dowlatabadi, 1996; Rothman and Robinson, 1997; Schneider, 1997).

Scenarios
To estimate a range of future greenhouse gas emissions, the IPCC convened a Special Report on Emission Scenarios (SRES) (Nakicenovic, et al 2001). Unfortunately, the very innovative IPCC SRES storylines, while encompassing a broad range of emissions estimates, did not attempt to assess the likelihood of any of the storylines they produced. This makes it very difficult for policymakers to perform risk-management exercises, since risk is classically defined as probability times consequences, and without probabilities (and their attendant subjective confidences) it is then up to the decision makers to guess what they think the policy analysts thought was the likelihood of various scenarios. Clearly that is a poorer basis for risk management decision-making–whether at federal, regional or stakeholder levels–than an honest assessment of the likelihood of various projections from the expert assessment community.

In addition to estimates of the subjective probabilities of emissions scenarios, similar subjective probability estimates of climate sensitivity must also be made–and were, again, not attempted by IPCC in its Third Assessment Report (TAR). Thus, this workshop will look to consider the joint probability of probability functions for emissions and climate sensitivity–the joint result of which would be a probabilistic estimate of climate changes in, say, 2050 or 2100, from which risk-management decisions could be considered. Fortunately, there have been a number of such joint probabilistic analyses in the literature recently (e.g., Forest et al, Wigley and Raper 2001, Schneider, 2002, Webster, 2003). However, a broader analysis of the underlying uncertainties will be a feature of the next IPCC assessment. This workshop is designed to provide a basis for that kind of analysis and will go one step further by considering a specific case study region: California–which will entrain many more consumers of this scientific assessment. This will entail downscaling assessment (e.g., Easterling and Mearns etc) as well as other methods to do cross-scale analyses (e.g., Harvey 2002 and Root and Schneider, 2003).

Climate models
It is known with a high degree of confidence that existing global circulation models estimate that temperatures in California will continue to rise with the increased concentration of greenhouse gases in the atmosphere. However, although less confidence can be attached, some of these models suggest that precipitation levels will increase well above historical levels, while others estimate a small but significant decrease in precipitation levels. This poses a challenge for scientists and policy makers because it is difficult to design robust (e.g., Lempert and Schlesinger, 2001) adaptation measures that would work under these widely divergent scenarios. At the same time, little attention has been paid to the estimation of the performance of the global circulation models regarding the large-scale oceanic and atmospheric features that determine California’s climate. Models that perform adequately for the California region may or may not show less divergence with respect to precipitation levels for California.

It would also be of interest for California to develop climate projections based on the outputs of global models that “properly” model the California region at larger scales, while at the same time, trying to assess probabilities to these regional climatic projections. At the moment most GCMs are typically as coarse as 300 km and therefore require downscaling models with resolutions of 50 km or less. However, new high speed computing efforts such as those lead by Philip Duffy at Lawerence Livermore National Lab have recently been demonstrated at resolutions of 50 km or less, but it is not yet known the extent to which they are useful to regional analysis. Sorting through the capabilities of various modeling approaches coupled with assigning probabilities to emission scenarios is at the core of this workshop.

The purpose of this meeting of experts will be to build momentum for a discussion on how the global scenarios, and their associated subjective estimates of likelihood, should be linked with the regional projections. California is an excellent test bed due to the relatively complex topography and meteorology and the influence of the Pacific Ocean on its climate. Moreover, the state is embarking on an ambitious research program, which is one of the first state-funded climate change programs in the nation. Thus, there would be high leverage in the implementation of the findings of this proposed workshop in the immediate future via the California effort, and then shortly thereafter as IPCC gears up for its Fourth Assessment Report (AR4). The California program is designed to complement national and international efforts to produce policy relevant research that the state will use in the design of mitigation and adaptation strategies. This can then serve as a model for other regions to perform assessments that fit their needs.

Key questions for producers and consumers
The design of the meeting will address the following four closely linked problems:
1. What are the range of possible SRES scenarios; how likely are they; how can subjective probabilities be assigned, and at what levels of confidence; what methods are most promising in such estimation; and how vexing is the lack of subjective probabilities for regional analyses for the California case study?
2. How well do GCMs do at producing meteorological variables at regional scales, and which of the various strategies to improve GCMs provide the greatest direct aid in improving knowledge of and confidence in the California regional modeling efforts and assessment strategies?
3. Utilizing California as a regional case study, how can the assessment and eventual reduction of uncertainties upstream, combined with improved downscaling models, better inform regional decision-making and resource management, and what are the stakes?
4. What elements of integrated analyses are essentially unknowable? Can robust strategies (Lempert and Schlesinger, 2000) be developed in spite of those aspects of analysis?

Outcomes
It is anticipated that while the meeting will have direct relevance to the work of the California Energy Commission Climate Program, the issues are of great interest to a broader community. Potential outcomes are 1) a clearer sense of how to improve GCMs and emission scenarios in such a way as to enhance their utility for regional modeling efforts from what is known today, 2) illuminating key barriers to the improvement of regional models themselves by identifying the unknown components and investing R&D into those amenable to discovering new knowledge, 3) applying 1 and 2 in the context of reducing California’s vulnerability to climate change, 4) communicating to the user/consumer irreducible ignorance –the unknowable, and 5) exploring the transferability of the California case to improved regional modeling elsewhere.

Anticipated products from the meeting will be a workshop report excluding printing costs, an environmental or scientific journal article, a popular press article, and a post session briefing on the key session findings hosted by the California Energy Commission.

Producer and consumer expertise
Expertise needed will include climate modelers & climate data analysts, oceanographers, economists, policy analysts, geographers, resource managers & practitioners, policymakers, hydrologists, risk analysts, resource managers and communication experts. Participants will include members of the US and international climate modeling centers covering both GCMs and regional modeling expertise, representatives of the SRES community, government labs and university centers. It is anticipated that members of the California Energy Commission and the Public Interest Energy Research Program will play a pivotal role in informing the meeting participants of California’s unique issues, work accomplished to date, and plans to utilize new information generated by this workshop.

In advance of the meeting, and particularly to aid in the development of concepts relevant to California as a case study, one or two discussion papers on this subject will be prepared with support of the California Energy Commission and distributed prior to the meeting.

To cover the range of disciplines needed, we anticipate a meeting size of approximately 25 participants with up to 5 from other countries. The format will follow Thursday arrivals, with 2 and half full days of session and Sunday afternoon departures. Funds will provide for planning, participant travel, meals, and lodging, archival video documentation, and report writing/graphic design/layout (not printing), and travel costs of the organizers.