Presentation at GRID/GVU Arendal 11 Jun 2007 CONNECTING GLOBAL CLIMATE SCIENCE, POLICY,

1. Presentation at GRID/GVU Arendal 11 Jun 2007 • CONNECTING GLOBAL CLIMATE SCIENCE, POLICY, • TEACHING AND OUTREACH • WITH AN INTERACTIVE JAVA MODEL • IN CONTEXT OF • RECENT DEVELOPMENTS IN IPCC AND FCCC • Ben Matthews • withJean-Pascal van Ypersele • Insititut d'Astronomie et de Géophysique, • Université catholique de Louvain, • Louvain-la-Neuve, Belgium • matthews@climate.be • www.climate.be/jcm

2. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm Recalling vision of 2001: (during early development of JCM in Copenhagen and Arendal) Personal transition: measuring air-sea CO2 fluxes in laboratory (group of Peter Liss, UEA)‏ also attending UNFCCC COPs (Geneva, Kyoto, Den Haag, Marrakech...)‏ => perceive lack of connection between climate science and policy => Develop simple interactive climate model as tool for global dialogue policymakers need to know sensitivity to options, not 'fatalistic' predictions => initial focus on flexible stabilisation scenarios (contrast to IPCC-SRES)‏ core science based (then) on IPCC-TAR useful to science-policy advisors (Denmark, Switzerland, Belgium...)‏ “the ultimate integrated assesment model is the global network of human heads”

3. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm Recent Development of JCM 5 in UCL-ASTR (see www.climate.be/jcm)‏ New adaptable structure/interface using Java 5 Update of core science from IPCC TAR => AR4 in progress More complex modules developed for specific research projects Has been applied to research applications e.g.: Stabilisationunder Uncertainty (remaining within EU 2C limit) Probabilistic Economic Risk Analysis (Climneg project)‏ Attribution of Contributions to Climate Change Past & future Land-Use Change emissions Aviation emissions of CO2, NOx, contrails and cirrus (ABCI project)‏ But still interactive / good for teaching explore the sensitivity to policy options, scientific uncertainties, risk / value assumptions, just by adjusting parameters with a mouse instant cause-effect response on linked plots from emissions to impacts easy to save model setups, plots, tables etc. available online, open source, documented (earlier versions were translated... update in progress) used for university courses in several countries Speed and flexibility useful for both interactivity and probabilistic / scenario analysis

4. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm Java Climate Model: History of Development

5. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm We still need a range of model complexities... Simpler models are still important, GCMs ( or even ESMs) can't do everything JCM defies the trend towards using only high-resolution GCMs, supercomputer networks but... “a chain is only as strong as it's weakest link” e.g .scenarios, impacts, communication + computing power didn't yet resolve uncertainty => still need probabilistic risk analysis whilst making more transparent the sensitivity to risk/value assumptions Research applications made JCM more complex, (GRID might say too complex for effective communication). Others say such models are too simple. But policymakers can't use GCMs, and want to create diverse scenarios If scientists don't give policymakers simple, flexible relevant tools,policymakers will create their own even simpler models (e.g. Brazilian proposal...) or “back-of-the-envelope” interpolations missing all feedbacks and nonlinearities Need to ensure quality of simpler models used for policy -relevant analysis...? (e.g. ACCC/MATCH process on attribution of contributions to climate change => recent meeting in Cicero)‏

6. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm A range of model complexities....

7. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm Synthesis by connecting reports? Examples from IPCC AR4: below: AR4 WG2 Table SPM-1: above: AR4 WG2 TS4: temperature as function of CO2 stabilisation scenario and time below: AR4 WG3 Fig 3.25 mitigation costs as a function of CO2eq stabilisation level

8. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm BUT making such synthesis based on single indicators can be misleading, for example: Climate Change ImpactsMitigation costs not just a function of Global Average Temperature level, CO2 concentration but also depend strongly on: socioeconomic baseline value assumptions in aggregation over space, time, sector & risk timing of warming, timing of investments, learning by doing regional effect of short-lived gases & aerosols mixture of gases, flexible mechanisms, etc.

9. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm IPCC Scenarios - AR4 WG1 concept that GCMs should do everything was inefficient way to compare scenarios => too few scenarios were run – 3 SRES are not enough! (simple model still used for others)‏ Policymakers need mitigation scenarios and to see the sensitivity to options (marginal effects)‏ => GCMs should parameterise simpler flexible models New IPCC Scenario Process towards AR5 (meetings in Laxenburg, Sevilla, Noordwijkerhout)‏ agreed that using special reports as a data interface between models too inefficient! => “new” parallel process concept to save time: define simple stabilisation scenarios in the middle of cause effect chain (CO2eq concentration / forcing)(at least three to cover full plausible (>likely) range and so GCMs identify nonlinearities in climate response and impacts)‏ GCMs => forward to climate, impacts, adaptation Socioeconomic (& Biogeochemical?) models => inverse calculation to emissions and mitigation Challenges of this approach:how to take account of cross-cutting feedbacks...? climate change => soil respiration, plant growth, methane release... climate change impacts => population, economic growth(when these are between separate models/processes)‏ Integrated models might do it better... .

10. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm JCM already demonstrated this approach: Example below from presention of Matthews & VanYpersele at WCCC 2003 Moscow, also to European strategy meeting Firenze Stabilisation under uncertainty: fixing a concentration or temperature (EU 2C) target: Defining the scenario by concentration or forcing spreads the cascade of uncertainty more evenly:

11. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm JCM can also be used to explore economic optimisation (Risk Analysis integrating over uncertainty)‏ - Belgian project Climneg II Make transparent the sensitivity to different ways of aggregating over... space (regions, intra-generational equity), time (discounting: intergenerational equity), risk (risk-aversion)‏ sector (comparing different types of impacts)Similar approach to Stern report But need better mitgation and impact cost functions (chain is only as strong as the weakest link)‏ => will return to this in new AR4-version

12. Java Climate Model, Live demonstration of the model: Note: the slides that follow were not shown at the side-event, they are just example snapshots for the online copy. Demonstrate webstart 1. 9 plots, show everything connected to concentration 2. stabilise temperature, change GCM => effect on emissions 3. core science – ocean layers, RF etc., cc feedback, AR4 GCMs 4. land use change, past and future 5. responsibility – brazilian 6. aviation emissions 7. regional emissions => regional impacts 8. interactions map and tree, also doc on click

13. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm

14. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm

15. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm

16. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm

17. Ben Matthews matthews@climate.beinteractive model:www.climate.be/jcm

18. Java Climate Model 2001-2007: Reflections Making interactive model tougher than making papers! Classic process model => papers : (One task at a time)Modeller sets assumptions, fixes model, runs once (can be slow), selects best data, explains results in sequence Interactive model: (Multiple applications)User changes assumptions, model adapts quickly, user selects any data, should be self-explanatory, in any order Also slower to expand... 100s of adjustable parameters => infinite combinations, impossible to check all Add new items => interactions grow expontentially => logarithmic pace of development... Funding for specialist research projects not overview / outreach (although “the chain is only as strong as the weakest link”)‏ => tools become more convenient for experts rather than for public / stakeholders Nevertheless... Good feedback, users appreciate JCM Fast & Flexible => can also iterate 1000s of combinations(e.g. to make probabilistic risk analysis more transparently)‏ Structure robust, modular, open-source, scope for expansion JCM was a “proof of concept” in 2001, now more complex... but no great breakthrough in Science-Policy interaction. Should we continue?

19. Coming soon: Anticipate AR4-based synthesis version (JCM6) by autumn 2007 Joint project with IVIG (Rio de Janerio) – global => regional policy. Apply to IPCC new scenarios processes (connecting, interpolating) Java-6 => scripting languages (demonstrations, automated analyses)‏ Structured open-source project ? ( “parallel processing...”) “The ultimate integrated assessment model is the global network of human heads” Experiment and adapt JCM – for both research and outreach www.climate.be/jcm let's work together to improve the science <=> policy interface