Wp2

Tyndall˚Centre for Climate Change Research

Integrated Assessment Models

Mike Hulme

March 2001

Tyndall Centre for Climate Change Research

Working Paper 2

Integrated Assessment Models Notes from the joint Tyndall Centre – PIK – ICIS meeting on Integrated Assessment Models Potsdam, 7 March 2001

Mike Hulme Tyndall Centre for Climate Change Research School of Environmental Sciences University of East Anglia Norwich NR4 7TJ Tyndall Centre Working Paper No. 2 March 2001

Introduction These notes provide a summary of the discussion and conclusions of a meeting on Integrated Assessment Models (IAMs). The meeting, convened by Professor John Schellnhuber, was an opportunity to provide early input into the newly emerging Integrated Assessment research theme of the Tyndall Centre for Climate Change Research. Eight representatives from the Tyndall Centre, four from ICIS (International Centre for Integrative Studies), and about 15 from the PIK (Potsdam Institute for Climate Impact Research) participated in the meeting (Appendices 1 and 2). The main aim was to bring together three established and emergent players in the European IAM theatre for intensive brainstorming on the way forward. Four major tasks need to be tackled in order to make substantial progress during the next three to five years, namely: (i) (ii) (iii) (iv)

generating concepts for the third generation of IAMs; identifying and realizing novel applications with existing and/or forthcoming models; clarifying the relationship between IAMs and Regional/Sectoral Simulators; and determining the division of construction and operation labour within the national, European and global IAM communities, respectively.

Most of the available time was actually spent on item (i), but the other items were implicitly taken into account — especially through the breakout groups (Appendices 3–5). The question of cooperation (item iv) received the attention it deserves as it seems rather impossible today to build a state-of-the-art IAM from scratch as a stand-alone institution — the splendid IMAGE story will not be repeated. So meta-modelling and distributive schemes are likely to dominate the medium-term future, at least. The workshop was kicked off by brief overviews of pertinent activities at the three research institutes involved. The following brainstorming phase was initiated by the convenor’s presentation of three different options for the design of third generation IAMs. These are not mutually exclusive, but each option has rather different implications for organisational structure and implementation. Their philosophies are also rather different (Appendix 6). The general discussion identified three crucial issues to be explored in breakout groups: Group 1: Group 2: Group 3:

IAM – Stakeholder interaction What new elements are needed for third generation IAMs? The architecture of a new (third) generation of IAMs.

Summaries of the deliberations of each group are provided in Appendices 3 to 5.

2

Conclusion The final plenary discussion reviewed the various concepts for third generation IAMs in the light of the breakout group findings. There was consensus that the modular approach within an international context (Option 3 in Appendix 6) is the only feasible one. This means, in particular, that an operational meta-model has to be created that is able (a) to specify the ‘demand side’ questions for IAMs to address (through stakeholders dialogues, etc.) and (b) to generate the appropriate ‘supply machinery’ as a virtual structure (through ad-hoc combination of already existing, properly processed and tuned simulation elements, etc.). There was also consensus, however, that this approach has to be enriched by delivering novel building blocks for IAMs that have been neglected so far (Option 2 in Appendix 6), and by venturing, at last, onto difficult applications terrain that can make all the difference for climate policy (for example, the consideration of singular events as triggered by global warming). The major practical conclusion from this brainstorming meeting is the overall impression that the Tyndall Centre, as a fresh and powerful force in the IAM arena, should take a lead in implementing the third generation entities. The success of this enterprise will crucially depend on the willingness of the overall IAM community to co-operate in the modular way (by providing, for example, original source codes). The Tyndall Centre therefore should seek formal collaboration with institutions such as ICIS and PIK, and possibly promote the establishment of an international IAM network or society with appropriate operating rules. The Tyndall Centre ‘Blueprint Project’ in Research Programme 1 (www.tyndall.ac.uk/publications/fact_sheets/it1_03.shtml) must take these perspectives into account, and may serve as a pilot venture.

3

Appendix 1 Agenda Joint Tyndall Centre–PIK–ICIS workshop on Integrated Assessment Modelling Potsdam, 7 March 2001

Session I • • •

Warming up: Brief Overview of Recent IAM Activities

ICIS: J. Rotmans et al. PIK: C. Jaeger et al. TC: J. Palutikof et al.

9.30 – 10.30

15 min 16 min 15 min

Discussion Session II

General Brainstorming: Towards European Integrated Modelling (EURIAM)

10.30 – 12.00

Task 1: Approach to 3rd Generation of IAM Task 2: Novel Applications Task 3: Relationship with Regional and Sectoral Simulators Task 4: Division of Construction and Operation Labour 12.00 – 13.00

Lunch Break Session III

Breakout Groups for Respective Tasks

Coffee Break Session IV

13.00 – 14.30 14.30 – 15.00

Concluding Discussion and Agenda Setting

4

15.00 – 16.00

Appendix 2 List of Participants International Center for Integrated Studies (ICIS): Marjolein van Asselt Kasper Kok Pim Martens Jan Rotmans

Potsdam Institute for Climate Impact Research (PIK): Ottmar Edenhofer Peter Frank (GMD) Hans-Martin Füssel Andrey Ganopolski Hermann Held Andreas Hoheisel (GMD) Cezar Ionescu Carlo Jaeger Rupert Klein Elmar Kriegler Maarten Krol Gerhard Petschel-Held Helge Rosé (GMD) Hans-Joachim Schellnhuber Frank Wechsung

Tyndall Centre for Climate Change Research: Peter Allen Mike Hulme Jonathan Koehler Jean Palutikof Sarah Raper Tim O’Riordan Simon Shackley John Shepherd

5

Appendix 3 Summary of Breakout Group 1: IAM – Stakeholder interaction

Summary The aims of the discussion were to: •

identify the role discussions with stakeholders should serve with respect to the creation of third generation IAMs;

•

analyse the nature of the ‘rules of engagement’ amongst modellers, amongst stakeholders, and between modellers and stakeholders;

•

discover better ways in which to integrate the perspectives of stakeholders into the hardware of IAMs, so that both are improved in terms of mutual understanding, and predictive capability.

On Stakeholders The group concluded that there were three categories of stakeholder: •

stakeholder as advisor: namely those whose knowledge, understanding or experience are vital to the validation and accuracy of the data being used for IAMs;

•

stakeholder as actor: namely those whose behaviour and outlooks are part of the models themselves for they are the players who shape the assessments;

•

stakeholder as user: namely those who could, or actually do, make use of IAMs for a variety of purposes — to provide perspective, to justify actions, to reinforce prejudices, to suggest fresh interpretations, to learn more (or many other possible reasons).

On ‘rules’ In general, the modeller uses quantitative rules for analysing information and forging assessments. For the most part, these ‘rules’ are based on rational principles of calculation. Stakeholders use many, many different kinds of decision rules. They may draw on belief, intuition, social consensus, received wisdom or happenstance connections.

6

Consequently, the manner in which knowledge is acquired and processed will differ, possibly greatly, between the stakeholders and the modellers. There is therefore more reason for greater understanding and empathy between the two sets of players.

Stakeholder knowing Various models of ‘knowing’ can apply to stakeholders. Some of these are based on rational economic (a neo-classical economic) principles. These include the formation of preferences, the development of expectations, the social context in which these are determined, and the operation of impediments, bounds, or constraints. Broadly this is the ‘bounded rationality’ approach. But understanding is also influenced by personality, by institutional setting, by moral outlook and by the influence of social networks and influential authority figures. In addition, ways of knowing are driven by ‘instinct’ and ‘intuition’, namely patterns of belief based on certainties that are located in mental maps of attitudes and outlooks. So the modeller may need to find ways of characterising behaviour and perceptions on the basis of these ‘soft’ processes of learning, and build these into the patterns of prediction in creative ways.

Co-evolutionary IAMs This suggests that IAMs may need a fourth generation of co-evolving relationships with stakeholders. This fourth generation would be set in the frame of innovative third generation models and modules, as described by breakout groups 2 and 3 (Appendices 4 and 5). But they would be designed specifically to interact with stakeholders, and hence be quite novel in the use of information, the display of assessments, and the characterisation of equity. For stakeholders as advisers the first generation of IAMs could widen the selections of stakeholder categories, and engage more deeply for validation and correction. This would burden and deepen the basis of stakeholder input, and ensure a more comprehensive generation of models. Alternatively, the fourth generation of models could go modular and thematic, looking at particular sectors such as health, tourism or water, or important strategies (e.g. carbon sequestration) and engage stakeholders as advisors in much higher ‘epistemic communities’, but much wider validation and guidance roles. The pursuit of both of these options is quite possible. For stakeholders as agents or actors the connectivities with modelling are altogether more complicated. There are no clear rules for engagement. There is a case for comparative micro studies of how actual actors (e.g. electricity policy makers, business managers and NGO activists) might react to particular models of outcome and impact on different groups locally, regionally and globally. There is also a case for looking more at those who gain from future climate change, compared to those who may lose (e.g. renewable energy firms as opposed to local fossil fuel based energy corporations and associated employees) to see how they would judge the images and outcomes of electricity futures. The aim would be intensive, small scale,

7

pilot studies, comparatively undertaken to build experience and confidence in the various kinds of partnership. Subsequent approaches may widen the frame of sectoral selection to bring in selective themes (e.g. water and agriculture, or sea level rise, coastal management and tourism). The point is to begin small, experimentally, and as a pilot study, and share experience through collaborative endeavour.

Stakeholders as users Stakeholders who use, or who might wish to use, IAMs come in various forms. One interpretation would be the level of need and involvement. A possibly typology might be as follows: 1. Strategic users: policy advisers, ministers, specialised ‘think tanks’. Integrated modular IAMs with clear ‘what if’ outcomes for various policy options. 2. Integrative users: policy analysts, international negotiators, senior NGO officials, business leaders. Interface modules of economy technology and society: thematic modules on particular sectors or policy issues. 3. Managerial users: community leaders, wide array of NGOs, business executives. Deliberative scenarios, virtual reality imaging, creative ‘what if’ analysis. 4. Educational users: citizens groups, academic generally, civic leaders, business groups, international bodies dealing with sustainability issues scenarios. Story telling and creative virtual reality. 5. Moral users: anyone. story telling, role playing, theatre, visualisations

8

Appendix 4 Summary of Breakout Group 2: What new elements are needed for third generation IAMs?

Questions addressed: 1. What new topical modules are needed? ↓ LINKAGES 2. What new methodological modules are needed? ↓ 3. What general guidelines and frameworks can be established for IAMs?

Question 1: What new topical modules are needed? Criteria:

Is it important / policy relevant? Has it been done? Is it sensitive? Can it be described scientifically?

Sectoral topics Tourism

fits the criteria e.g. 30% Austrian GDP

Aviation

a way to disaggregate complexity of tourism

Aquaculture + Fisheries

growing importance

Health

still first generation; issues of spatial scale and spurious precision; adaptation; cost-benefit analysis

Water

large body of research as basis for IAM already exists

Built Environment

link to pollution and health; urban heat islands; transportation

Coasts

introduces regional dimension

[Biodiversity

feasibility issue]

9

Non-sectoral topics Globalisation Extreme weather events and climate variability Abrupt climate change

QUESTION 2: What new methodological modules are needed? →

Agent-based modelling. Important for problems with strong human behavioural issues, e.g. tourism, health

→

Should be a stronger interaction with mathematical community, and to social scientists, and psychologists, to identify emerging and appropriate tools

→

Require tools capable of autonomous behaviour which can themselves adapt, e.g. neural nets, AI, cellular automata

→

Evaluation of tools is required

→

Adjoint approach to uncertainty analysis and for modelling abrupt climate change

→

There remains no good methodology for dealing with the multiple-scales of IAMs

→

Methodologies for modelling adaptation are also lacking. Monte Carlo-type approach? Or deterministic equation set.

QUESTION 3: What general guidelines and frameworks can be established for IAM? →

Meta modelling approach with ‘command’ structure

→

General platform, easily useable, multiple purpose according to the ‘task’ (‘task’ identified with help from stakeholders; Breakout Group 1)

→

Could modules be designed with the preceding point in mind?

→

Issues of spatial scale, time steps, module interface (Breakout Group 3). Needs very good communication and sharing between groups, to create common platform

→

Need to test methods to quantify human behaviour, such as agent-based modelling (Breakout Group 1).

10

Appendix 5 Summary of Breakout Group 3: The architecture of a new (third) generation of IAMs

Definition of a ‘model’ Before anything else, a discussion on the difference between a model and the actual implementation of that model proved necessary (Figure 1). A model can be defined as the simplified representation of something that is not a model.

REALITY Model of something that is not a model MODEL

IMPLEMENTATION

Figure 1. Definition of a (conceptual) model. In this sense, first, second and third generation IAMs are the same. The distinction between generations, and specifically the step from current second general to third generation IAMs, is more closely related to the way the conceptual model is implemented. Differences between second and third generation IAMs Monolithic whole INTERFACE

MODEL MODULE

MODULE

MODULE

MODULE

MODULE

Figure 2. Differences between second (left) and third (right) generation IAMs.

11

The third generation differs from the second in a number of ways, listed below and illustrated in Figure 2 above. Second generation: 1. Build a conceptual model of the system; 2. Model all components yourself (possibly reinventing the wheel and reprogramming existing routines); 3. Do your experiments (if time allows); 4. Present model outcome as assessment and find a question for your answers. Third generation: 1. 2. 3. 4. 5. 6.

Define the question; Build a conceptual modular structure; Check the existing library (module database) for modules that fit in your structure; Develop new modules and interfaces for non-existing modules; Play with replacing modules by others, or plugging in entirely new ones; Assess model results (for which plenty of time will be left).

In the second generation, modules are part of a monolithic whole, and thus constrained by the model in which they function. This severely limits the applicability and transportability of separate modules. Often, the modelling or at least reprogramming from other models of all components is very time consuming to such an extent that little time remains for the assessment. In the third generation, modules are separate entities which can communicate with others. To increase usefulness to other groups of interest, every module has an interface. The interfaces are built such that use of the modules by a range of potential users, modellers and policy makers alike, is enabled. Thus, duplication of (pieces of) models will occur less and less as the number of user friendly modules with interfaces increases. Additional advantages include the loss of almost all technical constraints. If interfaces are constructed well, modules will be platform-independent and modules can be run from any computer anywhere in the world. Furthermore, issues like (dis)aggregation, interpolation, and uncertainty analysis are facilitated. Thus, we can concentrate on thinking, developing new issues and assessment, instead of merely the technical implementation.

What is an interface (in third generation IAMs)? The success of the proposed modular structure strongly depends on the design of the interface. It should ideally be in simple digital form with a few clickable options (see Figure 3). The actual module will subsequently be run on a computer somewhere. Additionally, the interface should: • • •

provide a description of data in a language understandable by humans and by machines; it should be comprehensive; easy to extend.

12

INTERFACE OF CLIMATE MODULE

Temperature Unit 1

Scale A

Unit 2

Scale B

Precipitation ……. ……. ……. …….

Figure 3. Example of the interface for a climate module.

From second to third generation The following steps could be the first in the transitional process from second to third generation IAMs: • • • •

cut existing IAMs into modules; make an inventory of all input variables that, for example, the climate module may need, and all the output variables those modules can provide; make a climate module interface plus documentation; think about interesting, necessary, and sensible couplings between the climate module and other modules, which could, for example, involve a societal debate.

Conclusion GO MODULAR!

13

Appendix 6 Options for Third Generation Integrated Assessment Models Option 1: Building a show-case model Main characteristics: • • • • •

Comprehensive; Harmonized (particularly with respect to natural and socioeconomic aspects); State-of-the-art (at appropriate level of complexity); Self-contained (particularly with respect to data base); Validated and optimised.

Possible implementation: Joint (European?) venture, concerting distributed resources (human capital, source codes, data sets, computers, stakeholder discourses, etc.)

Option 2: Providing crucial modules Major Categories: • • • • • • •

Sectoral impacts (agriculture, settlements, health, tourism, etc.); External factors (technology, adaptation, life-style change, globalization, etc.); Systems analysis and dynamics (endogenous growth, extreme events, critical thresholds, LSDs, retardation and memory effects, etc.); ‘Humanization’ (stakeholder dialogue, micro-agents modelling, decision theatre, etc.) Policies (flexible mechanisms, equity, interactive regimes, etc.); Operation (uncertainty assessment, ignorance dynamics, validation, algorithmic stability, object-oriented wrapping, visualization, etc.); Integration principles (decision frameworks, valuation schemes, sustainability paradigms, etc.).

Option 3: Constructing a ‘Demand & Command’ system Basic idea: An operational shell (see computer science) that generates — from existing stock of models, data and challenges — virtual objects and services such as: • • •

Assessment objectives; Baseline scenarios and plausible driving-force futures; Special-purpose IAM types;

14

• • • •

Model intercomparisons; Specific modes of operation (e.g. inverse or iterative); Specific modes of integration; Co-production schemes (stakeholders, decision-makers, users, etc.).

Possible platforms: European Climate Forum; Integrated Assessment Society; Scientific community at large.

Combinations of options may be possible — consider, for instance, a dual strategy based on Options 2 and 3.

15

The inter-disciplinary Tyndall Centre for Climate Change Research undertakes integrated research into the long-term consequences of climate change for society and into the development of sustainable responses that governments, business-leaders and decisionmakers can evaluate and implement. Achieving these objectives brings together UK climate scientists, social scientists, engineers and economists in a unique collaborative research effort. Research at the Tyndall Centre is organised into four research themes that collectively contribute to all aspects of the climate change issue: Integrating Frameworks; Decarbonising Modern Societies; Adapting to Climate Change; and Sustaining the Coastal Zone. All thematic fields address a clear problem posed to society by climate change, and will generate results to guide the strategic development of climate change mitigation and adaptation policies at local, national and global scales. The Tyndall Centre is named after the 19th century UK scientist John Tyndall, who was the first to prove the Earth’s natural greenhouse effect and suggested that slight changes in atmospheric composition could bring about climate variations. In addition, he was committed to improving the quality of science education and knowledge. The Tyndall Centre is a partnership of the following institutions: University of East Anglia UMIST Southampton Oceanography Centre University of Southampton University of Cambridge Centre for Ecology and Hydrology SPRU – Science and Technology Policy Research (University of Sussex) Institute for Transport Studies (University of Leeds) Complex Systems Management Centre (Cranfield University) Energy Research Unit (CLRC Rutherford Appleton Laboratory) The Centre is core funded by the following organisations: Natural Environmental Research Council (NERC) Economic and Social Research Council (ESRC) Engineering and Physical Sciences Research Council (EPSRC) UK Government Department of Trade and Industry (DTI) For more information, visit the Tyndall Centre Web site (www.tyndall.ac.uk) or contact: External Communications Manager Tyndall Centre for Climate Change Research University of East Anglia, Norwich NR4 7TJ, UK Phone: +44 (0) 1603 59 3906; Fax: +44 (0) 1603 59 3901 Email: [email protected]

Other titles in the Tyndall Working Paper series include: 1. A country-by-country analysis of past and future warming rates, November 2000 2. Integrated Assessment Models, March 2001 3. Socio-economic futures in climate change impact assessment: using scenarios as ‘learning machines’, July 2001 4. How high are the costs of Kyoto for the US economy?, July 2001 5. The issue of ‘Adverse Effects and the Impacts of Response Measures’ in the UNFCCC, July 2001 6. The identification and evaluation of suitable scenario development methods for the estimation of future probabilities of extreme weather events, July 2001 7. Security and Climate Change, October 2001 8. Social Capital and Climate Change, October 2001 9. Climate Dangers and Atoll Countries, October 2001 The Tyndall Working Papers are available online at: http://www.tyndall.ac.uk/publications/working_papers/working_papers.shtml