“Given the inherent uncertainties in climate models, should climate mitigation and adaptation policy rely on the models?”
For the third role play the contributing roles were:
- Scientist 1 (Climate modeler)
- Scientist 2 (Model sceptic)
- Politician 1 (Decision maker of green-party)
- Politician 2 (Decision maker of a more conservative party)
We as scientific experts in the field of climate sciences argued for modeling. We argued that model results should be trusted for proposing mitigation and adaptation policies. Moreover, in order to address the problems associated with climate change, climate modeling is the best possible approach. On the other hand model sceptics argued that models are not the right tools for making mitigation and adaptation policies. Furthermore, they argued that model results should not be trusted because large uncertainties are associated with these results and due to this reason models have failed several times to predict future. They gave an example of 2007 financial crisis when economic models failed to predict the crisis.
We as climate modelers raised an interesting question, that if climate sceptics believe that model results should not be trusted then what other alternatives they propose?
Climate sceptics accepted that models can explain the trends of the past climate accurately but this does not mean that models can also project the future climate trends with accuracy.
Politician 2 (Ökologisch-Demokratische Partei) were on the same line with model sceptics. They believe in climate change but do not trust climate models for mitigation and adaptation policies. On the other hand politician 1 (Decision maker of green-party) supported us (climate modeler). They argued that model results should be trusted in order to design mitigation and adaptation policies. Politician 1 also argued that models do have uncertainties and they are not precise but if we do not use model results for policy making then what other methods should we use to take decisions?
Politicians 2 gave an example where the policies were based on climate models which lack the socio-economic impacts. The example was of Germany, who failed to meet its target related to the reduction of greenhouse gases by 2020. They argued that policies should focus on the well-being of German people and in order to meet the energy demands of the growing population Germany should rely on renewable energy. Politicians 2 also made an interesting point regarding the timescales. They argued that model projections have longer time scales while political parties, who have to take actions and design policy regarding mitigation and adaptation have less time.
We as climate modelers experienced difficulty in the whole discussion in providing arguments related to social and economic models. Model sceptics were giving examples related to the failure of economic models but our role was climate science modelers so we struggled regarding this part of the discussion.
- Politician 1 concluded that, they will keep listening and keep supporting climate modelers because it is one of the best way to design mitigation and adaptation policies.
- Politician 2 concluded that, policies should be based on observations, local environment and on the basis of the phenomenon that we know.
- We as Scientist 1 concluded that, rather than negating and rejecting the models completely we should take useful aspects of the models.
- Scientist 2 concluded that, they accept the model results for past trends but future projections are questionable.
For assessing the usefulness of models in projections of future developments, it is necessary to remind ourselves of the different types of uncertainties: there are epistemic uncertainties involved, e.g. insufficiently refined models, inadequate input data, etc. We would assume that these uncertainties have been minimized to the best possible degree, such that we can say that a projection is based on best practice.
Then there is the aleatoric uncertainty, which could be due to the true unknown factors: how will societies react to unforeseen perturbations, who will lead, which technological solutions or developments will appear, etc. This type of uncertainty can of course render results of future projections questionable. But should we completely abstain from such powerful tool? Modelers try to include such perturbations by stochastic noise, and the models are tested for their sensitivities against such unknown perturbations. So, even if we do not know such effects, we can — to some extent — incorporate their effects.
After all, our own personal, political, economical decisions are taken, based on an implicit modeling approach that we call experience!