Rethinking the science of climate change
Reported by Dr Carmen Smith, Engagement Coordinator, Centre for Science and Policy
In June, the Cambridge Zero Policy Forum met for the first of four reading group sessions, on Five Times Faster: Rethinking the Science, Economics, and Diplomacy of Climate Change, by Senior Fellow at the World Resources Institute, Simon Sharpe.
The book argues that limiting warming to 1.5°C means decarbonising the global economy five times faster this decade than we have managed over the last two decades, and this means changing the way we do science, economics, and diplomacy. This session focused on the first section of Simon's book and the role of science, and in particular, climate modelling, in our fight against climate change. Professor Emily Shuckburgh (Director, Cambridge Zero) provided opening remarks on this first section of the book.
In this section, Simon’s main question is whether the information produced by the climate modelling community is best suited to making policy decisions. He argues that because climate scientists predominantly focus on producing predictions of climate change under future emissions scenarios, there is much uncertainty and the risk of the most extreme potential outcomes, ‘the worst-case scenarios’, are often under-reported or ignored. Instead, he argues that we should be treating science as a risk assessment process with a strong focus on the worst-case eventuality.
A culture of prediction
Simon compares the way evidence-based decisions are made in the climate policy space with other policy areas such as intelligence, national security and healthcare. In contrast, climate science grew out of weather forecasting research, so there has been a strong culture of prediction, expanding timeframes out from the prediction of weather patterns. This tradition of plotting factors against time to say ‘this is how the world will be tomorrow’ differs from Simon’s suggested approach of ascertaining the likelihood of various risk scenarios based on current trends.
Climate risk assessments are less prevalent around the world than predictive models, with their focus being mainly on the risk of climate ‘tipping points’ being exceeded. These were described by Emily Shuckburgh in the Donald Rumsfeld categorization of ‘known unknowns’, which are fundamentally unquantifiable and therefore difficult to represent in a useful way to policy makers. Simon criticises scientists for avoiding challenging topics, such as the coupling of tipping points, in favour of a more conservative body of science based on evidence that can yield robust findings and conclusions. In avoiding these difficult topics, Simon questions the very basis on which the scientific establishment conducts itself.
The UK also conducts a national Climate Change Risk Assessment every five years and includes evidence from around the UK. However, this national study is criticised as being too detailed for use by policy makers in the UK, and Simon calls for a more concise assessment for policymakers, which highlights worst case scenarios: the question remains as to whether the scientific community in the UK would prioritise producing such a document.
A multifaceted problem
One example that Simon cites from the UK Climate Change Risk Assessment relates to the question of whether the Thames Barrier in London should be reinforced in the context of future climate change. This question has been answered well as it is a single and specific problem. A land management plan is developed to determine the minimum that government needs to spend to ensure that the Thames does not become unmanageable in the future. If sea level will rise a certain amount, the government needs to spend a certain amount e.g., £100bn over the next 10 years. Risk assessment is used to determine what will happen if £100bn is not spent i.e. will it cost more in the future? This risk determines the amount of money spent, and the problem is re-assessed in 10 years, with the aim to avoid the Thames becoming unmanageable.
The problem on a larger scale is that climate change involves multiple impending problems (flooding, migration, wildfires etc.) that are multifaceted and interconnected, making it difficult to conduct this kind of risk management analysis as it is fundamentally context-dependant. How do we measure our progress against resolving these multiple problems simultaneously and how do they affect each other? Interestingly, risk management theory in businesses categorises these as ‘systemic risks’, which multiple organisations contribute to, making them difficult to track and analyse for this reason. Furthermore, no single actor can really take ownership of them as it is difficult to draw a line around the problem, and therefore the solution remains allusive. Indeed, one of the major pushbacks of engaging in a risk assessment at COP is the question of how to decide which contexts are the most important and which countries would be considered a priority in this process.
The role of global average temperature
All global climate agreements are couched in terms of global average surface temperature. For example, the Paris Climate Agreement is based on keeping the temperature below an average 1.5°C increase. However, Simon argues that scientists focus too much on the average rather than on the probability band surrounding it, including the highest real risks that we are taking by not acting. For instance, although average global precipitation may not change, regional precipitation may change dramatically leading to floods and droughts. This is an example of why science has failed to provide concrete risk-based information to governments: when the discussion is framed as keeping the global average temperature below 1.5°C increase it obscures policy-relevant actions that can be taken at a local level. While scientists are trying to answer questions around what is going to happen to the climate in the future, policy makers and society are trying to answer specific policy-relevant questions such as, will climate change be fatal for the South East of England or for Eastern Europe? Although global averages are easy for science, they are not always the best guide for political action.
Interestingly, the relationship between climate modelling and economic modelling (the next section of Simon’s book) is precisely where global average temperature becomes pertinent, and the failure of climate risk assessments becomes clear. This is because most economic models use global average temperature as a proxy, although many of the impacts of climate change are not linearly related to global average temperature. It is a useful metric on which to base policy, as the hotter it gets the bigger the economic impact. However, this global proxy does not relate to the localised impact of increasing temperature, which can be very non-linear. We want to know how liveable certain areas of the world will be in the future, but scientists are so focused on the 1.5°C proxy that if economic policies are based on this, then we need to ensure that it is correct. It was noted that AI will be important in the future to understand fluctuations in global average temperature and how the economic system and other systems change in response to this across the world.
Risk at a national and international level
It was noted in the discussion that on a national level, countries conduct a cost benefit analysis with regards to the economic risks of climate change mitigation. As with other economically developed nations, China, for example, first engaged in the climate change discussion in 2013, and a decade later is more proactive internationally, however taking a leading role in combatting climate change is still seen to carry a risk of damaging China economically through incurring additional costs to the country. It is therefore important to consider the positive and negative consequences to international agreements of approaching climate change mitigation as a national risk assessment process.
On an international level, the existential risk from climate change is best communicated in the latest IPCC Report, where the Five Reasons for Concern are outlined. These are 5 categories of major global reasons for concern, tipping points being one of them, and includes expert opinions and evidence with a risk assessment colour rating of yellow for the lowest risk and purple for the highest. Compared to other similar reports, the IPCC report also has a very clear structure and way of communicating uncertainty, with one set of criteria based on quantifiable uncertainty, and another based on ‘confidence’, so any findings are cited as having high, medium or low certainty, and high, medium or low confidence. Emily Shuckburgh noted that this as a good example of Simon’s ideal risk assessment being undertaken at an international level.
During the discussion it was noted however, that the communication of these global risks at a national level through the media and by politicians is where much of the momentum is lost. Media commentators try to find ‘the story’, and often what hits the headlines is novelty, or in some circumstances a desire to not offend the public. But in doing so, what do scientists not communicate, and what do they fail to emphasise of the full risk profile? Simon argues that if we really want to make progress on this and rebalance communications from the scientific community, then the approach taken by IPCC report based on communicating a full risk assessment is the way to go.
Psychology and emotion in climate modelling
There is also a psychological component to risk assessment. During the discussion, the example was cited that if someone tells you that there is a 5% chance of having a fatal accident on the way to the office on your bike then it is likely that you will consider it a high risk and won't take a chance. Simon suggests that this emotional element of risk assessment, and the resulting action taken, should be more prevalent in climate science. However, given this emotional response, it was then noted how little many stakeholders care about some of the risk assessments being published, and in particular, how far into the future they need to care. This is because caring depends on how much you think your actions will mitigate the risk: to mitigate the risk of having an accident when cycling to work you can decide not to cycle and take away 100% of the risk of that happening.
When considering climate risks as an individual then, most actions will have a negligible impact on mitigating the overall risk. Even policy makers, as individuals, perhaps carry some of this psychological baggage and react in a conservative way, believing that there is not much that we can do when tackling some of the biggest global risks. As democratic governments are temporary, the policy world is far messier and more complex than the scientific world, with far greater trade-offs, and decisions that are not always well-informed or forward thinking. As governments are also less likely to be blamed for inaction than they are for taking action, there are strong incentives to maintain the status quo. The question then is even if scientists communicate risk more effectively to policymakers, how far will this change psychological biases in order to make progress. Can scientists communicate to politicians in a way that makes them think beyond this four year cycle? Notably scientists at the University of Cambridge have been working on the type of studies Simon is advocating for, however it is unclear how far the Cambridge science community can influence a change of thinking (and feeling) around risk within government.
Implications for climate modelling going forward
If the climate modelling community are to move away from predictive modelling and towards a culture of risk assessment, as Simon advocates for, some practicalities must be considered. It was raised that most climate models are currently built for prediction, working forwards, whereas risk assessment models work backwards, asking how likely it is for a particular outcome to occur. Predictive models can be adapted to risk assessment, but due to their different methodological aspects, building models that can do both is a computational challenge.
In the scientific community there is also a certain loss of information that goes into writing papers to be published, especially if several papers are written on one topic. When using a single database, this often requires different groups of researchers to have access to the original data in order for the model to maintain fidelity. This is important when building new climate change models that can conduct risk assessments based on prediction models. As a useful example, Ben Goldacre has done much work on this in the field of public health, reversing the loss of fidelity in going from a raw genomics database to reviewed papers by getting access to the original data.
Finally, it was agreed that a limitation of the scientific community is the tendency to work in silos. For example, research in the wine industry, into how climate will affect crops, receives significant funding in the UK. This has informed the planting of vineyards in the UK 20 years ago. It was suggested that this type of information could be used to inform research conducted by those studying the effect of climate change on barley and other crops. Similarly, it was suggested that local level actions to mitigate climate change should not be side-lined by international efforts such as the IPCC report. It is important for policymakers to work with people for whom these risks matter. This challenge will be taken up in the fourth session of the Reading Group on ‘what can be done’ when issues around co-production and policy evaluation will be discussed.