Hello everybody and welcome to C&F Talks. Today I have with me Professor Stephan Harrison, who's the Director of Climate Change Risk Management and Professor of Climate and Environmental Change at Exeter University. Stephan's going to be joining us at the Managing Physical Climate Risk Summit which is being held in London on the 10th of March. Stephan, welcome.

Nice to be here, thank you for inviting me.

Great to have you with us. Let's turn to our first question.

How useful at all is scenario planning for financial institutions and corporates in relation to physical climate risk?

Well, it's a very good question and if done properly they're extremely useful, because it engages people into the issue of climate change. It gets people focused on the fact that climate change is happening and that over time it's going to produce a whole range of physical impacts which all organisations, not just commercial ones, are going to have to take into account. The problem is how you construct a scenario.

How do you construct a reasonable and a realistic scenario for future climate change? What do they look like? So essentially this has to be done, I think, with this has to be a sort of a joint enterprise between commercial organisations and people who understand climate change and climate change scenarios.

All our scenarios are going to be based on climate models and unless you understand the ways which climate models work and the projections that they produce, I think that climate scenarios have to be treated very, very carefully because what you don't want to do is produce a climate scenario which is unrealistic or very unlikely and then make decisions based on that. So, I think it has to be done carefully.

Yeah. That really leads on to my second question.

In terms of using mathematical climate models, as you say, clearly you're expressing a sort of degree of caution as to how it's done. What are the most effective methods for those models and how can business leverage these methods and what are the sort of limitations?

I suppose a further somewhat related question, you know, climate change seems to be accelerating so how do you ensure that you have the range of assumptions which are realistic for businesses that to come to their decisions?

Well, I guess there are two ways of looking at that. First, and I'll talk about, the first thing I'll talk about really is the fact that we use sort of mathematical and statistical models in different ways. So, one of the things that businesses and organisations want to know is to what extent is that last flood we saw or that last windstorm we saw, how likely is that to happen again?

What do we mean by an extreme? How do we define what an extreme event is? And the way of doing that is we use probability theory in climate science.

We understand magnitude and frequency relationships, so how big event was it and how frequent was it? And we use things like extreme value analysis. So, these are the statistical models we use but they're full of, they're full of uncertainties and the uncertainties include how do we know that a large event is very unusual in the climate record?

And the only way you can assess that is by having a good understanding of what we call paleoclimate, the historical record of climate events. If you don't have, if you don't have that, if you don't know what the past looked like, you can't then say what, whether that event was unusual or extreme. So those statistical models that we need to understand and then we have climate model projections which are based on statistics, mathematics and physics and those, the uncertainties and those have to be assessed.

And I can, if you want me, if you want me to, I can talk about uncertainties in climate models but that then feeds back into the first question which is about scenario planning.

Yeah, so would I be right in saying if used properly, it's a useful guide but if not, you know, what's the plus or minus variance in these sort of models? What sort of range would you apply to that?

If you're talking about climate models then, you know, essentially the only way we can understand the future is by using projections but as the famous physicist Niels Bohr said, prediction is very, very difficult especially about the future. So, climate models don't really give us predictions, they give us projections. So, there's a, there's an important difference between the two.

A prediction is basically saying we pretty well know what's going to happen next year, next 10 years. A projection which a climate model produces is saying if you produce, if you produce this amount of greenhouse gas into the atmosphere and the atmosphere behaves like we expect it to, then these are the sorts of range of temperatures and precipitation changes we would expect. Now there are real problems with uncertainties though in climate models and those have to be communicated I think properly to end users.

Yeah, so people need to properly understand what they're doing, I suppose.

But one area that they're very focused on, particularly perhaps in the UK, is flood risk and to what extent do you integrate hydrological modelling and how does that feature in the climate risk management strategies that you recommend?

Well, absolutely. The interesting thing about flood modelling is it's, flooding is associated with changes in rainfall and the climate models are absolutely great for some things. They're great to understand what future temperature will look like but they're not very good at understanding what future precipitation will look like, especially at the small scales.

And the problem is that most organisations do want to know what about future flood risk but they're very, but those assessments, what the organisations want those understandings to be at the small spatial scale because that's where their infrastructure is or that's where their supply networks are, and the climate models are very poor at doing that. So again, we use a probabilistic assessment of future flood risk, and we use flood models. Now, flood models are great, right, so we use flood models all the time, but for an organisation, if they're interested in flooding at small scales, they can be quite expensive and quite difficult to run.

Yeah, yeah.

And they're looking for, in terms of the opportunities and really more the challenges for businesses as they try to use scenario planning and integrating climate risk into scenario planning, what's your advice to them in terms of how they should use these models?

Well, okay, that's, I could spend several hours talking about this, but I probably won't. One of the issues is that climate models give you a good understanding of the future climate, right, and the range of types of the sorts of events and the range of outcomes we might expect in a warming world. But most businesses, but all impacts in fact, are essentially not climate, really.

Many of them are associated with how the earth's surface responds to climate forcing. These are not, so understanding climate change is only part of the story. What you need to understand is how climate change will drive things like flooding, soil erosion, landslides and these are all issues which have many more degrees of freedom than the atmosphere has.

Yeah.

And so, we really need to integrate properly climate change with the responses for how earth's surface systems respond and that's a really difficult thing to do. The other thing I would say is that people tend to, when you're using climate models, you really should use as wide a range of climate models as you possibly can and that's called a sort of ensemble, that's the ensemble of climate models.

And what people tend to do, and I think this is wrong, is without thinking about it, use the mean of the ensemble, so the mean climate response for a whole range of models. And if you do that or if you just use one or two models, which also people tend to I think that leads you into mismanagement of future climate and misunderstanding of future climate impacts and that then produces scenarios which aren't really fit for purpose.

Maurice

So, when you're using these models, as you say, you know, people are interested in, for instance, knowing about local flooding, whether it's by chains or their factories or whatever. At the regional level, so if you can't have that degree of scrutiny at the sort of pinpoint level, if you like, can you, sort of, differentiate between regions, you know, what is the geographical area you can differentiate between when you're using these models?

Yes, yes.

Clearly you're interested in that.

Yes, routinely we use downscale climate models and assessments from downscaling, so you go from a large, the sort of a typical model of, say, a global climate model, a GCM, will produce projections of, say, 100, 150 square, you know, kilometres square. But those are, of course, much too broad, you know, broader set of projections for people who are interested in climate change in their local region. So, we can use downscaled climate projections and there are pros and antis for doing that, but those are routinely used and the European Union, we use those widely across the globe.

The European Union funded bodies have produced what's called CORDEX, Coordinated Regional Downscaling Experiments. So CORDEX type models can be used globally, and we use those globally. Again, we have the problem of relying on one, on a handful of models, though, and the only way around that, I think, is to use, is to assess the ensemble, the shape of the projections associated with a range of models.

If you just use one or two, then that is likely to lead you into a set of unrealistic scenarios.

OK, final question, because I know we're running out of time. So, when we look at this, are we talking about a mixture of an art and a science? How do we look at it?

Well, with all probabilities, we are looking at clearly hard science, right? So, the science of the atmosphere and the science of climate modelling is based on physics and physical chemistry. So, there's hard science behind this.

But when you're trying to assess what the future world is going to look like, then it's down to probabilities. But banks and organisations, financial organisations, use uncertainty all the time. So that's something which is...

Yeah of course. That’s what finance is, you know, the calibration of risk.

Yeah, of course. And so therefore, they're very well, very well prepared, really, and trained in using the uncertain future projections and treating probabilities in a proper way.

So, I don't think it would come as any surprise to businesses and many organisations that the climate futures are highly probabilistic and uncertain.

Yeah. So, in short, you need to know what you're doing and how you're doing it and understanding how applicable it might be, given the percentages. Stephan, thank you so much for sharing those thoughts with us. We're very much looking forward to hearing more at the conference.

For our viewers, that again, it's the Managing Physical Climate Risk Summit, 10th of March in London. Further information available on our website, www.cityandfinancial.com. We very much hope to see you there. Thanks, Stephan.

Thank you very much.