Has climate change made winter floods in the UK more likely?
We’ve been working recently with a group of scientists led by Oxford University and the Centre for Ecology and Hydrology, a national research centre for hydrological sciences, alongside other UK and European institutions, to investigate the possible influence of human greenhouse gas emissions on the risk of flooding. This research has just been published in the journal Nature Climate Change, where it will reach a wide range of scientists across the world. Of course this scientific paper is a technically detailed publication written largely for a specialist readership. Here we take a look at what this study means in broader terms for how we cope with flood risk in the UK.
The study concentrated on the floods in the Thames Valley in the winter of 2013-2014. The scientific team set out to answer the question “is climate change increasing the chance of floods like this happening?”, a question that was brought into focus when British Prime Minister David Cameron told Parliament “I very much suspect that it is” (a thesis echoed recently by the Met Office after flooding in Cumbria in December 2015).
So does the new paper support this hypothesis too? The short answer is yes.
Computer model experiments allow us to investigate the influence of climate change on flood risk
The study we worked on involved using sophisticated climate models to simulate the effects of greenhouse gas emissions caused by human activities since the industrial revolution. Those global changes were then fed through into more detailed models for the North Atlantic and UK, similar in some ways to the ones used for day-to-day weather forecasting, in order to look at the implications for the weather that we experienced from December 2013 through to February 2014.
In effect, the models were used to build something resembling a computer model “laboratory experiment” in which, within the virtual world simulated by the computer code, the effects of greenhouse gases could be removed from the weather in the winter of 2013-14. This allows us to study the difference between the world we live in now, where the climate has already been influenced by greenhouse gas emissions, and a hypothetical world in which the effects of greenhouse gases were held at pre-industrial levels.
The experiment was set up in a way that recognises and quantifies uncertainty about the climate and the models themselves. This analysis is done by running experiments called ensemble simulations, which in numerical modelling is rather like carrying out many repetitions of a lab experiment in order to account for the effects of slight variations or errors in things like experimental conditions, or in this case the specification of inputs to the models.
The paper shows that the unusually severe weather over southern England in winter 2013-14 was linked to particular, identifiable features of the atmospheric conditions over the British Isles. The computer model experiments show that these conditions have been made more likely by the greenhouse gas emissions than would have been the case in a pre-industrial world. These findings also revealed some new details about the way in which those emissions are influencing both temperature-related effects (warmer air holds more moisture) and circulation patterns (influencing the movement and persistence of rain storms). The results are in line with previous studies of the influence of climate change on flood risk and river flows.
But this study has also gone further than previous work by using additional models to translate the analysis of weather conditions into their effects on the risk of flooding to properties. Working closely with the Oxford and CEH scientists, we analysed data from detailed flood risk maps produced by JBA Risk Management to help estimate what a change in the likelihood of flooding attributed to greenhouse gas emissions means in terms of the number of properties placed at risk.
The headline estimate is that the additional flood risk stemming from climate change amounts to around 1,000 more properties in the Thames river basin being at risk of flooding in today’s world, compared with the situation if the atmosphere had pre-industrial levels of greenhouse gases. This is far fewer than the number of properties that could be placed at greater risk in the future, as the effects of climate change become more pronounced.
Confidence and uncertainties
So how much confidence can we have in these results?
In our analysis, climate models were employed not to make predictions about the future, but instead to help understand what has happened to our climate in the past, and what is happening now. All models are approximations, or simplifications, of reality, and so their outputs have to be assessed with care and cross-checked against evidence from observations, and theoretical or experimental knowledge from physical science. The approach used in the paper includes those checks, whilst the ensemble simulations allowed us to look at the precision of the estimates. The results showed that there remains fairly wide uncertainty about the influence of global climate change to date on flood risk, but, despite this uncertainty, the results from the model experiments say that it looks more-likely-than-not that the number of properties at risk of flooding has increased because of past and present greenhouse gas emissions for winter storms like those of December 2013 to February 2014.
Although the increase in properties at risk may seem relatively small compared with the total number of properties in the Thames catchment, and the uncertainty is rather wide, this balance of probabilities is still an important result. Another way of putting it is that although the effect of past greenhouse gas emissions is not yet strong enough to show up as an obvious trend in, say, peak river flows, the model experiments indicate that, on the balance of probabilities, we are already starting to experience some of their effects through a small increase in flood risk. That the effect is still relatively small, and is assessed as probable, rather than certain, is consistent with the fact that historical river flood flow measurements in the UK tend not to show consistent long-term trends.
The information we used to assess risks to properties in the Thames river catchment is based on very carefully prepared data and state-of-the-art computer models to predict how water flows over the floodplain. Even so, we had to make some approximations, which are discussed in the paper, in order adapt those data sets for use in the research. In their day jobs, working in JBA’s commercial businesses, our analysts would typically interrogate the same data in somewhat greater detail to work out the location and numbers of properties at risk of flooding. We could not do that in this research study because the ensemble climate modelling involved over 130,000 separate sets of results, which would simply have taken too long to process in this way. Like everything else in life, scientific research is also subject to constraints of time and cost.
However, our results give a reasonable first assessment of the increase in flood risk already attributable to greenhouse gas emissions, and it is clear, from a technical point of view, how this analysis can be improved in future experiments to add more detail.
What about Desmond, Eva and Frank?
Why has it taken until now to publish these results, after we have already had another sequence of serious floods in December 2015 and January 2016, this time centred on the north of England and in Scotland?
Part of the reason is the huge computational and data processing effort needed to run the type of modelling experiments described earlier. However this type of analysis is becoming more and more responsive. For example, initial assessments of the extreme rainfall brought by Storm Desmond suggests a similar “small but robust” increase in the likelihood of such storm events.
However, after a first set of modelling results are analysed it can be some time before results are published in a scientific journal. Scientific papers are generally subject to high levels of detailed scrutiny as part of a peer review process. This scrutiny usually triggers revisions and improvements to a paper, which, although it can be a slow process, helps to ensure that scientific findings are robust.
Another factor in this case is that the study of the winter 2013-14 Thames flooding combined several different elements in a new way by linking analysis of climate with models for the Thames river catchment and then for floodplain inundation and property risk. This kind of innovation takes time to be implemented, but then offers new insights.
Flood risk management in a changing climate
Finally, what do these studies mean for the way we deal with flood risk in the UK?
The Committee on Climate Change, an independent statutory body advising UK Government and Devolved Administrations, has already identified flooding is one of the four key areas that require urgent further action for adaptation.
Projections of future change, which inevitably involve some assumptions about future levels of greenhouse gas emissions and adaptation actions, indicate that flood risk in the UK will probably increase. In an independent review of the Environment Agency’s 2014 Long Term Investment Scenarios for flood and coastal risk management, Professor Jim Hall has noted that England’s flood defence systems can be adapted to cope with some projected changes, but this requires long-term investment at levels that are higher than recent funding.
The Committee on Climate Change has recently stated that if global warming is limited to no more than 2°C, as agreed in the recent Paris climate change talks, “then with additional flood risk management effort the UK should be able to avoid increasing flood risks and impacts”. But if approaches to flood risk management continue as they are today, then even if the Paris agreement succeeds in capping global warming to 2°C, there will be over a million residential properties at significant risk of flooding in the UK, compared with 860,000 now. (Here, “significant” means a 1-in-75 chance of flooding in any one year, which might sound like a small risk, but would translate to a 1-in-3 chance over a 30-year mortgage repayment period; see this factsheet from the US Geological Survey for further explanations of how to interpret flood probabilities).
One of the most important issues revealed by studies of long term flood risk in the UK is the changing nature of the economics of flood risk management, with a large number (around 0.3 million) of properties falling into a high risk category where community-funded protection may eventually become less affordable and harder to justify in purely economic terms. We may see increasing risks of occasional severe flood damage and disruption to land, properties and infrastructure such as roads, rail, energy, water supply and communications systems when flood defences cannot cope. More localised protection measures such as temporary flood barriers and flood-resilient building designs (sometimes called “property level protection”) could have an increasingly important role to play in helping people cope with flood risk.
When is action needed?
Recent research for the Committee on Climate Change says that increases in flood risk are projected as early as the mid-2020s, with 20% more residential properties being at significant risk (as defined above) by then if the global climate were to be on a course that would eventually lead to a 4°C rise in global average temperature by the 2080s. (For context, this figure lies in between the warming predicted if governments take no further action on climate change, 4.5°C, and if they follow current policies, 3.6°C).
This research into future projections of flood risk, alongside the paper we have just contributed to, reinforce the idea that more action on flood risk is needed. When put together, these studies contain evidence that there may already be detectable influences of climate change on flood risk, and that further increases in risk should be anticipated in the coming years, with a change in the profile of where the risk falls. There can be a long time required for policies on flood risk management to be implemented fully, with many decisions about flood defences taking years of planning, and affecting future outcomes over a lifetime or more. The evidence would appear to call for further action sooner rather than later.
Author: Rob Lamb, Director of JBA Trust