The fourth assessment report from the UNEP Intergovernmental Panel on Climate Change (IPCC) goes some way to predicting the impact of climate change on weather patterns.
It notes an increasing trend in extreme events observed during the last 50 years. It also says it is “very likely” that hot extremes, heat waves and heavy precipitation events will continue to become more frequent.
According to the most recent climate change scientific assessment reports of the joint World Meteorological Organization/IPCC, the warming of the climate system is unequivocal. “It’s not discussed anymore in the scientific community if climate change is happening,” said Jens Melhorn, head of flood group at Swiss Re, speaking at the RAA/IUA catastrophe modelling conference in London. “Climate change is here – it’s a fact.”
The IPCC points to a likely increase in tropical cyclone intensity with less confidence in a global decrease in tropical cyclone numbers. It says there will very likely be precipitation increases in high latitudes and likely decreases in most subtropical land regions. There is high confidence that by mid-century, annual river runoff and water availability is projected to increase at high latitudes and decrease in some dry regions.
But how accurate are these predictions? “It’s fair to say there are still significant uncertainties in future climate impact projections,” says Paul Bates, professor of hydrology at the University of Bristol and Willis Research Network member. “But the good news is if you look at how the IPCC reports have firmed up and constrained their statements about what’s going to happen, how likely is it and what’s causing it, those estimates are improving really quite rapidly given the scale and complexity of the problem.”
One solution to accurately predicting the current and future risk of natural hazards like flooding is to combine two different fields: catastrophe modelling and global climate modelling. Catastrophe models, developed for the benefit of the insurance industry, rely largely on historical records in order to predict risk levels. Global climate models, which are run and studied by scientists and academics for a wide range of stakeholders and policymakers, are simulated on supercomputers, which chart future climate and weather patterns.
“Catastrophe models are able to look at finer and finer details as they evolve,” says Tom Larsen, senior vice president at Eqecat. “As we understand more about climate change we want to leverage the catastrophe models and go forward.” Larsen is quick to defend the models, which came under intense criticism after Hurricane Katrina in 2005, but concedes “room for improvement is always there”. Compared to the global climate models, which are “all over the place”, he says the industry’s catastrophe models have more similarities than differences, and all have the same limiting factors. “We can make catastrophe models that are accurate but not necessarily precise.”
“Because catastrophe modelling is so highly reliant on the historical data record, it does not readily allow for changes in climate
Risk Management Solutions (RMS) is doing some work at the moment to combine its models with more predictive climate models. “The models that insurers use today are built to represent present day risk,” explains research analyst Nicola Patmore. “In our climate change practice at RMS we’re looking at pioneering the use of catastrophe models for understanding future risk.” The benefit of coupling catastrophe models with climate models is it can produce a lot more value in terms of actually quantifying future impact. “It’s something the academic community just hasn’t been able to do yet,” adds Patmore.
One of the limiting factors in the accuracy of global climate models is finding enough computer power. “The most limiting factor for the models is can you build a big enough computer?” explains Bates. “Once you drive the resolution of the models down you’ll then be able to make statements about local and regional impacts with more certainty.” Fellow Willis Research Network member Jane Strachan, who is based at the Walker Institute in Reading University’s Department of Meteorology, agrees. Using tropical cyclones as an example, she says the scientists are severely limited at present from studying these storms in a global context. “When you’re looking at extremes on a very local level it becomes more and more uncertain.”
But it is precisely these extreme events at a local level that insurers and reinsurers need to understand in order to better prepare for today’s and tomorrow’s peak risks. “How does the current climate affect the current risk is of the most importance to the insurance industry,” states Dr Peter Dailey, director of atmospheric sciences at AIR. In an effort to understand both current and future risk there will be a level of uncertainty, he adds. “Everyone needs to be clear that uncertainty is critical to the process. We’re simulating the effects of global temperature rather than simulating global temperature.”
“The biggest uncertainty is the uncertainty in the climate projections themselves,” explains Patmore. “What we have to do is take a scenario-based approach – we make sure we look at the outputs from multiple models and multiple runs from those same models so we produce a sort of ensemble – looking at all of the possible outcomes.” The cat modeller would then run various scenarios across the range of possible outcomes through its models to help determine the range of possible impacts in financial terms in the future.
Strachan is also concerned about uncertainty and how that is likely to be handled by the catastrophe models, as the two disciplines are connected. Ultimately though, progress is being made. “Making the needs of the insurance industry clear to the climate research community can really help to highlight and drive the right research,” she says. “Emerging areas of science should drive change in the catastrophe model approach.”
Helen Yates, editor of Global Reinsurance.
Cat models: Water world
â€œHigher temperatures fuel the hydrological cycles,â€ explains Swiss Reâ€™s Jens Melhorn. â€œIf you have more moisture in the air you have more rainfall.â€ Flooding is a useful case study for considering the future impact of climate change. It has also dominated recent headlines.
Two separate flooding events in the UK last summer cost in excess of Â£3bn. Heavy rains and flooding in China in June 2007 affected over 13.5 million people with more than 120 fatalities due to floods and landslides. Also in June 2007, flooding in New South Wales cost insurers in excess of $1bn. Bangladesh was hit in July and August by monsoon rains and landslides with Cyclone Sidr costing over 4,000 lives. Most recently, flooding in the US Midwest has caused damage to thousands of properties. Losses of up to $1bn were expected in Iowa alone.
Coastal flooding is one of the biggest concerns as exposures increase and sea levels rise due to global warming. As many as 150 million people in the worldâ€™s major cities could be reliant on flood defences by 2070 â€“ more than three times the 40 million people today â€“ as a result of climate change and urban development, according to a recent report from RMS. The top ten cities, which contain 60% of the total exposure, are from only three wealthy countries (US, Japan, and The Netherlands), with Miami ranked as top. Miami remains at the top of the 2070 rankings, with exposed assets rising from approximately $400bn today to over $3.5trn by 2070.
The annual probability of a 1-in-100 year event affecting one major city globally is as high as 74% and almost 100% over five years. â€œThe concentration of flood exposure in rapidly developing cities urgently underscores the need to integrate climate change implications into both national coastal flood risk management and urban development strategies,â€ said Professor Robert Nicholls, IPCC author and director of research at the University of Southampton. In the UK, using a conservative estimate of 0.4 metre sea level rise by 2050, a major coastal flooding event could cost up to Â£18bn according to the ABI, whereas today it would cost Â£6bn.
â€œLast year was a freak event,â€ says Swenja Surminski, policy advisor on climate change at the Association of British insurers, talking about last summerâ€™s UK floods. â€œBut the concern is that the type of freak events that donâ€™t really fall into these predictions will also increase.â€ The summer flooding was in fact the opposite of what would be expected in future. According to Professor Paul Bates, summers in the UK should become drier while winters will become wetter. He thinks the UK will see more flooding, but that it will be in the winter months. Professor Stuart Lane from Durham University believes last yearâ€™s flooding is more likely to be linked to a short-term (likely decadal) â€œflood-richâ€ cycle than to long-term climate change.
Modelling future floods
The challenges catastrophe modellers now face is to predict the current risk threat â€“ not the risk threat of 50 years ago. Because catastrophe modelling is so highly reliant on the historical data record, it does not readily allow for changes in climate â€“ whether they are caused by a naturally occurring cycle, or by climate change.
â€œThose past observations are not a reliable indicator of future flood frequency â€“ so letâ€™s say the flooding in 2007 looked like a 1-in-150 year event last summer â€“ but thatâ€™s based on what has happened previously,â€ explains Bates. â€œYou have tools at the moment to estimate current risks and those are exactly the things you need to estimate future risk,â€ continues Bates. Nicola Patmore, senior climate change research analyst at RMS, agrees. â€œThe most important lesson [from the 2007 flooding] is the need to have a better understanding of flood risk today,â€ she says.
In Europe, flood losses are rising by 12% per year (by seven percent if you account for inflation), according to Melhorn. He says there is an obvious reason for this, which is the increase in insured values, but insists that global warming is contributing. â€œFlood cover is under-priced because the models rely on data with below-average event frequency. They also make insufficient allowance for the fact the flooding often occurs in clusters.â€ He adds that Swiss Re is developing its own models â€œbecause we have the opportunity to change them. We donâ€™t have to wait for consultants.â€