Following another intense hurricane season, should catastrophe models be factoring in global warming?
As the waters are slowly pumped out of a devastated New Orleans questions are inevitably being asked as to what part – if any – manmade climate change has played. Both the frequency and intensity of hurricanes appears to be on the rise. But is this in the natural order of things or is global warming to blame? And if it is, should it not be factored into the industry's cat models?
According to Dr Eberhard Faust, head of climate risks in the Department of Geo Risks Research/Environmental Management at Munich Re, the more intense hurricane season is due to changes in climatology – but not just global warming. According to Faust we are currently in a “warm phase” which began in the mid-1990s. This is part of a natural cycle in terms of sea surface temperatures of alternating cold and warm phases – each one spanning decades. “On the one hand the high level of hurricane activity could be accounted for by natural decadal oscillations,” says Faust. “On the other hand scientific studies in recent months indicate that the observable warming trends in all ocean basins can only be accounted for by including anthropogenic [manmade] greenhouse gas emissions as drivers.”
Unfortunately, while some aspects of climate change can be measured and incorporated into the models, Faust is concerned that this will still not enable the models to predict individual events. “Such improvements pertain exclusively to middle-term to long-term statistics,” he explains. “Hurricane Andrew hit Florida in a neutral season in terms of the El-Nino/Southern-Oscillation cycle (when a high level of hurricane activity could not be expected). Therefore big hits can take place even if the respective middle-term climate characteristics are against high activity anomalies.”
The problem with current cat models, says Faust, is that they don't take the “multi-decadal variation” in activity levels – ie the cold/warm phases – into account because hurricane data only stretches back as far as 1900, not long enough to enable the models to map the cyclicality of the phases. “That is why present-day models seem to have no convincing performance in terms of predicting frequencies of loss amounts as seen in the 2004 and 2005 seasons.” Faust says the added problem with Katrina is that the storm surge – something that hasn't happened with a mainland US hurricane in nearly half a century – was not factored in. “At present this catastrophic feature is only roughly accounted for by the industry's models.”