Predicting Hurricanes In a Warming World Words

Predicting Hurricane Risk in a Warming World

*This article highlights one catastrophe modeling firm's approach to predicting hurricane risk in the face of global warming.*

In recent years, devastating hurricane events like Andrew and Katrina have increased media attention on higher hurricane activity due to climatological factors, including global warming. The catastrophe modeling community is aware of the buzz, and while some among it question the influence of climate change, others believe the warming trend signals a new regime—one that calls for similarly radical changes to hurricane modeling methodology.

From AIR’s perspective, the debate is not whether climate factors have an influence on hurricanes; scientists tend to concede that they do. If there is any debate at all, it is on how to provide insurance companies concerned with deriving reliable results from catastrophe models with the latest science, particularly that of the influence of elevated sea-surface temperatures (SSTs) in the Atlantic.

The Atlantic basin has been in an active phase since 1995. This phase is characterized by above long-term average hurricane activity. Some scientists, like those at the National Oceanic and Atmospheric Administration, associate this with the warm phase of the Atlantic Multidecadal Oscillation (AMO), a climate signal measuring the change in North Atlantic SSTs. The AMO is just one of several measures of Atlantic sea-surface temperature anomalies, but other measures indicate warming trends now, too.

At AIR, climate scientists have been monitoring the relationship between global warming and Atlantic hurricane frequency since the early 1990s. They stay abreast of the latest scientific literature on the AMO and other climate signals, evaluate related findings, and conduct research of their own. In 2005, AIR meteorologists undertook new and extensive analyses of the link between elevated SSTs in the Atlantic and landfall frequency and location—a research effort that is still ongoing. The result was a near-term hurricane catalog released in April 2006.

Unlike AIR’s standard catalog, based on over a century’s worth of hurricane data and more than 20 years of research and development, the near-term version relies solely on hurricane statistics from 2004 and 2005—seasons of notable Atlantic intensity. But despite general consensus as to the impact of warming seas in these years, standard use of a near-term catalog inspired by recent storm trends remains controversial. That’s largely because researchers have not yet translated their finding regarding a jump in Atlantic hurricane intensity into information useful to insurance companies—a percentage increase in foreseeable insured hurricane losses on land.

To date, the primary focus of investigation has been on basinwide storms, leaving landfall trends unexplored. If hurricane risk is to be assessed with a high degree of confidence, however, the relationship between hurricane activity in the Atlantic, landfall activity, and regional insured losses will require significant additional investigation. AIR meteorological teams are making great strides on this front.

Meanwhile, in the face of uncertainty around the near-term view—the methodology of which is currently undergoing rigorous review by peers—AIR maintains that a short-term model offers a valuable perspective from which to manage hurricane risk. The issue then becomes how to offer this untested model to insurance companies.

When it was first released in 2006, AIR made the scientific judgment to keep the near-term catalog separate from its standard version. By doing so, it did not impose a new model with high uncertainty on its clients, but instead allowed them to choose for themselves just how they wanted to use the model’s new information in their decision making processes.

This year, AIR will make the same provision—offering the near-term catalog as a supplement to its long-term model, rather than a replacement. Interestingly enough, the long-term model is scrutinized, too. It has occasionally been characterized as “backward looking” because it incorporates data from 100+ years, but this judgment requires inspection. The purpose of using historical data is not to reproduce the past, but to better estimate what may happen in the future. Along those lines, AIR’s standard catalog provides clients with scenarios that could happen in upcoming storm seasons even though they have not happened before—scenarios that include both more intense storms and years with more landfalls than have been historically seen. Ideally, nothing should happen that the model does not account for, and the AIR model for U.S. hurricanes has performed very well in this regard.

.MGW.