The 2012 tornado season started with an early roar, and, as of mid-April, the losses continue to mount. To date, there have been more than 400 preliminary reports of tornadoes in the U.S. this year, far higher than the seven-year average of 268 twisters for the same time period, according to the U.S. National Weather Service.
Following an astonishing $26 billion in insured losses from tornadoes in 2011, several questions persist: are these horrific storms increasing in frequency or do recent events represent merely an anomalous blip on the weather radar screen? And is Canada on the front lines in terms of vulnerability to severe convective storms?
There is no scientific evidence directly linking tornado activity to global warming or climate change. In fact, weather observers such as Environment Canada and the National Oceanic and Atmospheric Association (NOAA) say there are yearly averages for tornado frequency in Canada (about 80-100) and the United States (roughly 1,000-1,200) – the two countries that face the highest risk from severe convective storms. Other factors, such as improved observation practices, more widespread Doppler radar and population growth, may contribute to perceptions of increased severe storm activity.
“If you look at the history, over the last 60 years, there has not been an appreciable increase in the number of tornadoes that have happened,” says Jose Miranda, director of client advocacy for Eqecat risk modeling firm. “I think what you may see is a longer tornado season or a wider geographic area of exposure for tornadoes, but even with these, they are not necessarily caused by climate change.”
Instead regional weather patterns are often the culprit behind a wild tornado season. For example, reinsurance broker Guy Carpenter observes that one of the main factors behind increased twister activity in 2012 is the above-average sea surface temperature in the Gulf of Mexico, resulting from a mild winter. This warmer water generates additional heat and moisture to the air passing over it, which then collides with cold, polar air coming from the north. It is the middle latitudes in North America, between roughly 30 and 50 degrees North or South, where the environment is right for tornadoes, hail and severe thunderstorms.
Other factors, such as the track of the jet stream in North America, the Bermuda high (a semi-permanent, subtropical area of high pressure in the North Atlantic Ocean) and La Niña /El Niño cycles, can also affect the formation of tornadoes.
“Over time, there are annual spikes and dips, but overall, the number of tornadoes recorded each year remain within a narrow band,” notes Glenn McGillivray, managing director for the Institute for Catastrophic Loss Reduction (ICLR). “There are times where huge outbreaks occur, and these can throw off an annual total.”
Last year and the first four months of 2012 have been clearly one of those times. And, while the long-range pattern for tornado activity may not have increased, the insurance industry’s appreciation of the damage from severe convective storms has certainly changed.
“An issue for the industry is that the losses in 2011 went beyond primary companies to reinsurers and into Lloyd’s as well, “ notes Matthew Nielsen, model product manager for Risk Management Solutions (RMS), a risk modeling firm.” So you are starting to see severe convective storm losses like tornadoes compared more to hurricane-type exposures.”
Tornadoes were not typically considered one of the larger risks for the insurance industry from a single-event loss viewpoint. Before 1994, there was only one tornado that caused $1 billion in insured property damage. However, from 1994 to 2008 there were 11 such events, including two in 2008, according to Munich Re.
In 2011, there were six tornado events with over $1 billion in insured losses in the U.S., including two that exceeded $7 billion, Swiss Re noted in a sigma report. With the early start to tornado season this year, there has already been one severe convective storm in the U.S. in March that caused over $1 billion in insured losses.
These figures don’t include the tornado that ripped through southeastern Michigan in the very unusual time period of mid-March, resulting in at least 20,000 insurance claims with payouts expected to exceed $150 million. While there is no official tornado “season,” there is a peak period for historical tornado reports in a given area. Tornadoes can happen any time of the year and several have struck in Canada as late as November or December.
However, the NOAA states there is a general northward shift in “tornado season” from late winter through mid summer. The peak period for tornadoes in the U.S. southern plains, for example, is during May into early June. On the Gulf coast, it is earlier during the spring. In the northern plains and upper Midwest, which can extend into parts of Canada, it tends to be June or July. Many observers consider the outbreaks of March 2012, which also included parts of Alabama, Indiana, Kentucky and Ohio, to be an early start to tornado activity in North America.
It’s not just the early tornado season in 2012 that has caused concern in the insurance industry. Some research firms also suggest that the traditional geographic boundaries for tornado hotspots are shifting.
CoreLogic Spatial Solutions, a California-based information and analytics company, states that, according to historical data, the frequency and severity of storms is much more widespread than commonly perceived. In fact, of the top ten U.S. states with the most tornadoes from 1980-2009, only three are in the traditional “Tornado Alley” – the region of the U.S. Midwest that includes Texas, Oklahoma, Kansas, Nebraska, Colorado, the Dakotas and Illinois. The possibilities of tornadoes moving east, south – and even north – are much higher today. Some of the most severe weather outbreaks of 2011 involved states such as Arkansas, Alabama, Tennessee and Virginia.
“The apparent increase in the number of incidents and shift in geographic distribution of losses that occurred last year in the U.S. called the long-held notion of risk concentration in Tornado Alley into question, and is leading to changes in risk management policy and procedure,” says Howard Botts, a vice president with CoreLogic.
Certain areas of Canada face a greater risk of tornado activity, according to Environment Canada. These include southern Ontario, southwestern Quebec, Alberta, and a band stretching from southern Saskatchewan and Manitoba through to Thunder Bay. The interior of British Columbia and western New Brunswick may also be vulnerable to severe storms. The most severe tornado recorded in Canada was an F5 storm (see sidebar on Fujita Scale) in Elie, Manitoba in June 2007 – an event scientists once speculated would never happen in Canada.
Ontario has seen several damaging tornados in the past few years. A tornado that hit the Leamington area in June 2010 resulted in $120 million in insured damage, while another that tore across the Vaughan area north of Toronto in August 2009 cost $88 million. In August 2011, a tornado with winds of 300 km/h touched down in the town of Goderich, Ontario, killing one person. Insured losses reached $110 million, according to Insurance Bureau of Canada.
While twisters often touch down in remote, sparsely populated regions, the main concern for the insurance industry is the danger and damage an F4 or F5 tornado could pose to a major metropolitan area. Nielsen says estimates from RMS show an F4 tornado that hit Toronto could cause $4 billion in insured losses.
“You have a much more concentrated urban population, with high property values,” Nielsen adds. “The same holds true if you look at Western Canada. Calgary, Edmonton and cities in Manitoba and Saskatchewan have grown
rapidly. If you have a severe convective storm in those areas now, the risk to property is much greater.”
Insurance and reinsurance companies are looking more carefully at risk modeling for severe convective storms, which involves using a risk platform that includes “stochastic” storm sets to estimate the effect of certain scenarios on a book of business, according to sources.
“Insurers and reinsurers are increasingly interested in our convective storm model,” Miranda says. “Tornadoes are a highly localized, high frequency type of occurrence, Insurers can look at a number of variables, such as their market portfolio, geographic exposures, census data, insurance take-up rates. They can apply these to our model and determine what effect a severe tornado would have in a specific region.”
“We are seeing more inquiries from primary companies,” notes Nielsen. “They rely on their claims experience to look back 5-15 years to get some understanding of their exposures. However, our stochastic risk models can provide a much larger and richer view of data over a longer period of time.”
For independent adjusters, the risk of tornadoes must be incorporated into disaster response plans, particularly in Ontario and Alberta, according to Jim Eso, vice president, national property & casualty for Crawford and Company (Canada) Inc.
“Adjusters need to be able to mobilize quickly to areas that have been severely damaged since a change in course of only a few miles can mean the difference between a tornado spinning relatively harmlessly out into the countryside or hitting a metropolitan area,” Eso notes. “Windsor, Ontario, for example, has been hit by tornadoes four times in the last 75 years but is in the most active tornado region of Canada, meaning numerous ‘near misses’ happen every year.”
One example of a tornado hitting an urban area was the devastating storm that ripped through Edmonton in July 1987. The F4 tornado killed 27 people, injured more than 300 people and caused more than $330 million in property damage at four major disaster sites.
“We are coming up on the 25th year since the Edmonton tornado so it is very much on our minds here,” says David Riddell, president of Edmonton-based Canadian Claims Services. “The reality is that in this part of the country, we have to be aware of the risk that a tornado could cause a lot of devastation, particularly if it hits a major urban area. The key issue with a tornado is localized damage and the need for a quick response.”
Extreme localized property damage is one of the direct consequences of severe tornado outbreaks. The severity of damage can create multiple obstacles for adjusters, according to Eso.
“To adjusters on the ground following a tornado, they will be dealing with a smaller number of more severe losses and most likely with insureds who have lost everything, which makes it a very challenging situation,” Eso notes. “Adjusters may find that the local infrastructure is severely damaged and that local accommodation is not available as displaced homeowners will often be filling local hotels. Transportation, accommodation and communication equipment must be part of the adjuster response plan.“
For severe convective storms, the principal issue in loss frequency and severity comes down to the quality of building structures, particularly roofs, according to McGillivray.
“There is much work being done by such groups as ICLR, particularly at Western University, into how and why buildings fail in wind storms and what can be done to prevent wind damage,” McGillivray says. “The key appears to be in keeping the roof on. If you keep the roof on, you stand a better chance of keeping the walls up. You can also mitigate mould damage, and you keep debris out of the wind stream. Debris causes more damage downstream, and is responsible for a large number of the deaths recorded in severe wind events.”
To improve building structures, the ICLR is working on better roof and wall connections, improved nails and different nailing patterns and use of hurricane straps. New research shows that something as basic as missed nails in roof sheathing can reduce the capacity of the roof to withstand uplift by wind in a severe convective storm.
For Eso, investments in better building structures and construction techniques are solid approaches to vulnerable storm regions in Canada. “Improved building technology, such as better roof tie down systems, high quality roof shingles and even impact resistant glass and siding, in particularly tornado-prone areas would be an effective risk management option,” he says.
Another primary concern with tornadoes is adequate warning systems to prevent widespread loss of life and injury. McGillivray suggests that there may be an element of “cry wolf” when it comes to mass notification systems.
“There are places in the U.S. where hurricane sirens ring almost all day long due to severe weather warnings,” he says. “The concern here is that people tune them out after awhile. This was reported to be the case on many occasions with the outbreaks in 2011 and this year. Warnings are only of value when people heed them and do what they are supposed to do.”
McGillivray notes that there are several jurisdictions in North America experimenting with other media, such as texting warnings to mobile devices. The National Weather Service in the U.S. is also piloting more dire weather warnings in offices in Kansas and Missouri. The “impact-based” warnings are expected to be more direct and shocking, including phrases such as “Complete Destruction of Entire Neighbourhoods is Likely.”
For the tornado outbreak that hit Kansas and Oklahoma in mid-April, the NOAA issued an advance warning more than 24 hours in advance of “severe thunderstorms capable of producing tornadoes.” The early notification sent on April 13 was credited with limiting the number of fatalities and injuries in the widespread storm system.
For Riddell, these notifications represent a major leap forward in handling tornado outbreaks. “I think now with technology, weather forecasting and social media, there is more advance warning of severe weather events like tornadoes, “ he says. “I have the weather channel app on my phone and I received updates on severe weather in real-time last summer. It is much more the case now than it was 25 years ago.”
While news of tornado outbreaks will likely continue throughout the summer and cause much speculation about increasingly severe weather events, some point to another potential reason for high losses – the rapid expansion of urban sprawl in vulnerable storm regions.
“I think it is just as relevant (to ask) the question of whether continued population growth and construction into tornado-prone areas is leading to a higher claim frequency when a tornado does occur,” Eso says. “This is similar to what some have suggested has contributed to increased flood and water damage to homes in the last few decades.”