On the following pages, see a hypothetical severe thunderstorm outbreak in the U.S. with a loss exceedance probability of 1% (100-year return period) to the industry.

The Scenario

It is early spring when a frontal system brings cool air from the Rocky Mountains across the Southern Plains. When this cold front collides with the warm, humid air ahead of it, drawn northward from the Gulf of Mexico, it produces a severe thunderstorm outbreak that lasts eight days, traveling from Texas up to Michigan.

Wind-driven hail up to 5.25 inches batters much of Texas and part of Alabama. A total of 74 tornadoes touch down during the outbreak—in Louisiana, Alabama, Mississippi, Georgia, and Florida up to Arkansas and then into Indiana, Ohio, and Michigan. One-hundred and fifty-seven straight-line wind events result also.

Hail, tornado, and wind events in the outbreak scenario

(Source: AIR)

Texas, the worst affected state in this scenario, is the second most populated U.S. state after California. The four counties where the hailstorms cause the most insured loss are part of the Dallas-Fort Worth-Arlington metroplex—the largest metropolitan area in Texas and the fourth largest in the U.S. The four counties have a population of nearly 6 million.

Hail causes damage mainly to roofs, windows, and siding, and is sometimes followed by damage to contents caused by water entering through openings made by hail. Common materials used to make roofs in this area are asphalt, metal, tile (clay or concrete), and wood. Hail can also cause moderate to heavy damage to automobiles.

Residential houses in the affected areas are typically of wood frame construction; this, and the fact that residential construction is usually non-engineered, make them more vulnerable to high tornado winds and wind-borne debris. Commercial buildings are, on average, less vulnerable than residential structures or automobiles, but can exhibit widespread roof and cladding damage, which also leads to water damage due to rain leaking through cracks and punctures on building envelopes.

Estimating the Impact

AIR estimates this scenario would cause insured losses of $13 billion across 18 states.

The map below shows the distribution of loss by county. Texas suffers the majority of losses (85%)—with Alabama and Tennessee representing a distant second (9.2%) and third (2.3%), respectively.

Insured loss by county

(Source: AIR)

Four northern Texas counties—Tarrant, Dallas, Collin, and Denton—which comprise Fort Worth, Dallas, and suburbs of Dallas, and one county in northern Alabama, Madison County, which contains Huntsville, are hardest hit.

Hail is by far the biggest driver of insured loss, at 85%. Tornadoes cause 13% of insured losses, and straight-line wind 1.5%. Very large hail up to 5.25 inches in diameter causes extensive damage to some roof coverings. Roofs made of asphalt shingles exhibit damage from fractures, punctures, or spalling, which is the chipping, fragmenting, or flaking of the roofing material. Damage occurs primarily on a roof's windward side where hail has the most direct impact, and older asphalt roofs sustain more damage than newer ones. Age is less of a factor for tile roofs, but the windward side also exhibits more damage than the leeward side. Metal roofs sustain moderate to severe denting but no fracturing or puncturing. And cedar shingle roofs show fractures and punctures, allowing precipitation to enter. Older cedar roofs sustain more damage than newer ones.

Modeled insured losses by sub-peril and line of business

(Source: AIR)

Hailstorms associated with this severe thunderstorm system also damage the siding of homes and commercial property, and break window glass, giving another port of entry for water damage to the interiors of buildings and their contents.

Hail in this area also damages numerous automobiles by breaking their windshields and denting their exteriors. High wind, large hail, and tornadoes cause locally severe damage in scattered areas across the rest of the affected states. Much of the damage outside of the top five counties occurs in smaller towns and rural areas.

Are You Prepared?

This scenario is just one example of the extensive and widespread damage a severe thunderstorm could produce in the U.S. and in the Dallas-Fort Worth area. Note that while losses from hail dominated in this example, different breakdowns of losses by sub-peril are possible. The insured loss for this event has a modeled annual exceedance probability of about 1%. This is not an extreme tail event; far greater losses are possible.

A few catastrophe modeling best practices can help ensure that your model produces the most realistic loss estimates:

• Collect detailed information for the properties that make up your portfolio—including location and all primary building characteristics such as construction type, occupancy, building age, and height—and a true replacement value. Secondary features will help refine the loss results and help you get closer to the true vulnerability of the individual buildings.
• Roof response to hail is dependent on secondary features, such as roof age, pitch, covering, and attached structures. Roof coverings can also be classed by their hail impact resistance. Other secondary building features that affect a building's vulnerability to hail are: glass type and percentage for windows and window protection as well as wall siding type.
• Be aware of potential losses from non-modeled sources such as extreme flooding during some thunderstorm events, fire following, and hazardous material leakage and associated cleanup costs.
• Finally, consider your loss ratio, or how your estimated losses compare to the “total insured value” in each affected area. While your losses may at first appear to be high, the loss ratio they reflect (even in the most severely affected region of this scenario, the loss ratio is less than 5%) should be entirely consistent with an infrequent but thoroughly plausible catastrophic event.

No model can predict what the next catastrophe will be or when it will occur. The fundamental uncertainty makes it all the more important for companies to use catastrophe models to prepare for such losses. The careful analysis of model results can help risk managers prepare for many contingencies—thus ensuring that scenarios like the one presented here will not be entirely unexpected.