Hurricane Ike

Post Landfall 1 | Summary
Posting Date: October 21, 2008, 11:00:00 AM

Hurricane Ike Advisory: Latest Observations Hurricane Ike was the third hurricane to make landfall in the U.S. this year, preceded by Hurricane Dolly in late July and Gustav just two weeks prior to Ike. While Ike’s intensity at landfall was comparable to Gustav (both estimated at sustained winds of 110 mph by the National Hurricane Center), it was a much larger storm with a larger radius of hurricane force winds. In addition, the Texas coastal exposure directly subjected to storm surge from Ike was more concentrated than in Louisiana where Gustav made landfall, despite Louisiana’s generally lower elevation. Since Ike made landfall, AIR meteorologists have been actively analyzing additional meteorological data, including wind observations directly and indirectly associated with the hurricane, while AIR engineers have completed an extensive post-disaster survey to provide a detailed view of damage to individual properties and entire neighborhoods. This document summarizes the latest findings of AIR's post-event analysis of Hurricane Ike.

Hurricane Ike Offshore Post-storm analysis of Hurricane Ike’s path through the Gulf of Mexico was performed by AIR meteorologists using data from the NOAA Doppler weather radar to more accurately define the storm’s characteristics in the period prior to landfall. GIS analyses of these data allowed for estimation of hourly ranges of eye diameter and radius of maximum winds (Rmax), providing AIR information needed to more accurately determine the wind field associated with this storm. (Note that Rmax is not always provided in National Hurricane Center official advisories at the time of the event.) During the 12-hour period prior to landfall, Hurricane Ike in fact displayed different characteristics from those observed at the time of landfall. While still out in the open Gulf, Ike had a very broad circulation and did not exhibit a well-defined eyewall. The National Hurricane Center also reported a pronounced disparity between central pressure and wind data, with Ike’s central pressure being consistent with a much more intense hurricane than dropsonde and flight-level wind data was suggesting. As the storm neared the coastline, the circulation tightened up, Rmax decreased somewhat, and the eye became more organized, indicating that Ike was intensifying. Despite the narrowing eyewall, however, hurricane force winds extended an impressive 120 miles from the center. Given AIR’s initial assessment of offshore risk at the time of the event and the latest detailed revisiting of Ike’s offshore evolution, AIR believes that the loss scenarios posted on the ALERT website are an accurate reflection of the damage and ground-up loss to offshore rigs. Over the coming months, as more claims information is collected and analyzed, we will continue to assess the performance of the U.S. Hurricane Model for Offshore Assets. Because insurance terms and other direct and indirect contributions to offshore insured loss (e.g., business interruption) are highly complex, total insured loss estimates are less certain. A summary of AIR’s estimated industry loss range for offshore assets is included in the Summary Loss Table in the concluding section of this document.

Hurricane Ike at Landfall Soon after Ike made landfall along the Texas coastline, AIR deployed teams of engineers to survey the damage. What was observed in the field was consistent with output from AIR’s damage functions for U.S. hurricane. Mobile homes suffered moderate to, in some cases, extensive damage; single family homes suffered light to moderate damage, primarily to shingles and sheathing, but rarely compromising structural integrity.

In general, commercial structures suffered less damage than did residences, and the more engineered the property, the lower was the observed damage ratio. Some commercial structures—in particular small restaurants and gas stations—saw moderate damage as a result of their frequent use of light metal. Well-engineered high-rise structures suffered damage primarily to glazing—a damage pattern similar in nature to what was observed in New Orleans after the passage of Hurricane Katrina in 2005. Assessing storm surge damage can be elusive. From damage observations, it can be difficult to distinguish what portion of the overall damage is due to wind versus surge. Further, direct measurements of storm surge are relatively scarce. Estimating industry impacts when surge is a significant component of the total loss is especially challenging because it requires an assumption—often a highly uncertain one—about commercial take-up rates. For the final set of ALERT scenarios posted for Hurricane Ike after landfall, losses due to storm surge accounted for, on average, about 10% of the estimated total industry loss. The central (or expected) scenario, which indicates an estimated industry loss of about $10 billion, includes about $1.2 billion due to storm surge. However, it is important to bear in mind that this assumes a 10% commercial take-up rate, as well as 10% of residential surge losses ultimately paid under wind policies. To give a better sense of the sensitivity of loss estimates to these assumptions, the table below provides a full breakdown of losses for the central scenario. Under the assumption that 10% of surge losses will be paid on both residential and commercial exposures, the estimated AIR industry loss is $9.6 billion (8.4B wind + 1.2B surge). If we instead assume a 100% take-up rate for commercial policies, industry losses would increase to $13.9 billion ($8.4B wind + $0.75B residential surge + $4.7B commercial surge).

ALERT Posting L00 -- Central Scenario (Scenario #6) in US Dollars

	Wind Loss Estimate	Surge Loss Estimate	10% Surge Assumption
Residential	4.3B	7.5B	0.75B
Commercial	3.5B	4.7B	0.47B
Auto	0.6B	<0.1b>	< 0.1B
Industry Total	8.4B	12.2B	1.2B
Industry Estimate Including 10% Surge Assumption = 8.4B + 1.2B = USD 9.6B
Industry Estimate Including 100% Commercial Take-up = 8.4B + 0.75B + 4.7B = USD 13.9B

Note that the estimate for this scenario including demand surge, assumed to add an additional 5% to the total for this level of industry loss, is about USD 10.1B under the 10% surge take-up rate assumption.
While there is considerable uncertainty surrounding commercial storm surge take-up rates, CLASIC/2 does offer users the ability to code which policies cover surge losses and which do not, so individual insurers should be able to get accurate storm surge loss estimates provided they code their exposure data properly. Specifically, CLASIC/2 users should be aware that storm surge coverage may be a sublimit and thus have different deductibles and limits than the wind or fire cover. In CLASIC/2, users can explicitly code which policies cover surge, and thus apply 100% of the surge loss to that property. CLASIC/2 users can also enter the policy terms, including any sublimits, to handle the surge component as well as the wind component. However, for policies that include surge coverage, simply applying 10% of the surge could easily lead to underestimated losses. Finally, AIR’s post-disaster survey teams have returned with a large volume of high-resolution data from Hurricane Ike. These observations, combined with the soon-to-be-released National Hurricane Center’s tropical cyclone report on Hurricane Ike will be evaluated and analyzed. At this point, however, it appears that the AIR U.S. Hurricane Model performed well both from both a hazard and vulnerability perspective. A summary of AIR’s estimated industry loss range for losses associated with Ike’s landfall are included in the Summary Loss Table in the concluding section of this document.

The Effects of Ike Inland The effects of Ike well inland of landfall have been the subject of significant interest and even debate. Preliminary estimates issued by ISO’s Property Claim Services unit indicate loss of as much as $1.1 billion in Ohio and other, albeit smaller losses in states across the Midwest. There are two components to Ike’s damaging wind footprint inland: one, the direct effects of Ike’s status as a tropical cyclone; the other, the indirect effects that occurred after Ike had taken on extratropical characteristics and its remnants combined with a pre-existing extratropical system over the Midwest. The direct effects of Ike’s path inland from Galveston—and of the sort associated with all tropical cyclones’ inherent dissipation over land—occurred over inland Texas and north to parts of eastern Oklahoma and Arkansas. It is not unusual for a hurricane to produce significant wind damage well over 100 miles inland from the coast, and this inland penetration of hurricane risk is well represented in AIR’s U.S. Hurricane Model. The ALERT scenarios released at landfall indicate significant losses in northeastern Texas and parts of Arkansas, and are in line with expectations for a storm of Ike’s size and intensity. While Ike dissipated at a rate fairly typical for a Category 2 hurricane, the sheer size of the storm led to greater inland penetration than might be expected of a more compact system. Ike’s indirect impacts appear to have resulted from the remnants of Ike combining with and enhancing a moderately strong extratropical cyclone (ETC) that had been stationary over Michigan at the time of Ike’s landfall. The ETC linked up with Ike’s leftover energy to produce strong winds in the Midwest region on the 14th and 15th of September.

While it is unlikely that the ETC alone would have produced the high winds that were observed in Ohio and elsewhere, it is equally unlikely that the remnants of Ike would have produced them without the presence of the ETC. That is, these strong winds cannot be solely attributed to the extratropical transitioning of Ike—a common stage in a hurricane’s life-cycle as it moves into the mid-latitudes. In fact, this particular situation was quite unique in that it had characteristics of tropical (Ike), mid-latitude (ETC), and convective (severe thunderstorm) systems, as is laid out in the following narrative of events. On September 12, while Hurricane Ike was still in the Gulf of Mexico tracking towards Galveston, TX, an extratropical cyclone was centered over northern Michigan (denoted by an “L” in Figure 1), with a cold front (the dark blue line) extending from Iowa to Kansas. In association with the frontal boundary, several tornados were reported in Missouri and Kansas. That same day, tornados near the border of Louisiana with Texas were reported in association with Hurricane Ike. While severe thunderstorms are not uncommon along cold fronts, they are relatively rare in the early months of autumn.

Figure 1. Relative locations of Hurricane Ike and an extratropical cyclone (and its associated cold front) over the Midwest on September 12.

On September 13, the day Hurricane Ike made landfall along the Texas coastline, the center of the ETC relocated further south along the front, which was moving slowly to the east. Meanwhile, a stationary front (alternating red and blue lines) was positioned along the Great Lakes into New York and New England (Figure 2). A line of severe weather, including several reports of tornados, occurred in association with the stationary front along the southern border of Michigan. On the same day, there were at least 15 reports of tornados in Louisiana and Arkansas as Ike moved inland. Thus on September 13, these two distinct and well separated systems—the ETC and Hurricane Ike—both produced severe weather and damaging winds.

Figure 2. On September 13, Hurricane Ike comes ashore and the center of the extratropical cyclone relocates further south.

On the 14th of September, Ike’s sustained winds dropped to below 40mph and the storm was downgraded to a tropical depression. At this point, the National Hurricane Center (NHC) stopped tracking the system, leaving the end of Ike’s track located at the Arkansas border with Missouri. Meanwhile, the ETC re-centered itself north of Michigan as its cold front moved further to the east. Indeed, in the final advisory issued by the NHC on Ike, forecasters noted the approaching front, which at the time was associated with winds actually stronger than Ike’s: SURFACE OBSERVATIONS INDICATE THAT IKE WEAKENED TO A TROPICAL DEPRESSION DURING THE PAST SEVERAL HOURS...WITH 25 TO 30 KT WINDS AND HIGHER GUSTS OCCURRING WELL TO THE SOUTHEAST OF THE CENTER. THE SURFACE DATA SHOW A COLD FRONT IS APPROACHING IKE...WITH AN AREA OF 25 TO 35 KT WINDS DEVELOPING BEHIND THE FRONT FROM SOUTHWESTERN MISSOURI ACROSS NORTHWESTERN ARKANSAS INTO EASTERN OKLAHOMA. By the time the remnants of Ike reached the Missouri border, as shown in Figure 3, the ETC and its associated frontal system was producing strong winds over Indiana, Ohio, and Kentucky. It was on this day—when the center of what remained of Ike was still at some distance to the southwest—that the strongest winds associated with the ETC were observed, with the highest gust of 64 knots (74 mph) recorded in Clinton, Ohio near Akron. Despite these locally strong gusts, there was no severe thunderstorm activity reported on the 14th, indicating the absence of either tropical or severe thunderstorm characteristics.

Figure 3. On September 14, as the remnants of Hurricane Ike were centered on the border between Arkansas and Missouri, the highest winds were recorded in Akron, Ohio.

On September 15th, with the remnants of Ike now fully combined with the passing cold front, the ETC moved further east across northern New York, and continued to produce gusty and at times damaging winds over parts of New England. By the time it was over, the ETC, which was fortified by Ike’s remnant energy, had produced strong and damaging winds in a band from Illinois east to New York, and south to Kentucky and Tennessee.

Figure 4. By September 15, the remnants of Ike had fully combined with the passing cold front, which produced gusty winds over parts of New England.

Again, neither the ETC nor the remnants of Hurricane Ike are likely to have been able to single-handedly produce the intense and damaging winds across the Midwest. Rather, it was the unlikely—and unlucky—confluence of these two systems. It is AIR's position that the winds, damage, and losses directly associated with Hurricane Ike as it moved from the Texas coast to Arkansas—at the point at which the NHC stopped tracking Ike as a tropical cyclone—are accurately reflected in the ALERT posting of September 13, the day of landfall. In fact, this portion of Ike’s footprint highlights an important point AIR has been making for many years, namely that hurricanes often produce significant wind damage well beyond the high impact zone on the coast. Ike was still a hurricane well into Texas, and did not dissipate to tropical depression strength until it had reached Arkansas, more than 400 miles from its landfall location in Galveston. As for the strong winds that were observed in states like Ohio and Indiana on the 14th and 15th of September, AIR scientists have carefully evaluated the set of circumstances that produced the significant losses in the Midwest. How one might model this rather unique and rare confluence of events, however, is anything but straightforward. AIR considered an appropriate application of its suite of models, which together account for total year-long hazard, namely, the U.S. Hurricane Model, the U.S. Severe Thunderstorm Model, and the U.S. Extratropical Cyclone (Winter Storm) Model, each of which comprises a seasonal and regional climatology of severe weather events. Ultimately, it was determined that the most appropriate approach was to use the U.S. Extratropical Cyclone Model to generate an accurate depiction of the wind footprint across the Midwest. The decision was prompted by the fact that Ike had lost its tropical characteristics and the ETC, by the 14th, was no longer accompanied by convective (severe thunderstorm) activity. A summary of AIR's estimated industry loss range for the interaction of Ike's remnants with a pre-existing ETC disturbance is included in the Summary Loss Table in the following section.

Summary of Insured Losses Associated with Ike In summary, the effects of Hurricane Ike—from its unusual evolution over the Gulf of Mexico to the interaction of its remnants with a pre-existing disturbance over the Great Lakes region—were broad and far-reaching. In order to estimate the overall impact, clients can take advantage of the range of scenarios available through AIR's ALERT service, which represent the results of runs of AIR's modeling suite: the AIR U.S. Hurricane Model for Offshore Assets, the AIR U.S. Hurricane Model, and the AIR Extratropical Cyclone Model. The following table summarizes the estimated range of insured loss for the industry as a whole taken from the latest ALERT postings using each of these three models. It is important to emphasize that the actual impact of each component described in this table will vary for different books of business, especially those with a distribution of exposure that deviates significantly from that of the industry.
AIR's goal is to provide clients with the most accurate means of estimating the total impact of this highly complex and far-reaching event. Depending on how one aggregates loss, this table can be used to estimate combined impacts, for example, from all three components: offshore, onshore, and Ike's remnants combined with the pre-existing ETC.

Summary of Insured Loss Estimates (Billion USD)

Component of Estimated Loss	AIR Model	Region Affected	Industry Insured Loss Range (USD Billions)
			Low	High
Offshore	U.S. Hurricane for Offshore Assets	Gulf of Mexico	0.9	1.8
Onshore Hurricane*	U.S. Hurricane	Texas, Louisiana, Arkansas	8.2	12.2
ETC + Ike Remnants	U.S. Extratropical Cyclone	Ohio, Illinois, Indiana, Kentucky, Missouri, Arkansas, Pennsylvania, New York, Tennessee, West Virginia	1.9	3.0
Combined†			11.0	17.0

* Assumes 10% of commercial and residential storm surge damage will be paid. See “Hurricane Ike at Landfall” section for more details on model assumptions and alternative estimates. † Appropriately combining losses from this table depends on how one defines the event. Historically, offshore losses have been excluded from industry estimates for Gulf hurricanes. So in order to compare industry estimates for Hurricane Ike with estimates of Hurricane Rita's (2005) reported onshore loss, one would only consider only the Onshore Hurricane range shown above, that is, from USD 8.2 to 12.2 billion. The combined loss shown here indicates the overall industry impact of Ike, including damage to offshore assets, damage to onshore properties produced at landfall, and damage in the Midwest states produced by Ike's remnants combined with a pre-existing ETC. The combined range of USD 11.0 to 17.0 billion is the sum of low and high estimates, by component.

Analyzing Ike in CLASIC/2 and CATRADER In addition to the offshore and onshore Hurricane Ike CLASIC/2 and CATRADER event sets already available on the ALERT website, AIR is also providing CLASIC/2 and CATRADER event sets that will allow clients to quantify the impact of the ETC – Ike Remnants on their portfolios. In order to model these new event sets, users should be sure to select the "Winter Storm" peril from the analysis options dialog box. This can be accomplished either by selecting the "Wind" peril and leaving all sub-options selected, or by specifically selecting the “Winter Storm” peril in both CLASIC/2 and CATRADER. Please note that the onshore, offshore, and ETC event sets each contain a unique number of scenarios. The scenario numbers are independent across the three event sets, and users should not combine losses based on scenario ID numbers in order to assess their total loss. Users who wish to combine losses from any or all Ike-related postings should run each event set separately, and should note the high and low loss estimates produced by each full event set. The summation of the high and low losses from each modeled event set will provide a total range of losses related to Ike. In reviewing the calculated range of total loss, note that it is possible that very limited exposure cases, such as the case of a multi-location policy spread across states impacted by both the onshore component of Hurricane Ike and the inland ETC – Ike remnants with a single policy-level limit and deductible, may lead to an overestimation of modeled losses in CLASIC/2. The ETC which has been linked to the remnants of Hurricane Ike has several unusual characteristics including the production of hurricane force winds in the Midwest US leading to potential losses in the state of Ohio in excess of USD 1.0B. While ETCs are not uncommon in the Midwest, even in the autumn months, they typically produce low-level though sometimes widespread damage. Thus, identifying “like events” in the stochastic event set is difficult. Further, while there are several stochastic events in the US ETC model with significant damage in states like Ohio, most of these events are heavy winter snowstorms, i.e. “blizzards”. Thus, to preserve the realism of “like stochastic events” selected from the US ETC model for this somewhat unique scenario, the search was restricted to those events which: 1. Produce at least USD 800M of insured loss in the state of Ohio 2. Include at least 95% of Ohio’s loss from the wind sub-peril (not snow or freeze) 3. Include a reasonable distribution of loss in adjoining Midwest and Northeast states 4. Produce a total US insured loss of at least USD 1.5B, generally consistent with ALERT scenarios Having searched the US ETC 10K stochastic catalog with these restrictions, one event (Event ID 47133) was identified with all of these characteristics.

Post Landfall 1 | Downloads
Posting Date: October 21, 2008, 11:00:00 AM