One year later: Lessons from Hurricanes Harvey and Irma

By Mark Bove

Hurricane Harvey made landfall along the central Texas coast as a Category 4 storm on August 25, 2017, ending a 4,323 day “drought.” During that drought, no major hurricanes (those rated Category 3 or higher on the Saffir-Simpson Wind Damage Scale) had made landfall in the United States, making Harvey the first major hurricane to come ashore since Hurricane Wilma raced across South Florida in October 2005. Unfortunately, the next major hurricane drought in the U.S. only lasted a mere 16 days as Category 4 Hurricane Irma made landfall in southern Florida on September 10. In less than three weeks’ time Harvey and Irma caused a combined $60 billion dollars in insured losses.

What did the insurance industry learn from Hurricanes Harvey and Irma? Here are a few of the key findings:

Strong wind codes work as intended and should be implemented nationwide

In 1992 Hurricane Andrew left behind $15 billion in insured losses; Florida responded by strengthening state building codes and code enforcement several times over the following decade. In the summer of 2004, when Florida was hit by four hurricanes in a six-week span, it was apparent that new homes built to these stronger building codes performed much better in high winds than pre-Andrew homes--even in areas of southwestern Florida that experienced wind gusts in excess of 150 mph during Hurricane Charley. Further evidence of the positive impact of Florida’s new building codes was seen when newer construction in other states impacted by the hurricanes of 2005 did not fare as well as those located in Florida.

However, following the events of 2004 and 2005, an open question remained regarding the apparent success of Florida’s wind building codes: Was the observed better performance among new construction a function of the building codes themselves or a function of the building components that had yet to experience any deterioration or fatigue due to the extremes of Florida’s climate? Getting an answer to that question would take another 12 years…and the arrival of Irma. 

Hurricane Irma made its first landfall in the Florida Keys, one of Florida’s longest-inhabited regions, where a majority of the residential building stock was built pre-Andrew. Many older, wood-frame homes in the Lower Keys were damaged or completely destroyed as Irma passed overhead. However, newer construction, much of it now utilizing masonry walls like the rest of peninsular Florida, sustained significantly less damage. Even in cases where modern homes had lost siding or soffits due to winds or debris, the overall building envelopes tended to remain intact, preventing wind and water from causing further interior damage. Similar loss patterns were also seen with modern construction in the region where Harvey came ashore in Texas, where some local municipalities require wind resilient construction.

However, it was Irma’s second landfall in Collier County, home to Naples and Marco Island, that most clearly demonstrated the success of Florida’s wind/building codes. Unlike the Florida Keys, over half of the residential buildings in Collier County were built post Hurricane Andrew. This large percentage of wind-resistant construction provides the county a level of “herd immunity” from windstorms since wind resilient homes are less likely to be damaged or destroyed, and it limits the amount of wind-borne debris that could impact or damage other resilient-construction. The effect of more widespread resilient construction was confirmed – despite wind gusts of up to 140 mph reported at Naples Airport, wind damage in the county was relatively light and disproportionately affected older, less resilient construction.

Marco Island also provided further evidence that the building codes are working. Coincidentally, both Hurricanes Wilma (2005) and Irma (2017) made direct landfall on the island as Category 3 storms with 115 mph sustained winds, allowing for a near-exact comparison of building damage between the two events. As it turned out, Irma’s damage patterns were very similar to those observed after Wilma: damaged roof-ridge tiles and pool cages. This demonstrated that building performance in high winds on Marco Island hadn’t deteriorated much over the past 12 years, and that building codes are driving wind resilience, and not the age of the building itself.

The wind impacts of Harvey and Irma further prove that stronger wind building codes are effective at protecting lives and reducing damage from wind events. Given this track record of success, resilient wind building codes should be implemented in every U.S. hurricane state, as well as those states at risk of  tornadoes and extratropical windstorms.

Everyone lives in a flood zone

Although it was a Category 4 hurricane at landfall, most people don’t think of wind when they remember Harvey; this is because the storm made landfall in a fortunate spot -- a lightly populated area with limited property exposure. Instead, Harvey is remembered as a rain producer of biblical proportions, inundating a large swath of southeastern Texas with up to 60 inches of rain in just five days, leading to one of the worst flood events in U.S. history.

In addition to the record rainfall totals, there are other factors that lead to the severity of Harvey’s flooding in southeast Texas. Some of these factors are natural, like local geology and terrain impacts, but most are due to the population growth in the Houston metropolitan area. The widespread and rapid development of new communities and roads in the region have drastically reduced the amount of rain-permeable surfaces, resulting in increased runoff into local rivers and bayous. In addition, many developments in Houston do not raise the elevation of homes above the local terrain, increasing loss potentials. In the most extreme example, several residential communities were built inside a pair of large reservoirs west of Houston that were designed to hold flood waters during heavy rainfall events, literally putting the risks where the excess rain is supposed to go.

Despite repeated major flooding events in Houston over the past 18 years, most of the flood loss from Harvey was not insured. According to analysis by the AP, overall, the take-up (purchase) rate of flood insurance in Harris County, Texas, is less than 20%; most of that coverage is located in FEMA A or V zones, regions with a 1% chance (or greater) of flooding annually and where obtaining a mortgage requires proof of flood insurance. However, Harvey’s rains were so severe that it is estimated that about 41% of homes affected by the Harvey’s floods were outside of FEMA high risk flood zones, according to a study by Bloomberg and Rice University. In these lower risk areas, many homeowners opt to forego flood insurance because their mortgage doesn’t require it or they don’t perceive themselves at risk from flood. (Link: https://www.bloomberg.com/graphics/2017-fema-faulty-flood-maps/)

The truth is that every property has some risk of flooding, even those located outside of FEMA’s Special Flood Hazard Areas. While many locations may not have a risk of river flooding, the risk of flash flooding is present everywhere and often overlooked. Flash flooding, resulting from torrential rains, needs no river or creek to occur, just more water than the land and topography can handle at once. We need to change the perception that flood risk has geographical boundaries or limits. Flood risk is everywhere, and work needs to be done to mitigate that risk, whether by better land use practices, elevating or buying out homes that already exist in flood plains, or by closing the flood insurance gap with businesses and consumers.

The worst storm surge impacts aren’t always near the location of the hurricane’s landfall

As Irma approached the southwest coast of Florida, meteorologists were worried that the storm, which was forecast to closely parallel the coast, would create a devastating storm surge in the densely populated and low-lying coastal areas that stretch from Naples northward to Tampa. However, a combination of a more inland storm track and a weakening storm core prevented a major storm surge event from occurring in these areas, and only minor surge flooding was reported in these regions. Ironically, instead of experiencing record surges, the Gulf Coast actually experienced record levels of low water before Irma’s passing as easterly winds prior to landfall pushed water out of to sea.

While the west Florida coast was spared from a major flood event, persistent easterly winds caused by Irma’s westward track through the Turks & Caicos and Cuba had the opposite effect on the state’s east coast, causing widespread coastal flooding from Miami to Jacksonville, 350 miles to the north of where Irma was making landfall.

In fact, the highest storm surge levels from Irma occurred near Jacksonville (7.8 feet in Fernandina Beach) in the northeast corner of the state. Why was the surge so high in this region? Three key factors influenced the magnitude of flooding in northeast Florida. The first factor was the aforementioned easterly winds associated with Irma that piled up water along the coast. In addition, the concave shape of the Florida coastline in that region funnels water towards northeastern Florida, causing a local maxima in surge heights. Third, Irma was also a prolific rain producer, causing the St. John’s river that flows through downtown Jacksonville to rise. But as coastal flooding prevented the St. John’s from discharging into the ocean, the river instead overflowed its banks, causing extensive inland flooding in the city.

This is the second time in the past decade where a combination of long-duration easterly winds and the  shape of the coastline has caused a major surge event; the other event was Superstorm Sandy’s impacts along the New York Bight in 2012. But unlike Sandy, Hurricane Irma never made a direct landfall anywhere near the location of its worst flooding—an unusual situation that might not be contemplated by catastrophe risk models. Irma’s unusual storm surge also highlights the continuing need to examine the probability and range of surge events that could theoretically occur from tropical cyclones (and persistent antecedent winds) to help better capture the full range of potential surge outcomes.

These are just a few examples of the many lessons that hurricanes Harvey and Irma can teach us. Despite being similar in intensity, each storm uniquely impacted the human environment with regard to wind and flooding. As insurers and catastrophe risk modelers continue to explore the claims data from these events, many more lessons will reveal themselves.