Through the Eye of the Storm: A Hurricane Katrina Retrospective

November 7, 2025 / 9 min read

20 years after it hit the Louisiana coast on August 29, 2005, Hurricane Katrina remains a watershed event for the (re)insurance industry. In this article, Verisk leaders who were involved in shaping the company’s response share their recollections of the event. What follows is a tale of their diverse experiences of the event and how it shaped them.

satellite-image-of-hurricane-Katrina

The calm before the storm: Hurricane season 2004

The context for Hurricane Katrina begins in the prior year, 2004, which had an exceptionally active and destructive hurricane season. That year saw 15 named tropical cyclones, including nine hurricanes. Further, four major hurricanes battered Florida within a span of six weeks. What made 2004 such a devastating year for homeowners, insurers, and reinsurers was not just the severity of these relentless storms but also the cumulative nature of losses. In 2004, the industry was reminded that hurricanes, whether isolated or arriving in succession, can be genuine “risk of ruin” events.

Catastrophe modeling, first developed in the mid-1980s and adopted more broadly after Hurricane Andrew in 1992, had already matured into its second decade when Katrina made landfall. Verisk Extreme Event Solutions (EES) was then known by its original name, Applied Insurance Research (AIR) Worldwide, and its in-use catastrophe modeling platform was Classic 2. While the industry at large had grown sophisticated enough to license and use catastrophe models in-house, it still relied very heavily on catastrophe model vendors such as AIR to help insurers understand the causes of loss from catastrophic events.

Then as now, before and immediately after every major catastrophe, carriers found it essential to develop robust loss estimates. Consequently, as Katrina brewed over the ocean waters, the first reaction of most at AIR was to prepare to execute on planned catastrophe event response protocols to support the anticipated influx of client questions and requests.

Preparing for landfall: Katrina’s approach

Katrina formed as a tropical depression over the Bahamas on August 23, 2005. It made landfall in Florida as a weak Category 1 hurricane and further weakened as it passed over Florida. Farther westward, the warm Loop Current in the Gulf caused it to rapidly strengthen by August 28. Katrina was among the most powerful Atlantic storms on record, with winds exceeding 170 mph. The entire northern Gulf Coast from Louisiana to Western Florida was in the crosshairs of this monstrous Category 5 storm.

Roger Grenier was AIR’s Head of Software and Client Services. He, like most others in the industry, viewed 2005 as another active hurricane season, similar to 2004. There was no special premonition or heightened anticipation before Katrina’s landfall; the team focused on routine preparedness and monitoring.

Jay Guin headed the Research team and was responsible for providing the pre- and post-landfall industry loss estimates. EES has a long history of putting out accurate industry loss estimates following major hurricanes, going back all the way to Andrew. In response to the clustered events of 2004, the company implemented a new probabilistic event response system before the start of the 2005 hurricane season.

Katrina was the first major test of this system, presenting unique challenges due to its rapid intensification and large size. Jay felt significant pressure to maintain the company’s reputation for accuracy in industry loss estimates and was concerned about both the analytical framework’s robustness and the potential scale of damage as Katrina approached Louisiana.

Katrina made landfall on the morning of August 28, about 50 miles southeast of New Orleans. By then, it had weakened from its peak Category 5 strength to a very strong Category 3, and New Orleans appeared to have escaped its full fury. In an industry conditioned to contextualize catastrophes based on precedent, it was held to be weaker than Hurricane Camille of 1969, which made landfall as a Category 5 to the east of where Katrina hit the Louisiana coast. The predominant view at that time was that New Orleans had “dodged the bullet.”

Peter (Pete) Bingenheimer had been with AIR for about eight months when Katrina struck, working as an internal modeling analyst principally supporting a primary carrier with a large residential portfolio and handling ad hoc projects. Initially, since he and his peers with client support responsibilities were distant from the broader corporate response, their awareness was limited to what was more generally known on that fateful Sunday.

The levees break: Storm surge devastation revealed

Katrina produced one of the most catastrophic storm surges in U.S. history. As the hurricane made landfall, the shape of the coastline channeled the storm surge and caused it to reach heights exceeding 25 feet in parts of coastal Mississippi, surpassing previous records set by Camille. In some locations, surge heights were measured at more than 28 feet. This surge overwhelmed levees and flood defenses, especially in New Orleans, where the failure of the levee system led to widespread and prolonged flooding. Approximately 80% of New Orleans was underwater at the peak of the disaster.

flooded-suburban-neighborhood-in-New-Orleans

As the catastrophe unfolded, Jay, in his role as Head of Research, was called to step into the limelight as the company’s spokesman in his first experience of being interviewed on national TV. His dominant memory of the event is the intense pressure he experienced in that role. The grace under pressure that he demonstrated at that time marked him as a reliable spokesman for the company on the most sensitive issues.

The biggest challenge for Jay and the company, however, was how to estimate the flood losses. To address this, Jay turned to a bright scientist he had just hired—Boyko Dodov.

Boyko (now Senior Vice President, Research) joined the company as the first flood modeler and was four months into his job when Katrina struck. He was very new to the field, having come from postdoctoral research with no prior experience in catastrophe modeling or the insurance industry. He recollects feeling intimidated and uncertain about his new job and responsibilities as he was thrust into a highly specialized area of applied science . When the enormity of flooding became apparent, Jay asked him to develop industry loss estimates. That, according to Boyko, was the moment he found his purpose. Here was a monumental test of the scientific knowledge he had worked so hard and long to develop, and the enormous relevance of his task needed no further explanation. He got his sleeping bag, camped out in the office for a week, and delivered.

Lacking tools that are now taken for granted, he resorted to clever improvisation. NOAA had just released aerial images from its survey of post-Katrina flooding. Boyko georeferenced the images to manually delineate flood boundaries and overlaid them on digital terrain models (DTM) to estimate flood depths. This approach produced a high-resolution flood footprint within about a week after the event, which was remarkably close—within 5%—to the eventual loss estimates published by PCS (Property Claim Services).

Reconciling reported vs. modeled losses

Boyko’s intense workload lasted until he produced the flood footprints, and then it went back to normal. However, for Pete and Roger, the work in helping clients understand their losses was just beginning. The confounding factors that caused the unprecedented losses became clearer only after many long days in subsequent weeks.

Peter’s work focused almost exclusively on the portfolio of the national carrier he was managing at the time. The modeling process involved running event sets, testing scenarios, and conducting sensitivity analyses with different model settings. A significant challenge was the complexity of simulating flood dynamics, particularly involving Lake Pontchartrain and the levee system, which were not easily captured by out-of-the-box models.

Client inquiries were generally aimed at understanding the modeling approach and its limitations rather than challenging the results. There was a shared appreciation of the event’s complexity and the inherent difficulties in producing accurate loss estimates. The wind component, although significant, was overshadowed by water damage, and demand surge factors were recognized as inadequate to fully capture the event’s reality at the time.

Exposure coding soon became a major issue. There are several well-known examples of how the riverboat casinos in Biloxi were miscoded at that time, and the post-Katrina analyses highlighted as never before the importance of accurately characterizing exposures. Roger could see examples of how miscoded exposures were causing client portfolios to give skewed modeled losses. He and his team also had one other major epiphany.

A significant insight from the Katrina response was the realization that business interruption (BI) losses could be substantial even when physical damage to properties was minor. Clients reported large BI claims that were initially inconsistent with modeled loss estimates. To understand this, Roger and a colleague began calling hotels to assess damage and operational status. They found that many hotels experienced prolonged closure due to the overall regional shutdown, leading to legitimate BI claims unrelated directly to physical damage. This highlighted the complexity of claims adjustment and the limitations of existing physical damage triggers in insurance products.

Katrina’s lasting lessons for catastrophe modeling

Katrina was a watershed moment for catastrophe modeling, teaching humility in the face of nature’s complexity and highlighting the importance of incorporating physics-based models over empirical ones. The event exposed gaps in understanding storm size, multiple concurrent hazards, business interruption complexities, and demand surge effects. It also prompted a shift toward improving model science, exposure data quality, and client education on model limitations. The aftermath sparked debates on climate change impacts, as Katrina and subsequent hurricanes challenged prior assumptions about hurricane activity patterns.

flooded-streets-of-New-Orleans

People are the true strength of Verisk EES, and they rose to the occasion. Their recollections provide valuable insights into the technical, operational, and human aspects of catastrophe modeling during one of the most significant natural disasters in recent history, emphasizing how Katrina reshaped the industry’s approach to risk assessment and communication. The response to Katrina was a collective effort, and many of those closely involved have since made a name for themselves. The four stalwarts whose recollections have been captured here are now senior leaders within the business unit as it advances toward a vastly ambitious set of generational updates to its platform and flagship Atlantic Basin Tropical Cyclone model in 2026:

Jay Guin is a recognized trailblazer in the catastrophe modeling domain and continues to nurture top industry talent within and outside EES as the Head of Research.

Peter (Pete) Bingenheimer is now the Global Head of Business Development for EES.

Roger Grenier now leads Program Management for Verisk Synergy Studio, the generational update of the EES catastrophe modeling platform.

Boyko Dodov now leads the Research team for “The Next Big Thing” at EES—a globally correlated stochastic event catalog for all atmospheric perils

Access is now available to all existing clients via the EES Client Portal

Access now