Hurricane Katrina: What if it happened today?

By Marc Lehmann and Geoffrey Saville, Willis Towers Watson

Hurricane Katrina caused the deaths of an estimated 1833 people, wreaked havoc to an entire region and left a scar on the American psyche. Catastrophes of a similar scale could happen at any time, so what can policy-makers, local communities, private sector organisations and the insurance industry do about it?

On the morning of August 29th 2005, Hurricane Katrina struck the Gulf Coast bringing with it sustained wind speeds of between 100–140 miles per hour. It eventually engulfed a 400-mile-stretch of the coastline causing almost 2,000 deaths and more than $100 billion worth of damage. High winds destroyed beachfront towns across Mississippi and Louisiana, leading to the evacuation of several hundred thousand people.

After the levees in the low-lying areas of the city of New Orleans were breached, more than 80% of the city was flooded by the subsequent storm surge, which reached over 30 feet in places. The storm was at category 3 intensity, when it made landfall, but it could easily have been more intense, as only hours before it was at category 5 strength. The power of this large and violent hurricane led to massive storm surge and huge waves, which exceeded the design limits of the levees protecting the city.

Many residents in New Orleans were trapped, with some stranded on rooftops without power, food and water for days. The immense disaster response requirements led to difficulty in executing the rescue effort, with some claiming the Federal government was slow to help many areas. This led to chaos, mayhem and looting, even for the several thousand who sought refuge in the city’s Superdome (the nominated emergency shelter at the time) and the Convention Center.

A year after these horrific events first unfolded, New Orleans’ population was still only half its pre-storm size and most of the city’s hospitals were only partially operational – partly because only half of the city’s doctors were available. Power was restored but at only to two thirds of its original capacity. Many working class Americans deserted the city without homes or jobs to return to, and the rebuilding process was slow and painful.

It was widely reported after the storm that many homeowners had assumed that their insurance against hurricane damage would include damage caused by flooding, but this was not always the case. Hurricane Katrina’s most damaging aspect was related to storm surge rather than wind. Insurance companies in Louisiana paid an unprecedented $11 billion in damage to houses, but in most cases this did not cover the damage that was deemed to be caused by the rising waters. This highlights the importance of homeowners fully understanding the policy wording when taking out household insurance. Many uninsured losses had no way to be repaid, leaving thousands of people without homes or livelihoods, and only State or Federal aid to help. Since 2005, a year that also saw the Gulf of Mexico coastline also battered by Hurricanes Rita and Wilma, the Federal Emergency Management Agency (FEMA) have provided more than $15 billion to affected hurricane States, for reparation and public works projects to rebuild communities, business and cultural redevelopment, and efforts to enhance resilience.  More than $6.7 billion was provided to over one million individuals and households to aid recovery.  For a disaster of such magnitude, it requires response from both the insurance industry and government to help the communities that have suffered the greatest loss.

Managing future risks

Hurricane Katrina was the largest and third strongest hurricane to ever make landfall in the U.S. It has often been referred to as the most anticipated disaster in modern American history. Before the event the FEMA categorised New Orleans and San Francisco as the two cities in the U.S. most exposed to extreme natural disaster. And on the eve of Katrina's landfall, the National Weather Service warned that the hurricane would cause “human suffering incredible by modern standards”. During the preceding century, hurricanes flooded New Orleans on five other occasions – in 1915, 1940, 1947, 1965 and 1969.

Back in the early 1990s, following Hurricane Andrew and the Northridge earthquake, catastrophe models were developed to help the insurance industry better assess the likely losses that they could face for insured asset portfolios that were exposed to natural hazards. Today, these probabilistic models are used widely across the insurance and reinsurance markets and can simulate a range of possible hazards – such as earthquakes, storms and floods – to provide estimated property damage and business interruption losses for a range of probabilities (expressed in the industry as return periods).

In 2005 following Hurricane Katrina however, there were strong indications that the catastrophe models that were being used did not perform as anticipated and actually estimated the losses incorrectly – largely due to the unexpected and unmodelled impacts of flooding and storm surge following the failure of the levees. This was a real world scenario that the models had not taken into account. Since Katrina, however, there have been many improvements in the way that catastrophe models are used and applied which allow for better loss estimations. These improvements include:

  • the collection of more extensive claims data to help strengthen the relationship between insurance losses and the characteristics of severe events;
  • simulation of Katrina-style catastrophe events to help us understand the range on uncertainty in the results;
  • the introduction of climate variability and change features;
  • and the use of more accurate property data.

Extreme historical events such as Ivan, Katrina, and Ike turned out to be hurricanes with much greater storm surge depths than anticipated for events of their respective landfall Saffir-Simpson wind intensities. In response to this, the catastrophe modelling community started to rethink the way they model the impacts of storm surge which in turn led to a two-year project to find ways to improve modelling the risk. One significant outcome of this project was to hydro-dynamically model the build-up of surge over the entire lifecycle of a hurricane event, rather than equating a storm surge depth to an event's wind intensity at landfall.

In addition to introducing more improved storm surge modelling techniques recent advancements also include developing methodologies to model the failure of flood defences that currently protect the New Orleans Areas. This can simulate the probability of levees failing and then calculate the potential losses for the event based on the storm surge intensity. The methodology allows the simulation of hurricane Katrina in 2005 using the old levee system to recreate the actual event and compares that with a similar scenario applying the defence systems as they are today.

Outside the insurance industry, similar tools are also being used today to help companies understand their financial exposures in different parts of the world and make more informed decisions around their risk management decisions, whether that is to set risk retention and insurance limits or to aid risk mitigation strategies and improve operational resilience.

But catastrophe models are not the only tools that help companies assess disaster exposures to their business, nor should they be because of the limitations of models in terms of the types of peril and geographies covered as well as the inherent uncertainties that exist around loss estimates. For a comprehensive approach to catastrophe risk management these models should be considered in combination with other tools and services, such as actuarial modelling, GIS (Geographical Information Systems) hazard mapping techniques, risk engineering, and the establishment and testing of appropriate business continuity planning and emergency response procedures to address both pre-loss preparedness and post-disaster recovery. A recent study for example showed that models are not always appropriate for assessing the wind and storm surge impact for single sites – having been involved in the assessment of a large commercial facility we saw it became evident that an onsite risk inspection of the facility was required to better understand the assets at risk as well the adequacy of the local defence systems. The risk engineering visit concluded for example that maximum impact to the facility following a major storm is expected to be due to surge flooding of the lower basement areas knocking out business critical electrical and mechanical infrastructure and resulting in significant business interruption. Whilst this phenomenon is generally not captured well by the desktop models, the onsite analysis was able to supplement them by providing more accurate loss estimations to assist more informed decision making around appropriate insurance cover, as well as offer solutions to mitigate future risks.

Research is also being done on linking the information about the hazard itself to damage and loss. The link between losses and hurricane intensity has also been investigated through two Willis Research Network (WRN) fellows, Dr Jeff Czajkowski at Wharton Risk Management and Decision Processes Center and Dr James Done at NCAR. They have been looking at how factors other than just wind strength, traditionally used as the main measure of hurricane intensity in the insurance industry, impact damage and loss. Two storms that made landfall in the same area were studied, and chosen because they contrasted each other in many physical characteristics apart from the fact that they reached similar maximum wind speeds. Factors such as size, duration and changes in wind direction during a hurricane approach are shown to have significant contributory influences when assessing storm impacts.

Scientific efforts have also helped our understanding of risks and application of new technologies. For example, long-time WRN partners at NCAR are keenly aware of the kind of failures that led to such catastrophic impacts. During Hurricane Katrina, a combination of failures led to an overall impact that was far worse than the sum of each individual breakdown. The need for a different approach has become clear. By setting up the Engineering for Climate Extreme Partnership (ECEP) NCAR have the opportunity to create a synergistic framework to provide key tools for the public and private sectors to improve their quantification of risk and build resilience towards extremes of weather and climate.

10 years later…

In the ten years since Hurricane Katrina, much reconstruction has taken place to protect New Orleans and its surroundings from future storms and surge flooding. The levees protecting the city have been rebuilt and new storm pumping stations have been installed that are designed to evacuate flood waters into nearby reservoirs and lakes.

Engineers have increased the levee height and replaced many of the old levee system's concrete I-shaped walls with T- and L-shaped walls, which consist of steel support beams that extend as far as 65 feet (19.8 meters) below sea level, which should provide better support if the levees fill to near capacity during a storm. Work has also taken place to regenerate and improve the wetlands, which can act as a natural buffer against hurricanes to slow down storms, reduce wave size and decrease the energy of storm systems. Aside from enhancing the physical protection around the city, other measures were implemented including changes to policy and the improvement of evacuation plans, communication systems and alert information. All of these measures have served to enhance the area’s resilience to natural disasters.

Officials have pointed to the successes of the evacuation for Hurricane Gustav, which hit the region in 2008, as evidence that the city is far better prepared for the next storm. During Hurricane Gustav 97% of citizens evacuated on their own, and 21,000 evacuated with the city's assistance during the storm. An estimated 1.9 million residents evacuated the broader southern Louisiana region before the storm landed on September the 1st 2008, only three years after Katrina dealt such devastation. With more effective evacuation procedures in place, worked continued on reconstructing the levee system around New Orleans. When Hurricane Isaac arrived in 2012, the levees were put to the test. It made landfall as a category 1 hurricane, but even though the wind strength was significantly lower than Katrina, it was still a large and slow-moving storm which led to a significant storm surge, approaching the levels experienced during Katrina. The new levee system worked as planned, averting widespread flooding in protected areas. It has been estimated that had the improvements not been made, the resulting flood event would have been similar to Hurricane Betsy in 1965, which caused approximately $1.2 Billion in damage. However, as a word against complacency, Isaac did cause significant damage to homes and businesses outside of the levee protected areas.

There’s little doubt that the new protection systems make the region more resilient against another hurricane of Katrina’s ferocious strength, but this is not to say that an even stronger storm won’t hit the area in future. Protecting the city of New Orleans from the worst of storms is next to impossible but the city lives on, learning from catastrophes in the past to make it better prepared for the future.

This blog is based on “What have we learnt from Hurricane Katrina, 10 years on?” by Marc Lehmann, Strategic Risk Consulting team at Willis (published in “Resilience Magazine, Issue 6: Data Vision” with additions by Geoffrey Saville, Willis Research Network Atmospheric Hub Leader.