Road to Resilience: Surveying Damage after Hurricanes
By Sarah Bobby, Tim Johnson, and Karthik Ramanathan
Whether in the aftermath of a Sandy, a Harvey, or a Michael, the main objective of post-hurricane damage surveys is to gain as much knowledge as possible about these typically infrequent yet destructive natural catastrophes. Surveys can be critical to assess the built environment’s response and subsequently used to continually improve catastrophe models.
Each survey has unique objectives, ranging from building performance evaluation and footprint validation to modeling assumption confirmation and evaluation of the secondary features of the building stock. Many surveys are joint collaborations, providing a broader perspective of each event and helping us to consider variables we may not have taken into account before.
When preparing for a damage survey, it’s important to understand what aspect is unique about the event to learn from it most effectively. For example: Sandy (2012) gave an excellent opportunity to study storm surge in a densely populated metropolitan area; Harvey (2017) demonstrated the destructive inland flooding that can accompany events that have stalled inland yet are close enough to the coast to draw in moisture from the sea or ocean (see Figure 1); and Michael (2018) gave the first opportunity to study a design wind-speed event (see Figure 2). This last was of particular interest because major building code reforms were implemented in Florida after Hurricane Andrew in 1992; the Florida Building Code was adopted in 2002 and has been continually updated ever since.
What information do we leverage from damage surveys?
The goal for each damage survey of a U.S. hurricane differs based on the characteristics of the event and where landfall occurs.
Some of the most important questions that need to be answered involve the hazard extent and intensity:
- How far do the damaging winds and storm surge extend?
- Is the damage consistent with the modeled wind speeds and/or surge depths?
- Are there anomalies not captured in the hazard footprints that can be explained?
While media attention tends to focus on the hardest-hit areas, it can be difficult to gauge the full extent of the storm and the impact to communities farther away from the spotlight without being on the ground. Presurvey evaluations using satellite imagery, media, and social media reports can help narrow the scope of potential survey sites.
When conducting surveys, it’s important to look for regional trends. Secondary risk characteristics such as roof shape, roof cover type, opening protection type, and wall cover type can vary by region. Regional construction trends as well as installation practices can be better recognized after conducting damage surveys.
Building performance and modeling assumptions
In many cases, the hurricanes are unprecedented in some way; for example, the affected regions may never have experienced an event of that type and strength, and only theoretical simulations were available to quantify their impact. Classifying the building stock for hurricane-prone states offers a theoretical expectation of how the building stock will perform in the event of a future hurricane. An actual hurricane tests how the model performs and helps determine if the assumptions are valid. A damage survey helps reconcile model expectations with ground truth.
The enforcement of building codes is another important factor in determining vulnerability. Consistent failure modes or poor construction practices can be verified in the field. The requirements for opening protection or lack thereof can be validated as well. Michael (2018) is a good example of a hurricane striking a region that had reduced the building code’s window protection requirement and suffered considerably. A survey following Florence (2018) reinforced understanding of the window protection requirements in North Carolina, which differ from those of bordering states. Harvey (2017) gave more insight into the variation in construction quality and how this can affect structural vulnerability.
Building stock in a given state represents the evolution of construction practices over the last 100 years or more at various points in time. The checkerboard of buildings illustrates the implementation of several building codes, construction, occupancy, and height groups; and there’s an expected performance level associated with each. Very rarely do we see a design-level event that allows us to evaluate the performance of building stock across many wind speeds, time bands, and multiple damage states; but Michael (2018) was one such example.
Important insights through collaboration
Collaborating with other researchers or industry professionals on a damage survey is mutually beneficial because each discipline brings a unique perspective. Catastrophe modelers aim to show the potential loss associated with an event; research institutions may be more interested in new mitigation efforts or very focused components; and brokers and insurers may seek to gain insights into the risk they write and parameters to reduce loss potential. In some cases, collaborators want to survey a very specific line of business.
Natural disasters serve as a litmus test for building code effectiveness and often act as catalysts for significant code overhauls. In a similar fashion, damage surveys serve as an impetus for catastrophe model enhancements because they provide insights gleaned from the field. The 2004–2005 hurricane season in Florida provided a treasure trove of granular data on the vulnerability of commercial lines. It also provided the perfect natural setting to test the impact of the building codes introduced in the post-Andrew (1992) and International Codes era, particularly the first edition of the Florida Building Code.
After Hurricane Sandy unleashed its fury on dense commercial areas of Manhattan and brought to the forefront the complex nature of surge damage to these building assets, we applied our findings to build storm surge damage functions for our U.S. hurricane model that reflect a component-level framework for buildings and contents that integrates primary and secondary building features. We were also able to grasp the long tail that events of this type have on the business interruption losses to commercial lines of business. Findings after Hurricanes Harvey and Florence served to enhance AIR’s understanding of the damage that precipitation-induced flooding associated with hurricanes can cause, and we anticipate that our U.S. hurricane model will be updated in the summer of 2020 to include precipitation-induced flooding.
When AIR conducts damage surveys, one element that’s difficult to quantify is the impact to the people we encounter. Some are homeowners who have lost everything yet are grateful to be alive; others consider themselves fortunate to have fared better than expected. We encounter some who do not have insurance because they can’t afford it or don’t really understand the risk. The lesson for us is that hurricane risk involves people and their lives. While we aim to make models that reflect the risk, we can promote resilience and use the models to highlight the insurance gap. Being on the ground after an event reminds us why we do what we do.
Sarah Bobby, Ph.D., is a senior engineer at AIR Worldwide, a Verisk (Nasdaq:VRSK) business.
Tim Johnson, Ph.D., is a senior engineer at AIR Worldwide.
Karthik Ramanathan, Ph.D., is assistant vice president and principal engineer at AIR Worldwide.