Blog

The first steps into a house are important ones. When it comes to assessing flood risk from hurricanes, first-floor elevation can be a key factor. Agencies such as the U.S. Federal Emergency Management Agency (FEMA) rely on data such as first-floor elevation in their damage modeling Hazus program.

A new dataset of pre- and post-hurricane damage and first-floor elevation for homes and buildings in a Southwest Florida city was awarded a 2025 DesignSafe Dataset Award, which recognized the dataset’s diverse contributions to natural hazards research.

“This dataset is unique,” said Mehrshad Amini, an assistant professor in the Department of Civil and Environmental Engineering at the University of Rhode Island. “We gathered very high resolution and component-level damage and first floor elevation data for close to 3,400 buildings in Fort Myers that were impacted by Hurricane Ian (2022).”

Daniel Cox and Andre Barbosa in the College of Engineering, Oregon State University; and Sebastiao Appleton Figueira at Stantec co-published with Amini the award-winning dataset: PRJ-5700 | Virtual Damage Assessment and First-floor Elevation Estimation: Application to Fort Myers Beach, Florida and Hurricane Ian (2022). The dataset is publicly available on the NHERI DesignSafe cyberinfrastructure.

“This dataset is the largest first-floor elevation data set that is publicly available for buildings, to my knowledge,” Amini added. “After the quality assurance process and quality control, our first-floor elevation estimates are more accurate compared to a national inventory, like the National Structure Inventory, which is publicly available.”

What’s more, the researcher’s dataset shows that hurricane damage to buildings is strongly correlated to distance from shoreline, foundation elevation, and construction age—important variables for consideration in future studies.

Virtual damage assessment

The guiding concept behind the dataset is virtual damage assessment, which is akin to conducting a post-hurricane damage assessment without physically being on the site.

During the data-gathering process, Amini and team trained undergraduate engineering students to perform the damage assessment. Training included selecting building coordinates, assessing building characteristics such as number of stories, checking if the building was washed away or not and if not, then navigating through different imagery resources, conducting damage assessment, reporting selected resources used for virtual damage assessment, and documenting damage observation and challenges such as whether the building view is blocked by debris.

“Our results show that trained engineering students can perform damage assessment as reliably as experts, while also yielding faster results at much lower cost,” Amini said.

The dataset team collected publicly available street-level drone imagery and NOAA satellite imagery, high water mark hazard data collected by the US Geological Survey and NHERI Structural Extreme Events Reconnaissance network (StEER), Homeland Infrastructure Foundation-Level Data, and more.

Data challenges

The team encountered data challenges in trying to do damage assessment at the community scale, covering as many beachside buildings in Fort Myers, FL. as possible. The challenges included inconsistent image quality from a lack of footage and data after Hurricane Ian.

“NHERI StEER was the best resource that we used for the damage assessment,” Amini said. For some regions that were not covered, his team went to the field and set up a GoPro camera atop a car to get another layer of footage to ensure coverage of all buildings.

Component-level damage data

The dataset they collected includes component-level damage ratings on a building’s roof, wall, foundation, openings, building type, foundation type, and overall building damage state—all separately categorized.

“One of the important components is information about first-floor elevation that’s useful for damage prediction models such as in FEMA depth-damage functions,” Amini said. That’s because that first floor determines flood risk, as it’s the lowest point of entry for rising flood water.

He pointed to another example in Florida called an Elevation Certificate, where new buildings need to document the first floor elevation data that’s planned with an engineer or land surveyor before a permit is allowed for a house.

DesignSafe—a reliable platform

“DesignSafe provided us a reliable platform to store and share our large data set,” Amini added. “Also, DesignSafe supports version control that gave us the ability to make sure the data is updated and remains as accurate as possible. This ability to share detailed and accurate data to the research and engineering community is very valuable.”

“DesignSafe also provided useful feedback and instruction on curation of the dataset to maintain and manage it for maximum use by researchers, model developers, and practitioners working on hurricane risk, resilience and damage prediction,” Amini said.

He added that he hopes the dataset will be used to validate and improve community-scale damage and loss models, such as FEMA Hazus, In-Core, and R2D. The data can be used to develop new probabilistic damage models, called fragility functions, that can help predict damage to buildings due to surge and wave effects.

“We also hope for its use for AI or machine learning in the future by researchers focused on using the latest advanced technology for damage prediction,” Amini added.

Research results

Amini and team published results from their DesignSafe-supported virtual damage assessment research in the journal Natural Hazards Review in February 2025.

“I have been in different conferences and workshops, where researchers are using this dataset to better model inland hazard characterization and damage. For example, many hazard models that use advanced hydrodynamic modeling do not include the existence of buildings in their models. Our data set is helping researchers study these effects,” Amini said.

Resilience planners are also taking note of where Amini met with a town planner in Fort Myers during the study to ensure they address the community’s needs and concerns, as well as share the dataset with them.

Said Amini says, “Good data makes for better models, and better models save lives and reduce costs. We still need more data to make sure we cover all building types and all kinds of different built environments in coastal regions.”