The National Hurricane Centre’s hurricane database HURDAT2 has information going back more than 175 years, but a meteorologist has warned that it is not necessarily accurate.

“If you are making assumptions that we have a 100 per cent complete, accurate database back to the 1850s, you are mistaken,” Christopher Landsea told the ILS Convergence 2026. “There are a lot of technological influences in there. It has a lot of warts in it because it is not perfect.”

Dr Landsea is chief of the tropical analysis and forecast branch at the National Oceanic and Atmospheric Administration’s National Hurricane Centre.

As part of the Atlantic Hurricane Database Reanalysis Project, Dr Landsea and his team re-evaluate historic storms including some that have impacted Bermuda.

For decades, the September 21, 1922 hurricane was considered one of the most powerful storms to hit Bermuda.

According to Royal Gazette reports from that time, the cyclone passed directly over the island, killing a sailor and a child. It caused widespread damage to buildings and farms and hit the island with an 8ft storm surge, flooding coastal homes and submerging Flatts Bridge.

For a long time, HURDAT2 listed the storm as a category 4 on the Saffir-Simpson scale, with winds as high as 130mph.

After analysing reports from the 1920s, Dr Landsea in 2009 downgraded the historic hurricane to a category 3, making it the fourth strongest storm to hit Bermuda.

Dr Landsea and his team were also responsible for upgrading Hurricane Andrew from a category 4 to a category 5 storm, a decade after it hit Florida in August 1992.

That reassessment changed how rare a comparable landfall in southeast Florida appeared to be and also how people thought about potential losses.

HURDAT2 data is used by forecasters, researchers, engineers and emergency managers, but also by insurers to estimate risk.

“We use the data at the National Hurricane Centre to validate our model forecasts as well as the official predictions,” Dr Landsea said. “It is used to look at climate changes, climate variability on decades, year-to-year and long-term trends.”

The meteorologist said one of the “neatest” things is finding new historic tropical storms or hurricanes. He explained that decades ago, meteorologists would wait before naming a storm. It was also easy to miss smaller storms if they did not impact land.

“For example, in 1970 we found a storm that is not in the database, but had a ragged eye for two days as it moved over the North Atlantic from east to west,” he said. “That does not happen unless it is a hurricane. Based on this, as well as reports from a couple of ships, we were able to add this as a newly discovered hurricane.”

His team uncovers around three to five new historic storms a year.

Dr Landsea explained that some inaccuracies came from the limits of earlier technology and the methods used to record storms.

For example, hand-drawn storm tracks digitised in the 1960s produced a systematic error: the final six hours of many storms appeared to show their speed doubling. Other problems arose because older records lacked precise landfall information. Hurricane categories were assigned using different measures over time.

Dr Landsea said the database was getting better over time.

The aim is to make the record more transparent and dependable.

“The more that you understand some of the works and the issues, the better it is for your industry as well,” Dr Landsea said.