NOAA maintains North Atlantic tropical cyclone data and history in a catalogue known as HURDAT. For a while now, scientists have reanalyzed data in the catalog in consideration of a wide range of historical information and scientific advancements to improve data quality and completeness.
Such reanalysis projects often result in changes to a storm's "Best Track," the authoritative scientific agreement of where the storm traveled and its characteristics along the way. These updates may change the historical or scientific significance of a storm. For instance, reanalysis may result in a drop or increase in central pressure, an increase or decrease in storm winds, and thus the storm's Saffir-Simpson category and perceptions of the hurricane record.
The Landsea et al (2004) reanalysis of the 1992 Hurricane Andrew provides a great example of this.
Until the reanalysis took place, Andrew was on record as a Category 4 landfall. Thus, between 1900 the time of publication (which can be extended to today) there were 2 category 5 hurricane landfalls in the US mainland on record (Camille and the Labor Day Hurricane of 1935).
The Andrew reanalysis project resulted in the scientific re-categorizing of the storm as a Category 5. Overnight, the probability of a Cat 5 US mainland landfall increased by 50%!
Recently, Kieper et al published a very captivating account of the reanalysis of the 1969 Hurricane Camille life history in BAMS. The article mixes oral history from affected residents and scientists with considerations of virtues and limitations of historical data collection methods.
Hurricane Camille has long fascinated scientists and the meteorologically inclined. The storm's intensity increased remarkably rapidly and caused widespread devastation. It has long been considered the strongest storm to make landfall.
However, the reanalysis project resulted in adjustments demoting Hurricane Camille's first place status as strongest to affect the US. The reanalysis resulted in downgrading the storm's peak intensity of 165 knts to 150 knts and scientists deepened historical central pressure data from 909mb to 900mb.
Those interested can find several other adjustments to the historical record in the fascinating article.
When data is rare such as, the case with hurricane events, changing understandings of the past and thus, expectations for the future, is a paramount difficulty.