Hurricane Florence and Forecast Confidence
The Twitterverse exploded the week of September 3rd and did you notice something odd? Were you expecting to be overwhelmed by endless tweets of potential landfalls locations? Instead, many were preaching about the massive inaccuracy of weather forecast models for hurricanes still a week away.
As a meteorologist, this struck me as odd. The information I was looking at certainly suggested that landfall in the Carolinas was more than random. Do we see a change towards a more cautious approach to the normal deterministic approach? Or were people in denial that such a storm would hit during such a quiet hurricane season?
Long-range weather forecasting is much more than throwing darts. Many that say weather forecasts beyond five days away is unreasonable and akin to guessing. As someone who routinely issues forecasts beyond five days, I can promise there is some skill from the weather forecast models. You just have to know how to use it.
Probabilistic weather forecasting, using weather ensembles, creates a likelihood of weather events working. It’s kind of like playing the odds. You can improve your odds of winning at Blackjack by counting cards but aren’t guaranteed to win every hand. The same is true for forecasting; you won’t get every long-range forecast right, but you can include your odds using ensembles.
Instead of dismissing long-range weather forecasts and denying potential landfall victims of much-needed lead time, maybe a new approach would help. On the other side of the hurricane-laden Atlantic Ocean is a different approach to issuing weather alerts. A weather warning matrix, used by the UKMet office, combines the probability of high impact weather events with the severity of the weather.
This approach for Hurricane Florence may have painted a different picture for those on the Carolina coast. The matrix would have shown the low probability of an extreme event over a week before landfall. This approach would be more helpful than a casual comment that landfall is unlikely. Each successive forecast would have then move Florence up the matrix scale, and those in the path could act in an appropriate escalated manner.
Let’s hope and pray that those on the Carolina coastline have received enough notice to keep themselves safe. Hopefully, the prediction of future storms will use a method that provides useful information for timelines that are historically called guesswork.