Hurricane Laura made landfall early Thursday morning near Cameron, Louisiana with maximum winds of 150 mph. Hurricane Laura is the first Category 4 or 5 storm to make landfall in southwestern Louisiana during the period of record dating back to 1851, according to Colorado State University hurricane expert Phil Klotzbach. Unfortunately early images out of the region are consistent with what you would expect from a storm of this intensity. Because of its intensity and forward movement, it was still classified as a hurricane well inland at the time of writing. One thing that caught my eye was a Tweet from Professor Jay Cordeira that said, “NHC’s (National Hurricane Center) 3.5 day forecast for #Laura landfall was off by ~1 km. wow.” While there is no official confirmation of this observation, it is worth exploring.
Cordeira, a meteorology professor at Plymouth State University, told me by email, “I downloaded the forecast track/best track shapefiles from NHC and plotted them in QGIS….The ~1 km was derived from an approx. 0.01-degree longitude difference in the two lines at landfall.” If you don’t speak “metric,” 1 km is roughly 0.62137 miles. In other words, the National Hurricane Center got the exact location of landfall correct to within less than a mile according to his analysis. Nate Johnson, director of weather operations for NBCUniversal’s Owned TV Stations Group, added, “The NHC’s accurate track forecast for Laura wasn’t a one-off….In 2018, the 5-day forecast for Hurricane Florence’s eventual landfall point was off by less than two miles.” I also remember the European model hinting at Hurricane Sandy making a hard left into the northeast U.S. well beyond a week in advance.
Speaking of models, there are early indications that the American GFS model and some of the regional models like H-WRF performed very well with Hurricane Laura. However, the “Euro vs GFS” thing is not of interest to me because it is overblown in my opinion. Indeed, the European (”Euro”) model is better based on jargon-laden metrics used within our field of meteorology. However, the American model is still a world-class model, and both have their moments of success and failure. Forecasters at the National Hurricane Center use all of the available models, and their success with Laura three days out illustrates something that is well known in atmospheric sciences circles. Hurricane track forecasting has steadily improved over the past 50 years.
The graphic above reveals that forecast error has been reduced over time. In fact, a 3-day track forecast today is about as good as a 1-day forecast in the 1970s. In case you are having trouble deciphering the graphic, let’s explore the yellow line at the endpoint (2019). The graphic shows less than 100 nautical miles in forecast error at 3-days (72 hours). At 1-day (24 hours), there is less than 50 nautical miles in forecast error. This certainly places the 0.61 mile number in perspective. These advances are reflected in an ever-narrowing cone of uncertainty also though University of Miami meteorologist Brian McNoldy warns that “the cone tells you nothing about impacts, hazards, or the size of the storm.”
NBC’s Nate Johnson went on to say that Laura and Florence are shining examples of improvement in track forecasts going back decades. He closed by noting, “There is some year-to-year variation, but the trend lines point toward steady improvement at all forecast time steps.” The forecasts on Sunday morning, as Laura crossed the Dominican Republic and Haiti, were the first to shift west to a landfall location near the Texas/Louisiana border according to Professor Cordeira. He also pointed out, “With the slight exception of two forecasts overnight Sunday into Monday that brought Laura into central LA, every subsequent forecast for landfall location had very small forecast error.”
Accurate track forecasts allow emergency managers and policymakers to make sound decisions that save lives. According to NOAA, “A 20 percent reduction in track errors should reduce the average warning area by at least 10 percent (34 miles), saving the public more than $45 million over 5 years.” Additional information from dropsondes, satellites, and aircraft help to resolve atmospheric patterns that steer hurricanes. Unfortunately, intensity forecast skill still lags track forecast skill. The physics and underlying processes associated with intensity changes are not as well-represented, but a newer generation of models like HWRF shows potential for bridging this critical gap. Most forecasters clearly understood the potential for Hurricane Laura to rapidly intensify and it did.
As a disclaimer, this Professor Cordeira’s analysis is unofficial. Let’s wait for the official assessment from the National Weather Service, but it is certainly a great discussion point (and hopefully saved lives).