If you follow the NHC forecast cone, it seems that in many cases, the center of the NHC cone near the end of the forecast period – e.g 4 or 5 days out – is often very large so sometimes the center of the cone is missed by the center entirely. For the first 24 hours, the track is usually pretty accurate although “pretty accurate” can mean “only” a 40-50 mile miss. Even for 48-72 hours the cone is often much narrower.
The reason why is quite simple: widely divergent hurricane models with no way to figure out which one will be correct except in hindsight. For example you often read something like this:
“The track forecast is still highly uncertain at days 4-5, with the GFS and ECMWF positions about 200 nautical mile apart by 96 h.” So by 4 days, the difference between two of the better models is 230 miles. By 5 days it could be 280 or 300 miles or more.
Now think about how far 230 miles is – more than the difference between hitting Havana and Miami. Nearly Miami to Daytona Beach Florida. More than the distance between Boston and New York. More than the distance between Naples and Tampa. More than Charleston, SC to Jacksonville, Florida. Just 50 miles more than Jacksonville, FL to Atlanta. Only 80 miles more than the distance between Houston, TX to New Orleans, LA. Nearly the distance between Santa Barbara to San Francisco, CA.
Why does this happen? NHC often uses the middle between the models that they use so you might have one off to the far right of the cone and another to the far left with others in that area too. In many cases in hindsight one model is clearly correct, and one clearly wrong, almost a winner take all scenario. In those cases the “middle of the models” ends up being wrong either way.
The NHC does the best they can and the cone shows the projections for the center of the storm and the cone is pretty good about getting it, but the cone is extremely wide. For example, looking at 5 days out the cone can be 400-500 miles wide (e.g. looking at one now, it is around 430-450 miles wide). That is a huge swath of land. For a large storm, impacts can be felt throughout the cone of course even if it is “only” tropical storm force winds.
We often joke, “well, city X is right in the middle of the cone 5 days, they should be safe.” And this is why: one model ends up being right, the other wrong, the average/middle of the two is often the worst projection.
The models will improve with time, more real time data, improved computational power etc. and they are certainly better than they were 20 years ago or 30 years ago. Andrew’s projected impact was 50-70 miles off 24-36 hours out 30 years ago, it has improved since. Remember, this is an inexact science so the prudent course is to look a the cone and prepare. Even if you are near the edge of the cone 4-5 days out, chances are that it could shift to include your area.
As always, be safe. Run from the water and hide from the wind.
Recent Comments