The cone of doom, destruction and general mayhem

[SkeetoBite]

Concrete examples are always good too. Out of curiosity: what's the largest 5-day landfall error that there's been?

The 5 day cone is currently an area 310.2 nautical miles from the center point of the 5 day forecast plot. This means that the current 5 day cone is 620.4 miles wide across the center point. The size of the cone for each forecast plot never changes. In cases where the cone appears larger, the cause us due to a slow moving and possibly random path where severl cones overlap.

The National Hurricane Center uses the average error for the past 10 years to create the cone of error. As stated by wxman57, the NHC indicated that the center of a given storm will remain within the cone are 60 to 70 percent of the time.

Concrete example
The best example of a storm deviating from the "line" at landfall but remaining well within the cone of uncertainty was hurricane Charley-2004. As the folks in Tampa prepared for the worst case scenario, the folks in Port Charlotte and Punta Gorda were breathing a sigh of relief. We all know what happened next.

Never focus on the "line" in a hurricane forecast map. The cone of error represents where the center point of the storm will be 60 to 70 percent of the time. There is of course the remaining 30 to 40 percent of the time that the storm strays outside the cone of uncertainty. Hurricane watch and warning areas always extend well beyond the cone of uncertainty due to the difficultie in predicting tropical cyclones.

Resources
Official 10 year Forecast Error Data Table (NHC) Requires Adobe Acrobat Reader
National Hurricane Center Forecast Verification page Everything you ever wanted to know about forecast verification

[CronkPSU]

I love it!!!

[f5]

remember don't pay attention to the skinny black line a storm is not a point as we saw with Katrina&Rita

[wxman57]

Concrete examples are always good too. Out of curiosity: what's the largest 5-day landfall error that there's been?

It's not too uncommon to have a 5-day error point that may be over 1000 miles off. Happens a lot with storms that are stalled then they start shooting off in one direction or another, or with a storm that moves a lot faster than initially forecast (like Charley last year). And such large errors on low-confidence forecast tracks are the reason why the cone gets so big at 120 hours. Currently, the NHC has no way to indicate forecast confidence by increasing or decreasing the size of the cone.

[AJC3]

Concrete examples are always good too. Out of curiosity: what's the largest 5-day landfall error that there's been?

It's not too uncommon to have a 5-day error point that may be over 1000 miles off. Happens a lot with storms that are stalled then they start shooting off in one direction or another, or with a storm that moves a lot faster than initially forecast (like Charley last year). And such large errors on low-confidence forecast tracks are the reason why the cone gets so big at 120 hours. Currently, the NHC has no way to indicate forecast confidence by increasing or decreasing the size of the cone.

I remember some of the 3-day forecasts for Alberto (2000) having absolutely huge errors. In fact, upon looking it up....

http://www.nhc.noaa.gov/2000alberto.html

"There were two periods where the official forecasts were far worse than average. The first was from 0000-1800 UTC 8 August, where four consecutive 72 h forecasts had errors greater than 500 n mi. This was due to the official forecasts calling for a west-northwestward motion while Alberto actually turned northwestward at 16-20 kt. The second was from 1200 UTC 11 August to 1200 UTC 12 August, where five consecutive 72 h forecasts had errors in excess of 600 n mi, including one of 939 n mi. This was due to the official forecast calling for a fast east-northeastward motion though the period where Alberto actually began to loop. Figure 4 shows the guidance for 0000 UTC 12 August. Note that the guidance is somewhat split into two clusters - a faster and somewhat further north set and a slower set and somewhat further south set. The official forecasts followed the former set, while the latter, though not perfect, was more accurate."

[abears]

I am glad to see something like this. I agree that others should talk about it too. After getting some of Katrina and a touch of Rita, I hope the media will focus more on the cone than on the center. Thanks to all of you. Since I take the time to read this site, I wasn't surprised when we needed to leave for Katrina.

Keep up the great job!

[simplykristi]

That's awesome!

Kristi

[tampaflwx]

If you look at the official NHC 5-day forecasts for Katrina and Rita several days before landfall (look at around 4pm Wednesday for Rita and around 5am Friday for Katrina), these are similar time frames from landfall to what we are at currently with Wilma. (in fact, the 4pm setup with Rita could be exactly the same if Wilma strikes the US on saturday morning/afternoon). If you look closely, you will notice that in each of these forecasts, the actual landfall is at the extreme outer section of the cone. Katrina shifted much more west than expected, and Rita much more east.

Keeping Charley in mind from last year as another reference, it would seem likely if not expected to see a large shift in most of the models starting right now. My intuition tells me that the models will take a more northern component considering that they will notice the weakness in the trough that is currently expected to be very strong in guiding Wilma. In addition, we are already seeing the storm employ a more westerly movement towards the Yucatan, indicating a broader arc and therefore more northerly landfall on Florida. This viable shift in models certainly warrants consideration and could impact the Tampa region more directly, IMHO.

[tampastorm]

Interesting post. Everything I read is saying the more the west component the more south it ends up. How do you figure a more north landfall? Thanks