Question About NHC Probability Cone

[wxman57]

I know that the NHC probability cone represents a running 10-year average error, not any specific confidence level in the current forecast track. But what is the percentage chance that the track will fall within this cone? I believe that the cone represents a 60% chance that the storm will track wihtin its boundaries based upon that 10-year average error. Is that correct? The graphics on the NHC web page would seem to indicate this, since the average 72 hour error is just over 200 miles and the error on this graphic below for 72 hours is right around 60%:

[JonathanBelles]

math is not my forte

[Derek Ortt]

I'll send an e-mail to a friend of mine who works at NHC

[WxGuy1]

NHC forecast tracks of the center can be in error; the average track forecast errors in recent years was used to construct the areas of uncertainty for the first 3 days (solid white area) and for days 4 and 5 (white stippled area). The historical data indicate the entire 5-day path of the center of the tropical cyclone will remain within the outer uncertainty area about 60-70% of the time. There is also uncertainty in the NHC intensity forecasts. The intensity forecast chart and table below provide intensity forecast and intensity forecast uncertainty information.

That would seem to indicate that the cone indicates 60-70% probability level.
From the bottom of any "Coastal Watches/Warnings and 5-Day Cone" graphic (such as at http://www.nhc.noaa.gov/refresh/graphic ... shtml?5day )...

[wxman57]

Well, I know it's either a 60 or 70 percent chance already. I THINK it's the 60 percent region that they're drawing.

[caneflyer]

The cone is drawn from, and exactly corresponds to, the long-term average errors. It just HAPPENS that the cone, so drawn, encompasses the entire track of the storm 60-70% of the time. (Perhaps a range is given because the numbers in the Atlantic and Eastern Pacific are different.)

[wxman57]

It has to be drawn either for 60% or 70% - it cannot be drawn for a range and have a solid outer boundary. It does make a difference which one. I'm almost certain ithat the outer boundary is the 60% percentile, but I just want to be sure before making other assumptions.

[vbhoutex]

It has to be drawn either for 60% or 70% - it cannot be drawn for a range and have a solid outer boundary. It does make a difference which one. I'm almost certain ithat the outer boundary is the 60% percentile, but I just want to be sure before making other assumptions.

If I am remembering the discussion correctly when Frranklin was here in Houston for the hurricane conference a few years ago it is the 60% that is being shown. hopefully derek's friend will clear it up for us.

[caneflyer]

No, you are still missing the point. The size of the cone is determined SOLELY by the average forecast errors at each time period. It is not drawn with any particular percentage in mind.

After you have the cone, you can then go back over the period of record and determine how often the actual track fell completely within the cone. There is no unique percentage that must fall out of such an investigation, but rather the resulting percentage will be determined by the distribution of errors. (If all the forecasts had the exact same error, for example, then the percentage would be either 0% or 100%, depending on how you wanted to count exact matches on the cone boundary.)

There is no reason why the cone in the Atlantic (drawn from the Atlantic average errors) couldn't happen to encompass 68% of the tracks but the Pacific cone (drawn from the Pacific average errors) encompassed only 62% of the tracks. I'm not saying these are the percentages, but it COULD be this way, and so the caption would generalize these results by giving the range of 60-70% and this would roughly apply in both basins.