Thoughts on Seasonal Hurricane Forecasts

[MWatkins]

This isn't specific to anything this paticular person has "forecasted", but here are my thoughts on these type of predictions.

1. It's very hard to take them seriously. The best forecasters in the world can't tell you 3 days from landfall with any degree of certanity where a storm will hit let alone 3 to 6 months out.

2. These type of forecasts never seem to "threaten" unpopulated areas or even smaller cities.

3. As someone mentioned in another thread, there seems to be a high correlation between the location of the forecaster and predicted landfalls, or in some cases, subscriber bases.

4. There is a varing degree of science in some of these outlooks, but typically, they tend to be heavily reliant on Dr Gray's forecast. Much of the same, basic level reasoning is present in allmost all cases.

5. They are deterministic, which is an inherent forecasting flaw when dealing with weather at any time, but especially in mid-long term predictions.

6. Most are never verified unless something forecasted actually happens. Talke a look back to any forecast from last year from this or any other person and see if they've determined error rates and verified them out. If you happen to find one, examine the methodology and see if the verification stands up and makes scientific sense.

There are a rash of these starting to come out and I will address this tonight...and this is not the first time I've come out against these as anything other than interesting reading on message boards.

That is also why you never have nor ever will see me create a seasonal forecast. We simply do not know enough about these things to do so accurately with a deterministic forecast...and I surely don't.

One of the things that sets the CSU forecast process apart is the fact that it is probabilistic...which again, is the only way I believe this can be done, and even then, I have some doubts...

Let me close by noting there is a big difference between some of the message board forecasts and those posted on various web sites. I have no problem at all with the message board members putting out there for discussion...that's the whole point of message boards. IMHO the line gets crossed when they start showing up on websites along with just enough content to give them implied authority.

MW

[gatorcane]

Yes, I agree very well-said.

[skysummit]

Excellent! It's one thing to throw numbers up as a guess of just how many storms we'll have in one season, but to begin pinpointing locations as strike areas is just obsurd...especially when you're a supposedly professional organization. Thanks for that post!

[jasons2k]

MWatkins, overall I think very well-stated.

A few points to add:
1) Although these are far from perfect (or even useful) today, they lay a basis for future research and refinement. 100 years ago we couldn't forecast anything 6 hours out for the most part, so some of this work has merit in that it is laying the foundation for future generations of forecasters.

2) Yes, most of these never get verified, but some organizations actually do. JB from Accuwx does issue a point-based scorecard based on his predictions. Whether or not it's scientific is up for debate...

[CHRISTY]

Looking forward to listeing to your show tonight.

[BayouVenteux]

It's worth noting that this year, for the first time, Klotzbach and the CSU group will issue a landfall forecast of sorts with their June update.
Not sure if it's an additional forecast format or simply a tweaking of the existing one. Anyone know?

7.2 Forthcoming Revised Landfall Prediction Scheme

We have recently been investigating the potential predictability of steering current patterns likely to be present during the upcoming hurricane season. No individual or group can accurately predict exactly where or when a particular storm will make landfall months in advance; however, we have found that using a combination of our NTC forecast and several April-May steering current predictors, we can improve our landfall probability scheme considerably. We are currently working on documentation of this revised landfall prediction scheme, and it will debut with our 31 May update of the 2006 hurricane forecast.

Excerpted from the April '06 Extended Range Forecast
http://hurricane.atmos.colostate.edu/Forecasts/2006/april2006/

[Scorpion]

A broken clock is right twice a day...

[hcane27]

MWatkins, overall I think very well-stated.

A few points to add:
1) Although these are far from perfect (or even useful) today, they lay a basis for future research and refinement. 100 years ago we couldn't forecast anything 6 hours out for the most part, so some of this work has merit in that it is laying the foundation for future generations of forecasters.

2) Yes, most of these never get verified, but some organizations actually do. JB from Accuwx does issue a point-based scorecard based on his predictions. Whether or not it's scientific is up for debate...

... Hurricane Alley actually publishes a graphic for verification ... visibly demonstrating just how inexact they can be .....

[GeneratorPower]

Even a blind hog finds an acorn now and again.

[Air Force Met]

well I forecasted 23 named storms last season and I was very accurate. Take that NOAA!

Do it again this year...and again next year...and again in 2008 and you can stick it in someone's face. Until then, nobody's impressed.

My unit is on the 19th annual hurricane forecasting competition and out of the 10-15 of us (since it started), somebody has nailed it every year (and we ahve several categories...named storms, hurricanes, majors, highest cat, # in the gulf and number hitting Texas). We make it about more than just the number of storms. Somebody usually always beats Dr. Gray...unless he happens to be right. I've won 4 times (including last year) and come in 2nd another 6. We have excellent tropical forecasters and they aren't always right....and I am going to go out on a limb and bet you aren't going to be right either. If you are...maybe Dr. Gray should retire and you can put out the forecast.

How many are you forecasting this year?

[Steve]

Wait until next May and he will tell you.

/bring back the Midwest Hurricane Center, PLEASE!!!

Steve