Seven-day hurricane warnings on the horizon

[HurricaneJoe22]

http://www.sun-sentinel.com/sfl-flbfore ... 4811.story

[RL3AO]

IMO we are not ready yet for 7 day forecasts. But I guess this wouldn't happen for a while. If they do go with the seven day forecast, I think it should be path only and not intensity. Maybe in a decade the models will be more capable.

[Derek Ortt]

basing making 7 day forecasts off of how model NCEP does...

oh boy! I may even recommend the EURO before I recommend model NCEP

[KWT]

models do seem to be getting better with the track but they still make large mistakes sometimes and its always going to be down to the human interpretation of what model is right...sometimes storms do what the outlier was showing, I remember Katrina diving SW with only the GFDL really showing that occuring but because it was the outlier the NHC went with the rest of the guidence. As it happens that probably shifted the landfall a good deal westwards 3-4 days later.

As for strength, it would be shocking if they tried to do that, even at 5 day range they can be hopelessly wrong with that.

[CrazyC83]

I think beyond 5 days there should be cones of uncertainty drawn on the maps, but no points. Although I'd be most concerned about oil prices with that with investors incredibly frightened nowadays (oil prices routinely rise $4-5 for mere invests these days).

[JonathanBelles]

I didnt look too much past 3 days before, looking at day 4 and 5 as "if this happens." I dont think we are ready quite yet for 7 days. Another decade.

[HURAKAN]

With our current technology a seven-day forecast is plain stupid. I don't know why there is a talk about a seven-day forecast when you can't guarantee that a 1-day forecast is correct.

It would just look bad for the hurricane experts when the public notices the giant errors.

[vacanechaser]

I didnt look too much past 3 days before, looking at day 4 and 5 as "if this happens." I dont think we are ready quite yet for 7 days. Another decade.

well thats really what the article said.. not saying they are going to roll it out this year, but in the future... will be something to watch for i guess.. but to use the GFS!!! lets hope they re-work that model first..


Jesse V. Bass III
http://www.vastormphoto.com
Hurricane Intercept Research Team

[Andrew92]

Quite frankly, I would rather focus on trying to do what we can to improve intensity forecasts than create seven-day warnings.

Seven-day warnings have high margins for error on not only the intensity, but also the track. But we'll still kind of have an idea of what may happen in those long-range forecasts as long as we have the five-day forecasts.

But we still don't have the best skills in the world in predicting intensity. As is, we miss intensities but quite a bit even in just a 24-hour forecast. We just simply don't know what makes a storm bomb at certain times, like Charley before it made landfall, Opal in the GOM, or even Andrew a couple days before its first landfall. I shudder to think what the margins of error would be if we continued these forecasts for seven days.

So, IMO, no seven-day forecasts should be made. Five days is good enough as is, because it still gives a little time to prepare if people haven't made their full preparations, especially if it looks like a storm could be headed their way.

-Andrew92

[RL3AO]

I think the advisories should be cone only. You can try to tell people to not focus on the center line all you want, but they are going to. It would be much better if you eliminate it all together.

[CrazyC83]

I think cone and intensity for advisories with curved lines for timing - no points as they are deceptive.

A 7-day forecast means little for us, since we seem to be able to predict (with little skill mind you) 10 or more days ahead thanks to models...

[Category 5]

This is the worst idea I've ever heard. We have enough trouble 5 days out so to go to 7 days and to use a lousy computer model is the icing on the cake. This has disaster written all over it. I hope that they drop this idea.

[MiamiensisWx]

This is a very iffy idea.

For what it's worth, short-term forecasts are very difficult. Even improvements will not change the fact that models should not be taken as the gospel (ad verbatim) more than six or seven days out. That does not imply that the model guidance is useless; they simply should be utilized properly. For example, does Model X properly analyze the upper-air environment, low levels, incoming Pacific vort max, et al features? If it does not adequately handle one feature, it should be compared with meteorological observations and other models in order to determine the most realistic handling of the synoptics. That is the only way to discern a plausible solution for a storm path and possible intensities.

Even the NHC illustrated some failures in 2007. For example, when Humberto stayed further offshore and "unexpectedly" strengthened to 70 mph under favorable conditions, a hurricane watch was not issued despite explicit wording in the advisory that winds "could approach hurricane force" along the Texas coast. Obviously, Humberto continued to strengthen to a strong Category 1 hurricane as it struck High Island, Texas. This error (lack of a prudent hurricane watch) occurred within a few hours of landfall. One forecaster stated that the forecast didn't anticipate a hurricane because the forecasted positions (read: "black line" of the cone) indicated the TC would come ashore earlier than reality. That really puzzled me, since the NHC always places emphasis for the public (and themselves) to "focus on the cone, not the line." The TPC always adhered to that philosophy prior to Humberto. Could Humberto have been a rare case where humans assumed machines, models, et al could solve the finer details of a forecast, thus leaving no need to focus on the cone (versus the line) within a short distance of land?

I really hope that was not the case. What do you all think?

Max Mayfield and other noted experts would have classified Humberto as an excellent case of the potential "nightmare" scenario. A TS rapidly intensifies to a hurricane prior to landfall. The forecasters rely too much on the models and "black line". The combination of that factor and the rapid intensification leads to scarce warning time and the lack of a hurricane watch. The storm strikes a populated area. In the end, the public (already skeptical of the NHC) finds a legit reason to complain, changing their loyalty to their own personal "weather skills" and other meteorological companies such as AccuWeather. Although AccuWeather certainly has a right to exist and they have made some good calls, the public's willingness to place trust into a "questionable" private company could lead to problems down the road. For example, AccuWeather never called Humberto's intensification better than the TPC's poor performance, but they could receive a disproportionate portion of unwarranted trust in this case.

Friends, my last paragraph is the potential nightmare scenario.

NOTE: No AccuWeather bashing was done in the production of this post. I respect AccuWeather's right to exist, but they have made several questionable actions.

[Blown Away]

I'd love to see the 7 day error cone, I'm sure it would be funny.

[wxman57]

Our clients in the NW Gulf sometimes take as much as 7 days to prepare for an impact (time to shut down/evacuate). These giant semi-submersible platforms frequently sailing out of Ingleside, TX need 7-10 day windows (no tropical activity) in order to proceed. Therefore, we've been preparing 7 day track forecast for the past few years for these special clients only. In fact, we provide these track forecasts for tropical waves/disturbances and estimate development chances along the way. The engineers at the companies demand some kind of track and an estimate of development chances and earliest possible arrival time of TS winds.

But these clients have very much to lose (billions of dollars) if an unexpected storm hits. They cannot afford to take chances. Most of our clients would find less value in a 7-day track as few companies have plans that call for actions that far out. I can't imagine the NHC producing 7-day tracks in the near future. Maybe 10+ years from now. 7-day tracks would be of no use to the general public due to the large margin for error.

[HURAKAN]

It could look like this!

[OuterBanker]

Wx, I assume that your work for the industry is discrete and for the industry only to avoid a catastrophe. Not for public release as to avoid spikes and speculation. I'm afraid that if the NHC releases a seven day cone that takes in the GOM it will cause the news media to speculate the potential catastrophe and create undue price hikes.

[canegrl04]

This is an idiotic idea. What has changed since last season to make them think it would be more accurate or justified?

[HURAKAN]

This is an idiotic idea. What has changed since last season to make them think it would be more accurate or justified?

The article says in about a decade or more. My opinion is that if you can't make an almost perfect 24 hr forecast, what would make you think that you can make an accurate 168 hr forecast?

[senorpepr]

well thats really what the article said.. not saying they are going to roll it out this year, but in the future... will be something to watch for i guess.. but to use the GFS!!! lets hope they re-work that model first.

Let's hope they DON'T re-work the GFS. GFS is a great model for other uses. As for as I am aware, the GFS was never intended to be a predominate tropical cyclone model. Don't retool the GFS for tropical work--there are plenty of other models to use. Let us that use the GFS for other business stay uninterrupted.