wind alert forcast accuracy at your location

[Toby]

Cumbuco, Brazil: add about 4 knots

[edt]

What model do you guys find to be the best or most accurate.

Now that's the question you want to know the answer to . . . I think it matters where you are geographically there doesn't seem to be one "best" model. Some models seem better tuned for particular locations, and in my area we have two or three different models just for the great lakes it doesn't give predictions outside this area, Hawaii has its own models only for the islands, there are some european specific wind models. It seems the specific models are better if you can use them because for one thing they offer a lower resolution. If you have 10k model it won't see any objects smaller than 10k if you have a .5 degree model it won't see anything smaller than 56km it will entirely miss bays, lakes and only be good for a very general forecast.

But it's not the web site you want to focus on it's the models which the websites are only front ends for. Some web sites do have proprietary models, then it makes sense to talk about that particular model, but even with web sites with proprietary models, sometimes a public domain model will work better.

Last of all . . . all the models don't work very good more than three days away. That's just where weather science is right now.

[Bille]

...
Last of all . . . all the models don't work very good more than three days away. That's just where weather science is right now.

Yes, three days Max for accuracy ! But that's all they forecast for on the winds. They'll
try to add if it's gonna rain with a %, but more than a day or two out and the % always
changes.

Here's the one i use for Mohave ; works Good for summer with the S - SE winds, but
really Sucks for the North winds we get in fall & winter. When they call for S - SW, it's usually
S, but it switches earlier than when they forecast for just the South winds.

Does this sight work with any accuracy in your aria ? ---- :

http://forecast.weather.gov/MapClick.ph ... ijjsX_-EuR


Bille

[dyyylan]

Whatever the default for the area on windguru is seems to be fairly accurate during the season here. During summer when we rely on the power of jesus christ to give us wind, the only thing that is accurate is the direction. and using local knowledge of the direction you can estimate chances of it being rideable some time in the afternoon.

windalert's quicklook is a little optimistic here, but it's not too far off. Usually the numbers aren't 100% but it does give you a good idea of when it will be strongest, when it will drop off, etc.

[bnthere]

South Padre Island Texas is pretty on. Long range forecasts are generally conservative, predicted wind strengths improves towards short term. this is an easy spot tho, its almost always windy.

[BWD]

I'm unlucky to live in an area with weak summer thermals mixed with frontal winds and and frontal winds only the rest of the year.
But luckily I can make the best of this by looking at the models and real time sensors near my main spots and ranging out a ways in several directions.
Synoptic forecasts and larger scale pressure drive the wind, while models and local pressure observations help predict speed and direction, and the real time wind speeds help you choose which model to follow.
All the models can be right at times, provided what the model "thinks" will happen, actually happens.
The current observations give the best insight to choose which model the weather will approximate over the next 12-36 hours, and to timing, which is often off by 8 to 24 h from the models.
From there as they say, you pays your money and you takes your chances...
In other words, it can be a horse race, the models are like conventional wisdom and early odds, the local observations are like watching the paddock and the walk to the gate.
You can hit big money occasionally with odds, but you stay ahead by watching the horses....
And human forecasts?
They are the Bob Costas of meteorology, small and expensive but usually worthless!

[Laughingman]

I'm unlucky to live in an area with weak summer thermals mixed with frontal winds and and frontal winds only the rest of the year.
But luckily I can make the best of this by looking at the models and real time sensors near my main spots and ranging out a ways in several directions.
Synoptic forecasts and larger scale pressure drive the wind, while models and local pressure observations help predict speed and direction, and the real time wind speeds help you choose which model to follow.
All the models can be right at times, provided what the model "thinks" will happen, actually happens.
The current observations give the best insight to choose which model the weather will approximate over the next 12-36 hours, and to timing, which is often off by 8 to 24 h from the models.
From there as they say, you pays your money and you takes your chances...
In other words, it can be a horse race, the models are like conventional wisdom and early odds, the local observations are like watching the paddock and the walk to the gate.
You can hit big money occasionally with odds, but you stay ahead by watching the horses....
And human forecasts?
They are the Bob Costas of meteorology, small and expensive but usually worthless!

Nailed it!

[william_rx7]

A couple thoughts:
- Wind Alerts are typically on current conditions, not on forecasts. Do you really get an alert on that it's going to be windy 3 days for now?

- The accuracy of wind forecasts decreases exponentially over time. Most big models are 80-90% accurate in the 12-24 hour period, and accuracy falls off fast after that. Anything beyond 1 or 2 days is little more than a guess.

- I agree with the others that there isn't a significant difference between models. They are all created by giant government organizations (NOAA, Environment Canada) with billions of dollars of equipment, thousands of meteorologists, super computers and satellites.

- It's ridiculous to think that a service with a couple dozen wind sensors on poles 20' off the ground can compete with satellites and super computers.

- Weather forecasting is big business. If someone could truly demonstrate a significant and consistent improvement in forecasting, they'd be working for the Pentagon, or some commodity exchange, not a couple thousand guys looking to catch their next session.

- Most major forecasts models are only generated and distributed 2 times per day. Most websites hide the out of date forecasts, giving the illusion that they are changing every hour, but they aren't.

- It's important to compare a forecast with the current conditions, both in direction, and speed. If current conditions match the forecast, it's a good indicator that there wasn't a "surprise" that invalidated the 12 hour old forecast. The speed and path of a big weather system is hard to predict, everybody knows it's coming, but no one knows exactly when.

- Use the "big" events to determine what's next, not just the time. For example, you might see a forecast that predicts a big blast at 5:00, then flat after that. If the blast comes in early, it's likely to leave early too. In my experience wind models do a better job at predicting the trend than the exact timing of the events.

[edt]

nam updates 4 times a day, WRF 4 times a day, HRW 3 times, my local great lakes forecast updates 4 times a day, I think the normal amount of updates is every 6 hours not twice a day. It is very easy for a shoestring operation to get really nice forecasts a lot the modeling software is public domain you stick it on a computer for instance the NAM or GFS you can put on your computer. Computer today are so fast you can actually run one of these models on your laptop -- this would have been impossible 10 years ago. The reason you can get better forecasts than the ones available at the major web sites is they they have to forecast the whole world and their resolution is not very good. If you run the model yourself you can set the resolution at 1 kilometer and run it just for your local launch so it will pick up great regional detail not available in the regular forecasts. There is huge difference in the models, I know for instance that the great lakes models GLCFS or the regional WRAMS are way better predicting wind for me than any of the big national models like GFS or WRF. It's possible at your launch all the models are the same but not for me.

It's true weather forecasting is big business and I'm not sure why so much is free to download and run on your computer. I guess people make money on the value added stuff like videos of hurricanes and pictures of the sun with sunglasses on or smart phone apps?

[william_rx7]

So NOAA needs more laptops?