NOAA Probabilistic Wind Charts

[Dockhead]

Conachair, one of our best forum geeks, turned me on to these in the course of a different discussion:


NAEFS Probabilistic Wind Speed Guidance


This looks to me like an exceptionally valuable tool for strategic passage planning/departure planning.


Is anyone actually using this? Any tips or observations?

[TheOffice]

Nice! Of course, the issue is always accuracy of long range forecasts. A pretty depiction of bad data is still bad.

[Dockhead]

Originally Posted by TheOffice

Nice! Of course, the issue is always accuracy of long range forecasts. A pretty depiction of bad data is still bad.

But this is not at all a "pretty depiction of bad data". On the contrary -- unlike normal forecast information, this takes into consideration the inherent quality of what is being presented. That is the whole beauty of this -- instead of presenting the best guess as to what the exact conditions will be, this presents the RISK of conditions above a certain wind speed. It is a whole other layer of sophistication.


Of course the evaluation of risk also involves quality -- how well do we know? Can we know, given the high degree of chaos inherent to weather? But we can for sure know it a whole lot better than what are the specific conditions a week from now. And I like that -- because I am less interested in the unknowable specific conditions, than I am interest in the risk of dangerous (or uncomfortable) weather, when making long range passage plans.


I'm just interested to know if anyone has used these, and how accurate or useful they found them to be.

[barnakiel]

Awesome.


I think it can be next to impossible to tell how to interpret this, without digging VERY deep into what the model does.


Cheers,
b.

[estarzinger]

Originally Posted by barnakiel

I think it can be next to impossible to tell how to interpret this

B, It is not that complicated. It is multiple model runs (eg "an ensemble") with random small perturbation to initial conditions and model coefficients. And the color coded probabilities are the percent of the runs exceeding the selected wind strength.

Note: This is basic to ensemble forecasting, which is where the pro forecasting world is today.

In essence, it captures the 'chaos' probability in the initial condition. Which is a good fraction of the total forecast uncertainty.

However, there is a second source of error/uncertainty, which is fundamental model error - eg there are in fact fundamental modeling differences between GFS and EU and neither capture full real-world complexity. So, this metric (probably) understates the uncertainty.


Dock, I hope you got to see some polar bear cubs but did not have to shoot any adults. (unfortunately a German cruise ship's guides shot one in Spitsbergen recently in lets say a situation of desputable urgency).

[wsmurdoch]

With no greater than 25 kt winds shown off the Florida and Georgia coasts, I am guessing that it ignores thunderstorms.

[barnakiel]

THX

I will have to dig deeper & learn more.

My probability background is in options/futures and portfolio analysis. There is some overlap with meteorology but some things differ quite a bit.

Thanks for sharing. Good to see there is more wx related info becoming available these days.

I work with wx data daily but not in the forecasting field just only in wx routing for adventurers (mostly at sea). Good to see forecasting services moving on trying to do a better job. Much needed.

barnakiel

[Pelagic]

Thanks Dockhead, I like this because it offers in one presentation, the "Trending Analysis"over an extended period.

Comparing this daily with changes in Subsequent presentations for the same updated timelines is very easy to see changes.

Together with your own forecasting, it will give you a far more subtle feel, for what to expect.

[Nicholson58]

Thanks, nice link.

NOAA has surprisingly useful presentation of massive data regarding weather. An especially great value considering the cost. The more time you spend digging through their sites the more crazy stuff you find

[estarzinger]

Originally Posted by Dockhead

Is anyone actually using this? Any tips or observations?

Oh, I see that I did not actually directly comment on DH's question.

I actually, personally, don't use this specific chart much. Because I don't really care too much if the wind is 10 kts different than the point forecast. (note: for offshore use, for those who do like it, it is possible to get this same data in grib format - it is originated in gribed format)

The main uncertainty I care about when routing is system development and track, and there are ensemble uncertainty charts which (I feel) show that better than this one - like (for one example) the ensemble spaghetti charts.

[thannever]

"I'm just interested to know if anyone has used these, and how accurate or useful they found them to be." "This question really skirts the reality of how probability works. since the service is predicting the probability of low probability events, the experiences of small numbers of boaters in relatively low numbers of tests will not tell anyone much. some unlucky cruiser will experience something they hoped to avoid and will think it doesn't work, while most of us will get something like the forecast or less and perhaps think it does.
this is not an opinion on whether it in fact is accurate. I will probably start looking at it myself. thanks for posting about it.