Facebook

Login

Support Sailonline

If you haven't already - join the SAILONLINE YACHT CLUB!

Please also consider making a donation - all amounts are greatly appreciated!

Board » General Discussion » Weather - Brain Storm

Page: Previous 1 2 3 Next

Hi ij and Bonk,

The picture based on WRF definitely looks a lot better and more realistic than the course GFS based map. GFS is known to be very poor when it comes to the winds in coastal areas.

What do you think about using ECMWF, which is 0.5x0.5 and updates twice a day? It's global and I think as a non-profits they would let us use it on a free license if we ask.
Updating every 12 hours is also easier for a lot of SOLers with daytime jobs, or with wives that complain about their hobby that keeps them out of bed too often ;-)

The quality difference between GFS and ECMWF is easy to see on windy: https://www.windy.com/?-6.375,154.841,7,i:pressure,m:dwIakqh

The winds over and around the islands are much better in ECMWF.

Other global models like ICON or Arpege could be considered too. They are also updated every 12 hours, but higher resolution than GFS: https://opengribs.org/en/gribs

Windy has a great article on the pros and cons of the most widely used WX models: https://windy.app/blog/what-is-a-weather-forecast-model-guide-on-forecast-models-all-around-the-world.html
Many numbers you see on those lists are not correct. ECMWF updates 4 times per day but only to +90h on every second run (Similar to some other forecast runs with other models).

Also, they explicitly mention not-for-profits in their license selection guidance so you're just dreaming and there's little point in trying to ask anything from them.

--
i.
Of the available forecasts, these regional ones certainly seem interesting for higher than GFS precision:
- ICON-EU
- AROME
- NAM conus & maybe caribbean nests

For global models, I don't think there are freely available alternatives with enough incentives to make it worth replacing GFS, and many have clear disincentives such as non-even forecast lengths for different times.

Without at least seriously trying out with GFS 0.25 x 0.25 first, I wouldn't spend too much effort on the other global models that are freely available. (ECMWF is not freely available, if it would be, it would certainly seem attractive enough).

In practice, the blocker for any high-res wx is the lack of tiled wx support which requires too large modifications before some other blockers are removed first so it's not going happen over a weekend (I've definitely made some progress towards that direction but it's a big task with rather limited help so far). Even the WRF weathers which we've not been able to use for some time were somewhat big data sizewise for anything else than those small sprints (but certainly made the races much more interesting, IMHO).

--
i.
Also, they explicitly mention not-for-profits in their license selection guidance so you're just dreaming and there's little point in trying to ask anything from them.
--
i.

Is it possible to get access to the ECMWF model for internal use only on the server while using the GFS model in the gribs provided to users?

It seems to me that this would fit the licencing requirements for the not-for-profit use of the real-time ECMWF data.

It would also be closer to reality in that the underlying model used on the server would be different to the forecasts available to users.

There may also be a need for a new SOL rule that makes it illegal to use commercial paid-for products not available to all users for free.

I think this would make for some more interesting racing.
Most IRL races certainly have a rule about data used having to be freely available to all once the race start. But that doesn't mean you can't use Expedition, for example. It means you can't use PredictWind routing, or other professional services. So, dtayls, what did you have in mind with the rule? Just cover for-pay weather services? I assume no one's using that now, but you think if the server was using something that was made unavailable through the rules, it would be more like real life? I'd agree. But, the counterargument is that we're not actually living on the boats so you need some measure of predictability.
VIbex: The key to my idea is to use an underlying weather model for the simulation which powers our virtual boats that is different from the predictions available to SOL competitors.

I agree when you say "we're not actually living on the boats so you need some measure of predictability" but I don't think that means we need to be able to 100% predict the weather.

When the weather is very stable and highly predictable then the models will likely be very closely aligned. In those cases it might be safe to sail very close to a mark or an island, but when the weather is less stable you might choose a wider safety margin over absolute performance.

As for the question of using 'rules' to control which models can or can't be used, there is always going to be someone who wants to cheat.

Ideally, I like the idea of using something like a pre-release beta version of the ECMWF model not generally available except to researchers, if such a thing exists. Another alternative might be a psuedo-random merge of multiple model GRIBs ... This could be a fun project :-)
dtayls:I like it!
There's no differentiation for non-profits. The ECMWF licensing is using the very same conditions for non-profits and commercial. From https://www.ecmwf.int/en/forecasts/accessing-forecasts/licences-available:

"Provide a service using real-time ECMWF products

If you wish to provide a commercial, non-commercial, or not-for-profit service to third parties using ECMWF real-time products, you will need to accept the terms of the ECMWF Standard licence and pay for the associated data and handling costs."

I don't see any way around paying, regardless of non-for-profit or not (research, as expected, doesn't allow providing any services and it would be really twisting the rules anyway).

If a "secret" wx is desired, it's relatively easy to create one with WRF, just some parameter tweaks that remain secret are required. But then we will find out the WRF is volatile enough in itself so the secrecy loses its useful/meaning (one cannot plan long-term route with wrf even nearly as reliably as with gfs).

A "secret" wx would really just force people to sit in front of computer whole the time, especially in low TWS conditions, because twa steering becames a joke for racing purposes (or not to participate at all which is hardly useful goal to aim at ;-)). ...It gave me another idea though, maybe the platform should allow truly crewing the same boat with different usernames to share watches (different from private boat sharing that is of course possible even today). Then it would be a bit less solo mission to keep an eye on the wx.

--
i.

--- Last Edited by ij at 2020-09-29 19:48:35 ---
While they do indeed say "you will need to accept the terms of the ECMWF Standard licence and pay for the associated data and handling costs." what if that cost was $0 or the equivalent in Euros? I think you will find that their licensing model depends very much on how the data is to be used. The fact that SOL is a not-for-profit, that the data would not be made available for download to end users, and that we would not require 24x7 support might all have a big impact on the quoted price. Maybe someone should contact them and request a quote.

As it stands in SOL we are given perfect weather information for each 6 hour window which is a little unrealistic. What's more, some people know the idiosyncrasies of how the SOL server interprets the data and are therefore able to use that knowledge to their advantage. Is this really fair? And given that we allow this and the use of routers then we'd might as well allow full autopilot control of our boats!

If there is instability in in the weather then I would expect the underlying weather to reflect that and probably differ from the predictions. Those who understand the weather the best will generally make the best decisions and may even have a chance at winning over those who can drive the tools to perfection.

Ideally the underlying weather model used in the SOL simulation is based on the most accurate available. But some people may have access to this unless we 'rule' it out. So maybe we could also introduce some variations in to areas where the model is unstable. Again, this could be fun :-)

Page: Previous 1 2 3 Next

Please login to post a reply.

Races

Next Race: 00d 00h 00m


Current Races:

Tall Ships Races 2025 - Le Havre to Dunkirk


Welcome to the first of four virtual Tall Ships Races in European waters which are concurrently being organized in-real-life by Sail Training International . The course for this first race is between the French ports of Le Havre and Dunkirk; circa 125nm in Sailonline’s magnificent 90m Barque.
NOTE: Starts and Finishes in tall ships racing are always offshore to avoid conflict with shipping and shipping lanes; online and real-life may not match exactly.
Race #1923
INFO by brainaid.de
Barque 90m PARTICULARS
WX Updates:
0430 / 1030 / 1630 / 2230
Ranking: TS - SYC
Race starts: Jul 07th 16:00 Registration Open!
▶ Flash
GO TO RACE

Red Eye - Key Lime Pie-gatta 2025

Opposite the Bahamas at the tip of the Florida peninsula, we find the Florida Keys, a honky-tonk arpeggio of islands, where another SOL classic, the Key Lime Pie-gatta, so-called after the famous local condensed-milk-based tart speciality, awaits to be navigated once again. We’ll race its “Red Eye” 180nm in very-American Santa Cruz 52s, ignoring such artefacts as bridges, causeways and piers joining the keys one to the other and to the mainland!
Race #1846
INFOby brainaid.de
Santa Cruz 52 PARTICULARS
NAM_AWIP WX Updates:
0245 / 0845 / 1445 / 2045
Ranking: SYC - RED
Race starts: Jul 02nd 12:00 Registration Open!
▶ Flash
GO TO RACE

Sail Around Turkey 2025


Welcome to the 11th running of one of Sailonline's most iconic races, where our fleet races the length of the Turkish coastline from Hopa (the easternmost harbour on Turkey’s Black Sea coast) to Iskenderun (the easternmost harbour on Turkey’s Mediterranean coast). It’s a 1,372nm race and aboard our classic, 1st decade of the 21st Century, version 2 IMOCA 60, you should be able to get around the course in less than a week. That is if, of course, once you get out of the Black Sea you don’t stop to enjoy the sights or a BBQ or two in the Golden Horn, or on the Marble Islands, now land-locked Troy or Gallipoli or any of the myriad Aegean islands, that you shall have to carefully navigate past.
Race #1930
INFOfrom brainaid.de
IMOCA 60 v2 PARTICULARS
WX Updates:
0430 / 1030 / 1630 / 2230
Ranking: SYC
Race starts: Jun 27th 15:00 Registration Open!
▶ Flash
GO TO RACE

Go to race archive

SYC Ranking

  1. Sailonline Yacht Club Member CriticalHippo
  2. Sailonline Yacht Club Member KaSToR
  3. Sailonline Yacht Club Member WRmirekd
  4. Sailonline Yacht Club Member FreyjaUSA
  5. Sailonline Yacht Club Member rafa
  6. Sailonline Yacht Club Member Kipper1258
  7. Sailonline Yacht Club Member rumskib
  8. Sailonline Yacht Club Member BRENTGRAY
  9. Sailonline Yacht Club Member Patrick70119
  10. Sailonline Yacht Club Member CollegeFund

View full list

Series

Mobile Client

SYC members have the benefit of access to our mobile/lightweight web client!

The mobile client