(Game wind = Zezo wind ) ⍯ (Grib winds = Windy.com)
Post by iconPost by BooBill | 2021-03-12 | 15:10:50

Currently, shortly after the start of the eNordstream race, I'm seeing huge discrepancies between the game wind (which matches Zezo expectation) and the expected wind displayed in Wind.com (GFS model) and a grib that I downloaded from VLM (file: gfs_NOAA-2021031206.grb).

In the grib, I'm looking at 10m wind map with a resolution of 0.5 x 0.5 and 3h step.

The difference is really quite large, 18.3knts @ 235 in game, 15.4knts at 234 in grib.

Am I looking at the wrong grib, or is there a data problem somewhere?

commenticon 6 Comments
Post by iconPost by Hardtack | 2021-03-12 | 16:14:35
The game and zezo both use the 1° GFS model.
If you want to look at the original data, this is a good starting place: https://nomads.ncep.noaa.gov/
The GRIB files used by Cvetan's router are also available for download: http://zezo.org/grib/gribv1/all.grib

Post by iconPost by BooBill | 2021-03-12 | 18:02:45
Hey, that's better. Lining up now. Thanks again, Michael.
Post by iconPost by BooBill | 2021-03-13 | 01:44:11
One more question. In the NOAA grib filter, which elevation (10m above ground or sea level) and which variable is the wind? There's like dozen of different variables you can download. Some of the labels make sense, but I can't figure out which one is the wind used in the game.
Post by iconPost by zezo | 2021-03-13 | 10:05:52
10m winds.

But the game uses particular interpolation formula that you probably won't find elsewhere. And also rounds the data a bit. So you won't get exactly the same numbers, even for the actual data points.
Post by iconPost by Hardtack | 2021-03-13 | 10:34:32
The files contain component wind, in UGRD and VGRD. Data is only provided for the 10m above ground level I think, at least that's the one I'm using.
Post by iconPost by BooBill | 2021-03-13 | 13:48:38
Thanks guys. So, in http://zezo.org/grib/gribv1/all.grib it appears you are splicing together the fresh partial update with the previous forecast, so that it's always the full 16 day forecast that you are using. Is that correct Cvetan? I think you said earlier you switch from the old after a set period (1hr?) to the new (without interpolation), till the end of what's downloaded, then back to the old to the end of the 16 day period.

That works great for me. What I'm looking to do is replicate as close as possible what you and the game are doing for the short term, then apply some alternative models for the medium and long term to look at alternative strategies.
Topics list
Copyright 2009 by ZEZO.ORG. All Rights Reserved.