Zezo forum:

Post by defiant11 | 2015-03-15 | 21:19:49

Hi Cvetan,
With the new NoAA format, the 16 days routing and the summer time adjusted, could you please re-confirm the wind update schedule based in UTC.

Best regards

11 Comments

Post by RickW | 2015-03-16 | 02:04:03

All of the times are posted on the VR forum; Current Races, VOR page and the string for the Leg 5 briefing. Can't miss it. Predicted wind are 30-45 minutes later in Zezo charts as I recall.

Post by Bimmer | 2015-03-16 | 12:58:44

VR is now set to start tuesday 17-03 21:00 CET.

Post by zezo | 2015-03-16 | 15:48:50

Fixed

Post by zezo | 2015-03-16 | 16:36:09

On the subject question (all times UTC):

Update starts at 03/09/15/21 :50 with about 24 hours of forecast.
Then updates at 4/10/16/22 :00, :15., :30, :45 at which time the first 168 hours should be complete.

The forecast up to 384 hours is updated at 11:25/23:25. Only two updates/day because the long-term data comes in 12-hour increments.

Post by defiant11 | 2015-03-16 | 16:57:17

Thanks Cvetan, Clear.

Best regards
Def.

Post by Sir_Petrus | 2015-03-16 | 17:50:32

Please confirm if I am right. Just for the sake of those (like me) who wants to get out of bed in the middle of the night, but my 63 years old body starts fight my 10 years young mind.
So, pdates will be at:

03:50 ; 09:50 ;15:50 ; 21:50
predictions to the fore coming 24 horas.

04:00 ; 04:15 ; 04:30 ; 04:45
10:00 ; 10:15 ;10:30 ; 10:45
16:00 ; 16:15 ; 16:30 ; 16:45
22:00 ; 22:15 ; 22:30 ; 22:45
predictions to the fore coming 168 hours. (I don't understand this part)

11:25 e 23:25
predictions to the fore coming 384 hours.

Post by RickW | 2015-03-17 | 03:03:33

Sir_,
Then my mind must be 5 years. Where is the Whiskey?

I think there was something like this posted after the NOAA reformat. It was confusing to me. So, to condense this; the 24 hours come out on :50 and the next 144 hour predictions come out over the next 50 minutes. This is basically the 4/10/16/22 UTC updates. Pass me a glass of that juice. Right? I know this is 7-days

I am trying to get my pee brain around this. The first post of the Briefing for Leg 5 with the predicted wind times is wrong then; not UTC but CET, Cvetan gets then loaded on the schedule above. I think I have this right. Humor me with another glass.

Beyond 7-days comes the Game 12 hour wind cycle that gives us a look at 16-days.

This was helpful to lock in learning. Thanks my Knight Petrus.

Post by defiant11 | 2015-03-17 | 03:50:03

Hey Rickw,
The wind update in VR and from Noaa is not the same...Welcome to Zezo!

Good race to all!

Post by RickW | 2015-03-17 | 05:06:26

Thanks, I needed that !!! Goodnight.

Post by karriv | 2015-03-17 | 09:53:10

So this is my understanding on the matter:

1. Zezo and VR use the same source for winds, that is NOAA, i.e. Zezo loads wind directly from NOAA, not from VR. Thus update times for Zezo and VR have no direct relation
2. The wind model at NOAA gets updated in pieces, first days first, later days later
3. The first piece (first day) is ready at NOAA and updated to Zezo at around 3:50 (and also 9:50, 15;50, 21:50)
4. Consecutive pieces are ready and updated every 15 minutes after that so that the complete 7-day model is updated at 4:45 (and similarly for other updates)
5. The 16-day model is updated twice a day

On a separate note, regarding to waking up at the middle of the night, please note that NOAA doesn't always update as per the planned schedule. And if there is no update at NOAA, there can be no update at Zezo either.

Post by zezo | 2015-03-17 | 18:21:35

1. Yes, I get data directly from NOAA. VR may use a third party to fetch the data (they used to years ago, not sure about the current setup)

2. Yes, NOAA data is spooled one file a a time, 3 hours worth of weather data taking 2-3 minutes in real time

3. Small correction - first update at 3:40, then 3:50, 4:00, 4:15, 4:30 and at this time we have about a week of weather.

4. There are no incremental updates past the 186/192 hour mark, just one big update one hour later.

NOAA has two types of possible problems

1) Delayed model run. This rarely happens, maybe 2-3 times/year

2) Web server timing out. This happens sometimes. I guess there are few load balanced servers and one of them goes down.

Both problems can delay the download schedule, and if the delay is big enough VR may use a weather data from the previous model run (like 18Z+12 instead of 0Z+6)

This does not happen too often and there is no automated check to see if data is in sync. It's good idea to compare the numbers and drop me a note if you see differences in the weather. Then I try to find the grib used by VR and apply it here. There was just one occasion when the VR data was so out of whack that I could not find the corresponding GRIB file.