Surf Forecasts not updating

thermalben's picture
thermalben started the topic in Friday, 14 Jun 2019 at 4:13pm

We're experiencing some issues with our surf forecast model, and as a result they haven't updated since Wednesday night. 

The reason for this is that the underlying data source (from NOAA) was unexpectedly changed mid-week, and it took a while for us to find the original developer who set this up for us (its rare that the wave model requires on-demand maintenance - we've experienced this kind of problem just a couple of times in the last ten years).

We have made changes to the code, but it's a tedious process to check if it works, because we can only verify every six hours when the model updates. So far we haven't been able to correctly patch it, but we hope to have it fixed ASAP.

Sorry for the inconvenience - we're doing everything everything we can to get it back up and running. 

thermalben's picture
thermalben's picture
thermalben commented Saturday, 15 Jun 2019 at 8:20am

Still haven't managed to find the correct patch, we're continuing to debug. Hopefully we'll have it fixed today. 

thermalben's picture
thermalben's picture
thermalben commented Saturday, 15 Jun 2019 at 4:12pm

Have applied a different patch, hopefully this will do the trick. 

thermalben's picture
thermalben's picture
thermalben commented Sunday, 16 Jun 2019 at 6:27am

Yep, worked. Back to normal now... almost.

There's just one problem.. due to the wave model not running for 48 hours, it's lost some of the initial conditions (when starting up the wave model from scratch, it actually takes 5+ days to become accurate, because it doesn't know of any pre-existing swells).

As such, the next couple of days are disproportionally small, especially in locations like Victoria where we're expecting plenty of surf today and tomorrow, because any swell energy 'generated' over the last 48 hours is unknown to the model. See our Torquay data point as an example (surf should be in the 3-4ft range today):

Sorry the inconvenience. 

surfiebum's picture
surfiebum's picture
surfiebum commented Sunday, 16 Jun 2019 at 5:18pm

Sadly it still matches the sunny coast forecast

dannyz's picture
dannyz's picture
dannyz commented Sunday, 16 Jun 2019 at 7:09pm

all good mate, technology is amazing for us most of the time, the odd hiccup here and there is understandable!

Kai_Harmsworth's picture
Kai_Harmsworth's picture
Kai_Harmsworth commented Sunday, 16 Jun 2019 at 9:14pm

Mistakes happen.... Always good to judge an organisation by the way it reacts to mistakes/breakdowns in service - You guys have nailed it.
cheers for the heads up

mjh

nicodisco's picture
nicodisco's picture
nicodisco commented Sunday, 16 Jun 2019 at 9:14pm

Good job guys (y)

Patrick0710's picture
Patrick0710's picture
Patrick0710 commented Sunday, 16 Jun 2019 at 9:59pm

You still haven’t fixed me being a shit surfer!

Wood's picture
Wood's picture
Wood commented Monday, 17 Jun 2019 at 10:02am

Indonesia Reports are still wrong correct ?

backpointboy's picture
backpointboy's picture
backpointboy commented Monday, 17 Jun 2019 at 10:07am

Victorian Report is still broken

thermalben's picture
thermalben's picture
thermalben commented Monday, 17 Jun 2019 at 10:17am

Most AU forecasts should be back on track by Wednesday. 

Indonesia (and other distant locations) are a few days behind - the weekend is when the models should be back to normal.

In fact, the Indonesian WAMS sum up the situation really well - check 'em out for Ulus here: https://www.swellnet.com/reports/indonesia/bali/uluwatu/wams

Specifically, the period charts.

As of Sunday night (see below), because the wave model had been restarted only the day before, it wasn't aware of any pre-existing swells. It was starting from scratch - a becalmed Indian and Southern Ocean (which is theoretically impossible, however this how all wave models start at some point). The only action is S/SE of Madagascar, which is showing 24 hours worth of swell generation from a polar low/front combo.

By Wednesday, you can see the leading edge of this swell - the first one generated since the model was switched back on - pushing up towards WA, and half way through to Indonesia (notice the lack of long period SW swell through the northern Indian Ocean, but some small trade swell pushing through to the Maldives and the east African continent).

And then by Saturday, the leading edge of the first long period SW swell is into the Indo region (with more trailing behind), and the model is back to normal again.

carpetman's picture
carpetman's picture
carpetman commented Monday, 17 Jun 2019 at 10:23am

Good stuff Ben.

Are you expecting any differences in output with the new FV3 model? improved accuracy?

thermalben's picture
thermalben's picture
thermalben commented Monday, 17 Jun 2019 at 10:34am

TBH I haven't had a chance to read up on it as I've been snowed under. However, each version upgrade usually results in small improvements in accuracy. I'll check out the release notes this week to see what they say.

That being said, the 'accuracy problems' with wave models are unlikely to be massively improved by small tweaks in wind source data input IMO. Without going into specifics, there's a couple of other area of interest at my end (whether they can be improved upon remains to be seen).

Ryan Hanley's picture
Ryan Hanley's picture
Ryan Hanley commented Wednesday, 19 Jun 2019 at 10:43am

Ben, is the forecast fixed yet ?

thermalben's picture
thermalben's picture
thermalben commented Wednesday, 19 Jun 2019 at 11:18am

Depends where you are. Most of the East Coast is getting close to normal now, southern coasts (WA, SA, Vic, Tas etc) are a few days away and Indo looks like it'll be back into the swing of things early next week. 

theblacksheep's picture
theblacksheep's picture
theblacksheep commented Friday, 21 Jun 2019 at 7:30am

Doesn’t seem like the Gold Coast is fixed?

thermalben's picture
thermalben's picture
thermalben commented Friday, 21 Jun 2019 at 7:54am

In what way? Looks fine to me.

reallylongdreads's picture
reallylongdreads's picture
reallylongdreads commented Friday, 21 Jun 2019 at 11:55am

My surf-skill forecast is trending down. Any models you can suggest I look at on instagram to help my motivation?

thermalben's picture
thermalben's picture
thermalben commented Friday, 21 Jun 2019 at 12:11pm

Personally, I recommend just about anything that's not on Instagram.

Ben Harding's picture
Ben Harding's picture
Ben Harding commented Friday, 21 Jun 2019 at 12:53pm

Is dbah south facing enough to cop the 6ft SF as modelled tomorrow morning Benny?

thermalben's picture
thermalben's picture
thermalben commented Friday, 21 Jun 2019 at 1:25pm

Our model overcooks short range S'ly swells (this has been documented before).

As per Wednesday's notes, there probably won't be much change from what I said: "The models are pushing 4-6ft for Saturday (south facing beaches in Northern NSW) but I reckon this is slightly overcooked, I feel it’ll be smaller around 3-5ft. North of the border we’ll see small waves across the outer SE Qld points around 1-2ft, with bigger (though very wind affected) 3ft+ surf at exposed northern ends. Expect slightly smaller surf on the Sunshine Coast."

Also, to be super pedantic, the model is showing 4-6ft for the Gold Coast's south facing beaches (4-6ft is different from 4ft, 4-5ft, 5-6ft, 6ft and 6ft+). 

Ben Harding's picture
Ben Harding's picture
Ben Harding commented Friday, 21 Jun 2019 at 1:48pm

All good and thanks man!

Kit Wilson's picture
Kit Wilson's picture
Kit Wilson commented Friday, 21 Jun 2019 at 6:52pm

Do you know when it will be fixed ? Magicseaweed sucks

jonnybamford's picture
jonnybamford's picture
jonnybamford commented Saturday, 22 Jun 2019 at 9:33pm

whats mornington peninsula doin for the next few days?

theblacksheep's picture
theblacksheep's picture
theblacksheep commented Saturday, 22 Jun 2019 at 11:14pm

Just that it has got the red banner at the top, saying the graphs are wrong?

thermalben's picture
thermalben's picture
thermalben commented Sunday, 23 Jun 2019 at 6:58am

Error message has been removed from the top of the forecast page (has been running fine in AU since Tues, but some OS locations i.e. Indo were a few days behind because of the longer swell window).