venta: (Default)
venta ([personal profile] venta) wrote2010-12-28 11:45 am
Entry tags:

And I'm beginning to freeze, I've got icicles hanging from my knees

Recently, a colleague who comes from a family of meteorologists has been explaining to me how the only reliable weather forecaster is the Met Office. People like the BBC lift their forecasts directly from the Met Office's predictions. Metcheck, he tells me, get their data from the Met Office, but use their own models and are unreliable.

I actually think Metcheck is more reliable than the the Met Office. On Sunday night, having seen the TV weather forecast, we hastily changed our plans and on Monday we fled south before the threatened snow arrived. The Met Office said snow, snow, snow. Metcheck said pshaw, just a bit of rain.

An sms from the mother this morning confirms that Metcheck was indeed correct. Bother.

Anyway, I'm ignoring meteorologicalTim for now and trusting Metcheck.

The forecast for our postcode on Wednesday is as follows:

Metcheck forecast for Wednesday, showing glitch

In case that image-compression was too much, here's a closeup:

Closeup of forecast, showing -206C

Eek. Time to panic-buy blankets.

[identity profile] fractalgeek.livejournal.com 2010-12-28 12:16 pm (UTC)(link)
Perhaps they ought to get some of the Pluto express scientists over to see what happens when an atmosphere freezes out?

[identity profile] vicarage.livejournal.com 2010-12-28 12:39 pm (UTC)(link)
I expect Metcheck, like most independent organisations, using US global data released under FOI rules, rather than UKMO data processed using the software what I write. They will have different forecasters looking at different model output, produced on coarser grids. I doubt they run their own models, just some postprocessing.

[identity profile] venta.livejournal.com 2010-12-28 06:30 pm (UTC)(link)
Well, I was definitely told that, unlike most places, Metcheck do run their own models. It is entirely possible the person who told me didn't know what he was on about, though :)

[identity profile] vicarage.livejournal.com 2010-12-28 06:32 pm (UTC)(link)
To run a decent model you need a top500 supercomputer. I don't know if they have one or not.

[identity profile] venta.livejournal.com 2010-12-28 07:29 pm (UTC)(link)
Hmm. I think Wikipedia suggests (http://en.wikipedia.org/wiki/Global_Forecast_System) that they do not. (Although they are presumably using a different model from the Met Office.)

[identity profile] motodraconis.livejournal.com 2010-12-28 03:10 pm (UTC)(link)
Woah! That's going to take more than a few electric blankets to sort out.

[identity profile] lathany.livejournal.com 2010-12-28 03:21 pm (UTC)(link)
Heh.

[identity profile] feanelwa.livejournal.com 2010-12-28 03:55 pm (UTC)(link)
I'm entertained that the colour code has actually wrapped around to call it hot, too.

[identity profile] feanelwa.livejournal.com 2010-12-28 03:55 pm (UTC)(link)
Colour code/number representing colour, you know what I mean.

[identity profile] venta.livejournal.com 2010-12-28 06:29 pm (UTC)(link)
Ah, yes, I hadn't realised that. I'd just interpreted it as "red == extreme thing you need to notice", but I think you're probably right that the bit of code that does the colouring thinks it's very hot.