Forum menu
I use a mix of yr.no, met office and BBC. BBC is by far the worst. I also look at the rain radar and windly. On the day, it's all about the rain radar really.
Be interesting to know what the op does for work and how often he **** up.
Bomb Disposal
Terrible choice for someone who's colour blind
Surely it’s just safer to assume that it’s raining in Wales.
SOP at Greenman Festival at Crickhowell for the weather: Is it raining? Looks around, yes/no. Will it stop raining? Looks over shoulder, yes/no. Delete as applicable.
I used to use WeatherPro and paid a subscription for hourly forecasts, then Apple bought Dark Skies, tweaked it and now I get forecasts at least as good as WeatherPro and I don’t have to pay a subscription anymore. What I always do, though, is check the precipitation radar for hourly and 12 hourly because that gives a much better idea of what the weather is going to do. FWIW, Apple uses Met Office data for U.K. forecasts, data sources for everything in the Weather App can be easily found in the link right at the bottom of the main app page.
I use a mix of yr.no, met office and BBC. BBC is by far the worst.
The BBC now use data from DNT Group GMBh, parent company who produces the WeatherPro app, or used to. I also use Windy sometimes as well, but mainly I stick with Apple’s Weather app, it’s at least as accurate as any other, mainly because it’s local forecast is from locally sourced data, rather than European or American, and the BBC do show the difference between the European data and American when it comes to the weather, much of which is coming across the Atlantic, following the Gulf Stream and the Jet Stream.
Of course, you could always use a piece of seaweed hanging outside, or see whether cows are lying down or standing up.
Red skies in the morning and at night are actually quite accurate, along with high cirrus clouds.
Yer takes yer choice, etc…
It was dry with me until 1530 in South Wales as predicted, armed with that knowledge I went out for a ride at 1100, damp but no rain.
As some have said I checked with two apps the Met App and Accu Weather App to confirm.
BBC is by far the worst.
The daily "Weather for the Week Ahead" on iPlayer is pretty good. Bit more detailed than anything else.
Google Weather on my phone seems pretty accurate too most of the time.
Problem with places like Wales, Lake District etc is it can be chucking it down in one valley and bright sunshine in the neighbouring valley.
I’ve noticed MO has been less reliable recently.
Me too. For rain at least, still pretty good for temps and winds. Although they're not the sort of binary thing you notice like raining/not raining!
Look at the pressure charts, available in the met office app, or elsewhere. Learn to interpret them.
Look at the rainfall radar if it’s the next few hours you’re bothered about.
Don’t rely just rely on anything that simply gives the ‘happy sun’ and ‘gloomy cloud’ emojis.
Pack a raincoat for Wales, whatever the above says. And imo don’t drive for hours to walk the dog!
It’s a forecast, that’s why it’s called a forecast.
AccuWeather seems to be pretty much spot on for me / my location but I'll also look at Ventusky as it's possible to look at a few weather models and see where they overlap and where they diverge.
Funniest thread I've read in ages.
Do people really need/want spoon feeding this much?
"...but the app says..."
It's February, in the UK, weather is complicated, make your own mind up.
Terrible choice for someone who’s colour blind
I'm bloody colour blind (massively so) but when I chose to work I work with wood. Not allowed in the paint shop on my own.
I live in North Wales, where I live it’s been dry all day.
Missus was visiting her family in Anglesey and it was pouring down with really low cloud. Her drive home was through just about every weather system we have. We certainly don’t bother with weather apps here. The mountains and the sea mean that the weather isn’t exactly predictable.
Be interesting to know what the op does for work and how often he **** up.
I create mobile apps!!
actual chance of rain if I go out for a ride = 100%
actual chance of rain if I stay in = 0%
I use putting on my waterproof jacket on a bike ride as the most reliable guide to the likely level of precipitation.
The greater the time spent putting my waterproof jacket on during a bike ride the more likely the rain will stop when I have finally got it all zipped up and I'm ready to go again.
I did a 200km audax in Wales today. It was forecasted to be mostly dry. It wasn't. I put my waterproof coat on.
There's a lot of ill-informed opinion on this thread.
Tldr: some people aren't as clever as they think they are and it's very complicated
There’s a lot of ill-informed opinion on this thread.
Tldr: some people aren’t as clever as they think they are and it’s very complicated
You've hit upon the STW Near-Universal Post - one that can be applied to almost any thread on this site.
Cop out of epic proportions.
You're demanding the impossible and getting all wound up when it doesn't happen. Honestly this is offensively stupid.
The Met Office are as good as it gets. If you think you can do better, crack on. Why not let us all know how you plan to improve the current state of the art in forecasting?
It’s the UK. The weather changes on a whim, just deal with it
Unfortunately the weather doesn't look at the forecast 😕
At this time of year, it doesn't take much of an atmospheric difference to turn grey and overcast into mizzle or outright rainy. (Source: Pissed it down all over me on a supposedly 'dry' ride yesterday in the Dales).
This Feb has been really unusual in terms of temperature, which means there have been some unusual movements in our weather systems, and you can tell it has messed with their models a bit. 5-day Met Office has actually been slightly more accurate than BBC around here (BBC has basically forecast rain for the next three days non-stop, and it hasn't done that).
I generally look at all of them and take an average. Or look out of the window in the morning.
Have a look at the forecast map. Sometimes there's rain moving through all day. All it takes is a ten mile error in the path of the edge of the weather system that is hundreds of miles to make the difference between rain and sun in a specific location. The Met Office seems to always get the general set up right, just that sometimes rain misses or hits a specific location.
You need to look at the maps and preferably listen to the video forecasts that explains what's going on. If you just type in a location and look at the icon you're always going to be disappointed in a temperate location like ours.
My closest weather station is about 200 meters lower elevation than where I live. Often get forecasted cloud or rain, but actually get mist and fog. I guess I’m in the cloud. Go further up the mountain and it’s a sunny day.
We were forecast rainmageddon yesterday afternoon…..not a drop.
I was at Brenin yesterday, never rained once, where were you hoping to go?
Weather forecast models do include current climatology
As an alternative view, my phone popped up "expect rain at about 9pm" the other day, sometime around 5pm.
9:05pm, it started raining.
The ability to forecast rainfall with an accuracy of 5minutes, several hours out, to my precise location is a chuffing miracle of modern science and technilogy. occasionally the models are a bit innacurate, or the weather system is skirting you rprecise location. They are rarely totally wrong.
Plus as above, its february, in Wales. Assume rain.
There’s only a 3% of rolling a double 6 and my brother did it twice!! How can that be right??
Statistics just can’t be trusted these days
Reminds me of this:
"Bomb Disposal
Terrible choice for someone who’s colour blind"
I now want to start a band called The Colour Blind Bomb Disposal Unit
(apparently the Weather Girls has already been taken......)
The thing that gets me is the number of times the met office forecast will be right the day before, and often agree with the bbc forecast, and then on the day their forecast changes to something that turns out wrong, whereas the bbc forecast remains the same as yesterday and turns out to be correct. You would think that the forecast would get more accurate as the time horizon approaches.
And the Met Office spent a ton of money on a new Cray not that long ago, seemingly so they can also run everybody else’s models for comparison, as you see on their deep dives on YT.
It looks like Googles AI forecasting is looking promising and has outperformed conventional forecasting recently, maybe they should concentrate on providing lots of data points for it and see how it goes.
I remember Dark Skies being uncannily accurate for the weather in Horley, Surrey on the day I got married - like saying ‘rain in 14 minutes, stopping 11 minutes later’, so we got the jazz band inside and sure enough rain arrived in 14 minutes, and stopped 11 minutes later. Seemed to be able to do that precision a lot then. Wasn’t so good in recent years for some reason.
There are a lot of new AI models out there and some of them are demonstrating good "skill", i.e. they work well, for 7-14 day forecasts. Look on the ECMWF website for more details. A couple of companies are claiming very high levels of accuracy if you pay them lots of £. It could be smoke and mirrors or excellent, time will tell.
The Met office doesn't run other peoples models in that way as it is computationally expensive. The take model outputs from other agencies like GFS, ARPEGE and model those data sets. They also look at lots of reanalysis data too, to perform hindcasts to assess model skill/suitability.
Weather forecast models do include current climatology
Let me expand upon that... weather forecast models use current datasets as inputs to their calculations, this includes amongst other things sea surface temperature, windspeeds, precipitation and so on. They implicitly represent current climatology as the current observation data points includes any changes to date , e.g. look at North Atlantic basin sea surface temps, they have been rising for a while now (outside of El Niño/La Niña).
Weather forecast modelling does not include climate forcing due to the short time frames, so any complaints about the assumptions of stationarity of the climate aren't that relevant. They do not predict future climate as they are not that forward looking.
Re: AI forecast models, some good results are coming out but AI forecast models will not replace the standard ones any time soon. Expect to see hybridised models in the near future. Like any new tech they need iterations to become really useful and the current ones do not capture the extreme weather we can see such as storms. The future might look very different though
What does all this mean? We live on a big wet windy rock in the NE Atlantic expect lots of weather and sometimes the forecast isn't perfect
Meteo de France expalined that in the past they had a regional specialists who checked over the weather forecasts to confirm they were plausible. Cuts mean that's gone down to one for the whole of France so the computer predictions are mostly unchecked and go out raw, with some forecasts that an experienced human would call bollocks on.
@Edukator - they probably thought that as the forecast models improved, they could use less of the experienced (i.e. expensive) staff and save money.
Most of the utility companies did that here in and around the Millennium and quickly found out that they needed to get them back in to tell them the stuff that wasn't in computer models or documented. A lot of people in their mid-50s were suddenly earning good consultancy fees 🤦♂️