If it's going to rain within the next 90 mins, this very British AI system can warn you

Computer scientists at DeepMind and the University of Exeter in England teamed up with meteorologists from the Met Office to build an AI model capable of predicting whether it will rain up to 90 minutes beforehand. Traditional forecasting methods rely on solving complex equations that take into account various weather …

[Sgt_Oddball]

By the internet gods!

We might have an actual use case for "AI" unique to a problem the UK faces...

Imagine when this on general release, we can finally be free of the tyranny of having the washing rained on. Our forgetting your umbrella.

Finally a utopia for all in Blighty.

/sarcasm

Still one can dream.

Coat icon because until then you never know when it's going to rain.

[DJV]

Re: "you never know when it's going to rain"

I'd say the best way to find out is to look at the BBC weather - if they say it's going to rain, it won't - if they say it will be sunny, it won't be - I suspect they are staffed by people who used to work for Gartner.

[Tom 7]

Re: "you never know when it's going to rain"

I live by the met office radar. https://www.metoffice.gov.uk/public/weather/observation/map Zoom in on home and set it running. 95% chance of taking the dog out for a walk for an hour and not getting wet. We're at 500' here so can get wet even when the radar says there is none, and the uplift can make rain where non existed 1/2 hr before but its the best I've managed so far!

[tip pc]

Re: "you never know when it's going to rain"

I use the rainalarm app on iPhone.

It. Uses weather radar and tells you if your current location will get rain in x time.

[Anonymous Coward]

Re: "you never know when it's going to rain"

Gold

[gfx]

Re: By the internet gods!

I usually look on buienradar.nl or weerslag.nl the last hour animation gives an indication, the prediction not so much. For the uk i saw this map https://www.netweather.tv/live-weather/radar

[big_D]

Re: By the internet gods!

I'd like even the daily forecast or 3 day forecasts to be anywhere near accurate.

We seem to live in a small pocket of no-rain. Our immediate area has been dry for several years now (well below historical levels or rain and well below surrounding areas currently).

Our water table has dropped by about 4M over the last 10 years. We have a well in the garden that we used to use to water the garden - the previous owners of the house even attached the pump to the house system and used the water for showers, washing machine and hot water. But no more. It dried up about 5 years ago and we've drilled 3 new holes since then, but, instead of the old bore hole of 4M, we've been down to the bedrock at between 9M and 10M and we have maybe a 20cm water column.

The local weather forecast often says rain. We get rain maybe 10% of the time it is forecast - the surrounding area gets the rain maybe 80% of the time it is forecast. The next town over (15KM) experienced flooding in the Spring, we didn't even get any rain!

Even the local weather forecast in the weather app is usually wrong when it says that we will have rain, especially heavy rain.

[jmch]

Don't need AI for this!

Just 1 line of code:

If location = UK then rain_in_next_90_minutes = true

(I'm sure the AI could be very useful for other locations, however!)

[Anonymous Coward]

Re: Don't need AI for this!

Reminds me of when visiting Skye years a ago a local person explained the "alogrithm" for predicting rain.

"If you can see the Cullins then its going to rain, if you can't see the Cullins then its raining already"

[Doctor Syntax]

Re: Don't need AI for this!

The more general form is "If you can see the $LocalHill then its going to rain, if you can't see the $LocalHill then its raining already"

There's a reason why East Anglia is drier than the rest of the UK.

[Tom 7]

Re: Don't need AI for this!

The Suffolk Mountain Rescue team were some of the hardest drinkers I've eve met.

[IGotOut]

Re: Don't need AI for this!

I love the Scottish phrase "Summer is the best day of the year".

[Tom 7]

Re: Don't need AI for this!

I remember a headline in an Aberdeen paper - 'Aberdeen sizzles at 66'.

[Doctor Syntax]

Re: Don't need AI for this!

I think the Lake District equivalent would be "Summer is the best time of day".

[Rich 11]

"AI" is the solution

I did like their practical assessment of success, ie ask the users.

Since I only usually care about the rainfall levels in my immediate vicinity and at times when I plan to step outside, I find that the standard trick of hanging up an old pine cone still mostly does the job. I want to know whether to carry a jacket when I leave for work, when I come back from work, or when I want to go shopping. These days, of course, it's no longer necessary to also hang a pine cone up at work and keep a spare jacket there.

Success rates may vary, just like with any other forecasting system.

[Paul Kinsler]

Re: the standard trick of hanging up an old pine cone

How does this work? Is it that you first go outside to hang up the pine cone, notice what the weather seems to be doing, then go back inside to adjust your garb appropriately? :-)

[Anonymous Coward]

Re: the standard trick of hanging up an old pine cone

Absolutely. This was invented centuries ago before we had windows.

(on the off chance you're not joking and you've never been near a wood, pine cones react to moist levels in the air and close when it's too moist to protect the seeds they hold)

[monty75]

Re: the standard trick of hanging up an old pine cone

Like a scrotum in a cold swimming pool

[Anonymous Coward]

Re: "AI" is the solution

We use a pebble in our part of Scotland. Get a pebble with a hole in it and tie it outside your bedroom window. Check it when you pull back the curtains in the morning.

If it's wet, it's raining.

If it's white, it's snowing.

If it's fuzzy/indistinct, the haar (North Sea mist) has come it.

If it's swaying gently, it's windy.

If your window is cracked, it's very windy (and you hung the pebble too close)!

If it's dry and not moving, it's frozen.

NB: there's no reading for fine and sunny as that's so infrequent as to be considered necessary.

[TRT]

Re: "AI" is the solution

I've seen a similar one with a bit of string.

String wet - raining etc...

String gone - hurricane.

[Schultz]

Re: "AI" is the solution -- But for what problem?

90 minutes forecast of rain sounds a bit unambitious to me. Where I live, the weather forecast quite reliably predicts rain (down to the hour) some 12-24 hours ahead. If I look at the weather radar, I might be able to extrapolate some 90 minutes without much training. Is this a case of an AI solution searching for a problem? Did they find a prediction gap at 90 minutes before rainfall that they can address?

I expect a lot more solutions of this kind in the near future: all that AI money requires reports to be written and success to be demonstrated. I am sure the researchers will spend the required time to find -- and solve -- problems you have never been aware of.

It's fashion in science, it's a recurring phenomenon and it always plays out by the same script. If you can read about the importance of some scientific field in your daily newspaper, you can take a bet that the important work has already been done. By that time, the serious scientists already laid the groundwork, solved the hard problems, and convinced the (generally skeptical) community that their work is the best thing since sliced bread. The fields then blossom, fertilized by generous financial support to the legion of less-innovative scientists that walk the well-trodden paths and plow the fields of yesteryear's innovation. A decade later you end up with a lot of Dutch tulips, and sometimes a turnip.

[Anonymous Coward]

Re: "AI" is the solution -- But for what problem?

Depends where you live. Suspect may not be the UK based on what you say about weather.

[Flywheel]

Huh

It's Yorkshire. I can look out of the window and can 99% guarantee it's going to rain today. And tomorrow.. and the day after that ... :-)

[LybsterRoy]

Re: Huh

As a Yorkie born and bred I can fully endorse that comment.

I now live in the Scottish Highlands and you don't ask if it's going to rain you simply ask when it's going to rain. Always carry a coat with a hood.

[Beornfrith]

Re: Huh

I wouldn't have pegged Lybster as being a particularly wet place, but then you're not far from Wick and it's bizarre ability to attract any and all rain to it...

[Mike 137]

Objective performance?

"with lead times from 5–90 min ahead. Using a systematic evaluation by more than 50 expert meteorologists, we show that our generative model ranked first for its accuracy and usefulness in 89% of cases against two competitive methods"

In all fairness, I've only read the abstract, but i'd really like to know the method's absolute false positive and false negative rates, rather than just its "accuracy and usefulness" rating compared with other methods.

Apart from which, given a 5 minute lead time I reckon any observant human could have a pretty reliable judgement about whether it's going to rain. At the 90 minute lead time, any observant human familiar with a given geographical area might also do pretty well. The most interesting report would therefore be how much better than an observant and informed human this might perform.

We seem to be constantly looking for ways of replacing the capacities of competent humans with complex machines. If successful this might well lead to a general loss of human competence, and there's no logical reason why this could not eventually degrade the competence of the population of creators and trainers of the machines. However, so far each machine at best has only a single narrow specialised skill, so we'd need an awful lot of them to replace the entire gamut of competences of a single competent human.

[Anonymous Coward]

Re: Objective performance?

But your single expert human can only look out of one window at once, and is probably not lookin out the window anyway because they are reading El Reg.

Agree with your general rant about the usefulness of AI though....

[Chris G]

Re: Objective performance?

When I first moved to Spain, I had a one on one Spanish conversation teacher as I am rubbish in classes.

My teacher's husband was the boss at the local met' office, on meeting him I asked about the forecasting success of the system and it's difficulties, as we lived on an island.

His answer was that the crucial part of forecasting after reviewing all available data, including that from satellites, was to look out of the window.

[Anonymous Coward]

Re: Objective performance?

to look out of the window

A technique that Météo France could do with learning, since they are quite capable of predicting bright sunshine all day even when it's been raining hard for the last hour.

[TRT]

Re: Objective performance?

I can tell you that the rain in Spain stays mainly in the plain.

[Chris G]

Re: Objective performance?

@TRT

And I can tell you that it bloody well doesn't! I live at 417 metres above sea level in the foot hills of a national park, my water deposit collects water from the house roof (20 cubic metres), it is the end of the summer and ought to be empty but is full up to the overflow.

We have had a few thunderstorms during the summer that produced hail big enough to make me nervous about my solar panels, especially after my Spanish neighbour showed me pics of what a hail storm did to his car, it was a write off! Hail stones bigger than golf balls.

[katrinab]

Re: Objective performance?

If I want to know what the weather will be like in London in 90 mins time, I look at what it is like in Swindon right now. That seems to work most of the time as weather tends to move from west to east.

[Mike Lewis]

British Weather

They're still waiting for it to stop raining more than ninety minutes so they can test it.

[Empire of the Pussycat]

Only three rules needed

It has rained.

It is raining.

It will rain.

[thondwe]

Radar

Err - Surely you get the same level of accuracy just by looking at the Rainfall Radar - a.k.a. looking out a bigger window?

[Peter Prof Fox]

Helpful solution to DIRECTION of rain

I'm rain averse. But if looking at a metoffice rain radar I can't tell what direction the rain is moving.

SOLVED! By the power of less than 200 lines of good old javascript you can see for yourself in a so-obvious-no-wonder-the met-office-aren't-interested way.

Here is my bench-top prototype. https://vulpeculox.net/misc/jsjq/rain/index.htm

[Dr Paul Taylor]

Re: Helpful solution to DIRECTION of rain

Is that moving back-and-forth or forth-and-back? Arrows (for the wind) would be clearer, and simpler to program.

[Anonymous Coward]

Re: Helpful solution to DIRECTION of rain

Quote "so-obvious-no-wonder-the met-office-aren't-interested"

Were you looking for something like this?

https://www.metoffice.gov.uk/public/weather/observation/map/#?map=Rainfall&zoom=5&lon=-4.00&lat=55.01&fcTime=1632937500

Hint, it's on the Met Office main website.

AC as I work there.

[BeefEater]

Re: Helpful solution to DIRECTION of rain

I have a link to that Met Office facility in my bookmarks and always check it before going out on my bike.

This process is way more accurate than looking at the actual weather forecast.

My only observation is that the radar seems to record a level of rain (drizzle) that is barely noticeable when standing in it, so it is a bit too pessimistic.

Strangely, I received two notifications from Google yesterday that it would rain in my locality in 2 hours time, and both were accurate (within a 10 min window). Have they rolled this out already?

[DS999]

Re: Helpful solution to DIRECTION of rain

I use an app called Accuweather here in the US. Its main screen will tell you if it predicts rain within the next hour, but I still find it is better to view the animinated radar screen and make my own guesses. Whatever prediction algorithm it uses for the next hour is not very "Accu", and in any case most of my rides are longer than an hour.

Occasionally my educated guess is wrong, but during the summer when it is hot and humid I look at having an unexpected soaking while on my bike as a good thing. Cools me down and gives the bike a free wash! Kind of missed that this year as it has been abnormally dry - only got caught in a downpour once.

Once we get to this time of year though I will avoid a ride if I think there's even a chance of getting wet. When it is no longer quite so hot and humid, getting rained on while biking is just cold and miserable (plus I'll be wearing a shirt and the back of it gets really dirty from all the crud the rear tire throws up!)

[Peter Prof Fox]

Re: Helpful solution to DIRECTION of rain

What a surprise that as a Met Office person you can't see what's missing from your example. Those patches or rain could be moving in any direction. What use is that for predicting the next hour or so? None at all. Hence the two snapshots compared. Everything is fine at the Met Office...

[MikeGH]

Have they heard of darksky?

[Dr KB]

Other Options

Or an Apple Watch - Other watches exist, but might or might not have this functionality.

[monty75]

Re: Other Options

Apple bought Dark Sky. Presumably for this use case.

[Doctor Syntax]

"weather forecasts can be good or bad in lots of different ways; perhaps one forecast gets precipitation in the right location but at the wrong intensity, or another gets the right mix of intensities but in the wrong places, and so on. "

It sounds like a job for quantum computing.

[Binraider]

I'm sure the kilowatts of power to run this model justify the much simpler approach of looking at the cloud, wind, and air pressure state.

Sure, accuracy varies a bit, but the 80/20 rule applies. Nobody needs to know that there's a 82percent chance of rain at 18:57PM. (+-18 Percent, +-90 minutes).

If your problem is that sensitive to the weather maybe you shouldn't be making it out of paper, outside.

[Yet Another Anonymous coward]

>Nobody needs to know that there's a 82percent chance of rain at 18:57PM

Except for outdoors events, sportball and airports

[trevorde]

Benchmark to beat

https://www.bbc.co.uk/news/av/world-24713504

[Plest]

Hmm, here's my solution.

I went up the woods and got some pine cones for free off the ground....I've saved myself £2m quid in Uni research grants to know when it's gonna rain!

[heyrick]

What!?

Satellites, loads of sensors, radar, supercomputers, AI, billions of banknotes, and decades of experience...

...and the "accuracy" is 90 minutes? That's worse than a rural granny.

[IGotOut]

I've a similar method.

Look out window... Done.

[sharpwolverine]

Hey, I can do the exact same by looking out windows.

[harmjschoonhoven]

@sharpwolverine No Windows, Apache inside.

[WolfFan]

You’re in bloody England. In September.

There will be rain in the next 90 minutes. Period.

[robert lindsay]

I remember this being proposed in the US

When I worked for The University of Oklahoma in the early 90's. The idea was greeted with abject horror, by the science team. FWIW one of the members of that team became Trump's Science Advisor. They deserve each other, is all I will say.

[Precordial thump]

Australia north of 20°S

Pr(rain) = 3 < month < 11 ? 0 : 1

[Cuddles]

If the rock is wet, it's raining

So the "AI" looks at radar maps of the current weather, and based on that makes an educated guess at what the weather will be doing in a hour's time. This appears to be almost exactly equivalent to looking out the window and trying to guess how bad the clouds are currently looking.

[Colin Bain]

The question of good does not seem to include concept that it actually predicts the rain if I read this right. Merely that compared to current models it is better than them. I haven't met a model that really works junked they set in front of a camera, so a total waste of time really.