Jump to content
Lightning
Local
Radar
Hot?

Recommended Posts

Posted
  • Location: Home: Chingford, London (NE). Work: London (C)
  • Weather Preferences: Winter: cold and snowy. Summer: hot and sunny
  • Location: Home: Chingford, London (NE). Work: London (C)
Posted
PHYS.ORG

Being able to create accurate weather models for weather forecasting is essential for every aspect of the American economy, from aviation to shipping. To date, weather models have been primarily based ...

 

Posted
  • Location: Home: Chingford, London (NE). Work: London (C)
  • Weather Preferences: Winter: cold and snowy. Summer: hot and sunny
  • Location: Home: Chingford, London (NE). Work: London (C)
Posted
WWW.TELEGRAPH.CO.UK

As extreme weather events increase in number, an AI programme developed in the UK promises to save countless lives

 

Posted
  • Location: Islington, C. London
  • Weather Preferences: Cold and snow, cool and wet.
  • Location: Islington, C. London
Posted (edited)

What with it being noted how AI has set Google back on its emissions targets on the power to run all these machines, to the point of them being higher than they were in 2019, I can't help but feel this AI push will end up doing more harm than good... Especially ironic in the topic of weather and enviromentalism.

Edited by LetItSnow!
Posted
  • Location: Perth / Boulder, Colorado
  • Weather Preferences: Warm and sunny!
  • Location: Perth / Boulder, Colorado
Posted

 LetItSnow! The costs of running an AI model in terms of energy and time are many orders of magnitude less than a traditional NWP model.  The cost of training is high, but it's not clear how often (or if at all) this needs retraining.  If these models really have captured the underlying physics (and it is appearing in an increasing number of papers it has) then retraining/fine-tuning will be infrequent.  The existing AI models haven't been retrained at all, and their forecasts produce use about 10,000 times less energy.

  • Like 3
  • Insightful 1
Posted
  • Location: Islington, C. London
  • Weather Preferences: Cold and snow, cool and wet.
  • Location: Islington, C. London
Posted

 Earthshine Oh, well, good then. But AI in other areas has proven to be a problem. That's why I wondered.

Posted
  • Location: Leighton Buzzard, Central Bedfordshire
  • Weather Preferences: Just take whatever is offered.
  • Location: Leighton Buzzard, Central Bedfordshire
Posted

 Earthshine I'm impressed with them, they seem to be verifying very well indeed.   

Posted
  • Location: Ireland - East Coast
  • Location: Ireland - East Coast
Posted

 LetItSnow! Its possible that AI will increase efficiencies in production such that AI will cause a reduction in carbon output. That's what I am expecting. Obviously there is a ramp up to get the data centers operational, but the application of AI is in it's infancy.

Posted
  • Location: Islington, C. London
  • Weather Preferences: Cold and snow, cool and wet.
  • Location: Islington, C. London
Posted

 Downburst From what I can see, the emissions generated to having to store AI is a problem but could be mitigated if ran solely off renewable energy. Doesn’t help there’s been a frenzy with AI generated images creating not just those problems but moral ones too. I think AI, like its creators, is imperfect and nuance needs to be had when talking about it. It’s up to us whether it ends up being a helpful tool or a terrible tool. 

Posted
  • Location: Leighton Buzzard, Central Bedfordshire
  • Weather Preferences: Just take whatever is offered.
  • Location: Leighton Buzzard, Central Bedfordshire
Posted

 LetItSnow! I would say at the moment AIFS is doing very well and will probably take the place of the existing models..  

Posted
  • Location: Home: Chingford, London (NE). Work: London (C)
  • Weather Preferences: Winter: cold and snowy. Summer: hot and sunny
  • Location: Home: Chingford, London (NE). Work: London (C)
  • 3 weeks later...
Posted
  • Location: Home: Chingford, London (NE). Work: London (C)
  • Weather Preferences: Winter: cold and snowy. Summer: hot and sunny
  • Location: Home: Chingford, London (NE). Work: London (C)
Posted (edited)
WWW.TELEGRAPH.CO.UK

AI tool could be able to show heat differences between one end of a street and the other

A research team led by the Met Office used data from previous heatwaves in London to develop a machine-learning tool that could predict temperatures on a 100-metre grid. The results were 225 times more detailed than its standard forecasts. The study found that machine-learning methods of forecasting urban heat improved prediction of air temperatures by up to 11 per cent compared with the original weather forecast data.

Screenshot2024-08-13at18_40_52.thumb.png.adbd790737cdd858711022af77c379e4.png

 

Edited by danm
  • Thanks 2
  • 4 weeks later...
Posted
  • Location: Home: Chingford, London (NE). Work: London (C)
  • Weather Preferences: Winter: cold and snowy. Summer: hot and sunny
  • Location: Home: Chingford, London (NE). Work: London (C)
Posted
BLOG.METOFFICE.GOV.UK

Reporting on progress of the Met Office and Alan Turing Institute AI for Numerical Weather Prediction project.  The UK's weather is notoriously changeable, prompting everything from casual...

 

  • Thanks 1
  • 3 months later...
Posted
  • Location: Home: Chingford, London (NE). Work: London (C)
  • Weather Preferences: Winter: cold and snowy. Summer: hot and sunny
  • Location: Home: Chingford, London (NE). Work: London (C)
Posted
WWW.ECMWF.INT

Representatives from three European meteorological institutions – Météo-France, the German National Meteorological Service (DWD), and ECMWF – discuss how artificial...

 

  • Like 1
  • 1 month later...
Posted
  • Location: Lancing
  • Weather Preferences: Sun, Snow and Storms
  • Location: Lancing
Posted (edited)

Looking forward to these developments in the AIFS this year.

WWW.ECMWF.INT

A tour of the year past and some previews of the year to come.

Screenshot_20250119_021338_Chrome.jpg

Edited by Mcconnor8
  • Like 2
  • Thanks 1
  • 5 weeks later...
Posted
  • Location: Lancing
  • Weather Preferences: Sun, Snow and Storms
  • Location: Lancing
Posted
WWW.ECMWF.INT

ECMWF’s new Strategy from 2025 to 2034 sets out anticipated progress in science, impact and organisation and people. It emphasises the need for...

The ECM centre are making developing the AIFS and other machine learning weather ventures their top objective for the next 10 years. 

The first operational version of the AIFS goes live on the 25th February with many improvements over the current experimental version.

  • Like 2
Posted
  • Location: Hessle
  • Weather Preferences: Cold Snowy Winters, Hot Thundery Summers
  • Location: Hessle
Posted (edited)
On 17/02/2025 at 10:18, Mcconnor8 said:

The first operational version of the AIFS goes live on the 25th February with many improvements over the current experimental version.

Given I rate the current AIFS highly this will be very interesting.

I've noticed that when the AIFS first came out, the long range charts were very flat synoptically. Now you can see that has improved visually and it did a good job with Eowyn.

Edited by Derecho
  • Like 2
  • 3 weeks later...
Posted
  • Location: East Lothian
  • Weather Preferences: Not too hot, excitement of snow, a hoolie
  • Location: East Lothian
Posted

Artificial Intelligence and Numerical Weather Prediction  I went to the Royal Met Soc talk last night in Edinburgh (there were also 200 online watching) 

Quote

Prof. Kirstine Dale is the Met Office's Chief AI Officer (CAIO) and Principal Fellow for Data Science. As CAIO she is charged with embedding AI in the Office’s core business: initially focusing on the use of AI in weather forecasting through leadership of the ‘AI for numerical weather prediction’ (AI4NWP) programme.

Talking about how this AI model will run alongside the NWP models and the humans. Someone asked about model characteristics, how we know about the strengths /weaknesses of GFS/ECM models etc because we know how the physics works. She spoke about Black Box models- where the internal decision making processes are not transparent, not easy to understand - so hard to know how the model got to its prediction and how to improve. It is only as good as the data sets. It struggles with extremes, with intensity such as in Storm Ciaran because it doesn't have experience of extreme data. The model charts try to 'smooth' which is no good for fronts , and they are very important to us forecasters.

It was a whistle stop tour of what is happening within the Met Office, which was a shame as I'm sure the audience wanted to hear more and there were plenty of questions from the floor and online. Such as about the interface to generating text output as the MO has a lot of written forecasts currently. 

Afterwards, chat about - the UK needing more school leavers and graduates skilled in maths, physics, data science, coding. (There aren't enough teachers in these areas,) STILL not enough girls and young women choosing these paths.

 

Could contain:

Could contain:

Could contain:

  • Like 2
  • Thanks 2
Posted
  • Location: Edmonton Alberta(via Chelmsford, Exeter & Calgary)
  • Weather Preferences: Sunshine and 15-25c
  • Location: Edmonton Alberta(via Chelmsford, Exeter & Calgary)
Posted

 Jo Farrow

On 12/03/2025 at 03:25, Jo Farrow said:

Afterwards, chat about - the UK needing more school leavers and graduates skilled in maths, physics, data science, coding. (There aren't enough teachers in these areas,) STILL not enough girls and young women choosing these paths.

my son who is graduating from University in the summer with a degree in maths and further maths etc..applied to the met office was told there was a hold on hiring graduates

Posted
  • Location: East Lothian
  • Weather Preferences: Not too hot, excitement of snow, a hoolie
  • Location: East Lothian
Posted

 cheeky_monkey Was that because they do a recruitment trawl at a certain time of year, autumn I think. And their opportunities are advertised at that point, or did he try within that?

Posted
  • Location: Edmonton Alberta(via Chelmsford, Exeter & Calgary)
  • Weather Preferences: Sunshine and 15-25c
  • Location: Edmonton Alberta(via Chelmsford, Exeter & Calgary)
Posted

 Jo Farrow yes he has been in contact with the met office since his 2nd year

Posted
  • Location: Lancing
  • Weather Preferences: Sun, Snow and Storms
  • Location: Lancing
Posted

This thread on X is interesting, "Aardvark" is able to produce forecasts comparable or better than NWP models using an order of magnitude fewer observations as starting data.

  • Like 1
  • 2 months later...
Posted
  • Location: Home: Chingford, London (NE). Work: London (C)
  • Weather Preferences: Winter: cold and snowy. Summer: hot and sunny
  • Location: Home: Chingford, London (NE). Work: London (C)
Posted
WWW.FT.COM

Rapidly advancing technology is helping meteorologists to make more accurate and detailed forecasts even further into the future

 

The AI revolution changing how we predict the weather

Rapidly advancing technology is helping meteorologists to make more accurate and detailed forecasts even further into the future

https://archive.ph/SHtH6

  • 1 month later...
Posted
  • Location: Lancing
  • Weather Preferences: Sun, Snow and Storms
  • Location: Lancing
Posted

AIFS Ensemble is now open source and can be ran yourself in minutes with very little computing power!

 

  • Like 2
Posted
  • Location: Near Darlington
  • Location: Near Darlington
Posted

 Mcconnor8 Very interesting, have various computational stuff here but nothing with a GPU that can handle it. Will have to sort that!

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...