Blessed Weather

Indeed. I rather suspect they will perform very well. I have to admit I'm not sure I'm clear about this question. From what I can ascertain the current crop of leading AI models includes: GraphCast (Google), Pangu-Weather (Huawei), FourCastNet (NVIDIA), AIFS (ECMWF), Aurora (Microsoft), MetNet-3 (Google), and Aardvark (University of Cambridge). Although I did spot that the Cambride model is sometimes referred to as simply Cambridge AI. Is this the model you were wanting stats for? I hope he doesn't mind, but I've tagged @Mcconnor8 in on this particular question as he's a regular poster on verification stats and may be able to help us.

So who's going to see the most storm action overnight tonight and through Saturday? For a change the models are in reasonable agreement about the tracking, as shown in these 06z forecasts of rainfall accummulation up to midnight Saturday. The models shown are Arome, Arpege, ECM, NMM and UKMO: Source: https://www.meteociel.fr/

BlueSkies_do_I_see The Model Verification thread has some data:

June 2025 The run of anomalously dry and sunny weather experienced through the Spring months has continued into June here in our Region, even with a number of brief unsettled and thundery interruptions (more on this below). According to the Met Office the month was provisionally England’s warmest June on record, with a mean temperature of 16.9°C (2.5°C above the long-term average). Areas of southeastern England saw provisional mean temperatures over 3°C above average, and many counties including Greater London, Essex and Kent recorded the warmest June on record. The UK maximum temperature for the month was 33.2C recorded at at Charlwood (Surrey) on the 21st. Here's the Met Office anomaly charts for the month, versus the longer-term 1991-2020 climate average: Source: https://www.metoffice.gov.uk/research/climate/maps-and-data/uk-actual-and-anomaly-maps There were a couple of notable thundery breakdowns to the otherwise dry and sunny month, the first on the evening of the 13th which ran into the early morning of the 14th: Lightning strikes 24 hours to 07.30 14th: The second spell occurred on the 22nd June and whilst the associated heavy rainfall caused some significant flooding in localised parts of Kent it was very much a 'Kent Clipper' affair. Lightning strikes 22nd: To complete the picture, the Met Office chart for overall strikes recorded during the entire month: Finally, here's the regional summary of the month's weather as recorded by Roost Weather: Source: https://www.roostweather.com/

Staying on Bodensee in Germany for a few days. This photo captures a thunderstorm approaching Meersburg over the lake on Monday 7th July.

Here's the current Netweather leader board at 16.05, so probably the peak yet to come. You can keep checking here.

Enjoying a lovely holiday in Seefeld in the Tyrol atm. Tonight's sunset a real stunner:

Today's high temps leader board here on Netweather. Phew, what a scorcher.

To make a change from bashing the models I thought it appropriate to congratulate them for their performance in accurately forecasting the arrival of the high pressure this coming week. I made the opening post of this thread on the 1st June by showing the GFS and ECM AI forecasts for the 16th. Here's how they performed. The first chart is their forecast from over two weeks ago, the second chart is today's 0z opening chart (affectively the 'actual' today). That is pretty impressive! GFS ECM

Sprites Indeed. I couldn't agree more. It's normally my 'go to' hi-res model and usually performs well. For some reason in its 0z run on the Friday morning it made a very poor job of forecasting the track of the storms for later that day, showing them completely missing us, staying off the coast and out in the North Sea. Here's the cumulative rainfall forecast from that 0z run which nicely illustrates the fail. Remarkable for such a short time ahead. Source: Meteociel archive https://www.meteociel.fr/modeles/arome.php?archive=1&ech=3&mode=101&map=0

For the record, here's what happened with the storms from last evening and overnight. Using the hi res radar calculation of rainfall accumulation and the lightning archive we can see that the event turned into an 'east coast hugger', first clipping Kent (who saw the best of the action by far) before rumbling its way up the coast. Cumulative rainfall: Lightning Fri evening: Overnight into Sat: Source: Netweather Extra Here's the screen shots I captured on my phone: 11.30 00.30 So how did the models do? On the previous page I posted Friday's 0z run forecasts from a selection of them and I'm giving the winner's rosette to NMM and the wooden spoon to Arome.

This morning's 0z runs showing 23.00 Friday to 05.00 Saturday. There's better consensus amongst the various models, although still some variance, even at only 15 hours or so out. I think it's clear that folk near the East Coast stand the best chance of seeing some storms sometime between 22.00 tonight and 02.00 on Saturday. Not looking so good for the west of the Region: NMM with East Coast hugger: Arome going for Kent clipper: UKMO best track for most: Arpege not bad for many: ICON with East Coast hugger: Finally, some stills from other models: ECM largely in the clipper camp: UKV has shifted from yesterday's 'barely a clipper' to now be 'not bad', although those near the East Coast again in the best location: 22.00 Midnight Sources: https://www.meteociel.fr/modeles/ and Netweather UKV output.

Vortex3929 Yes, let's hope it's not a clipper. I'll do a multi-model comparison in the morning and we'll see if there's some consensus about the tracking by then.

Hum. So the latest UKV run (the 09z) even more pessimistic about the Friday evening storm, with it now barely clipping Kent. The other models I quoted above are still sticking to their forecasts in their 06z update. One of 'em is going to have egg on their proverbial face. Latest UKV forecast for Fri evening:

Will it, won't it? Friday evening into early Saturday continues to show the most promise for those of us who enjoy a storm. (Sorry to those who don't). The Arome and NMM models pretty much in agreement with track, timing and available instability, as shown in the following animation of rainfall and MUCAPE (Most Unstable Convective Available Potential Energy) for the period 22.00 Fri to 05.00 Sat: Arome: NMM: Source: https://www.meteociel.fr/modeles/ However, this morning's UKV seems to deviate somewhat from the other models and has the threat further east, turning it mostly into a 'Kent Clipper' affair (twice). No animation, but a couple of stills: 20.00 rainfall lightning density midnight rainfall lightning density Source: Netweather UKV output Place your bets!