If Wishes made Weather

The Arome is showing this formation of stronger winds over Anglia the system pulls away as well:

Thinking about that, I wonder if there isnt a requirement for two systems. One where a general alert of dangerous weather is given out some days in advance (like the NOAA system which we seem to have achieved here with the blanket orange warning) and one with a more localised quick reacting system when there is a supercell or blizzard moving in your direction. I have seen this in the States as well as in Germany (where some places still have a siren system as well as more modern mobile alerts and radio and TV interruptions.

Same sort of nonsense going on here. I have had conversations with two colleagues today who were completely unaware of the storm. Most seem though not to put two and two together as its so very long since we experienced anything like what is forecast so they think it will be simply troublesome with a few bins blowing over.

This is very interesting. I would have hoped though that it would be more of a mechanical thing where if a pre-determined threshold was reached or confidently predicted then a red warning would be issued.

And whacks the 130kph swathe right through from the south coast to Lincs and through London and Anglia

And I see that the modelling is now raising alerts across the other side of the North Sea. This is the ‘unwetter’ (unweather = bad weather) position in Germany. All of northern Germany under a red level alert for wind up to. 130kph and costal areas in excess of that. On thing to note on the warning is that in some (possibly all) regions (sort of like our counties) once alert levels reach a certain level then civil defence contingency plans like pulling in part time firefighters to man reserve fire trucks and calling in the THW are activated - so its more than a citizens be aware and take action system.

I noticed this development yesterday on the 18z models. It appears this morning that both the Arome and the Arpege show this quite distinctly: The other models are indicating it as well, but in a less defined way, for example the WRF: and here the GFS which sort of indicates the possibility showing almost a ‘tail’ of stronger winds into the northern Home Counties and southern Anglia: I would hazard a guess that all models are seeing a possibility - perhaps now at this timeline almost a certainty - of something, and that some of the models are defining it more clearly. Either way for my location near Stansted I am taking the current MetO 70mph maximum gust indication on their app as the lower end of the envelope and would not be surprised to see 10 mph added to that.

There are some signs of this intensification whilst passing over this area on the other models as well (albeit more subtle).

With ref to Nick’s comment above, Arpege develops this region of vicious winds to the north of London as the storm moves through:

Slightly further north or not, it seems not to make much of a difference. Sort of like does my roof land in my neighbours garden, or their neighbours. Quite simply a swathe of the southwest, southern and middle England, Wales and some parts of the north are going to see 80+ mph winds and much of that is inland. Looks quite ferocious around this neck of the woods in the northern Home Counties / Anglia region - and this has been showing despite subtle track differences for three days or so.

And look at where they bunch (higher probability of occurrence) - its about the top third.

Yes - astonishing. Taken at face value these charts put circa the lower half of the UK under 70 mph gusts, and almost all of the lower third under 80 mph+ (i.e. in the 130kph+) bracket. And as people have noted, it is the third that is both the most densely populated and least experienced with this sort of event.

So we now appear to have 4 models - GFS, UKMO, UKV and NMM showing surface gusts of more than 80mph across a wide swath of the country as the storm moves through. On shore north Wales looks horrendous. The met office app though is consistently showing lower than this at least around the area of Cambs, Herts, Bucks Essex Norfolk where I have checked. I would have thought though that now with some tracking differences taken into account, then if you are sitting in the middle of the lighter pink or in the case of the UKV darker red/white areas, then this will be concerning.

Well this would be pretty much a wipeout for a line south from the South West to the Wash. I have seen that the estimated gusts though in the metoffice site have been ca 20% lower than the ones that even there own chart has been showing all day and yesterday - so maybe nothing as bad as this?

That is good and clear. Interestingly, the bit in northern Germany that is showing the highest snowfall amount is the Harz mountains, the highest of which (called the Broken) is 1142m above sea level. Just a little less than Ben Nevis. It is famous for just getting loads of snow (or rain depending on time of year), with the maximum measured snow depth of 3.80m. If you look at a topographic map the Harz are the only really high ground between the Atlantic and the Urals.

... and by virtue of that is (I think disproportionally owing to its weighting) dragging the mean up anyway.

A question for someone who has a good technical understanding of the models. I understand that the op run is produced at higher resolution than the control. So its data output should have a higher predictive value. In calculating the mean value charts then is the op run ‘discounted’ so that every member of the set is treated equally (giving an even mean), or if not, the mean must surely always have a bias to the op run sort of built into it. Many thanks anyone

There is a difference - the angle of the higher pressure zone to the north between Norway and Greenland. On the one chart, the high pressure is elongated sort of towards Iceland. This allows a linking toward the large mid Atlantic high - and I believe better blocking from systems coming in from the west.

Sorry - heading west!

This is fun - the low pressure developing in the stream over southern Poland and heading East

It’s a bit difficult getting excited about the charts and potential snow depths that are being indicated at the moment on the charts (if they ever happen). Being a kid who was in secondary school in the ate 1980s in the ‘north’, it was about waist deep drifts - not the 5-10 cm levels being progged for next weekish.

Surely to have a warning then the expected weather needs to be a sort of minimum percentage away from the average. Anslight covering of snow and a -3c frost (as forecast for Stortford) is within the parameters of absolutely normal weather in winter. Why a warning at all. In July are we to be warned for a dry day at 17c?

Hello- on the models. Perhaps it might’ve just help to think of them like this. What I am going to write is a massive simplification but here goes: - the models are not conscious and do not work out right or wrong answers. They are just a set of equations run by computers that start with numbers and generate outputs that are other numbers. These output numbers are then turned into charts that we humans can read. - the input data is a set of measurements of temperatures, pressures, humidity, wind speeds and direction (and maybe some other things as well). - these input data has no inherent predictive value. It’s just numbers. The predictive element comes in by the programmers choosing which equations to use to change these numbers into other numbers. So a simple and false example. Imagine that at a certain place it is known that if the temperature is 20c at 8am, then 80% of the time in June the temperature will be 30c at midday, then the programmers may use an equation to provide an output that is has a value of 30c for the midday chart. Of course it will be wrong 20% of the time, but it is still reasonably accurate. The input data may be said to have a high predictive value. - now, this is a little more complex. At the moment with the SSW, these sort of assumptions don’t hold so well. Also - bearing in mind that millions of calculations are done to create the charts - some of the predictive assumptions built into the models ( i.e. which calculations to use) conflict. This all means that the high predictive value that can be put on the data normally no longer holds true. It is said to be more entropic. - a further aggravating factor is that time. The models I believe use some of the output data as input data for their next chart. Thus some of the output that creates the midnight chart (say) goes into creating the chart that is for 6 hours later, and of course when this output is already not very predictive, then anything based upon it will be even less predictive. There is a point often called F I where the outputs are of no predictive value (in terms of saying what the actual weather will be). They are just for indicating what, given the input data and the laws of physics, is possible. Not necessarily probable. Hope this helps a bit and my apologies to any mathematicians reading this.

I have noticed this sort of cyclical nature in the outputs. Maybe something to do with simple diurnal variation in the input data. Probably suggestive of taking a sort of average view of outputs over a period of a couple of days rather than suggesting that any one chart is predictive of itself.

Does anyone know if the models either add new data or use different calculations to produce the different frames in any particular run, or are the different frames simply a reiteration of the same calculations based upon the outputs that produced the previous frames? I suspect a blend of both, but would like to know as it would explain one or two things about how much predictive value any particular chart can have. Many thanks