Josh Dorrington

I know you might be concerned about the carbon footprint of big AI, but the thing you have to understand is that if *we* don't use the energy output of a small country on training these models, then someone else will just come along and do it for a miniscule fraction of the cos- oh hang on a sec...

Some ensemble members are even entertaining 5-6 sigma IVT anomalies over England sometime between the 24th-30th, indicating early possibility of a true extreme event (but not something that is currently likely)

Keeping an eye on the end of Jan, ECMWF currently predicts low pressure over the East Atlantic from the 24th onward. There's a lot of uncertainty in position and detail at this lead time, but a look at the Z500 precip precursor for southern england shows a high chance of bad weather on the way:

Check out joshdorrington.github.io/DominoWeb/s2... for more details! Too much to tweet about here!

This is well captured from the flow precursor perspective: the 'north norway' and 'sicily' Z500 precip precursors represent two halves of the current tripole pattern and are both are well over 2 std deviations from normal.

You might not guess that the weather in sunny sicily has much in common with Tromso up at the top of the world. But at the moment the same tripolar structure of geopotential height over Europe is bringing heavy rain risk to both regions, where a breaking ridge is augmenting low + high lat lows.

Further afield, I've noticed that Tunis and the surrounding area are in for a (relatively) rainy week. Strong cutoff low activity developing by Monday, coincident with a low-latitude wave train passing over North Africa:

If you're in the south of the UK, you might be enjoying some cool clear weather now, but the month looks set to turn: in general there are some weak mean signals towards cyclonic flow over the UK, with a larger-than-usual number of ens members predicting strong atmopsheric rivers.

Oh yeah, I promised some hot takes. Well as you may have noticed from the plots above, West Norway looks to be on the wet side right until the end of January. Good job I have multiple raincoats now!

I am finding these precursors very useful for quickly monitoring the extended range forecast, and quickly internalising the uncertainties and forecast evolution of rainfall-favouring weather. Why not check them out yourself for your region of interest?

joshdorrington.github.io/DominoWeb/s2...

We can also see this with the precursors, if we look at typical patterns of integrated vapour transport (IVT). We see clearly that this IVT precursor metric picks out the three peaks in the current forecast, each corresponding to the predicted passage of an atmospheric river over the North sea:

OK, so winds are one thing, but as we all know, you need moisture for rain. The large-scale flow may be setting the stage for precip, but is anyone ready to 'perform'? Unfortunately for Bergen, yes, we have three atmospheric rivers predicted to hit on Mon, Tues and Sat :( See these IVT EFI plots:

In fact we can see this even more clearly if we look at the upper level zonal wind precursor: these strong westerlies are exactly the sort of weather which bring rain to west Norway.

But from the 13th to the 20th the index flips positive, indicating a shift towards potentially more rainy conditions. In this case this looks to correspond to a building of high pressure over western Europe, with strong mid-level westerlies in the North.

Let's take today. There is currently a big ridge extending up to Iceland, keeping West Norway cool and dry: somewhat opposite to our Z500 precip precursor above. If we look at the current ECMWF forecast for this Z500 precursor index, we can see that the index should stay negative for the next days.

To make this insight into the synoptic drivers easier to use, we can produce precursor indices for each precursor pattern. Just like an NAO index, a precursor index has mean 0, std dev 1, and summarises the current Euro-Atlantic flow. ⬆️values = ⬆️ xtreme precip chance, ⬇️ values = reduced chance.

(These patterns are obviously not raw composites, but have been filtered to isolate significant and high amplitude flow features. See doi.org/10.1002/qj.4622 for methodological details.)

If we turn to ERA5 reanalysis, and look at composites of the weather associated with heavy DJF precip in West Norway, we find an NAO+ like dipole of geoopotential height, strong westerly winds over the North Sea, and southerlies over the UK and norway -- that is, a strong, SW-NE, tilted storm track.

Lets take an example. What weather patterns accompany heavy (>90th percentile of daily, area averaged) precip over West Norway in winter?

If you've heard me talk about my work recently you'll know I'm interested in synoptic precursors of rainfall across Europe. What weather drives heavy rainfall in a region, storms, atmospheric rivers, rossby waves, etc? These precursors can be more predictable at long lead times than precip itself.

A new year, and my first one living in Bergen! Finally getting settled, and found time to get some operational weather products running that I've been working on for aaages.
Here I'm going to give a brief explainer on these 'precursor products', and present some hot-takes on Jan precip in Europe. 🧵

A photo of Løvstakken, a mountain in Bergen , on a sunny and snowy day with a snow cloud breaking over it.

Yes Bergen rains a lot, but when it doesn't, it can look like this 😍. Don't need a textbook to study orographic precipitation here.