forecaster

Meanwhile.....

Good point, sorry, brain meltdown. I could get away with it though.....mid-December 2009 was "before the cold set in" in 2010

EC32 tries to suggest a somewhat more NW flow for the last week of November, but not a strong signal. Prior to that, +ve heights nearby or to our south.

Here is the pattern just before the cold set in that month:

That was very interesting, thanks.

That type of pattern was identified as a key driver in the 2010 winter. There are some similarities with the current pattern, but also some differences. Nothing ever quite fits the conceptual model perfectly.

Just a very quick example....correlating MEI against 1000mb temperature does not detect any strong relationships in our part of the world. Similar story for 500hPa heights:

Do we have any idea who he works for? I know he said "multinational". Why is it a secret? Is this the same Stewart Rampling? (http://www.countrysideconsultants.co.uk/profile.htm)

I think "Aleutian High" typically refers to a pressure pattern in the stratosphere.

It seems plausible. But "locals" all over the world will always claim that their spot is somehow significantly warmer or significantly colder, or significantly different in any other way, to the "official" records. These claims are often exaggerated but sometimes contain grains of truth.

This is an astonishing image to see on the CFS because it's off the 10 day averaged run. Would not be surprised to see this on an individual run, but these massive anomalies on the 10 day are amazing.

I am usually defensive of the EC32, but the run from 3rd October valid for next week did predict higher than normal pressure just south of Greenland (extending N), with a weak signal for lower than normal pressure to our NE. I don't think that will verify too well. As it currently predicts a continually unsettled spell though, you would expect periods like that are closer to climate norm and easier to predict.

I suspect that is the case. Subtropical highs like living off the west edge of continents (i.e. Europe for the Azores High). If the Gulf Stream "shut down", perhaps we would end up with a weaker but similar current, forced by persistent winds around the Azores High. SST's would be lower, so perhaps we would have similar weather patterns but cooler resulting temperatures. Lots of "perhaps" there!

What are we looking at here? Degrees from normal? What level?

Interestingly, it was close to the time of Tropical Cyclone EVAN in the SW Pacific. The models for some time had difficulty getting a grip on that feature - it seems plausible that this could have been a distant source of error for the flow in the NH.

Decided to have a look at this famous failed easterly from last year. This and the pages following: http://forum.netweather.tv/topic/75166-model-output-discussion-12z-04122012/page-26 Not an experience many people want to revisit I imagine!

See attached. This winter uses a newer version of the model though.

Just out of interest, I have attached the Met Office forecast for winter 2010/2011 and for winter 2011/2012 issued in the corresponding Octobers. The model has apparently improved since then. I think it's version 5 now. As you can see, for both these winters it picked the general pattern nicely.

Thanks, I just hope I haven't made any mistakes! I am yet to find anywhere freely available online which breaks down the compositions of the waves. An exception is the FU Berlin stratospheric page which shows you the Wave 1 and 2 amplitudes. I think the NCEP stratosphere site gives you Wave 3 also. Beyond that, I'm yet to find anything.

1) Pretty much. And fourier analysis allows you to get an idea of what the composition is. 2) I always find it really difficult trying to manually visualise for a given chart what the component waves might be. It might not be possible at all....however, you can think of it in a simple way just using two different waves:Imagine in your latitude belt you have just two waves. One of them is Wave 2 and the other is Wave 7. So very distinct waves. Wave 2 is perhaps stationary, very persistent whereas Wave 7 moves in the westerly flow, and by itself would just look like a standard mid-latitude zonal pattern with lows and highs. Where the Wave 2 ridge and Wave 7 ridge are in phase, there would be "constructive interference" (from standard wave theory). The resulting ridge would therefore be more intense than under other circumstances. Likewise, where the troughs are in phase, you would probably see a deep low. In areas where the phases don't match up, you would see destructive interference. So for example, a Wave 7 ridge beneath a Wave 2 trough would perhaps result in a fairly weak area of high pressure. And a Wave 7 trough + Wave 2 ridge would probably see the resulting low pressure system not being deep or not forming a closed low at all. Apparently, in the past it used to be possible for forecasters to improve on NWP by being aware of the long wave patterns in the short term. If you know that you're under a Long Wave ridge, then you would naturally expect resulting medium wave (troughs) and shortwave (troughs) moving into your area to weaken somewhat. I suspect these days this isn't really possible in the short term (due to sophistication of modern NWP), though would be an interesting experiment. In the long term it may be a different matter, and is somewhat akin to how John Holmes approaches long range forecasts using the 500hPa anomaly charts.

Height fields at a given surface are superpositions of many different waves on top of each other. You can pick out the individual waves by doing something called fourier analysis, which has applications in many different fields. Once you do that you will end up with an idea of the different contributions from waves 1, 2, 3 etc. all the way up to about 9. What does this mean? The numbers are the "wavenumber". This means that in a given belt of latitude, a wave with wavenumber 1 will have one peak and one trough around the entire world. This is called wave 1. Likewise, a wave with wavenumber 2 ("Wave 2") has two peaks and two troughs as you go around the globe within this latitude belt. And so on. As you can probably see, as you increase the wavenumber, you are cramming more waves in as you go around the world. So the distance between each peak and each trough gets less. This is another way of expressing the relationship between wavenumber and wavelength. Wavenumber is essentially the inverse of wavelength. Long waves are usually referred to as those with wave 4 or below (aka Rossby waves or planetary waves). It is possible for these to be stationary, even in a mean westerly flow. With wavenumber 5 and above, you get into "synoptic waves" and these usually move with the mean flow (i.e. for us usually towards the east). The paper you referenced has the same idea (in terms of wave theory, wavenumber is a consistent concept), but is talking about Kelvin Waves. These occur in the atmosphere and the ocean - probably best known in the tropical atmosphere as an eastwards moving wave of about 40 knots. Usually when meteorologists talk about wavenumber they are talking about Rossby waves. Your question is definitely worth asking because I suspect that around the internet some people throw these terms around without getting a real physical understanding of the terms they're hurling around.

It snowed on Boxing Day 2004 in South Wales

I don't think a "no signal" forecast is as useless as you're claiming, and it happens frequently enough in all the models. e.g. the CFS shown below. Even the best models have huge areas of the globe covered with "no signal". It might mean that there truly is no signal (intrinsic), or it could mean the model is not good (practical). Hard to know in individual cases. I think what we hope, is that when a model does go for a signal, that it comes off well. The majority of seasons come in close to seasonal average, so for a model to predict "no signal" more often than not might be a sign that it is well calibrated.

Rather tricky to get meaningful signals from this when it has the whole world within 1C of average!