Long Range Forecasting - Fact Or Fiction

[DAVID SNOW]

Long range forecasting is and always will be a game of luck.

The skill in forecasting is really to minimise the amount of luck you need and then play the percentage game.

Some are better than others at minimising this, such as GP and to be fair also the METO I think.

But at the end of the day all long range forecasts are a mixture of luck and knowledge, the more knowledge the better but none of them IMO can reduce the luck component below 50% for the UK.

If an organisation keeps predicting Warm summers and Mild winters, sooner or later they will be caught out,and end up looking a bit silly

[weather eater]

When it comes to long range forecasts I think we have to manage our expectations, we are a small island on the edge of a massive ocean and a large land mass, that in its self makes any LRF difficult. We are also a small country and given that most of our weather crosses in a west to east fashion, its also a narrow one. Because of that it is possible to make a good general forecast and still get it wrong for us. For me GP is the person I pay most attention to when it comes to LRFs, he appears to have the best understanding of those overall patterns and external drivers on NW at this point of time. One of the problems that faces those attempting long range forecasting is that the drivers for those overall patterns start at a micro level, a level that we don’t have the instruments to read, its never going to be easy. There is of course another way to look at it, and that is do we want accurate LRFs, how boring would it be if in October it was announced that the winter would be wet and mild and that we knew that forecast was a dead cert.

[Nick F]

With so many differing techniques and forecasts available from various forecasters, do you think there's science behind long range forecasts, or is it purely guesswork?

Depends which method is used really, there is the unconventional methods which aren't published that can be construed as guess work by some, but there are definately two of the three main methods of Long Range Forecasting which can be considered 'scientific'. Of the three main conventional methods for long-range weather forecasting: the least scientific method is using Analogues, which many may know is using past weather patterns and matching them to similar weather patterns that maybe occuring now to predict what may unfold in the future. The drawback on this method is lack of scientific analysis of the variables in the atmosphere at a current point of time which will tend to make past analogues that are close enough match be unlikely to be followed. And of course, you need enough close analogues to make a forecast viable! There is then the Statistical Method; whereby variations from average weather conditions can be modelled by using statisical relationship between a group of different variables, using such variables as SST anomalies and the extent of snow cover over landmasses to predict NAO for the following winter. MetO have tried to use the N Atlantic SST anomaly patterns in May to give an LRF for the following winter. This worked for a few winters, e.g. 2005/06 they were quite close to the mark, but others, like last winter they failed badly. Then there is the Dynamical Model method, whereby ocean-atmosphere interactions are looked at on a global scale using powerful supercomputer model simulations to mimic how the coupled-ocean atmosphere system behaves. The LRFs they produce though will still me limited, like the others, by the so many variables that will occur with time, but the use of ensembles may go some way in alleviating this problem. This is probably the way forward with regards to LRFs for weather agencies that have the money to have these expensive supercomputers, as models will undoubtedly get more powerful.

Do the often sensationalist and inaccurate stories in the press regarding lrf's damage the reputation of long range forecasting as a whole?

Certainly the media, and sometimes on weather forums, tend to take forecasts out of context, mis-quote, etc, to potray to joe public that a particular LRF is ‘a joke’, MetO seemed to have got particular stick for this recently, and not always wholly justified for their lambasting.

Does the level of detail in a long range forecast affect your view on it, for instance - do you think it's possible for a forecast at a range of 1-2 months or more to be specific about particular days or weeks?

But surely though the object of most long-range forecasts that will have any success is not to predict weather on day-by-day basis or even week-by-week basis, rather to forecast monthly depatures from the averages of temperature, rainfall, etc.

If you think there is some science behind long range forecasting, do you think there has been progression in recent years, and if so do you think that in the coming years there's a possibility that lrf's will become increasingly more accurate?

Certainly, in recent years, considerable improvements in numerical weather prediction with advent of faster and faster computer processing and development of sophisticated numerical modelling that long-range forecasting has improved for those agencies that rely on computer models for their LRFs. However, there is still is no universally accepted technique, as far as I am aware, of producing LRFs to any great long-term successfulness and I imagine that such forecasts are still limited with regards to all the data that they really require to improve. The level of confidence with any LRF based on science will always be somewhat restricted to probabilities, given the chaotic nature of the atmosphere with the increasing variability/uncertainty with time. And unfortunately, as others have mentioned, the media and joe public can have difficulties grasping the concept of ‘probability’ and rather have more ‘exact’ forecast!

[johnholmes]

.

I am absolutely no expert in lrf work; that I have done over a 3-4 year period shows that mixing synoptic scale inputs with teleconnections further out can reveal a score, over that time period of over 70% out to 10-14 days and above 65% out to 21 days.

Thus its certainly possible to give a general trend almost out to a month with some success. If we simply said that tomorrow will be the same as today for 3 weeks or more then a ‘score’ of about 30% might be expected. Thus one can see how important this area of lrf work is from the point of view of commerce and industry that require an overall idea of above/below normal, and by how much for max and min temperatures, wind strengths and precipitation.

Turning to what this thread is really about and that is seasonal forecasting at least for a season ahead or perhaps longer. One of the major successes in this time scale are the hurricane frequency predictions. Over the past 2 to 3 years they have, unless I’m very mistaken, been pretty accurate. This allows additional manpower etc to be economically made available through that time scale for the parts expected to bear the brunt of these vicious storms.

Of course what most on web sites are really after is the ‘holy grail’ of predicting the oncoming winter much more so than say a summer prediction.

Roger drew attention to how close in actual terms for SE England was the daft headline from UK Met about a B-B-Q summer. If only that term had been left out then they would have reaped far less ridicule than they did. Equally, I have no inside knowledge, but in their winter forecasts they did seem ‘hung up’ on sea temperature anomalies rather than taking this and the quite large number of other variables that impact on any seasonal forecast. Again Roger is correct in separating the improvement in synoptic scale forecast such that day 7 is about as accurate as day 1 was 50 years ago and similarly at longer time scales. You cannot connect seasonal predictions to synoptic scale models, no matter how huge their computing power may become. Its simply a different meteorological world.

I do believe that, rather off topic that slow, very slow, improvements in synoptic scale forecast will occur over the next 25-50 years. Equally that the rather grey area between synoptic and climate type outlooks, say days 14-30 will also improve as does our understanding of the energy exchanges between ocean and atmosphere, and hugely important our knowledge of what happens beneath the surface of the oceans and how that sometimes rises and at other times falls. So far we are barely scratching the surface with items like ENSO etc.

In terms of seasonal forecasting then I believe if we can get qualified professional meteorologists to work with NOT against highly intuitive amateur or at last none main stream climate/meteorologists, then we could see some quite dramatic improvements in this area.

On a parochial level some of the work by Roger Smith aided by BFTP along with the deep insight into the masses of teleconnections by GP (Stewart) and chiono and a couple of others, along with a truly deep and constructive spell of research into items like the CFS data and even on Net Weather we may get in 5-10 years, possibly longer, routine scores that are better than chance. That is around or perhaps even beyond 50%, with 33% being the expected value over time by saying this winter will be like last winter every year.

I’ve wittered on long enough but in all 3 time scales I see progress of some depth occurring through this century.

[Glacier Point]

Then there is the Dynamical Model method, whereby ocean-atmosphere interactions are looked at on a global scale using powerful supercomputer model simulations to mimic how the coupled-ocean atmosphere system behaves. The LRFs they produce though will still me limited, like the others, by the so many variables that will occur with time, but the use of ensembles may go some way in alleviating this problem. This is probably the way forward with regards to LRFs for weather agencies that have the money to have these expensive supercomputers, as models will undoubtedly get more powerful.

Not just the preserve of the super computer Nick, and the core of my message over the last 2 years has been that we absolutely need a non-model based methodolgy to describe the coupled ocean atmspheric system to put alongside the models, both in the 8-14 day, 30 day and intra-seasonal timeframes.

Back in 2005 Ed Berry (formerly National Weather Service) started to blog about the importance of understanding the fluxes, ebbs and flows associated with deep tropical convection. Then, in 2007, which will I think be seen as a seminal moment in long range weather forecasting, he and Klaus Wickman introduced the concept of the Global Synoptic Dynamicak Model which was subsequently revised to produce the Global Wind Oscillation.

For me this has advanced long range forecasting significantly.

It's a simple concept. The weather patterns driven by a range of forcing factors lead to a 'phase space'. Eight phases of the global circulation which oscillate according to the relative strength of the forcing factor (El Nino or La Nina) and feedback mechanisms which either destructively or constructively interfere with these processes. This is a subject for a different thread and not an easy one to convey in postings. However, I would recommend the back catalogue of Ed's blogs and daily monitoring of the ESRL PSD Map Room pages for the wind budgets which make the Global Wind Oscillation.

http://weatherclimat...k.blogspot.com/

http://www.esrl.noaa...st_maproom.html

The most strking aspect of the GWO is that the composites nearly always capture the longwave pattern. Therefore, get the forecast of where the GWO will be (through rigorous evaluation and monitoring of daily windflows, convection, torques, eddies and fluxes), and you stand a highly elevated chnace of getting the longwave pattern right. The composites that I often post are only drawn up from 16 days picked out for similar phase space. Just 16 days. Thing about it, how could such a chaotic system produce such similarity ? Answer - the system is not as complex and chaotic as you might think.

In the shorter long range, we cannot overlook the value of the CPC (NOAA) 8-14 day mean height anomalies as I find this generally a very good guide and more recently the ECM and GFS ensemble mean height anomalies which are the only useable model tool beyond t144.

Fact or fiction ?

Fact.

Edited October 4, 2010 by Glacier Point

[chionomaniac]

I guess Ed Berry is not allowed a blog site with his new job or do you know that he has one?

[TEITS]

For me it depends on the methods used and how this forecast is presented.

Now I do believe its possible to predict the likely weather patterns by using teleconnections and by using these its possible to predict whether temps/rainfall is likely to be above, below, or near average. The forecasts that in my opinion are nothing but rubbish is when someone predicts a white Christmas in October i.e recent newspaper articles. Anyone who goes into detail like this is just making the forecast up.

I will also add that whilst its possible to predict the general weather patterns, in my opinion detailed forecasts will always be restricted to 7 days. I believe this will be the case even in 20yrs time as supercomputers become even more powerful. The reason is simply due to the Kaos theory and although these models use ensembles to take this into account they will never overcome this problem.

[Roger J Smith]

I'm really happy with the general tone of this thread, I think we're getting a good range of opinions here but also a general feeling that something positive is already happening and even better things may happen soon. The GWO or MJO approach has intrigued me recently because I see some similarities between it and the time scale of some parameters in the solar system magnetic field model. Of course there should be connections if both theories are valid and they would not need to be mutually exclusive paradigms.

TEITS, I am more optimistic than you about overcoming the apparent chaos of weather events. We only think it's chaotic because we have not generally accepted a cause and effect hypothesis. When I see similar peaks in various index values from data sets in two different parts of the world (but similar climates) then I suspect that we are on the road to a predictive framework. That road is bound to be a long, twisting one with many difficulties and setbacks. However, I've gone through many of those already, so my task would be to pass off my findings to someone with more time left. I think people would be intrigued if not necessarily amazed by some of my computer files -- you don't need a supercomputer to crack 150 years of daily temperature records, I have done this on a 1986 vintage "Wang" that is now full to the brim with files that produce index values for both the 150 years of daily data that I have for Toronto, and for the 350 years of monthly anomalies of the CET.

By "index value" I mean the profiles of certain postulated "drivers" above and beyond the most obvious one that everyone recognizes, the Sun in its annual variation in apparent latitude. I've profiled each synodic and sidereal period of every major object in the solar system, and each one of them has a broadly similar profile but also its own individual characteristics. None of these profiles (when averaged over that long a period) is particularly large but when you synthesize ten or fifteen of them, their cumulative amplitude can be fairly large. There are similar and perhaps more predictive index values to be used in solar variation. All of this has been developing gradually and I recently updated my files to include the last decade, the main problem being that all of this has been a monstrous amount of work that almost overwhelms one's memory and mental capacity, and it would all be a lot easier if I had a staff of dozens of assistants and a salary because somehow I also need(ed) to work at something that kept three square meals appearing etc etc. Fred knows a bit about all of this ... and has a bit of the same problem. I would imagine we're not alone. Trying to stay on top of LRF and global climate issues, no matter what your foundation may be, is really a full-time mental challenge that can be very consuming. If we had a smaller planet with fewer regional climates, this sort of thing might be well within one person's grasp. But the subject matter is (I'm sure many here will understand this) staggeringly complicated. I don't claim that it's extremely difficult, because the difficulty is more in the complexity than some sort of math or physics challenge. I studied special relativity at one stage of my higher education, and believe me, that was mind-stretching and difficult to visualize. Long-range forecast concepts are generally speaking a lot easier to visualize. But we are nowhere near the advanced stage of 1920-era physics in climate science, let alone the 2010 version which must be well beyond what I studied back in the 1970s.

Even if I had an easier life than I do, it would seem beyond one person's capacity to "crack the code" entirely and I really think it will take a community effort once the right paradigms are accepted. There is just inevitably going to be a period of detailed "fleshing out" work once we get past the hurdle of paradigm definition. It's a lot like atomic scale physics as uncovered from the mid-18th century to the late 19th, I would suspect -- once somebody gets a basic theory in place, there will be many years of work required to discover the various atomic elements of a working model (like for example, if what I discovered about the Toronto temperature series were to be accepted as significant, then some poor sod in Australia will be earning his PhD testing out the Sydney data for the local version of the same theory, and another one in Buenos Aires, and another one in Tokyo -- that's going to be the likely future course of development of any working theory, because I suspect it will be very much of a case of applying general principles to regional climates while some other group is assembling all of the details and building a global model).

Whether anyone will really want to know in a lot of detail what the weather is going to do in six months or a year, remains to be seen. Perhaps the trends are all we really need to know. Economic impacts don't depend that much on specific dates of events. If the winter is to be snowy in France or Belgium, for example, they will want to have extra road supplies and winter weather products in place, but whether the storms are on 15 December or 15 January is less important in that planning. But some people might like that scale of precision.

Will there one day be books (on line or in a bookshop) with maps showing expected weather for a whole year ahead? I think there already are products like that, but they tend to have hit or miss results. If something like that becomes as accurate as even day 4 GFS output, they will start to become "reliable" but if they are more like day 8 or day 15, then they will continue to have supporters and detractors. Nobody would want to fly to the Moon in 1950s rocket technology, but people were willing to give it a try by the end of the 1960s. We are in that sort of situation here, but there's no telling how many years will be needed to get from early rocket science to modern rocket science. I'm guessing ten to twenty. As with everything else, money drives the train to some extent. If big money were put into this, it would speed things up. Putting most of the money into the computer side of things will have limitations, as I think GP was saying also, we need theories that work, then the computers have their place.

[Nick F]

Not just the preserve of the super computer Nick, and the core of my message over the last 2 years has been that we absolutely need a non-model based methodolgy to describe the coupled ocean atmspheric system to put alongside the models, both in the 8-14 day, 30 day and intra-seasonal timeframes.

Very true Stewart, though I imagine it can take an awful lot of your time to construct a LRF without use of computer simulation! Certainly your method of using tropical forcings and their effect on Global atmospheric circulations makes sense once you get a good grasp of the concept, though I imagine you have to take in account other forcing in the atmosphere, such as Stratospheric temperatures, etc? The non-model based methodology you use makes it easier for the weather enthusiast who an grasp the basic ideas, of themselves being able to get an idea how the longer range weather patterns may evolve in the Northern Hemisphere. And need not be the preserve of forecasters who have access to supercomputers, which makes it all the more exciting, as the ‘door-is-open’ for other outside the big international weather agencies, like yourself, to have a well-reasoned and perhaps successful stab at LRFs.

The most strking aspect of the GWO is that the composites nearly always capture the longwave pattern. Therefore, get the forecast of where the GWO will be (through rigorous evaluation and monitoring of daily windflows, convection, torques, eddies and fluxes), and you stand a highly elevated chnace of getting the longwave pattern right. The composites that I often post are only drawn up from 16 days picked out for similar phase space. Just 16 days. Thing about it, how could such a chaotic system produce such similarity ? Answer - the system is not as complex and chaotic as you might think.

I find that very intriguing that chaotic atmosphere concept can someway be overcome, though I still think that predicting a long-wave pattern, say 3 months ahead, maybe fraught by increasing variability with time. Unforeseen major stratospheric warmings or coolings, for example, may slow or speed up the long-wave pattern as it moves across the Northern Hemisphere which will have a knock-on effect down the line, which may override a tropical forcing signal? I presume that stratospheric warmings are less easily to predict/occur quicker than ENSO states? So a winter forecast produced in November predicting the long-wave pattern in February may still be the reserve of probabilities rather than stating it will more likely than not happen with any accuracy.

[Methuselah]

If an organisation keeps predicting Warm summers and Mild winters, sooner or later they will be caught out,and end up looking a bit silly

Which applies equally (perhaps even more so?) to those who routinely predict severe and longlasting cold...Quite a few spring to mind.

[knocker]

How would a computer model know when a butterfly might decide to flap its wings in Africa? Or when a packet of energy might form over the northern Gulf of Mexico to create the genesis of a winter storm? Now if those butterfly wings are actually subtle energy inputs from predictable astronomical events or geomagnetic variations, you've got something. If not, then all the effort in the world will fail to solve this riddle, it just won't be possible to model such things accurately.

Yet, if the large scale features are based on the energy set-up of external drivers, then it stands to reason that the variations within the large scale features would become predictable as second-order variables.

I don’t for one moment claim any great knowledge on this subject and it is probably very presumptuous of me to enter to enter the fray but if you don’t ask you don’t learn. If I’m reading this correctly you are saying the energy set-up of external drivers are predictable astronomical events or geomagnetic variations.

In another post (apologies for taking pieces out of context) you say “When I see similar peaks in various index values from data sets in two different parts of the world (but similar climates) then I suspect that we are on the road to a predictive framework.â€â€¦..and go on to say “By ‘index value’ I mean the profiles of certain postulated ‘drivers’ above and beyond the most obvious one that everyone recognizes, the Sun in its annual variation in apparent latitude.

So my question (hopefully not too stupid) is does your research bear any similarities with that of Dr Theodor Landscheidt that concentrated on long-range forecast of U.S. drought based on solar activity? In this, as far as I can ascertain, he linked various events such as ENSO, NAO, etc, with to cycles in the sun’s orbital motion around the center of mass of the solar system. He is in fact saying that as the relationship is based on astronomical data that can be computed, the forecast could be extended farther into the future. For more detail:

http://www.john-daly.../US-drought.htm

Edited October 5, 2010 by weather ship

[kold weather]

Very interesting thread, I think RS/BFTP are probably onto something quite important when it comes to looking at the sun/solar influences, I was a big doubter a few years back but having seen how not just those two but others using similar methods have done quite well against the average you'd expect for LRF certainly does open my eyes.

One thing I've not seen mentioned but really is just as key as any model IMO is experience. I really think thats underestimated at times. For example last winter I went mild due to the strong El Nino. I missed a couple of vital things which looking back now screamed -ve AO but because I'd never seen that pattern before I missed it. If such a set-up were to occur again I'd remember what happened last time and be much better placed to make the correct forecast.

[DAVID SNOW]

Which applies equally (perhaps even more so?) to those who routinely predict severe and longlasting cold...Quite a few spring to mind.

Both camps as bad as each other IMHO.

I guess the difference between them is, one organisation has millions to utilise on research and building super computers,and others have....................

A crystal ball.

[Roger J Smith]

Yes, there is some overlap with space weather research being done by Landscheit and others.

There are also unique concepts to the research. Fred and I met here on net-weather and discovered that we were pursuing some related research, which is why we have collaborated recently on seasonal forecasts.

I've just been reading on some other forum that a forecast has come out calling for the coldest winter in a thousand years in Europe. I'm not going to touch that one. If it was the coldest winter in 65 years that would be bad enough. A repeat of say 1684 might be a bridge too far considering the SST values now in place compared to the height of the LIA. The fact that the Baltic would take most of the winter to freeze over now would have its own effect on that scenario. But I think this forecast is based on the idea that the atmosphere may be in some previously-unknown high-variability phase that manifested itself with the unprecedented heat waves in Russia earlier this summer. On our side of the great pond, weather watchers were generally astounded by the snowfall depths achieved in Washington DC and Philadelphia last winter, some of these values had 500-year return periods. But the east coast heat wave summer of 2010, while quite anomalous, was basically on a par with 1980 and failed to break extremes set in 1936 or 1911 in some places. So the idea that we are entering some era of enormous extremes, while it may signal a period of high variability, may be somewhat oversold too.

[Yorkshiresnows]

<br />Very interesting thread, I think RS/BFTP are probably onto something quite important when it comes to looking at the sun/solar influences, I was a big doubter a few years back but having seen how not just those two but others using similar methods have done quite well against the average you'd expect for LRF certainly does open my eyes.<br /><br />One thing I've not seen mentioned but really is just as key as any model IMO is experience. I really think thats underestimated at times. For example last winter I went mild due to the strong El Nino. I missed a couple of vital things which looking back now screamed -ve AO but because I'd never seen that pattern before I missed it. If such a set-up were to occur again I'd remember what happened last time and be much better placed to make the correct forecast.<br />

<br /><br /><br />

Great post Kold and very much agree with your thoughts on BFTP and Roger's efforts.

I am ashamed to say that I thought they were frankly daft basing predictions on solar and lunar events.

Their forecast accuracy over the past couple of years (and in particular in regards to the position of the Jet stream) has made me eat my words and I very much look forward to their continued contributions to LRFing.

So much for us all to learn.

Y.S

[Phil UK]

A great new addition, Paul and thank you for it.

My thoughts on LRF?

Unless as mentioned here, have a crystal ball. Let's take for example, Summer 2007, 2008. "The hottest ever!" said the MetO.

Turned out to be the "Wettest ever!". (In Solihull anyway! ) Should never put the cart before the horse. If I said that tomorrow will be Friday 8th October 2010. I'd be right. BUT!! Ask me what the weather will be like in, say Devon. I couldn't tell you that. Ask me what it will be like in 14 days from now in the same area... Again, I cannot premise that.

Autumn in a sense, we tend to think of as wet and windy which speaks true of the south west and Scotland. But not neccesarily so. They say drought, I do doubt. They say wet, I would not bet.

Hope you get my way of thinking. In short, don't trust LRFs!!

Phil.

[mcweather]

Thanks Paul for this great talking point feature.

To my mind LRF's have improved but mainly in the circles where those who are doing it think outside the box somewhat. I,d put far more trust in the likes of GP, BFTP, Roger J Smith and Brian Gaze over on TWO than I do the Met Office. The first four on the list actually show the reasoning behind their forecasts for one thing.

I think what finally cooked the goose for the Met office seasonal forecasts was not the BBQ summer forecast but its failure to spot the that the coldest winter for thirty years was heading our way. Bearing in mind that the METO has spent millions and millions of pounds and has exponentially more resources as it disposal than any one here on Netweather or TWO. There is simply no way they should have failed to pick up on what forecasters elsewhere were seeing.

[Essan]

I think what finally cooked the goose for the Met office seasonal forecasts was not the BBQ summer forecast but its failure to spot the that the coldest winter for thirty years was heading our way. Bearing in mind that the METO has spent millions and millions of pounds and has exponentially more resources as it disposal than any one here on Netweather or TWO. There is simply no way they should have failed to pick up on what forecasters elsewhere were seeing.

I'm not aware that any other forecasters saw it though?

Yes, Joe laminate floori predicted a cold winter (as he does nearly every year) - but this was just coincidental as his reasoning behind the forecast proved to be hopelessly wrong.

Yes, PWS predicted a cold winter (as they do nearly every year) - but this, like their forecasts of record breaking temperatures every summer, is simply because such forecasts get published in the gutter press and thus bring publicity.

Britain is probably one of the hardest regions in the world to issue LRFs for. N America is easier, for example, because we know of certain connections with ENSO: hence we could predict that the winter Olympics in Vancouver would suffer from lack of snow, and that ironically they will likely get plenty this winter.

The problem with the MetO LRFs is that they were simply seasonal probability trends extrapolated from a global model - they weren't really forecasts at all. But they were sold as such and that led ultimately to their downfall. Interesting btw that GloSea4 is so far signalling an above average winter for Britain, though oddly also with hints of northern blocking (according to my unprofessional interpretation!).

[mcweather]

I'm not aware that any other forecasters saw it though?

Yes, Joe laminate floori predicted a cold winter (as he does nearly every year) - but this was just coincidental as his reasoning behind the forecast proved to be hopelessly wrong.

Yes, PWS predicted a cold winter (as they do nearly every year) - but this, like their forecasts of record breaking temperatures every summer, is simply because such forecasts get published in the gutter press and thus bring publicity.

Britain is probably one of the hardest regions in the world to issue LRFs for. N America is easier, for example, because we know of certain connections with ENSO: hence we could predict that the winter Olympics in Vancouver would suffer from lack of snow, and that ironically they will likely get plenty this winter.

The problem with the MetO LRFs is that they were simply seasonal probability trends extrapolated from a global model - they weren't really forecasts at all. But they were sold as such and that led ultimately to their downfall. Interesting btw that GloSea4 is so far signalling an above average winter for Britain, though oddly also with hints of northern blocking (according to my unprofessional interpretation!).

Hi Essan.

I take your point about the Meto extrapolating from their global model but that doesn't say much for the very expensive global model does it. Joe B and PWS did predict cold (and yes they often have eye to making a headline) but I didn't quote them in my post. The main point is that sensible forecasters with a zillionth of the resources of the Meto nearly all went for a cold to very cold winter.

As you quite rightly say Britain is one of the hardest places in the world to forecast for but that remains true for all the forecasters not just the METO. What will be interesting will be to see how the press including the broadsheets ( because they were quoting PWS this week, as well as the gutter press)treat players like PWS if the winter turns out to be mild and wet rather than cold and snowy. My hunch is that they will let them off comparatively lightly because unlike the METO they will have spent their own money to come to erroneous conclusions rather than the taxpayers.

[Essan]

What will be interesting will be to see how the press including the broadsheets ( because they were quoting PWS this week, as well as the gutter press)treat players like PWS if the winter turns out to be mild and wet rather than cold and snowy. My hunch is that they will let them off comparatively lightly because unlike the METO they will have spent their own money to come to erroneous conclusions rather than the taxpayers.

Well the media have run headline stories for the past 2 years about record breaking heatwaves in August as predicted by PWS - and they seem to have quickly forgotten them! There was also the snow in September ..... Much the same can be said of Piers Corbyn who likewise claims in his press releases to get it right and might mention an occasion when he did, but not all the other times when the media ran with predictions he'd made which turned out completely wrong.

As an aside, earlier this year the press completely misquoted out of context a PWS LRF forecast to imply they were predicting that winter would last until April (the forecast was actually for a cold spell at the beginning of March). Upshot of which is, whatever the source, don't believe the media interpretation. Check the original source to see what was actually said.

[mcweather]

Well the media have run headline stories for the past 2 years about record breaking heatwaves in August as predicted by PWS - and they seem to have quickly forgotten them! There was also the snow in September ..... Much the same can be said of Piers Corbyn who likewise claims in his press releases to get it right and might mention an occasion when he did, but not all the other times when the media ran with predictions he'd made which turned out completely wrong.

As an aside, earlier this year the press completely misquoted out of context a PWS LRF forecast to imply they were predicting that winter would last until April (the forecast was actually for a cold spell at the beginning of March). Upshot of which is, whatever the source, don't believe the media interpretation. Check the original source to see what was actually said.

I certainly agree with you on that. I gave up on trusting the national print media to report long range forecasts reliably and in their proper context years ago. The classic was the Meto forecast for the below average winter of 2005/6 which according to the national press meant it would be another 1963.

[Glacier Point]

I'm not aware that any other forecasters saw it though?

Yes, Joe laminate floori predicted a cold winter (as he does nearly every year) - but this was just coincidental as his reasoning behind the forecast proved to be hopelessly wrong.

Yes, PWS predicted a cold winter (as they do nearly every year) - but this, like their forecasts of record breaking temperatures every summer, is simply because such forecasts get published in the gutter press and thus bring publicity.

Britain is probably one of the hardest regions in the world to issue LRFs for. N America is easier, for example, because we know of certain connections with ENSO: hence we could predict that the winter Olympics in Vancouver would suffer from lack of snow, and that ironically they will likely get plenty this winter.

The problem with the MetO LRFs is that they were simply seasonal probability trends extrapolated from a global model - they weren't really forecasts at all. But they were sold as such and that led ultimately to their downfall. Interesting btw that GloSea4 is so far signalling an above average winter for Britain, though oddly also with hints of northern blocking (according to my unprofessional interpretation!).

Disagree, All of our forecast variables came off.

Impact of surging increase in angular momentum supportive of high latitude blocking. Check.

Impact of enhanced Brewer Dobson Circulation during the autumn resulting in increased ozone concentrations over the Arctic Stratosphere leading to high latitude blocking (supported by anomalous upwelling of the tropical stratosphere (adiabatic cooling) and anomalous downwelling of the arctic stratosphere (adiabatic warming). Check.

Tropical forcing located east of the Dateline supportive of a -NAO. Check.

Atlantic and Pacific SSTA supportive of -NAO signature. Check.

Low solar / east QBO combination supportive of high latitude blocking. Check.

The fact that we did not specify the degree of cold is a limitation, as is February's milder call. Perhaps the degree of the support for a strongly negative -NAO / -AO should have lead us to this forecast, but we got the variables right.

Itr should also be remembered that LRFs also extend to one month out. Remember early December when that longwave pattern was becoming evident and we went cold.

This also emphasises the importance of showing your working out. Methodologies must be set out in long range forecasts to enable independent assessment and verification otherwise you will always be open to criticism of 'lucking out'.

[Yorkshiresnows]

Disagree, All of our forecast variables came off.

Impact of surging increase in angular momentum supportive of high latitude blocking. Check.

Impact of enhanced Brewer Dobson Circulation during the autumn resulting in increased ozone concentrations over the Arctic Stratosphere leading to high latitude blocking (supported by anomalous upwelling of the tropical stratosphere (adiabatic cooling) and anomalous downwelling of the arctic stratosphere (adiabatic warming). Check.

Tropical forcing located east of the Dateline supportive of a -NAO. Check.

Atlantic and Pacific SSTA supportive of -NAO signature. Check.

Low solar / east QBO combination supportive of high latitude blocking. Check.

The fact that we did not specify the degree of cold is a limitation, as is February's milder call. Perhaps the degree of the support for a strongly negative -NAO / -AO should have lead us to this forecast, but we got the variables right.

Itr should also be remembered that LRFs also extend to one month out. Remember early December when that longwave pattern was becoming evident and we went cold.

This also emphasises the importance of showing your working out. Methodologies must be set out in long range forecasts to enable independent assessment and verification otherwise you will always be open to criticism of 'lucking out'.

Great Post,

Totally agree. Not sure what the previous poster saw of the run in to last winter, but your goodself, BFTP and Joe B were all singing from the same hym sheet.

Please keep up the great work.

Y.S

[ZONE 51]

It just shows how complex lrf is to do, since studying new elements that bring together a lrf (not just the basics) im realising just how hard and very complicated it can be, when your forecast goes out everyone would expect it to be as in that forecast, at times something is going to make a sudden change and disrupt your prediction, and i dont blame the weather forecasters that make the forecast when it goes wrong, they are experts, they look at lots of types of data from the atmosphere to the oceans, so many things interact with each other its amazing and very interesting. I believe sometimes that their is a sudden unusual interaction that catches us out.

Edited October 10, 2010 by nimbilus

[Essan]

Great Post,

Totally agree. Not sure what the previous poster saw of the run in to last winter, but your goodself, BFTP and Joe B were all singing from the same hym sheet.

But that's my point. They weren't. Joe B, for example, was forecasting a cold winter for completely different reasons. He just got lucky for once. I don't know what BFTP's forecast was.

GP's right though: this demonstrates very clearly why it's important to show your rationale.

If I predict a cold winter because there are more berries on the trees and you predict a cold winter because there are less berries on the trees and we get a cold winter, were we both right? Are both our LRF techniques verified? Or was it just coincidence and the number of berries has no relevance?