Jump to content

The Seasonal Forecast Thread


Recommended Posts

This is my annual attempt at prediction using solely statistical means.

I took the daily CET set, computed monthly averages, and then took out the trend, and adjusted for seasonal factors. For every year in this new set, I computed the Pearson correlation coefficient, r, from every year preceding it for the months Jan->Oct, and then I took the top ten coefficients to get a best analogue set of data. I then simply read off the temps from the original CET monthly set from the years suggested by the correlation analogue.

From that I summarised the data to get the following,

post-5986-0-62892400-1352903282_thumb.pn

Of intetest is months 11-3 at the bottom of that table, which gives the following summary,

post-5986-0-10589900-1352903291_thumb.pn

Here it is as a min/max chart,

post-5986-0-46001300-1352903298_thumb.pn

There is a statistical 95% chance of the CET for each month being within these ranges assuming this method is sound. Overall, it looks like January has the best chance of coldness but note the skew (minus is skewed to cold, plus is skewed to hot) in February. This set of data has a strong bias towards cold particularly for February, but even in December and January. I think that the warm(ish) November is panning out as shown. Nov and Mar are added as a control for confidence.

So nailing my colours to the mast - what's the values?

Nov: 7.32C (+1.17C)

Dec: 3.46C (-0.73C)

Jan: 1.80C (-1.60C)

Feb: 2.74C (-1.31C)

Mar: 6.46C (+0.93C)

An observant reader will notice that I have suggested that Nov, and Mar, will be outside the comfort of the 95% zone shown above. Also +/- are CET monthly averages over the entire set. Anyway, looks to me like a very bitter CET zone, especially January, and February next year, but should start seeing it in the models in about a week or so time.

And there you have it.

Posted Image

Edited by Boar Wrinklestorm
  • Like 2
Link to post
Share on other sites
  • Replies 1.2k
  • Created
  • Last Reply

Top Posters In This Topic

Top Posters In This Topic

Popular Posts

The above summer forecast verified reasonably well in most cases, the prediction implied a top ten outcome and in fact the summer was around 3rd to 5th warmest on record depending on which data you us

Detailed version -- highly variable winter pattern expected     As promised, here are some more detailed predictions based on my research. I have tried to assess all the information offered in vari

2014/15 Winter Forecast    My attempt at a winter forecast. Some of the theory is taken from a previous project i undertook at University. I have attached references that i have used/think may be of

Posted Images

If I had used this method last year (I didn't) then this would have been the results,

post-5986-0-50923700-1352907091_thumb.pn

And here are the numbers,

post-5986-0-88181600-1352907111_thumb.pn

So, it seems, this method seems to get the shape right, but not necessarily the magnitude of the temperature (at least by reason of this chart and a few others that I've done) Also has not a cat's chance in hell of picking outliers (like Dec 2010, for instance) Interestingly, all the predicted values seem undercooked (so be wary of my forecast!) and I think that could be because I am pattern matching without a trend, and, perhaps, I need to alter the analogue values by readding the trend back in since the actual values will include a trend. It's almost certainly an artefact of the Pearson correlation which only includes the mathcing of the ups and downs and ignores magnitude.

Not sure that that is quite as clear thinking as it could be. Leave it with me ...

Edited by Boar Wrinklestorm
  • Like 1
Link to post
Share on other sites

Here's my forecast for 2012/2013 with the trend added back in,

post-5986-0-51125300-1352908483_thumb.pn

or,

Nov: 8.26C (+2.11C)

Dec: 4.40C (+0.21C)

Jan: 2.74C (-0.66C)

Feb: 3.68C (-0.37C)

Mar: 7.40C (+1.87C)

Not so chilly as before, but colder than normal. Don't forget that the +/- are against the entire CET set, not the meteorological norm (1970-2000?)

It'll be interesting to see how that pans out (Nov looks to be a little on the high side)

Edited by Boar Wrinklestorm
Link to post
Share on other sites

My winter forecast has now been completed, i have included a pdf file.

--

ESS's - Winter forecast 2012/13 - December, January, February.

Main Factors:

La-Nina and El-Nino.

Sea surface temperatures.

Hurricane season.

Solar activity.

Long range forecast models.

Forecast for December - (updated from my Autumn and early winter forecast, the change to a more colder month)

Polar/Arctic air at times.

Highest rainfall - Western and Northern areas - Atlantic weather systems most likely to affect here.

Driest - Southern and Eastern areas - high pressure most likely.

Temperatures - below average - Track of weather systems/Jet Stream/Atlantic High placement.

Rainfall - slightly above average - Low pressure more frequent then high pressure until late month.

Snowfall - above average - I expect snow to fall more likely than rain.

A Cold month, strong Greenland high developing, North-sea/Scandinavian low pressure/Atlantic High placement.

As we enter this month i expect the Jet stream to be hanging around and not really moving to far south at this stage, i expect the JS to be across the Southern Coast UK or North France, the Jet Stream quite weak at times. Low pressure systems affecting all areas, a few of these Lows deep, some very wet and very windy weather around as these systems sweep through, mild air at times within the systems and cold air sweeping down on the back edges of some of the Low pressures, and with some cold days and nights i expect some sleet and snow in the North and East to low levels and the cold air at times pulled into Western and southern UK then snowfall here to-even to low levels. There is likely to be ridges of high pressure inbetween the weather systems especially in South UK with frost and fog, some severe frost is likely for most places under the clear spells, the severe frost most likely late month where i expect stronger high pressure across the UK. Strong High pressure moving in to affect most areas from around mid-month, i'm expecting the Atlantic to slowly settle down with far less activity reaching our shores, the Low pressure systems being blocked, although some fronts affecting the far North and West at times late month but even here things becoming more settled, the High pressure bringing with it much calmer and much colder conditions, i expect sunny cold days and frosts for most, it is possible that severe cold develops towards the end of the month eventually bringing in a northeasterly flow with snow showers.

Forecast for January -

A very cold month, strong Scandinavian High.

Arctic air/Elements of a Siberian winter.

Highest rainfall - West UK - I expect rainfall more likely in the far west.

Driest - SE UK (low confidence) - Most precipitation to fall in West and North.

Temperatures - Well below average - Scandinavian high/Atlantic block/Northern blocking.

Rainfall - below average, low rainfall amounts as snow is most likely.

Snowfall - Well above average - more likely heavy snowfalls than rainfall, most of the precipitation falling as snow.

A very cold and snowy month.

This month starting off with high pressure, severe frost and fog, very cold with snow showers especially for the east side of UK, becoming progressively colder from the NE, significant falls of snow are likely across many places, most likely in areas exposed in a north-easterly flow, the severity of this cold cannot be answered at this stage, but i must emphasise that such charactistics of such an expected synoptic setup do usually lead to severe cold weather. At times in between the cold spells there is likely to be less cold conditions, but still cold and mostly dry, this where the high pressure cells move placements. As we go through the month a second severe cold spell developing possibly a strong but more easterly flow is expected, this would bring severe cold and snow, with Eastern Europe expected to be in a freeze it is likely to be severely cold-even colder than the north-easterly early month across the UK.

A cold blocked month with frequent snowy spells from the n/ne/e, much of the snow falling from bands and showers of snow.

Jet Stream south of UK, possibly over Spain.

Forecast for February -

A cold possibly very cold month, Scandinavian High, Atlantic block until late month.

Temperatures - Below average possibly well below.

Rainfall - below average - low frequency of Atlantic weather systems affecting the UK, cold enough for snow most of the month.

Snowfall - Above average - much of the precipitation falling as snow, some significant snowfalls widely.

Low confidence at this stage, i am expecting continued blocking of the Atlantic for most of the month, high pressure, cold or very cold.

Snowfall from the n/ne/e, and some widespread severe frosts, the snow affecting many areas from the North and East, i expect the Atlantic to fire up late month with the Jet Stream moving back north, weather fronts hitting West UK causing snow to fall along the fronts as they move East, with milder air flows eventually gaining influence at the end of the month.

Winter 2012/2013 - becoming progressively colder and drier through December after Atlantic low pressures with heavy rainfall, gales, and some snowfall, mid-late December frosts and snow showers, leading into a severe cold January with frosts and snow, February continuing with quite the same situation as January, with the Atlantic doors opening late February with battleground snowfall events as the mild air pushes up against the gained cold block of late Dec/Jan.

below is my forecast in PDF

ess - winter forecast 2012-2013 - pdf document.pdf

ESS.

Link to post
Share on other sites

I find myself in general agreement with Boar W above but my current CET estimates are

DEC 4.8

JAN 5.0

FEB 2.0

MAR 6.5

and Feb has the greatest chance of going sub-1 C or even sub-zero from my analogue set. As I mentioned, the confidence on a transition from mild Jan to cold Feb is not high due to sparse analogues but this is what the research indicates.

In more detail, Dec likely to be variable with mildest days near first week and Christmas holiday, coldest around new moon 12th and following week. There could be a windstorm in that period as mild pattern yields to cold in stages. Sticking to earlier mild with rain or drizzle and fog Christmas forecast, stormy 27th-28th.

Jan forecast to be highly variable, some very mild days possible, much colder trend late in month, could resemble 1947 in general outline (adjusted somewhat later to conform to second-order timing factors). Windstorm risks around 10th-12th and complex winter storm patterns likely late in month.

Feb predicted to be very cold with good chance of widespread snow and some chance of epic winter "singularity" -- obviously nothing will likely match 1947 but this is in the analogue set. This is when I think UK and possibly Ireland could see snowfall disruptions (best chance, around 6th to 12th).

March likely to start cold and see rapid thaw, melting of any remaining snow, pattern change to very mild late in month.

  • Like 3
Link to post
Share on other sites

Another winter forecast:

MattHugo81

The seasonal EUROSIP Nov update is out and unfortunately (for cold fans) shows a zonal, mild, wet & windy winter period.

15/11/2012 12:21

MattHugo81: The EUROSIP model takes into account the ECMWF, UKMO, Meteo-France and NCEP. As a result all seasonal models point towards a mild/wet winter

Along with the ECWF and CFS all suggesting a mild wet Winter it does go to show that even if many of the factors influencing seasonal climate hint at a cold season other synoptics seem to suggest otherwise.

Interesting, and it will be good to review at the end of Winter to see if the forum experts out perform the pro-forecasts!

Edited by i'm dreaming of...
Link to post
Share on other sites

Another winter forecast:

MattHugo81

The seasonal EUROSIP Nov update is out and unfortunately (for cold fans) shows a zonal, mild, wet & windy winter period.

15/11/2012 12:21

MattHugo81: The EUROSIP model takes into account the ECMWF, UKMO, Meteo-France and NCEP. As a result all seasonal models point towards a mild/wet winter

Along with the ECWF and CFS all suggesting a mild wet Winter it does go to show that even if many of the factors influencing seasonal climate hint at a cold season other synoptics seem to suggest otherwise.

Interesting, and it will be good to review at the end of Winter to see if the forum experts out perform the pro-forecasts!

From what i've read and interpreted the Pros are only reading what these models are showing -MH is a classic example, whereas GP is using lots more research and analysis to interpret and predict. RJS has his own methodology.

From my limited understanding of GP's work it seems a bit ahead of our own Meto, for example, so i'm siding with GP.

Link to post
Share on other sites

From what i've read and interpreted the Pros are only reading what these models are showing -MH is a classic example, whereas GP is using lots more research and analysis to interpret and predict. RJS has his own methodology.

From my limited understanding of GP's work it seems a bit ahead of our own Meto, for example, so i'm siding with GP.

I have no clue what is coming this winter. GP's early thoughts are very interesting, as they are based on statistics and probability. In my opinion this is a more stable reasoning, from this time angle, than computer algorithms, for a sighting to what winter beholds. The clear conclusion is that this year there is a greater chance that we will have colder conditions than we have experienced of late. Unfortunately (currently) the computer models are saying something else. You know that saying about statistics, well lets hope this winter is not a case in point.

  • Like 2
Link to post
Share on other sites

I have no clue what is coming this winter. GP's early thoughts are very interesting, as they are based on statistics and probability. In my opinion this is a more stable reasoning, from this time angle, than computer algorithms, for a sighting to what winter beholds. The clear conclusion is that this year there is a greater chance that we will have colder conditions than we have experienced of late. Unfortunately (currently) the computer models are saying something else. You know that saying about statistics, well lets hope this winter is not a case in point.

The models are often all over the place and often take a while to pick up a trend. Sometimes they are consistent over a period of times.

And, of course, a computer model will only feed out data from the data that's been fed in.

Trouble is on here ( and on other weather forums) we have whole swathes of people who are led to believe computers are always right. They're not.

  • Like 3
Link to post
Share on other sites

The models are often all over the place and often take a while to pick up a trend. Sometimes they are consistent over a period of times.

And, of course, a computer model will only feed out data from the data that's been fed in.

Trouble is on here ( and on other weather forums) we have whole swathes of people who are led to believe computers are always right. They're not.

Whilst you have correctly identified garbage-in garbage out cliche, you've missed out the bit in the middle. It's like describing someone with ugly ears as having no brain. What happens in the middle is some sort of process, that, crucially, is designed and built by human beings; in many cases, particularly with meteorology, it's the very best that exist in the field; they design and build this software with the best data, and the best peer-reviewed research. It is, for all intents and purposes, representative of the pinnacle of the science that we know at the time.

Now, one could argue that therefore it's the expert's who can't know what they're doing; but that's plain wrong, too. The simple facts of the matter are that the atmosphere behaves in a non-linear chaotic fashion, where a small perturbation - missed by some buoy down the Southern Atlantic, perhaps - vastly affects the outcome, here in Britain, hours, days, weeks, even months ahead of time. This is exactly the reason why we have ensembles. Incidentally, have you ever checked the maximum and minimum temperature of GEFS two weeks out and found that the temeperature was outside these bounds? I bet they rarely, if ever, are. There is nothing wrong with the software; although to be deterministic you'd need an infinite amount of computing power.There's nothing wrong with input data; although to be deterministic you'd need an infinite amount of input data.

Of this, of course, the designers and builders of these systems know full well the shortcomings of numerical weather prediction; and incidentally, computer models do not pick out a trend, and care not one iota whether or not they are self-consistent. If you see a trend, that's the software just doing it's grunt work. Tomorrow the grunt work may even spit out something entirely different. You have confidence when you see a trend, because you believe, whether you know it or not, that since the trend is visible, that regression to the mean has overcome the chaotic nature of weather.

Computers always do as they're told, and are subsequently always right - if you put the same set of the data through the same computer model, you'd get the same result (in terms of computation, that is deemed 'right') But one has to know what you are looking at - GFS output, for example - in order to make sense of it, since, as you imply, looking at a chart for Christmas day from this mornings GFS 6z 500hPa chart, it may well be a correct forecast, it may well be a wrong forecast, but the simple fact is - we can't tell - but the computer will carry on grunting out those charts.

One might see that as an exercise in pedantry, and perhaps it is, but, I think, at this time of year, it is important to say.

Edited by Boar Wrinklestorm
  • Like 3
Link to post
Share on other sites

Another winter forecast:

MattHugo81

The seasonal EUROSIP Nov update is out and unfortunately (for cold fans) shows a zonal, mild, wet & windy winter period.

15/11/2012 12:21

MattHugo81: The EUROSIP model takes into account the ECMWF, UKMO, Meteo-France and NCEP. As a result all seasonal models point towards a mild/wet winter

Along with the ECWF and CFS all suggesting a mild wet Winter it does go to show that even if many of the factors influencing seasonal climate hint at a cold season other synoptics seem to suggest otherwise.

Interesting, and it will be good to review at the end of Winter to see if the forum experts out perform the pro-forecasts!

Taking a blend of all the forecasts/early thoughts and data available to view then I feel the outlook for the coming winter is very much an uncertain one with mixed signals - Very interesting times ahead. I'd favour a more traditional or rather typical winter for the UK, with no real weather type set to dominate for long periods unbroken, cold, snow & frost will feature but also interspersed with some less cold mobile westerly type periods taking a share of the limelight - with a final CET close to or just below average for the 3 months, with a general lean towards below average rather than above. If we do see a well below average CET month then I expect it to be balanced out when looking at the winter as a whole, similar to 2010/11.

I'd be very interested to hear Rogers thoughts as we now make our way towards proper winter Posted Image

Edited by Liam J
Link to post
Share on other sites

Taking a blend of all the forecasts/early thoughts and data available to view then I feel the outlook for the coming winter is very much an uncertain one with mixed signals - Very interesting times ahead. I'd favour a more traditional or rather typical winter for the UK, with no real weather type set to dominate for long periods unbroken, cold, snow & frost will feature but also interspersed with some less cold mobile westerly type periods taking a share of the limelight - with a final CET close to or just below average for the 3 months, with a general lean towards below average rather than above. If we do see a well below average CET month then I expect it to be balanced out when looking at the winter as a whole, similar to 2010/11.

I'd be very interested to hear Rogers thoughts as we now make our way towards proper winter Posted Image

See a few posts up?

Edited by Catacol_Highlander
  • Like 1
Link to post
Share on other sites

SATURDAY, 17 NOVEMBER 2012

<a href="http://globalcoolingnewiceageuk.blogspot.co.uk/2012/11/potential-for-some-of-coldest-and.html" style="color: rgb(170, 221, 153); text-decoration: initial; display: block; font-weight: bold;">Potential for some of the coldest and snowiest conditions in at least a century AT TIMES...

There is also the potential for some of the coldest and snowiest conditions in at least a century AT TIMES to be recorded in the December to January period of the upcoming winter. The January period is also slightly more favorable to experience the worst of the winter conditions in terms of snowfall and temperatures.

Snippet from snow risk dates and temperature forecast available (Produced and made available on 9th November 2012) @http://www.exactaweather.com/UK_Premium_Forecast.html

Link to post
Share on other sites

Hard to read the above post, but think it says what I was reading elsewhere ... funny thing is, I am alert to the chance that this February may the coldest in some time, possibly 27 years (1986 was sub-zero) and outside chance back to 1947 which would very likely remain coldest on record no matter what.

But the news story claims coldest Dec and Jan which are much more recent benchmarks of cold, so it would have to go some just to be coldest December in three recent cases or coldest January in four. I am not seeing much evidence for it, we'll be lucky to get either of these months below normal let alone coldest in a century. My money's on February.

Link to post
Share on other sites

Hard to read the above post, but think it says what I was reading elsewhere ... funny thing is, I am alert to the chance that this February may the coldest in some time, possibly 27 years (1986 was sub-zero) and outside chance back to 1947 which would very likely remain coldest on record no matter what.

But the news story claims coldest Dec and Jan which are much more recent benchmarks of cold, so it would have to go some just to be coldest December in three recent cases or coldest January in four. I am not seeing much evidence for it, we'll be lucky to get either of these months below normal let alone coldest in a century. My money's on February.

Its from exacta ( james madden) so the more unreadable it is, the better !!

  • Like 6
Link to post
Share on other sites

Its from exacta ( james madden) so the more unreadable it is, the better !!

Absolutely agree ,If he said it, its got about as much chance of coming to fruition as winning the lottery. Edited by TobyT
  • Like 1
Link to post
Share on other sites

I think those calling a mild December as part of there winter forecast may fall at the first hurdle, IMO a lot of what we though we knew could be called into question and need further tweaking Fred. The Strat for one is certainly not playing ball as to how we understand the effects of cooling. It does make fascinating viewing this year.

As for Roger's thoughts, again his call for a mixed month in December could be way off base due to unforeseen factors that at this moment in time we simply just don't understand. Off course winter hasn't started yet and Roger's forecast could end up being spot on.

  • Like 1
Link to post
Share on other sites

I think those calling a mild December as part of there winter forecast may fall at the first hurdle, IMO a lot of what we though we knew could be called into question and need further tweaking Fred. The Strat for one is certainly not playing ball as to how we understand the effects of cooling. It does make fascinating viewing this year.

As for Roger's thoughts, again his call for a mixed month in December could be way off base due to unforeseen factors that at this moment in time we simply just don't understand. Off course winter hasn't started yet and Roger's forecast could end up being spot on.

Possibly....but then again possibly not. A chilly start, which isn't certain proves nothing. I have to admit though and have said it that dec has been very hard to call and confidence isn't high and thoughts of GP and C etc are hard to dismiss...and indeed are not dismissed...we just have our methods. However, the very signal we had in 2010 IS NOT there this Nov/December...make no bones about that. I've more or less plumped for descending winter....so time will tell.

BFTP

  • Like 1
Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...