Much has been printed in the tabloids over the past month or so that we are in for a severe winter. Not a week goes by without the Daily Express splashing that the ‘Worst winter EVER is on the way’. Just this morning the same rag told us that three months of ‘exceptionally cold’ weather are due. On closer inspection the story elaborated the scene with quotes from James Madden of Exacta Weather, one of the ‘experts’ feeding these fantastical stories.

Quite how Mr Madden and other experts arrive at these forecasts is a bit of a mystery. The mystery has deepened further since I decided to crunch a few numbers and try to predict what is in store for the months ahead. Looking at data for this area stretching back over 130 years to 1881 I decided to calculate a seasonal average and arrived at a final figure using singularities – basically looking at the weather patterns we’ve had during October and November.

Many professionals would scoff at this method of pattern-matching, so I’ve incorporated a couple of other ‘now’ factors and taken on board current variables such as sea surface temperatures in the Atlantic and Pacific.

The figure I arrived at, taken as an average of the closest matching autumn periods, is a mean temperature of 4C with rainfall totalling 133mm over the months of December, January and February – that’s about a degree colder than average and 90% of average rainfall. The probability of a winter with a mean temperature of between 4C and 5C is 37% – the most likely outcome. With this in mind a winter in the form of 1986-87 is possible – though whether we would see the same extremes of temperature and snowfall that we experienced in January 1987 is open to question.

Here is a link of my method to predict the coming winter.

The Met Office, with all their computing power and expertise, seem to be hinting at something similar – ie slightly colder than normal. The opening couple of weeks to my forecast seem quite plausible, looking at tonight’s models. Look at the similarity in the synoptic charts for December 8th here.

Over the last few years the snow lovers among us have been spoiled after a run of very mild winters during the late Nineties and early Noughties that prompted climate expert Dr David Viner to utter the immortal words that one day “children just aren’t going to know what snow is”. It was only a matter of a few years before the words from the senior research scientist at the University of East Anglia seemed a bit hollow.

Mark Twain, born this day in 1835 , once uttered the famous phrase: “Climate is what we expect; weather is what we get” – there is always a chance we could suddenly lurch back to milder winters. This winter could spring a surprise, be it very cold or very mild. But judging what’s happened over the last 130-odd years – a 1962/63 or a balmy 1989/90 look very unlikely.

* Forecasting models use probability on any given outcome. Millions of observations are fed into the Met Office database (and other countries’ weather agencies) every day. Supercomputers then crunch through this data to give probable outcomes. With the volatility of the atmosphere it is not surprising that certainty of any outcome often falls away rapidly. Forecasting has improved greatly in the last 20 years – though anything the models churn out beyond three to five days should be handled with caution. Long range models can give *some* idea of general trends for the months ahead – but changing just one variable can vastly alter an outcome at the end of the run.

* *Over the past few months I have been collating data for the area around Wanstead. This data is freely available from the excellent Met Office library and is emailed via Excel spreadsheet. Rainfall stats include a near-complete daily archive, stretching from 1961 to 2003, from City of London Cemetery . Sadly the rainfall station, along with many others, ceased to supply the Met Office after cuts were made shortly after the turn of century. Prior to 1961 I have used monthly figures taken at the Greenwich Royal Observatory stretching back to 1881. Though this is 6 miles away the difference in temperature between the two areas would be miniscule compared with rainfall data and so can be used. I use my own stats for the period after 2003.