Probabilities for winter for the London area would be a more apt title for this post but that’s not very exciting compared with the hyperbole published almost daily by the likes of the Daily Express.
Much has been written about the sources of their forecasts over the past couple of years. Splash headlines that promise Armageddon Arctic conditions or Biblical blizzards never seem to materialise. So without boring you further I’ll move on to my own views on how I think the next 90-odd days will unfold.
Taking into account more sophisticated methods than my own there seems to be a signal for something colder than last year – though that’s not saying much given that 2013/14 was the 11th warmest winter on record with NO snow falling observed in this area.
Though it was abandoned by the Met Office years ago I have decided to base my predictions more on analogues this year. I have a lot more data at my disposal – the series I use now extends past 1881 back to 1797.
The figure I arrived at, taken as an average of the closest matching autumn periods, is a mean temperature of 4.2C with rainfall totalling 156mm over the months of December, January and February – that’s about a degree colder than average and about average rainfall – though I think it could be less given that December is looking relatively dry on the current model output.
The probability of a winter with a mean temperature of between 3C and 4C is 33% – the most likely outcome – though this low figure emphasizes the mixed signals this autumn. With this in mind a winter similar to 1984/85 is possible which saw a brief cold snap at the end of December together with a two-week cold spell that began during the first week of January. There was also a 10-day cold spell during February of that winter.
To try to add value to the above outcome I also had a look at the likelihood of this winter being as mild as last winter.
It has been noted in the past that mild winters often come in twos. I had a look back through the series to see if this was true.
The occurrence of two very mild winters in succession is 27/217 (12.4%). The occurrence of a very mild winter being followed by a very cold or severe winter is 16/217 (7.4%). So while, from these simple stats, another mild winter is more likely, it is not really high enough to consider over the average I found in the first calculation made from straight autumn statistics.
A final fact to consider is just how mild and wet this year has been. Every month this year, apart from August, has been warmer than average by an average factor of 1.2. If December continues warm this year could possibly end as the warmest on record. However, it is also possible that nature is about to redress the balance.
Here is a link of my method to predict the coming winter.
* 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.
You must be logged in to post a comment.