Does the weather have a memory?
by Philip Eden

There has always been an impression that certain types of weather recur at certain times of the year � indeed, the idea is enshrined in ancient country lore which contains myriad sayings and proverbs which tie climatic characteristics to the calendar.

But most thinking people would regard as preposterous the notion that the atmosphere can remember how it behaved on a particular date in previous years so that it can do the same this year. Put as starkly as that, the idea is certainly irrational and unscientific. But a gargantuan heap of statistical work over the last century and half has identified some significant tendencies to unusual weather at particular times of the year. These seasonal tendencies are called �singularities� � a word coined by the German climatologist, A.Schmauss, in 1938.

Work on singularities can be said to date back to Alexander Buchan's analysis of Edinburgh temperatures in 1869 in which he identified times of the year which were regularly warmer or colder than would be expected from the smoothed annual curve of mean temperature. These famously became known as -Buchan's cold and warm periodsö and were widely quoted out of context during the last century. More detailed and extensive work was conducted on the continent � notably in Germany, the Netherlands, Austria and Sweden � during the early-1900s, and the subject was taken up in the UK Meteorological Office by C.E.P.Brooks, J.E.Belasco, and H.H. Lamb from the 1930s onwards.

Rigorous statistical techniques were applied to daily sea-level pressure patterns over Europe and the north Atlantic over a period of 60 years. The end result was that several key periods were identified throughout the year when these synoptic patterns deviated markedly from the normal seasonal progression. The events certainly did not happen every year, nor were any exact dates set in stone, but more than 20 singularities which occurred in more than half the years of the analysis were detected in the British climate:

Event Avg. Start Avg. End Peak Frequency %
Early-Jan storms 5 Jan 17 Jan 8 Jan
Mid-Jan settled 18 Jan 24 Jan 21 Jan
Late-Jan storms 24 Jan 1 Feb 31 Jan
Early-Feb settled 8 Feb 16 Feb 13 Feb
Early-Mar storms 26 Feb 9 Mar 1 Mar
Mid-Mar settled 12 Mar 19 Mar 14 Mar
Late-Mar storms 24 Mar 31 Mar 28 Mar
Mid-Apr storms 10 Apr 15 Apr 14 Apr
Late-Apr rains 23 Apr 26 Apr 25 Apr
June monsoon 1 Jun 21 Jun 16 Jun
July heatwave 10 Jul 24 Jul 16 Jul
Late-Aug winds 20 Aug 30 Aug 28 Aug
Early-Sep warmth 1 Sep 17 Sep 10 Sep
Mid-Sep storms 17 Sep 24 Sep 20 Sep
Old Wives� Summer 24 Sep 4 Oct 29 Sep
Early-Oct storms 5 Oct 12 Oct 9 Oct
St Luke�s Summer 16 Oct 20 Oct 19 Oct
Mid-autumn storms 24 Oct 13 Nov 29 Oct
St Martin�s Summer 15 Nov 21 Nov 18 Nov
Early-Dec storms 24 Nov 14 Dec 9 Dec
Mid-Dec settled 18 Dec 24 Dec 21 Dec
Christmas storm 25 Dec 1 Jan 28 Dec

After a further sixty years these singularities are still identifiable on many occasions. Even in this era of high-tech weather forecasting, the list can still come in useful from time time: for example, when medium-range ensemble forecasts point in two contrasting directions the real atmosphere is most likely to follow the route closest to any relevant singularity.