Winter Early Or Late ?

[Roger J Smith]

CET analysis -- winter early or late ??

__________________________________________________________

Using just long-term average temperatures, I was able to see that February was milder than January for much of the 18th century, but colder in the 19th century. The trend then began to oscillate through the 20th century, but generally went back to a preference for January as the cold month. The epic winter of 1947 was a major exception, and so was 1956.

However, cold weather can hit in December too. Variations from January to February might tell us something about how winter comes early or late, but using December data establishes the trends better ... there is a good correlation of December and January temperatures, so an early winter is likely to be a cold winter in general.

An index of "lateness" was developed by comparing the 11-year running mean of December anomalies with the following Februaries. The index was calculated by subtracting the February mean anomaly from the December mean anomaly. If the value was zero, this tells us that February was no milder or colder, relative to average, than December was. A high value suggests that February was colder. A low (negative) value suggests an early bias to the season.

The following graph shows the results of this analysis. (Click on current if the graph does not immediately appear)

CET_LATE.xls

... from 1664 to 2003, since first period was 1659-69 and last period was 1998-2008.

---------------------------------------------------------------------------------------------------------------------------------

Note in each case that the December of the year cited on the graph was in the previous year. In the table, column A is the Dec anomaly for each year, followed by the Jan and Feb anomalies for the following year. That places the three data points in the same winter season. The first cell (A1) is blank because it falls in Dec 1658 before the series began.

The extremes of the lateness/earliness index are identified on the graph. I apologize that the labels are shifted to the right of the peaks and troughs, due to program conversion.

Here's a general summary of these trends:

The heart of the LIA cold spell showed a bias towards later winters. This makes sense when one considers that some of those winters extended well into March and that the ice margin around Iceland and Jan Mayen must have been quite far south in the late winter and early spring. Hence the peak in 1687 falls close to some very cold winters such as 1684 (remember these are 11-yr running means).

At no time before 1780 did winters generally come and go "early" according to this analysis. There were occasions when they came later, as with the peaks in 1750 and 1769. The milder spell around 1720-39 seemed to generate a slight shift to earlier winters, but not a very strong one.

Around 1780 it is clear that something shifted in the climatic background, and winters came earlier through several decades that were also notable for frequent cold winters. (the index does not measure the severity of winters, except that early winters tend to be colder than late winters -- January anomalies, which do not figure in this index at all, drive the bus as far as winter severity is concerned; it is a rare cold winter that does not have a cold January).

The bias to early winter "peaks" in 1792 and 1813, but really this whole period tends to early onset. The period is also fairly similar to the Dalton minimum in solar variability, although it began during the high activity cycle of the late 1780s.

By the 1820s the pendulum was swinging to late winters again. This trend dominated from about 1825 to 1865. A new development was emerging, the winter with only one cold month that was a February. or a particularly cold February, like 1855. It can be noted that this month was also very cold in eastern North America.

By the 1870s the early winter phenomenon was back, and it reached its zenith around 1880. Three obvious examples would be the winters of 1879-80, 1890-91, and 1894-95. These late, late-Victorian winters seemed to end rather suddenly when the climate warmed around 1896 (although February 1902 was also rather cold, it did not manage to dominate any 11-year periods).

The mild period from about 1896 to 1939 showed little bias either way and the few colder winter months that managed to appear were scattered through the three calendar months. Nevertheless, winters seemed balanced on January as their mid-point even though sometimes January was much milder than the other two months (as with 1916).

Another surge of later winters greeted the postwar Britain, but I suspect most of that peak in 1952 comes when 1947 and 1956 fall into the same grouping for two data points. Otherwise, the lateness index conceals the fact that winters suddenly became much colder around 1940. The timing didn't shift then, just the severity.

The slight "early" signal in 1962 is really a return to equilibrium around the cold winter of 1962-63 and through the decade, but eventually winters started trending later towards 1983. This is despite the fact that one or two notable exceptions could be found, such as 1981-82. In this colder period from about 1979 to 1987, it seems that the statistical trend was towards later winters, perhaps indicating that this was a response to the same signal as had dominated the LIA.

With the period widely thought to be "even larger teapot" or "global warming" in nature, winters shifted back to a more balanced timing and if anything started to fall early in the 1990s. This reflected a weak arctic vortex that could easily be shoved back north even by the feeble insolation increase of February.

Although using an 11-yr rm means that we have no data points on the graph after 2003, the trend of what is available in shorter bursts, shows a trend to later winters again. This is perhaps why the global warming or even larger teapot period has hung on for so long, perhaps the background signal is for cooling in general but some other variable has swung the cold signal to the unfavourable later manifestation. My speculation is that a winter could appear soon that swings back to the early signal and combines with the colder signal in the background of long-term climate to produce the long-foretold "old fashioned winter." It may very well be this winter. An early start would signal a better chance of below normal temperatures in January, which is what really seems to count in producing a memorable winter. Very few such winters fail to include a cold January, even if they are not dominated by it.

[North-Easterly Blast]

It is a very interesting fact that December was generally very cold in the 1960s, after being generally mild in the 1950s followed by generally cold Jans and Febs, then very mild again in the 1970s and 1980s. The 1977-87 era generally saw very cold winters, whilst December was often the mild month (1981 apart) followed by much colder weather in January and February. In the last two decades January and February seem to have been the major casualties of the even larger teapot era, whilst December actually got quite a bit colder in the 1990s after the mild 70s and 80s, followed by generally mild Jans and Febs. Although in the 2000s it does seem to have become more balanced although January and February have still seen the largest warming trend compared to the past.

It is a very puzzling fact to me that only January and February have been the casualties of the modern mild winters, whereas December appears to have changed very little, and even got generally colder in the 1990s

[Summer of 95]

It is a very interesting fact that December was generally very cold in the 1960s, after being generally mild in the 1950s followed by generally cold Jans and Febs, then very mild again in the 1970s and 1980s. The 1977-87 era generally saw very cold winters, whilst December was often the mild month (1981 apart) followed by much colder weather in January and February. In the last two decades January and February seem to have been the major casualties of the even larger teapot era, whilst December actually got quite a bit colder in the 1990s after the mild 70s and 80s, followed by generally mild Jans and Febs. Although in the 2000s it does seem to have become more balanced although January and February have still seen the largest warming trend compared to the past.

It is a very puzzling fact to me that only January and February have been the casualties of the modern mild winters, whereas December appears to have changed very little, and even got generally colder in the 1990s

From about 1990 to 2005 December was actually colder on average than February (and similar to January) over much of the country, which is contrary to what the textbooks say about a "maritime" climate.

Having said that, since 2005 December hasn't lived up to expectations, 2006 was a bit of a shock to the system not having any hint of winter before New Year (for the first time since 1994)- it was a common occurrence in the 80s. February, which was a write-off for much of the 90s as far as winter weather was concerned, has shown signs of recovery since 2000 with only one (2002) being on a par with 1990, 1995, 1997, 1998. January has for the most part not been worth bothering about since 1987, with only a couple of cold first weeks (in 1997 and 2002) and one cold last week (1996) going into bat for the Winter side in that month. March, however, has definitely been more dependable for wintry weather since the start of the millennium compared to the late 80s/early 90s when it hardly seemed to produce any snow at all.

[Weather-history]

As I've mentioned in a thread I posted on here a while back on how mild Februarys dominated a good part of the Victorian age.

http://www.netweather.tv/forum/index.php?s...mp;hl=1846-1885

5 out of 6 consecutive Februarys had a CET of 6.0 or over and it became such that December was a colder month than February over 30 year averages at one stage.

The February average for 1856-85 was as high as 4.7

[Roger J Smith]

Some evidence of the trends being discussed can be seen in the next graph, which shows actual values for Dec, Jan, Feb averaged over long periods (21-yr running mean or r.m.) and not monthly anomalies.

This separates out the three months to some extent because the long-term averages have been 3.2 C for Jan, 3.9 C for Feb and 4.1 C for December.

Have a look at the graph and you'll see some of the trends that NW members above were discussing.

CET_WINT.xls

I've said that December is the "top curve" but not all the way across the graph, December is the curve with data points shown as squares; February is shown as diamonds, and January is shown as a line but it's quite easy to identify as it stays lower almost all the time.

January is almost always the lowest of the three curves, but it does get tangled up with Feb and sometimes Dec during milder periods. When the weather turns colder, January usually responds and is by far the coldest of the three months. However, around 1880 December managed to win out as coldest month, briefly. Keep in mind these are averages for 21 years on the go, and therefore include all of two decades.

The trends for February and December show more variation. Although I've stated that December is the highest of the three curves, that's just in general terms, sometimes February is higher, for example, for almost half a century around 1800, and again around 1870, February was quite a bit higher than December. Remember, December is shown by squares, when diamonds break through to the top, that's February as the mildest of the three winter months.

In the modern period, December showed a fairly sharp rise in this 21-yr r.m. value a bit earlier than either January or February, which in more recent times have had very similar temperatures and have been warming up about at the same pace, although they remain cooler than December.

Looking back over the three and a half centuries, then, you can see that the character of winter in the U.K. (the CET region at least) has changed in many ways. First of all, it has generally warmed up, but notice how mild it was for many years around 1920 and even back around 1730. The cold winter of 1740 wasn't the heart of some long cold interval, but shows its mark on what was otherwise a relatively mild period. The colder January regime came a bit later, let's say 1760 to 1820.

There are times when both December and January were cold relative to normal, and February was mild. This was shown already in the early/late index.

It should be noted that the first interval on this graph is 1659-79 and the last interval is 1988-2008. The December in each grouping belongs to the next calendar year.

Edited November 16, 2008 by Roger J Smith

[damianslaw]

I much prefer the concept of a cold dec/jan combo as opposed to cold jan/feb combo, when we don't see a single cold spell in december i get very frustrated, love the cold in december it adds to the seasonal feel etc, by february people are in need of some sunshine and mild warmth.