On 2011-01-20 15:17:10 +0000, "Ian Bingham"
said:

"John Hall" wrote in message
...
These are CET daily mean temps in degrees Celsius for the 3 winter
months over the period 1772-2007 (2006 in the case of December).

Jan Feb Dec
1 3.29 3.92 5.11
2 3.30 3.86 5.02
3 3.18 3.99 5.05
4 3.38 3.99 5.08
5 3.23 4.07 5.14
6 3.36 4.11 4.80
7 3.18 4.23 4.59
8 3.21 3.98 4.36
9 3.35 3.93 4.17
10 3.43 3.92 4.27
11 3.28 3.80 4.18
12 3.04 3.71 4.37
13 3.28 3.63 4.49
14 3.31 3.72 4.49
15 3.19 3.97 4.39
16 3.34 3.89 4.32
17 3.34 3.74 4.42
18 3.30 3.73 4.38
19 3.44 3.76 4.06
20 3.29 3.73 3.93
21 3.48 4.14 3.94
22 3.56 4.33 3.87
23 3.50 4.45 3.82
24 3.51 4.40 3.77
25 3.39 4.48 3.44
26 3.46 4.45 3.41
27 3.62 4.59 3.41
28 3.54 4.50 3.40
29 3.72 - 3.64
30 3.79 - 3.59
31 3.96 - 3.44

I didn't bother with Feb 29th, as calculating the mean would have been
too fiddly to be worthwhile.

I've already commented on January in another post. February shows a
tendency to be a bit colder between about the 11th and 20th, giving some
credence to the Buchan cold spell (which the singularity work of Brooks
and Lamb I believe also identified). The last week is notably milder.

December shows a tendency for a relative cold period from about the 8th
to 12th. It then cools down again from about the 19th. There's no sign
of a post-Christmas mild spell, though, with the 25th to 31st being the
coldest part of the month. I suspect that the mean for the last 100
years only might show very different trends.
--
John Hall

Thanks for these figures, John. Very interesting. What strikes me
about them though is how un-smooth they are for an average of 235
years. For example, in the second half of February the temperature
rises from 3.73 to 4.45 in just 3 days. This rather suggests that the
winter temperature profile has changed during that period, and probably
several times. For example it changed after Buchan's time, resulting
in the poor fellow getting an undeserved raspberry from the
meteorological cognoscenti.

My 21-year mean temperature falls throughout December to a minimum in
the week between Christmas and the New Year, rises to the mid-January
maximum, then falls to the last week of January. It then rises to the
6th of February, after which comes a cold period from the 8th to the
19th which, as you say, roughly corresponds to Buchan's First Cold
Spell, but overshoots it by 5 days.
Ian Bingham,
Inchmarlo, Aberdeenshire.


I computed the three day running mean to even out day-to-day variation,
and plotted the running mean against winter day number (December 1 = 1,
February 27 = 89). Then as you expect you get a nice shallow U curve,
with the start of December significantly more mild than the end of
February. The trough of the curve is January 12-13, with means of 3.20
and 3.21.

There are two anomalies in the U though: a cold spell around December
10 and a warm spell around February 6.

The early December cold spell falls in a supposed Buchan warm spell,
and the February warm spell is followed by a Buchan cold spell, which
in fact isn't colder than you would expect, but is colder than the warm
spell immediately before it.

Those anomalies are still obvious if you use 5-day smoothing rather than 3-day.

Oh well. Back to work.

--

Trevor
Graphing in Lundie, in the Sidlaw Hills, 10m NW of Dundee, elev. 185m
Weather through www.trevorharley.com