Where has the winter cold gone?

[Stratos Ferric]

It's that time of year when I revisit the winter stats to explore the potential for the coming season. I'm sure I'll reprise some updated old favourites, but first up a new plot. This is in direct response to the chatter recently about the potential for sustained winter cold.

What the plot shows is that winter spells of cold weather (I've used the cut off of a mean of 0C per day) have become far less frequent. For ease of viewing I've smoothed the data a little, what's shown is five day rolling means, but it removes some of the spikiness in the data whilst not distorting the relative length and intensity of cold weather (the net effect will be to shorten the plotted duration very slightly). What is manifestly clear is that, as has been pointed out by a few people, cold is still possible (just) but is far less frequent, of shorter duration, and less intense in the modern climate than was formerly the case.

What can we say for this winter? Little other than that a prolonged spell of intense cold is unlikely.

[West is Best]

What is manifestly clear

Must just be me Stratos but the entire think looks the same to me from start to finish. Can't see any differences at all.

What can we say for this winter? Little other than that a prolonged spell of intense cold is unlikely.

Gordon Brown could do with someone like you at the moment ...!

[Stratos Ferric]

Must just be me Stratos but the entire think looks the same to me from start to finish. Can't see any differences at all.

Gordon Brown could do with someone like you at the moment ...!

WiB, if you're honestly suggesting you cannot see the difference in the pattern of the solid bars at the bottom of the plot then you are either more obtuse, or more stupid, than I thought. I don't believe you're the latter, so I'll assume the former. Of course, I might be being a bit harsh, so...this might help if, like WiB, the difference isn't apparent. This data is NOT smoothed, it uses actual days.

At least for once you're being honest about the political situation. Presumably your objective bandwidth is limited to one weighty matter at any one time.

[jethro]

I'm sure you're a very busy man but is there any chance you could this for the whole of the 20th century?

[Stratos Ferric]

I'm sure you're a very busy man but is there any chance you could this for the whole of the 20th century?

The reason for not doing [already] is because the plot would be so squeezed that it would be illegible [and it's also a cumbersome file to work with]. What I will do at some point over the weekend is produce the previous two blocks of 30 years as well. However, there will be a strong correlation between the occurrence of cold days and the overall CET / winter mean. If one were to produce a plot showing anomaly in cold days and anomaly in CET the polarity of each line (above / below the overall period average) would probably match fairly well.

ON this basis I would EXPECT that the 40s through 60s would have significantly more marked occurrences, whilst the 20s / 30s would have reduced levels, though, I suspect, still more than we have at present.

Edited October 5, 2007 by Stratos Ferric

[jethro]

Thanks SF, I'm sure it will make interesting reading. I also think it may highlight the differences fairly graphically between what the younger members on here consider to be a cold period and what those of us who are, somewhat longer in the tooth, consider to be a cold period.

[BrickFielder]

There may be a slight risk in using the data in this way with respect to the differences in night and day temperatures. Clearly mean temperatures have risen ,yet a large part of that is due to higher night time temperatures and more overnight cloud. Using the data to show cold spells and the relative likelihood of cold spells is OK but it does not necessarily reflect the likelihood of snow falling even if in all probability there is a common linkage to a root cause.

[Stratos Ferric]

There may be a slight risk in using the data in this way with respect to the differences in night and day temperatures. Clearly mean temperatures have risen ,yet a large part of that is due to higher night time temperatures and more overnight cloud. Using the data to show cold spells and the relative likelihood of cold spells is OK but it does not necessarily reflect the likelihood of snow falling even if in all probability there is a common linkage to a root cause.

True, but it's a decent proxy - although for that I would perhaps choose a t level slightly above zero. The problem is snow is nearly always marginal in the UK. About the only sure fire assessent with this data would actually be to make a case for periods when it couldn't snow, and prove the reverse of the starting hypothesis: i.e. that the period when it could NOT snow has grown significantly.

The other option will be to use just the minima, which perhaps is a good idea, so I'll do an assessment at some point that uses the same plots but on the basis of days with a minimum t below 1C. I'd expect the general pattern to be the same though.

[Chris Knight]

SF, Interesting charts, one question though. Your legend notes the months in the order JFMD, does that mean for 1957 you are using the data for J, F, M, and D of 1957, or the D of 1956, with the J,F,M of 1957 which is in the same winter, albeit not the same year?

[snowsure]

I appear to be missing something here; What happened 1970-1978? 8 years without cold? How about 1987-1991?

Why have you picked ranges 1962-1970, 1978-1987 and 1991-2006?

Again, apologies for being thick. Interesting post though. One that, I am sure, will continue to cause discussion.

[Stratos Ferric]

SF, Interesting charts, one question though. Your legend notes the months in the order JFMD, does that mean for 1957 you are using the data for J, F, M, and D of 1957, or the D of 1956, with the J,F,M of 1957 which is in the same winter, albeit not the same year?

The former: all months in one year together, rather than all months in one winter. Although, the way the plot works, the months stack vertically in each year, each one starting on the left of a column where the previous on finished on the right. Therefore, plot wise, the picture would hardly be changed if we moved December across one column: instead of being the highest plot in each year, it would become the lowest in the next - the overall shape of the curve is the same.

I appear to be missing something here; What happened 1970-1978? 8 years without cold? How about 1987-1991?

Why have you picked ranges 1962-1970, 1978-1987 and 1991-2006?

Again, apologies for being thick. Interesting post though. One that, I am sure, will continue to cause discussion.

I've chosen the periods that are coldest, as indicated by days averaging 0C, and deliberately excluded the spells with little cold: these are obvious to the eye. The intention with this plot was to help WiB with his little difficulty expressed higher up, where he was struggling to see whether anything much had changed. The selected periods, therefore, are those with runs of at least some cold. I shall let you draw your own conclusion.

[Chris Knight]

The former: all months in one year together, rather than all months in one winter. Although, the way the plot works, the months stack vertically in each year, each one starting on the left of a column where the previous on finished on the right. Therefore, plot wise, the picture would hardly be changed if we moved December across one column: instead of being the highest plot in each year, it would become the lowest in the next - the overall shape of the curve is the same.

Thank you. EDIT: It does appear to put your block bars out of order, as it were, in the first diagram, and I guess your second curve would fit more tightly too if D snuggled up against the following J instead of the preceding (or do I mean Following?) M. It may also change your annual curve slightly too. Just a thought.

Chris

Edited October 5, 2007 by Chris Knight

[Stratos Ferric]

Thank you. EDIT: It does appear to put your block bars out of order, as it were, in the first diagram, and I guess your second curve would fit more tightly too if D snuggled up against the following J instead of the preceding (or do I mean Following?) M. It may also change your annual curve slightly too. Just a thought.

Chris

On the first diagram it would move each of the red blocks right one column. Not really material in the overall assessment; you can't alter the fact that nowadays we just don't seem to get sustained runs of below freezing means.

[snowsure]

I've chosen the periods that are coldest, as indicated by days averaging 0C, and deliberately excluded the spells with little cold: these are obvious to the eye. The intention with this plot was to help WiB with his little difficulty expressed higher up, where he was struggling to see whether anything much had changed. The selected periods, therefore, are those with runs of at least some cold. I shall let you draw your own conclusion.

Perhaps a little naughty to include the 1963 episode; At 50 days, it seems a little bit of an outlier and skews the data. It suggests that the average number of sun zero winter days has gone from 18 to 14 yo 6.

However, a 5 or 10 year rolling mean plot demonstrates that we are in a period of rapidly decreasing days with a temp<0*C.

Does your sequence show the CET zone? Would we see the same with temperature records from, say, a more northerly area?

[snowsure]

Can't edit the above post of mine so this is an addendum:

Having looked at Sheffield (Here) I can confirm that the number of days with an air-frost has also decreased since 1957. The 5 year rolling average, 10 year rolling average and the raw data all confirm the point made by SF at the beginning. I, for one, am a little more despondent regarding the prospects for this coming winter.

[Stratos Ferric]

Can't edit the above post of mine so this is an addendum:

Having looked at Sheffield (Here) I can confirm that the number of days with an air-frost has also decreased since 1957. The 5 year rolling average, 10 year rolling average and the raw data all confirm the point made by SF at the beginning. I, for one, am a little more despondent regarding the prospects for this coming winter.

I'd always be guarded about making year on year extrapolations. Despite WiB's remarkable change of 'allegiance', and his pandering for October cold, an autumnal continuation of the summer cold could, from a purely statistical perspective, be the worst possible lead in to winer. Why so? Long runs of cold are virtually unknown nowadays, and - save for one caveat - a cold September - November period would have been expected to give way to a mild winter. Of course, there's no saying that that still won't happen, but purely by the numbers, having had a few cool months, a milder autumn is better for those hoping for a cool winter.

The forgoing caveat: that we haven't arrived suddenly at some point at which our climate is now cooling.

[Weather-history]

Here's a table I did on UKWeatherworld, last year (now updated), on the total number of days for each year where the CET daily mean was 0C or less.

2007: 2 so far

2006: 4

2005: 2

2004: 0

2003: 7

2002: 3

2001: 8

2000: 3

1999: 2

1998: 1

1997: 7

1996: 14

1995: 8

1994: 6

1993: 3

1992: 12

1991: 16

1990: 0

1989: 1

1988: 1

1987: 17

1986: 24

1985: 28

1984: 4

1983: 6

1982: 10

1981: 19

1980: 6

1979: 23

1978: 17

1977: 6

1976: 8

1975: 0

1974: 2

1973: 3

1972: 3

1971: 8

1970: 17

1969: 13

1968: 11

1967: 6

1966: 8

1965: 12

1964: 10

1963: 48

1962: 24

1961: 13

1960: 3

1959: 7

1958: 9

1957: 2

1956: 23

1955: 22

1954: 17

1953: 4

1952: 13

1951: 5

1950: 17

1949: 1

1948: 5

1947: 44

1946: 18

1945: 11

1944: 5

1943: 3

1942: 27

1941: 16

1940: 33

1939: 6

1938: 5

1937: 4

1936: 12

1935: 9

1934: 1

1933: 8

1932: 2

1931: 11

1930: 2

1929: 24

1928: 5

1927: 11

1926: 6

1925: 14

1924: 4

1923: 5

1922: 5

1921: 1

1920: 5

1919: 12

1918: 8

1917: 31

1916: 7

1915: 4

1914: 3

1913: 1

1912: 8

1911: 2

1910: 5

1909: 13

1908: 10

1907: 10

1906: 6

1905: 5

1904: 12

1903: 10

1902: 16

1901: 16

1900: 9

1899: 17

1898: 0

1897: 8

1896: 4

1895: 31

1894: 9

1893: 9

1892: 28

1891: 21

1890: 32

1889: 14

1888: 14

1887: 21

1886: 26

1885: 7

1884: 3

1883: 5

1882: 4

1881: 25

1880: 16

1879: 42

1878: 16

1877: 1

1876: 7

1875: 5

1874: 26

1873: 9

1872: 0

1871: 17

1870: 23

1869: 9

1868: 4

1867: 26

1866: 4

1865: 17

1864: 21

1863: 0

1862: 13

1861: 19

1860: 21

1859: 11

1858: 14

1857: 8

1856: 14

1855: 35

1854: 7

1853: 26

1852: 2

1851: 2

1850: 15

1849: 11

1848: 12

1847: 13

1846: 15

1845: 23

1844: 25

1843: 7

1842: 12

1841: 26

1840: 23

1839: 16

1838: 35

1837: 10

1836: 12

1835: 15

1834: 0

1833: 9

1832: 5

1831: 13

1830: 25

1829: 26

1828: 3

1827: 25

1826: 13

1825: 4

1824: 5

1823: 22

1822: 8

1821: 6

1820: 30

1819: 19

1818: 13

1817: 12

1816: 20

1815: 28

1814: 53

1813: 20

1812: 16

1811: 21

1810: 17

1809: 16

1808: 22

1807: 21

1806: 6

1805: 12

1804: 16

1803: 26

1802: 15

1801: 16

1800: 16

1799: 34

1798: 19

1797: 7

1796: 23

1795: 42

1794: 8

1793: 3

1792: 16

1791: 10

1790: 2

1789: 13

1788: 20

1787: 8

1786: 24

1785: 25

1784: 41

1783: 17

1782: 15

1781: 4

1780: 33

1779: 10

1778: 7

1777: 12

1776: 32

1775: 3

1774: 25

1773: 20

1772: 24

CET Ice Days since 1878 (CET maximum 0C or less).

1878: 8

1879: 11

1880: 4

1881: 12

1882: 1

1883: 0

1884: 1

1885: 0

1886: 2

1887: 3

1888: 0

1889: 1

1890: 14

1891: 8

1892: 9

1893: 5

1894: 5

1895: 15

1896: 0

1897: 1

1898: 0

1899: 3

1900: 0

1901: 4

1902: 2

1903: 2

1904: 2

1905: 0

1906: 1

1907: 2

1908: 2

1909: 0

1910: 0

1911: 0

1912: 3

1913: 0

1914: 0

1915: 0

1916: 0

1917: 7

1918: 0

1919: 0

1920: 1

1921: 0

1922: 0

1923: 0

1924: 0

1925: 1

1926: 3

1927: 7

1928: 0

1929: 10

1930: 0

1931: 0

1932: 1

1933: 1

1934: 0

1935: 1

1936: 1

1937: 1

1938: 3

1939: 0

1940: 8

1941: 5

1942: 6

1943: 0

1944: 1

1945: 5

1946: 3

1947: 22

1948: 1

1949: 1

1950: 1

1951: 0

1952: 0

1953: 0

1954: 6

1955: 3

1956: 9

1957: 0

1958: 0

1959: 1

1960: 0

1961: 4

1962: 5

1963: 15

1964: 0

1965: 0

1966: 2

1967: 1

1968: 0

1969: 1

1970: 0

1971: 2

1972: 1

1973: 0

1974: 0

1975: 0

1976: 0

1977: 1

1978: 2

1979: 4

1980: 0

1981: 5

1982: 8

1983: 0

1984: 0

1985: 12

1986: 6

1987: 6

1988: 0

1989: 0

1990: 0

1991: 4

1992: 1

1993: 1

1994: 1

1995: 2

1996: 4

1997: 2

1998: 0

1999: 0

2000: 0

2001: 0

2002: 0

2003: 0

2004: 0

2005: 0

2006: 1

2007: 0 so far

Number of double digit CET (10.0C+) maxima during the following winters (1st Dec-28th/29th Feb)

1878-79: 3

1879-80: 13

1880-81: 8

1881-82: 12

1882-83: 19

1883-84: 19

1884-85: 16

1885-86: 2

1886-87: 12

1887-88: 6

1888-89: 11

1889-90: 21

1890-91: 7

1891-92: 12

1892-93: 9

1893-94: 15

1894-95: 5

1895-96: 18

1896-97: 11

1897-98: 19

1898-99: 37

1899-00: 11

1900-01: 17

1901-02: 14

1902-03: 21

1903-04: 4

1904-05: 14

1905-06: 7

1906-07: 8

1907-08: 13

1908-09: 6

1909-10: 14

1910-11: 22

1911-12: 23

1912-13: 24

1913-14: 27

1914-15: 10

1915-16: 30

1916-17: 6

1917-18: 19

1918-19: 19

1919-20: 23

1920-21: 27

1921-22: 24

1922-23: 15

1923-24: 5

1924-25: 17

1925-26: 22

1926-27: 5

1927-28: 17

1928-29: 5

1929-30: 14

1930-31: 10

1931-32: 22

1932-33: 18

1933-34: 6

1934-35: 33

1935-36: 3

1936-37: 25

1937-38: 20

1938-39: 21

1939-40: 8

1940-41: 8

1941-42: 11

1942-43: 30

1943-44: 14

1944-45: 20

1945-46: 19

1946-47: 4

1947-48: 23

1948-49: 31

1949-50: 26

1950-51: 3

1951-52: 16

1952-53: 14

1953-54: 17

1954-55: 20

1955-56: 18

1956-57: 30

1957-58: 14

1958-59: 14

1959-60: 20

1960-61: 19

1961-62: 20

1962-63: 3

1963-64: 8

1964-65: 12

1965-66: 26

1966-67: 18

1967-68: 12

1968-69: 11

1969-70: 5

1970-71: 13

1971-72: 14

1972-73: 18

1973-74: 22

1974-75: 36

1975-76: 27

1976-77: 3

1977-78: 14

1978-79: 7

1979-80: 20

1980-81: 18

1981-82: 15

1982-83: 21

1983-84: 18

1984-85: 17

1985-86: 18

1986-87: 17

1987-88: 18

1988-89: 37

1989-90: 36

1990-91: 11

1991-92: 22

1992-93: 19

1993-94: 19

1994-95: 32

1995-96: 11

1996-97: 19

1997-98: 37

1998-99: 29

1999-00: 30

2000-01: 20

2001-02: 34

2002-03: 24

2003-04: 21

2004-05: 26

2005-06: 9

2006-07: 40

Edited October 6, 2007 by Mr_Data

[Stratos Ferric]

Here's a table I did on UKWeatherworld, last year (now updated), on the total number of days for each year where the CET daily mean was 0C or less.

2007: 2 so far

....2006-07: 40

Is there a record for the longest post ever submitted?

And unless I'm very much mistaken that data shows that yet another benchmark was exceeded last winter.

[J10]

Based on the excellent 0c figures provided by Mr Data, a chart showing the downward progress of the sub CET 0c days per year.

[J10]

2nd Chart, this time showing the rise in the number of days exceeding 10c, again based on Mr Data's excellent work.

[johnholmes]

interesting charts both those Julian.

I wonder what most climatologists were saying in the mid 30's with that upward blip in temperatures.

The more I look at basic data and read more on climatic changes the less sure I am on just what is going on, let alone why, and even less, what will/may happen in the future.

[pottyprof]

WiB, if you're honestly suggesting you cannot see the difference in the pattern of the solid bars at the bottom of the plot then you are either more obtuse, or more stupid, than I thought. I don't believe you're the latter, so I'll assume the former. Of course, I might be being a bit harsh, so...this might help if, like WiB, the difference isn't apparent. This data is NOT smoothed, it uses actual days.

At least for once you're being honest about the political situation. Presumably your objective bandwidth is limited to one weighty matter at any one time.

Is it me or is that plot flattening out??

[Thundery wintry showers]

Yes it is flattening out- and it shows a decline in the frequency of <0C days.

[Stratos Ferric]

Is it me or is that plot flattening out??

Your perception is clearly more acute than WiB's. Mr D's data, and Jacko's plot, are essentially the same analysis I produced for the last thirty years, extended back through the CET series but summed annually, rather than showing monthly values and the cumulative plot. This answers Jethro's query from further back.

[acbrixton]

Kevin/Julian many thanks for the data/graphs. I have broken the data down decadally showing average no. days per year with less than 0c, less than/equal to 5 days, greater than or equal to 10/20/40 days and will post the table when I have the strength!

What is striking though is how hugely anomolous are 2006-1997 and 1986-1977: the former is so far beyond recent warming trends as to raise the possibility that it may be a blip in an otherwise gently warming trend (cf. Arctic sea ice in 2007?); we will of course not know until we have data up to at least 2026.

As to the latter decade the cold seems highly unusual by 20c standards but also is cold by those of the second half of the 19c: the decadal mean of 14.3 days was only exceeded once in the 20c, twicw in the second half of the 19c, three times in the first half of the 19c. taking the decadal average of such days equal to or greater than 10 in each year (7) only 4/25 decades, all in the first half of the 19c/18c, exceeded this and no decade in the 20c equalled it.

regards

ACB