Before and after: temperature correlations before and with Australian greenhouse emissions – Part II

My previous post was concerned with finding out whether correlations between CET and temperatures over other parts of the Enriched World duirng the northern winter showed a clear correlation or anticorrelation that maintain themselves in the face of radical climate change due to land clearing, coal-fired power and extensive road building (exposed in yesterday’s Age as politically rather than economically driven) by Australian individuals and big businesses, and additionally by South Africa and the Gulf States.

This post will extend the study of the relationship between CET and temperature in the eastern US to cover the western US (Nevada and Washington State) and the Plains States (South Dakota), and I will do another graph for Barrow on the Alaska North Slope, although I did not calculate a correlation coefficient because reliable temperature data for the North Slope go back only to the middle 1920s, when land clearing in Australia was well under way and aluminum and titanium metallurgy which have produced a significant part of the greenhouse pollution emitted since then.

We saw in the previous post that winter CET was:

1. positively correlated (even excluding Australian greenhouse gas emissions) with winter temperature in Florida and the Ohio Valley division
2. marginally positively correlated (excluding Australian greenhouse gas emissions) with winter temperatures in Texas (south-central US)
3. not correlated with winter temperatures in Maine

Several winters noted previously (notably 1939/1940, 1962/1963 and with opposing anomalies 1948/1949) suggest negative correlations should exist between winter CET and winter temperatures in western North America. This is less well-known than the anticorrelation between CET and winter temperatures in northeastern (Nunavut, Nord-du-Québéc, Labrador, western Greenland) North America known ever since the North Atlantic Oscillation was discovered in the nineteenth century.

If we look at the warmest CET winters between 1881 and 1974 – excluding 1898/1899 and 1948/1949 which I have discussed in earlier posts – the warmest (1934/1935) is quite exceptional in following most of the opposite patterns to those very cold CET winters woudl imply to occur in exceptionally mild CET winters:

We can see that, although the warmest among the 105 winters between 1869/1870 and 1973/1974 in the CET series, the winter of 1934/1935 was, like the coldest CET winter of this era in 1962/1963:

• warm over the Lake Baikal region
• warm over Alaska and Western Canada
• cold over the northeastern United States
• warm over most of Greenland (although not on a 1921 to 1974 base period):

The second-warmest winter in this period, 1942/1943, also is problematic for these predictions as it was warmer than normal also over the southwestern United States and Greenland, and colder than normal over the southeast. In fact, all three months of the winter show some deviations from predicted patterns – February 1943 was warmer than normal over Alaska, where December 1942 was the coldest on record and January 1943 very cold indeed:

Global temperature anomalies for December 1942. This was the coldest December over Alaska since 1917, but began a very mild winter over the UK.

We can see, in effect, a pattern that would suggest a westward shift in the longitude of the Icelandic Low, so that the Northwest Territories rather than the Inuit lands of Greenland and Nunavut would be on the cyclone’s cold western side. A North American High strengthened and displaced southeastward might also produce such a pattern.

This pattern is a little akin to March 1963 after the coldest winter over Central England since 1740 – mild and very wet over the United Kingdom, but very cold and “Siberian” over Eastern Europe:

(Notice how March 1963 itself was colder relative to the global virgin mean than the northern winter of 1962/1963, for warm air advection into Nunavut and Alaska ceased, whereas cold air advection into northeastern Europe did not).

Only here do we see a very typical positive NAO pattern with cold weather confined to Greenland and lower latitudes under enhanced and poleward displaced subtropical anticyclones over North America and Europe. The anomalies in western North America, however, remain the opposite of what our previous predictions would suggest, along with observations in such winters as 1902/1903 and 1948/1949.

Here, in this third-warmest winter, we do see more or less the expected pattern, although it might be noted that in the conterminous United States the winter featured the third-warmest February between 1883 and 1975, and the fourth-coldest December since 1885.

This is somewhat, but not wholly, less satisfactory than the winter of 1924/1925, but positive NAO was a very strong trend in the early 1920s. It might be noted than in Maine and adjacent areas March 1923 was the coldest since 1880 and/or 1885 – indeed in Portland, Maine the coldest since before 1875.

This is basically satisfactory except that the anomalies are weak and were from month to month inconsistent – January 1957 was exceptionally cold in the Pacific Northwest but mild in Alaska where Fairbanks had its warmest January since the snow-drenched January 1937, whereas December 1956 had been mild everywhere south of the border but as expected very cold in the northern regions:

Global temperature anomalies for December 1956. Note the extreme cold over the normally cold regions of the Baikal and Amur basins, Alaska, western Greenland and most of Nunavut and Nord-du-Québéc, plus what was before last year (and would remain minus Australian and Gulf States greenhouse emissions) the warmest December on record over the eastern United States.

South Dakota (Great Plains)

Here we do not see any major correlation, either by Spearman rank (ρ) or Pearson product-moment (r). The white diamonds are almost randomly distributed, and a notable contrast can be seen with the cold winter of 1962/1963, which was a little above the virgin mean.

One does see a cluster of dark red diamonds in the upper right, which suggests a definite alteration from the virgin patter, and indeed the dark red diamonds do show an absence in the upper left. Oddly, however, for those who want to believe the relationship has been altered by Australian greenhouse gas emissions, there does not exist a white diamond anywhere close to the upper left of the graph, as if a natural limit to this possibility exists. The same is true of the light red diamonds, and seems to beg some sort of explanation.

The case of South Dakota, unlike those of the eastern and southern United States, does suggest a possible alteration in relationships due to Australian-made greenhouse gas emissions. However, a peculiarity in the blanks in the upper left (warmer than normal over South Dakota, colder over the United Kingdom à la 1941/1942 or 1986/1987) suggests that a case for relationships altering remains unproven.

Nevada (Southwestern United States)

With this graph for Nevada, representing the southwestern United States and especially the Great Basin (where temperature inversions in anticyclonic winters dominated by a Great Basin High can create extremely distinctive local conditions) there are two notable features of the virgin Spearman ρ and Pearson r correlation coefficients:

1. they are highly negative so that when Nevada is cold during the winter, the UK tends to be mild
2. much more distinctly different than in other plots
3. because the Spearman coefficient is less sensitive than the Pearson to outliers, this suggests that the natural correlation coefficient in winter temperatures is more strongly negative than the raw values

This prediction of a negative relationship would be expected from both the coldest and the warmest UK winters since 1880. Nonetheless, if we look at the light red diamonds, one detects a clear negative pattern still applying to this group of datapoints. The only exceptions noticeable are 1978/1979 and 1984/1985, which are when the influence of Australian-made greenhouse gas emissions was still not nearly as complete as it is now. These winters are analogous to the severe northern winter of 1916/1917, when easterly or northerly flow covered all of Europe for five months and North America for six:

This map coincides with the record cold European winter and spring of 1916/1917. Note that, contrary to the correlations we have seen in earlier scatter plots, the record cool was in the West and North rather than the Southeast.

If we look at the dark red diamonds, one noticeably sees nothing in the extreme top right as would be predicted if the natural relationship had changed or been eliminated completely. Even the dark red diamonds, upon repeated examination, give a negative line of best fit that may be limited by small sample size and the consistently increasing influence of Australian coal power, land clearing, and highway construction. More in fact that the virginally unrelated graph for South Dakota, Nevada does not suggest a fundamental change in the relationship of other regions’ mean winter temperature with the CET.

Washington State (Northwestern Conterminous United States)

This graph is similar to that above for Nevada in terms of correlation coefficient, if not in details of every year due to the influence of Great Basin Highs in anticyclonic winters (e.g. 1930/1931, 1954/1955, 1963/1964, January and February 1984) when it can be cold in the southwest but very mild in Canada.

Contiguous US temperature anomalies for February 1984 in ˚C. Observe the cold over the Great Basin in contrast to abnormal warmth over the northern and eastern regions

If we look at this graph carefully, there is no indication that it is any different from the Nevada graph except that the negative correlation is slightly less and that the winter of 2008/2009 appears as an outlier amongst the dark red diamonds:

Global temperature anomalies for northern winter 2008/2009. Amongst the dark red diamonds, this season appears as an outlier in the CET/Washington State relationship, being the coldest winter in both regions for thirteen years – yet was no colder than the pre-1974 mean

Barrow (Alaska North Slope)

As a final post in this second installment (I am unsure whether I will do a third to look at extratropical Asia where temperature records are not as long but are less well-known), I will look at the station of Barrow, Alaska, on the Arctic Ocean 510 kilometres north of the Arctic Circle. Data for Barrow – easily the oldest station on the North Slope – go only to 1923/1924, but given that in the previous post Alaska stood out for its conspicuous warm anomalies during cold UK winters, I was interested to do what graph I could.

Even with the strong impact of global warming, it is very easy indeed to detect a negative correlation within all three classes of coloured dot.

The top right and bottom left are almost empty apart from a few recent winters as the impact of man-made global warming intensifies, whereas dots are dense in the top left and bottom right. Likely the correlation coefficients between Barrow and CET winter temperature are more negative than those for Nevada, but I have not bothered to calculate them. The isolated white diamond is the winter of 1934/1935, whose full graph is given at the beginning of this post. Between 1922 and 1974 it was the warmest CET winter and the second-warmest in Barrow behind only 1941/1942. The evolution and cause of this anomaly will be examined from Barrow’s temperature graph shown below, although it might be noted that Barrow’s mean temperature troughs out at the tail end of February – indeed its warmest March, that of 2002, is as much as 11.4˚F or 6.3˚C colder than its record-warm February 1989 (during which Barrow was amazingly 0.02 inch wetter than superhumid Yakutat which had easily its driest month on record).

The winter of 1934/1935 is extremely famous in Alaska for Fairbanks’ only ever “brown Christmas” due to a powerful chinook wind melting all the snow cover during the second week of the month when temperatures rose to a phenomenal 58˚F or 14˚C for several days, followed when the weather turned seasonably cold by conditions too dry for snow. As we can see above, temperatures in Barrow did not reach such record-breaking levels (in fact, no day in December 1934 is record-warm in Barrow) but did average around 20˚F or 11.1˚C above normals for those dates, resulting in, as shown below, substantially above-normal temperatures:

In the UK, December 1934 was notorious for the frequency of rain, the mildness and lack of sunshine. Princetown in Dartmoor recorded 307 hours of rain during the month – in other words it rained for 41.26 percent of the time there, much more even than the long-term mean of such notorious wet places as Ketchikan (26 percent). January 1935 was cooler – though still above average – and dry in the UK, and as the graph above showed only marginally warm in Alaska, besides being exceptionally cold and snowy in Canada and cold throughout southwestern Eurasia due to easterly flow from the European block:

February 1935 is a month I noticed a great deal looking at temperature tables for the northern hemisphere because in huge areas of Siberia, Central Asia, and Canada it was an exceptionally mild month due to enhanced westerly flow and in Canada, chinook winds. The global anomaly of +0.27˚C above the 1880 to 1974 mean is not striking (January 1926 had a figure of +0.38˚C) but this figure is certainly higher for the northern hemisphere alone.

This fast westerly flow, of course, explains the temperature anomaly very well, the only surprising thing being how well the warm air entered Alaska. However, the pattern is consistent with a split jet stream, seen classically during the winters of 1943/1944 and 1991/1992, with heavy rainfall over southern Alaska and the southwestern United States and dry conditions over the Pacific Northwest:

The winter of 1946/1947 may appear to be an outlier, but in fact the cold occurred at different times – in Barrow during December and January, in England during late January and February. Barrow did not get below -47˚F or -43.9˚C on the day Snag, Yukon recorded the coldest temperature in North America away from the Greenland Ice Sheet, but whilst the UK was in a deep freeze, temperatures from 8 to 28 February in Barrow were near normal.

The winter of 2006/2007 shows the complete dominance of Australian-made pollution on the global climate – the only cooler than normal areas being in the Sahara where an expanded Hadley cell allowed more radiational cooling, typical of highly positive NAO episodes. The weakening of polar highs meant that hotter-than-normal air covered all the rest of the globe. This was the winter when loss of sea ice reached its most critical level – without a sign of sanctions against the worst (per capita) culprit nations who are too important to the global economy for the rest of the world to dare make them pay the vast costs of increasingly runaway climate change.

Conclusion:

Despite severe and increasingly rapid alterations to the climate from fossil fuel use and land clearing in a small number of oil- or mineral rich and largely arid nations around the Indian Ocean (Australia, the Gulf States and South Africa being the important culprits), most natural temperature correlations and anticorrelations do not seem to have disappeared. In most cases, if one subdivides into pre-1974, 1974 to 1997 and post-1997 using coloured diamonds, it is possible to see something approaching the same relationship in each colour of diamond.