Two “little ice ages” revealed by CET summer data

The past couple of weeks have been preoccupied by studies of UK and US precipitation and temperature data – often with the inclination to prepare things for publication here without doing anything on this line.

Yesterday and today, in between trips to visit my brother in Balaclava, I have prepared a comprehensive, tabulated analysis of the CET data as a follow-up to the work of Uncharted Territory on the famous anomaly of 1740 – the coldest twelve months in the CET record.

Mean annual Central England Temperature for each fiscal year from 1659/1660 to 2014/2015

As we can see from this graph of fiscal year Central England Temperature, there has been a general increase since the extremely cold decade of the 1690s, so much so that many recent years are off the map of what would be possible without Australian greenhouse gas emissions. There were exceptions in the 1740s and 1750s, the 1810s following a known but undocumented major eruption, and between the middle 1880s and middle 1890s following the eruption of Krakatoa. The major rise in the late 1980s no doubt coincides with the beginning of the gradual elimination of import tariffs on cars in Australia, previously so high (57.5 percent from 1978 to 1988) as to push up car prices and lower Australian greenhouse emissions.

Uncharted Territory’s work on winter CET has given me the incentive to look at summer CET over time, so we shall do this one next. This is especially significant as long-term ice fluctuations are much more influenced by summer than winter temperature – colder winters actually inhibit the growth of ice sheets as seen in completely unglaciated Siberia and Manchuria. Following on from Uncharted Territory, I have provided a 9-year and 75-year running mean on the graph:

Conclusions one might draw from the graph:

1. As with the fiscal year CET graph, there is a general rise in summer CET since the 1690s
2. There are numerous clear groups of cool summers such as:
1. from 1679 to 1700 (14 of 22 summers in lowest quartile)
2. from 1809 to 1823 (nine of fifteen summers in lowest quartile)
3. from 1839 to 1848 (seven of ten summers in lowest quartile)
4. from 1860 to 1862 (two of three in lowest five percent)
5. from 1879 to 1894 (ten of sixteen summers in lowest quartile)
6. from 1902 to 1924 (twelve of 23 summers in lowest quartile)
3. Before global warming intensified in the 1980s, there were noticeable fewer long runs of hot summers than of cool summers
• there were only two periods (plus a brief one from 1893 to 1901) with five or more “hot summers” (top quartile) in nine years
• these were as noted above all between 1771 and 1783 before the Laki eruption
• or in the late 1720s and early 1730s
4. It is notable that, before the 1980s, extremely hot summers never occurred close together apart from those of 1778 to 1781 – in contrast to the spells of cool summers in the seventeenth and nineteenth centuries
5. There are many cases of isolated hot summers during periods dominated by cool summers (1826, 1846, 1911) but except perhaps 1725 no instance of an exceptionally cool summer during periods dominated by hot summers
6. The 75-year running mean shows a notable peak in the latter half of the eighteenth century centred around 1770 (corresponding to the period from 1730 to 1810)
7. The nine-year peak of hot summers in the 1770s (from 1773 to 1781) was not rivalled until Australian road and coal industries took control of the climate
8. The 75-year running mean reaches an all-series minimum centred around the early 1890s following the eruption of Krakatoa

Thus, summer CET data suggest two “little ice ages”, one peaking in the late seventeenth century and one in the late nineteenth. I am too unfamiliar with summer temperature patterns elsewhere in the globe to know how general these trends are. Precipitation-sensitive glaciers in central Chile – a region ranking among the most severely affected by anthropogenic greenhouse warming – reached maxima late in the nineteenth century after a wet period from 1820 to 1905, which strongly suggests the width of the Hadley circulation reached its narrowest in this era. It is true that in other regions glaciers were already retreating during the latter half of the nineteenth century, though whether this reflects temperature or precipitation changes I am unsure.

We will now look at the spring CET graph:

Broadly speaking, the spring CET graph is not dissimilar to the summer CET graph above. Major differences:

1. The 75-year spring CET peak in the eighteenth century is a little later and does not reach above the virgin mean
2. Deep short-term sequences of cool springs occur in the 1740s and 1760s (culminating in the second-coolest spring in 1770) that are not followed by cool summers
3. There is no deep trough following the 1810 volcanic eruption due to several notably hot springs in 1811, 1815 and 1822
4. The minimum in spring temperature following the eruption of Krakatoa is easily the most pronounced since the 1740s.
5. The early-1840s summer temperature trough is a double trough in the spring season due to exceptionally cool springs in 1837 to 1839 and 1855.

If we look at the autumn season, we see a distinctly different pattern from the spring and summer seasons:

Here, unlike the spring and summer seasons, there is no maximum in the 1770s or 1780s. Rather the lowest trough in the entire 75-year running CET mean occurs at the same time as the maximum in spring and summer.

Apart from several hot autumns between 1818 and 1828, the sole significant maximum before the 1930s is during the first half of the eighteenth century – the 75-year maximum is centred upon 1743 and falls off rapidly due to a record cool autumn in 1786 and several other very cool autumns between 1782 and 1794.

Although it is also seen in the spring graph, there is a notable 1680s CET autumn upswing that however, relates only to a pair of succeeding hot autumns in 1680 and 1681.

A critical feature of the autumn graph is that it shows a much larger signature from anthropogenic greenhouse gas emissions than the spring, summer or winter graphs. Only three cooler-than-normal autumn seasons have occurred in the past forty years, and two of these are only just below the virgin mean. The distance by which record hot recent autumn seasons have exceeded previous records is also considerably greater than for the summer and spring seasons. This probably reflect the low British insolation and considerable seasonal lag during autumn. Absorption and release of heat by greenhouse gases probably has greatest impact during autumn not only in Britain, but just as importantly in areas – Siberia, the Arctic Ocean – from where cold airmasses must be advected into the UK to produce an unusually cool autumn like 1952, 1919, 1887 or 1786.

I will only briefly discuss the winter and extended winter (November to March) graphs as these have been done elsewhere:

The extended winter bears distinct similarities to the autumn graph above – only one November since 1994 has had a CET below the virgin mean of 5.97˚C or 42.74˚F. The major anthropogenic rise is earlier for the extended winter due to two exceptionally warm seasons in 1897/1898 and 1898/1899 (this last was however exceedingly cold in the United States).

Note the extreme cold in the US during the extended winter of 1898/1899, when Central England was having one of its mildest seasons since the 1730s with an average of 5.93˚C or 42.67˚F

In fact, only one extended winter since 1891/1892 has had a CET below 3.5˚C or 38.3˚F, as against twenty-three (two in eleven) between 1765/1766 and 1890/1891! The period from 1702 to 1739 is a near rival, but was not nearly so prolonged and had a decade of cooler seasons than anything seen over the past 120 years from 1716 to 1727. The actual winter CET series shown below shows the same pronounced trough in the 1810s as the summer one, though a scarcity of mild winters dates back to the early 1750s and those mild winters that did occur (e.g. 1778/1779, 1789/1790) were often dry and influenced by Saharan airflows – not as with conventional mild British winters by wet maritime Atlantic air.

A distinct peak in the 1860s and short-term troughs associated with prevailing negative NAO values in the 1940s and 1960s means a relatively steady 75-year mean since 1890 despite two notable severe winters that decade. The winter CET is similar to the autumn one except that its short-term peak is in the middle 1730s rather than late 1720s, and there is no peak but a continuing trough in the 1810s and 1820s.

Although explaining these peaks and troughs observed in CET data before 1974 is beyond the information I have at present, I do know that rainfall proxy data suggest that before the 1720s and from 1760 to 1820 the Sahel was wet and Southern California dry. This combination of high Sahel rainfall and low rainfall in SoCal implies a strongly positive Atlantic Multidecadal Oscillation (as in the 1950s and 1960s). The mild winters noted by Uncharted Territory for the 1730s suggest a highly negative AMO (perhaps more negative than during the late twentieth century), which is further supported by evidence of Sahel drought and reduced SoCal drought in that decade plus mild British winters during the negative AMO era between about 1905 and 1925. It also suggests a highly negative PNA, which would need verification in western North America (which did have many cold winters between 1896 and 1939) that would be very difficult since winter temperature does not directly influence growth or hydrology in most areas.