Debating the Little Ice Age

Here is a paper, Debating the Little Ice Age | Profs Morgan Kelly & Cormac Ó Gráda (@EconomicsUCD, @UCD_Research). Excerpts, underlines, italicization, et al. are on our own.

ABSTRACT:  This paper replies to commentaries by Sam White and by Ulf Büntgen and Lena Hellmann on ourThe Waning of the Little Ice Age: Climate Change in Early Modern Europe’.  White and Büntgen/Hellmann seek to prove that Europe experienced the kind of sustained falls in temperature between the fifteenth and nineteenth centuries that can justify the notion of a Little Ice Age. Neither of them adequately addresses the cogency of the anecdotal or statistical evidence presented in our article, especially with regard to the spurious peaks and troughs created by the smoothing of temperature series — the so-called Slutsky Effect.

In two related articles, “The Waning of the Little Ice Age” and “Change Points and Temporal Dependence in Annual Weather Reconstructions: Did Europe Experience a Little Ice Age?” we examined, respectively, the documentary and statistical evidence for a Little Ice Age (LIA) in Europe, finding little hard evidence to support the widely held belief that Europe experienced sustained falls in temperature between the fifteenth and nineteenth centuries. …

THE LITTLE ICE AGE ACCORDING TO WHITE      … the period between roughly 1400 and 1900, apart from a mild phase in the mid-1700s, was distinctly cooler on average than the centuries before or after”. ….  Figure 1…shows the probabilities of a good year, conditional on the previous year being good, and of a bad year, conditional on the previous year being bad. The eleventh and twentieth centuries at either end stand out from the rest. That the twentieth century has a higher probability of good winters and successive good winters than do earlier centuries is consistent with global warming.  In the centuries between the two, the probability of good or bad winters appears fairly constant, as does the probability of a bad winter being followed by a bad one. The probability of good winters following good winters is also fairly constant, except for that of the seventeenth century, which is nearly 20 percent higher than those of the surrounding centuries, despite being in the depth of the supposed LIA.  Note, however, that the credible intervals (or Bayesian confidence intervals) overlap with other periods. For summers, the probability of a good summer, or of successive good or bad summers, is fairly constant. The probability of bad summers is slightly lower in the twelfth and thirteenth centuries, and higher in the nineteenth century, but, again, the credible intervals overlap with other centuries. The point is by no means to deny the possibility of occasional clusters of bad years; in fact, the statistical article drew attention to the 1810s, the 1590s, and the 1690s, in particular.  …

Unfortunately, neither the tone nor intellectual level of White’s criticisms improvesthereafter.  After dubbing narratives linked to the LIA, such as the demise of the Norse Greenland colonies and English vineyards, “red herrings,” White devotes nearly half of his commentary to defending interpretations of them that are consistent with an LIA.  We view these narratives, mostly due to Lamb, as circumstantial evidence, not as “proofs” of the LIA. …

Norsemen and Others      In the documentary article, we argued that the paucity of hard data leaves room for several plausible but still nonfalsifiable explanations for the demise of Greenland’s Norse colonies. In addition to the LIA, we discussed six others that could account for the flimsy evidence available. The literature on the topic continues to accumulate.  Since our article went to press, a new study by Arneborg, Lynnerup, and Heinemeier also denies that climate cooling forced the hand of the colonists.  Their skeletal analysis indicates that the last colonists were neither stunted nor diseased.  “Perhaps,” summarizes Linnerup, “they were just sick and tired of living at the ends of the earth and having almost nothing but seals to eat.” Both seals and a change in climate feature in another article of 2012, this one by Dugmore et al., in which several factors that we also mentioned play a role: (1) increasing conflict with indigenous inhabitants, whom the colonists called “Skraelings” (“now the Skraelings have desolated the whole western settlement,” as one mid-fourteenth-century source reported); (2) the marginalization of Greenland when Norway began to shift its focus to the south and east; (3) the declining importance of the trade in walrus tusks; and (4) the tiny size of the settlements (a single Inuit raid in 1379 A.D. may have deprived the colony of 5 percent of its hunters).  Regular commercial contacts with Norway virtually ceased decades before the collapse of the eastern colony; the smaller western colony seems to have disappeared before any evident cooling in the supporting meteorological data.  Our basic point—which White obfuscates—remains that, apart from any climatic considerations, the settlement’s existence was precarious.  As Dugmore, Keller, and McGovern (whom we are accused of misquoting) concluded in their 2007 study, “One widely held view is that the impact of climate change, the failure of their pastoral subsistence base, and an inability to adapt were key factors in the end of Norse settlement in Greenland.  Alternatively, as we argue here, unfavorable economic changes and falling populations might actually have been the key factors in increasing the settlements’ vulnerability.”  We could not agree more.

London’s Frost Fairs      White’s view of the frost fairs is contradictory. On the one hand, he concedes that “no serious scholar” considers the two dozen frost fairs on the river Thames between c. 1400 and 1814 as “proof” of an LIA. On the other, however, he cites them as evidence of cooling, since the data “appear to predict” the seventeenth-century peak implied by Northern Hemisphere proxy trends … White also skirts around our statistical point that frost fairs were much more likely during cold winters in the seventeenth and eighteenth centuries than during cold winters in the preceding or following centuries.

Evidence of years when the river Rhône and Lake Constanz froze adds to the confusion since it suggests different chronologies …  In the case of the Rhône, the fifteenth century was one of the mildest centuries of the second millennium, whereas the fourteenth century was the coldest. The number of Seegfrörne (lake freezings) on Lake Constance (Bodensee), however, peaked in the fifteenth and sixteenth centuries. Neither series registers the seventeenth-century peak in the number of freezings championed by White.

The ambiguities inherent in these comparisons echo broader ambiguities about the dating of the LIA, which White exacerbates by proposing three distinct definitions of the “Real LIA” (1400-1850, 1310s-1810s, and c. 1580-c. 1710).  The second and the third definitions are intended to capture the “human dimension” and the “human experience” of the LIA, respectively, but they serve only to allow the historian’s tail to wag the climatologist’s dog.  The false precision of these dates, grounded in historical events, confuses secular trends and extreme years.  White conflates the two—“LIA climate fluctuations brought clusters of extreme events” (White, 349)—but the events could easily have occurred without an LIA, as shown above.  …

Hunting in the Snow      … In terms of content and iconography, Bruegel was following a well-defined “book of hours” tradition, traceable back to the Middle Ages; “Hunters” was one of a series of six calendar illustrations commissioned by Antwerp merchant Nicolaes Jonghelinck in 1565.  Two of the other illustrations, far from suggesting an LIA, have been described as a “radiant expression of high spring” and an example of “the flat glowing scenery [that] is pure midsummer.”

… the fading interest in this genre after c. 1675 owed less to climate change than to fashion, as the public wearied of gloomy representations of winter.  The weather, as depicted in landscape painting during the Golden Age, owed more to “the stylistic requirements of the market” than to meteorological reality.  Paintings were accordingly biased toward dramatic or fine weather and away from the humdrum grey clouds most commonly found in Dutch skies, then and now.  The stock-in-trade of Hendrick Averkamp (1585-1634), who has become synonymous with the LIA, was joyful winter scenes, played out almost always “in calm and stable weather conditions with stratiform clouds.”….

Population and Agriculture      Turning from symptoms to consequences of the LIA, Lamb and his followers repeatedly claimed that the impact of the LIA was particularly severe in the colder, marginal areas of Europe. As temperatures dropped, “grain cultivation [in Iceland] had to be given up”; “farms in many [of Norway’s]upland districts stood empty for hundreds of years”; and even in Denmark, “visitors to a royal wedding in 1406 reported much uncultivated, sodden land,” lamenting that “wheat was grown nowhere.”  We addressed such claims indirectly through an analysis of population trends and agricultural yields.  The ramifications of cooling should have been evident in demographic trends, particularly in the marginal parts of Europe.  On the contrary, the “perfect storm of population pressures and rapid cooling” asserted by White did not prevent the populations of Scandinavia and Switzerland from increasing their share of the European total…  Similarly, the cooling associated with an LIA should have resulted in diminished cereal yields, especially in the case of the more cold-sensitive grains. We could find no such evidence in the most comprehensive inventory of cereal yields available.

Nor is White’s gambit of highlighting extreme years — 1621 when the Bosphorus froze, 1658 when a Swedish army marched across the sound, or 1709 when French wine burst in its bottles — convincing.  We could equally invoke the winter of 1941/42, when the extreme cold had important ramifications for the outcome of Word War II; that of 1947, when ice floes were seen off the East Anglian coast, and the Dover-Ostend ferry service was suspended due to pack ice off the Belgian coast; or the “big freeze” of 1963, when the sea froze six km out to sea from Dunkirk, and a car could be driven across the frozen Thames at Oxford. But what would such conditions prove?

England’s Vineyards      In this case, too, White contradicts himself. On the one hand, he declares that our discussion of the demise of wine production in late medieval England is “irrelevant,” with “no bearing” on climate cooling. On the other, however, he states that studies based on grape culture that point to “cooler summers in early modern Europe” are a “recurring element in descriptions of freezing LIA winters and offer a good indicator of their severity”

England’s retreat from winemaking, which was a key part of Lamb’s classic case half a century ago, is now part of the conventional wisdom on the LIA.  Our case—that wine production was always a marginal activity in England, that the quality of English wine was inferior, that the trade between England and western France entailed both regions to select their comparative advantage, and that, therefore, arguments invoking the LIA are redundant—stands.

Glaciers      White’s rebuttal of our short discussion of growth and shrinkage of glaciers is the most confusing and contradictory of all.  Rather than confront our evidence of stasis before the nineteenth century… he invokes Groves’ unhelpful chronology, which times the main advances as “dating to around 1320, 1380, 1580 to 1610, 1690 to 1700, in the 1770s, around 1820 and 1850, in the 1880s, 1920s and 1960.” The implication that cooling lasted well beyond 1850 should have alerted White to the possibility (as we noted) that higher winter precipitation brought by mild and humid winters may also have played a role, making the connection between temperature and glacier length hardly straightforward.

THE LITTLE ICE AGE ACCORDING TO BÜNTGEN AND HELLMANN      Our response to Büntgen and Hellmann is less involved because they do not address anything that we wrote. In fact, they barely refer to us, even less to our arguments, except occasionally to re-assert that we are wrong. Were their article an exam, we would be tempted to respond, “Answer the question asked.”

In the statistical paper, we show that the four main documentary reconstructions of European weather over the past centuries do not reveal the trends or breaks that we would expect from a European Little Ice Age. Instead, the temperature series resemble white noise–independent draws from a distribution with a fixed mean and variance. In order to dispute our findings, Büntgen and Hellmann need to do one of two things—(1) to prove that the series that we analyze, which have been constructed by leading European climatologists, many of whom have co-authored papers with Büntgen and Hellmann, are wrong or (2) that our statistical analysis, in particular the powerful martingale difference tests that form the analytical core of the statistical paper, is deficient. Büntgen and Hellmann attempt neither of these strategies. Instead, they present a number of studies that purport to show systematic drops in European temperature during the past few centuries. But there are two problems with most of these studies, both of which Büntgen and Hellman ignore: They are largely based on tree rings, and the data are smoothed.

Climatologists have gone to considerable effort to reconstruct documentary weather series for Europe rather than using tree rings because tree rings are not a reliable proxy for weather in most parts of Europe. Tree rings reflect weather only at the limits of a tree’s geographical range where it is under constant stress due to aridity or cold. In Europe, this fact limits their usefulness to high mountains or northern Scandinavia — hence, the importance of documentary evidence. … …

The second difficulty with Büntgen and Hellmann’s series arises from the standard climatological practice of smoothing data. In the statistical article, we demonstrate that smoothing a white-noise series… leads to the appearance of spurious cycles–a so-called Slutsky effect… Although random, at least before the twentieth century, each series appears to show episodes of unusual cold. The Central European series is particularly relevant to Büntgen and Hellmann; it has a particularly cold episode in the late sixteenth century and other cold spells during the late seventeenth and early nineteenth centuries. What makes this graph relevant is that… the most damning evidence against us is the PAGES 2k Consortium reconstruction of European temperature.

However, this reconstruction is based on the standard Central European reconstruction that we analyze in the statistical paper and graph-smoothed… with additional tree-ring series from the Pyrenees, Alps, Balkans, and Scandinavia. The LIA episodes of deep cold in the PAGES 2k construction correspond to the spurious dips in our Central European series…

Once again, we emphasize that although the hazards of unthinkingly smoothing weather series is a central theme of our work on the LIA, Büntgen and Hellmann do not mention it once. Instead, they attempt to refute our findings with what is, in effect, a smoothed version of one of the main series that we analyze and show to be unchanged across the supposed European LIA.