A New Take on an Old Millennium

The subject of reconstructions of temperature variations of the past millennium has been discussed many times before on this site (see e.g. here, here, here, and here). Despite the apparent controversy, the basic conclusion–that the global and hemispheric-scale warmth of the past few decades appears anomalous in a very long-term context–has stood up remarkably well in many independent studies (see Figure 1).

Figure 1. Reconstructions of Northern Hemisphere temperatures for the last 1000 years (various colored curves) compared with instrumental record (black curve).
[source: Wikipedia] (click to enlarge)

This is not to say that all estimates agree in their details. Indeed, there is a fair scatter among the various published estimates. Some of these differences are believed to reflect differences in seasonality and spatial emphasis. Past summer, extratropical temperature changes appear, for example, to have have differed significantly from annual temperature changes over the entire (tropical and extratropical) Northern Hemisphere, and tropical Pacific Sea Surface Temperatures appear to have varied oppositely with temperatures in the extratropical regions of the globe. See for example the review paper by Jones and Mann (2004), in particular the discussion and references in section 5.3 therein. Some differences appear to be related to the particular method used to “calibrate” the proxy data against modern instrumental records. While the various methods can be tested with climate model simulations , it would arguably be more satisfying if inferences could be obtained in a manner which bypasses the difficult issue of calibration entirely, and also eliminates any need to establish the precise seasonality of information reflected by the various available proxy records.

This is what Osborn and Briffa have done in their article “The Spatial Extent of 20th Century Warmth in the Context of the Past 1200 Years”, which appears in the Feb 10 issue of the journal Science. The article uses a rigorous statistical methodology to re-examine the question of whether late 20th century warmth is anomalous in the context of the past 1200 years. This is done in a manner that does not require the explicit calibration of the proxy records. In essence, the authors have revisited a question posed earlier in a paper by Willie Soon and Sallie Baliunas (2003: see our previous discussion here), investigating whether or not evidence from past proxy records of temperature support the existence of past intervals of warmth with the widespread global scale of 20th century warming. The Soon and Baliunas (2003) paper was heavily criticized in the scientific literature (e.g. Mann et al, 2003) for failing to distinguish between proxy evidence of temperature and drought or precipitation, and for not accounting for whether temperature anomalies in different regions were contemporaneous or not.

Osborn and Briffa, by contrast, have carefully taken these issues into account. They make use only of those proxy records which demonstrate a statistically significant relationship with modern instrumental temperature records, and which were dated accurately enough that records from different locations could be compared against each other in a chronologically consistent manner. They then standardize the records and look for evidence of simultaneous relative departures that point in the same direction (i.e. “warm” or “cold”) using appropriate pre-set thresholds for defining a significant event (they try both one and two standard deviations). There is an important distinction between this careful statistical approach, and the selective cherry picking that is often used by contrarian commentators to misrepresent the available evidence. For example, it is possible to find evidence of significant warmth or significant coldness over literally any century-long interval in at least one of the 14 records used by Osborn and Briffa (see Figure 1 in the article). However, this alone tells us very little. What is of interest, instead, is whether centuries-long intervals can be found over which warm events or cold events tend to cluster.

Osborn and Briffa use Monte Carlo simulations to test the null hypothesis that a given number of simultaneous “warm” or “cold” events should simultaneously emerge among 14 such independent records from chance alone. Where they are able to reject this null hypothesis, they conclude that there is evidence of large-scale warmth or coldness, and they quantify its spatial extent. They also establish that their results are robust with respect to the elimination of any single record, and that the main conclusions are independent of the particular base period (e.g. whether they use the full interval AD 800-1995, or only the modern interval of 1856-1995 which overlaps with the instrumental record). While some of the caveats of some past studies are applicable in this study too (for example, the authors only use 14 proxy sites), the authors do attempt to address them. For example, they show that the modern instrumental record averaged only over their 14 sites captures the full Northern Hemisphere mean temperature variations remarkably well over the available (approximately 150 years) interval. The authors also examine the difference between the number of significant warm and cold events over time, and this tells a similar story.

Figure 2 (from Osborn and Briffa, ’06). Fraction of the records available in each year that have normalized values > 0 (red line), > 1 (light red shading), > 2 (dark red shading), < 0 (blue line), < –1 (light blue shading), and < –2 (dark blue shading), with the latter three series multiplied by –1 before plotting. The series are shown from 800 to 1995 and have been filtered to remove variations on time scales less than 20 years. [source: AAAS] (click to enlarge)

In each case, the authors find that the most widespread warmth by far is evident during the mid and late 20th century. The conclusion is not especially surprising, as nearly all previous peer-reviewed studies over the past decade find that late 20th century warmth is anomalous in a millennial or longer context. Indeed, the curve they produce (Figure 2)–with a modest negative trend over most of the past millennium ending in a dramatic positive 20th century spike–might be likened in shape to a certain implement used in a popular North American winter sport. But we digress…

It is not so much the conclusion, but the approach that the authors use to reach their conclusion, that is most important about this latest study. The authors take advantage of a very straightforward analysis of climate proxy data, avoiding the highly technical and arcane issues of statistical calibration and methodology that are so frequently seized upon by those who dispute that recent large-scale warmth is anomalous in a long-term context. This paper adds to the mounting weight of evidence that such warmth is indeed anomalous in at least a millennial context. We doubt that this, or for that matter, any study will silence the increasingly small but persistently vocal minority of contrarians who continue to challenge this conclusion. But to them, we offer the reminder that paleoclimate evidence comprises only one of many independent lines of evidence indicating a primary role of human activity in modern climate change. If the only line of evidence that remains in dispute pertains to estimated millennial temperature histories, then the case for denialism appears extremely weak indeed.

References:

Jones, P.D. and Mann, M.E., Climate Over Past Millennia, Reviews of Geophysics, 42, RG2002, doi: 10.1029/2003RG000143, 2004.

Mann, M.E., Ammann, C.M., Bradley, R.S., Briffa, K.R., Crowley, T.J., Hughes, M.K., Jones, P.D., Oppenheimer, M., Osborn, T.J., Overpeck, J.T., Rutherford, S., Trenberth, K.E., Wigley, T.M.L., On Past Temperatures and Anomalous Late 20th Century Warmth, Eos, 84, 256-258, 2003.

Soon, W. and Baliunas, S. Proxy climatic and environmental changes over the past 1000 years, Climate Research, 23, 89-110, 2003.