Where did IPCC-FAR, Figure 7.1c Come From?

IPCC 1990, Figure 7.1c is a Replica of the Lamb 1965, Figure 3a, "Analyst's Opinion"

In the IPCC's First Assessment Report, published in 1990, three schematics were published that were labeled, "schematic diagrams of global temperature variations." The third (c) schematic was labeled, "the last thousand years" with a dashed line that "nominally represents conditions near the beginning of the twentieth century." The graph is below.

IPCC-FAR 1990, Figure 7.1c
"Schematic diagram of temperature variations over the last thousand years. The dotted line nominally represents conditions near the beginning of the twentieth century."

I think it may be fair to say that the biggest mistake the IPCC has ever made was using the word "global" with reference to the (c) schematic, because that objectively is not the case. It was a replica of Hubert Lamb's reconstruction of Central England temperature (CET), where Central England is a 21 x 34 mi patch of land in the middle of England. Hubert Lamb originally published this in a paper from 1965.[1] The smoothed part of the (A) reconstruction was labeled "analyst's opinion" or annual temperatures (as opposed to summer or winter temperatures) smoothed to 50 year means. The last data point in Lamb's paper was for 1900-1950, so this cannot be understood to have any data past 1950.

Lamb 1965, Figure 3
"Temperatures (C) prevailing in central England, 50-year averages. A: year. B:high summer (July and August), and C: winter (December, January and February)."

The data for the (A) graph above is reproduced below, and these data form the basis for the adjusted and (what I'm calling) "raw" temperatures at the top of this post. The "analyst's opinion" (Lamb's opinion) normally falls between these two values, except for before about 1200 CE.

Almost 20 years, later, in Hubert Lamb's book, Climate History and the Modern World, Figure 30, p. 84,[2] Lamb published essentially a smoothed version of the same graph, where the top graph (a) was essentially the same annual "analyst's opinion" with uncertainties going back to about 1250 CE. The graph below again ends at about 1950. And this is the basis from FAR Figure 7.1c.

Lamb 1982, Fig. 30
"Estimated course of the temperatures prevailing in central England since AD 800. Probable fifty-year averages: (a) for the whole year; (b) for the high summer months, July and August; (c) for the winter months, December, January and February."

The IPCC used a hand-drawn replica of Lamb's graph for CET from his 1982 book with smoothing, removed the temperature scale, and called it a "schematic" for global temperatures, I guess assuming that the general shape of CET should be close to being representative of global changes. The IPCC's schematic probably ends in 1950 with its source in Lamb's book, but it's hard to be certain, and it doesn't really matter, since for the IPCC it's just a schematic.  In fact, in an appendix to an article in the The Holocene in 2009,[3] Jones said that two years later in an IPCC supplement, the schematic was dropped and "the need for more data that would allow for the spatial aspects of past changes acknowledged." There probably should have been an explicit statement that this schematic was not to be considered representative of global temperature variability. A comparison of Lamb's CET and IPCC FAR, Figure 7.1c is below.

The Red curve is from Lamb 1982. The Black is IPCC FAR 7.1c. Blue is a Smoothing of HadCET

To be clear, the curve from Lamb 1965/82 was good work for its time, but it lacked true proxy data (it's a reconstruction from historical records) and it contains no calibration to instrumental data prior to the beginning of the HadCET dataset (which itself is only reliable prior to 1772). And Lamb made no assertion that his reconstruction was representative of global temperature variability. This should be seen essentially as a qualitative reconstruction, even though Lamb's graphs had a y-axis with temperature scale. Jones[3] in fact writes:

Many in the palaeoclimatic community have known that the IPCC (1990) graph was not representative of global conditions (even when it first appeared) and hence the reference to it as a schematic. Lamb’s (1965, 1977, 1982) series has been used as one of the series comprising the NH composite developed by Crowley and Lowery (2000), representative of Central England. Various authors (eg, Farmer and Wigley, 1984; Wigley et al., 1986; Ogilvie and Farmer, 1997) have shown that such representativeness is only really the case for the instrumental part of the record from 1659 which is based on the well-known Manley (1974) series. Greater amounts of documentary data (than available to Lamb in the early 1970s) were collected and used in the Climatic Research Unit in the 1980s. These studies suggest that the sources used and the techniques employed by Lamb were not very robust (see, eg, Ogilvie and Farmer, 1997).

Jones concluded that Figure 7.1c was "locally representative (nominally of Central England) and not global, and was referred to at the time with the word ‘schematic’." A lot of confusion would have been avoided if in 1990 the IPCC had simply claimed that Figure 7.1c was a CET reconstruction that may have relevance to global temperature changes, but further research is needed to confirm this.

Case in point, as far as confusion is concerned, here is a graph from Tim Ball claiming to be European temperatures for the last 1000 years.

The caption for this graph says, "Temperatures for Europe plotted against the 20th century average. (Source: Based on diagram in 2nd assessment report IPCC 1995/1996.)" Except this is a hand-drawn replica of Lamb 1965/82 as smoothed by FAR 1990. It is not from SAR 1995.  Figure 3.20 in SAR 1995 was taken from Bradley and Jones 1993 and went up to the 1970s. It was a proxy reconstruction from 16 proxy summer temperature records for the Northern Hemisphere, and it only went back to about 1400 CE. The graph below from SAR includes both proxy and instrumental data on the same graph so you can see how well they agree/disagree.

IPCC SAR 1995, Figure 3.20

Ball's graph looks nothing like this graph from SAR; instead, Ball's graph is a replica of FAR 1990, Figure 7.1c with the temperature scale from Lamb 1965/82. Ball added the temperatures from 1950 to 2000 to the end of Lamb's data, which is unsourced and essentially just made up by Ball. If he had correctly continued to use HadCET, then by 2000 CE temperatures would be warmer than the peak of the MWP (more discussion on this point is here). Just to clarify this point, I took the curve from FAR scaled to be the same as Lamb 1882 along with HadCET. I set both to the same baseline and plotted HadCET as 50-year running means.

Clearly and unambiguously, if we assume the FAR/Lamb reconstruction was correct for CET, this shows that by 2000, CET was warmer than the peak warmth of the MWP. Ball couldn't show this, which is why he essentially fabricated the last 50 years of his graph, then he labeled it "Europe" instead of Central England to make it seem more comparable to legitimate multiproxy reconstructions (like those from Michael Mann and colleagues).

References:

[1] Lamb, H.H. (1965). The earlier medieval warm epoch and its sequel. Palaeogeography, Palaeoclimatology, Palaeoecology 1, 13–37

[2] Lamb, H.H. (1983). Climate: history and the modern world. Methuen, 433 pp.

[3] Jones, P.D. et al. (2009). High-resolution palaeoclimatology of the last millennium: a review of current status and future prospect. The Holocene 19.1, 3-49.