Here at Climate Audit, we occasionally try to solve mysteries that have vexed climate scientists for years. On a previous occasion, we helped UCAR locate the mysterious civilization of Chile, on another occasion the lost city of Wellington NZ and, most recently, helped NASA find the lost city of Cobija, Bolivia. Today we’ll help the climate science community identify the provenance of a graphic shown below, that was produced in 1990 by a mysterious organization known to insiders as IPCC.
At his blog last year, William Connolley, obviously impressed with then recent CA success in locating Chile and Wellington NZ, appealed for help in solving this outstanding puzzle, specifically challenging me to identify its provenance for the climate science community, even accusing me of being “curiously uninterested in the source [of this graphic], or lack thereof” and worried that I was “ignoring” the provenance of this graphic, leaving the wider community to fend for itself.
One of his readers, obviously of the view that it was not necessarily my responsibility to sort out this particular IPCC conundrum, asked Connolley why he didn’t just determine who the IPCC author was and ask him:
Where does the 1990 graph come from then? I presume the IPCC author didn’t just hand draw it? Who is the IPCC lead author responsible, and if he is still alive, couldn’t one just ask?
This seems like a logical method, but Connolley was uninterested:
AFAIK it was hand drawn – it looks like it was. Its certainly not a computer plot. Things were free-and-easy back in ’90 I suppose (thought as far as I know this is the only graph from the ’90 report that does not have a good source) -W
At wikipedia, Connolley expounded further on the lack of a clear source for this graphic:
A schematic (non-quantitative) curve was used to represent temperature variations over the last 1000 years in chapter 7. The vertical temperature scale was labelled as “Temperature change (°C)” but no numerical labels were given; it could be taken to imply that temperature variations of the MWP and LIA were each of the order of 0.5 °C from the temperature around 1900. The section specifically states recent climate changes were in a range of probably less than 2 °C. The 1990 report noted that it was not clear whether all the fluctuations indicated were truly global (p 202). The graph had no clear source (it resembles figure A9(d) from the 1975 NAS report, which is sourced to Lamb, 1966), and disappeared from the 1992 supplementary report.
To say that I was “uninterested” in its source was untrue. This is the sort of thing that usually interests me, but I can’t solve every climate science puzzle instantaneously and on demand. On an earlier occasion, I posted up an extended excerpt from IPCC 1990, which is not available to many readers. However, I didn’t then know the source of the IPCC 1990 graphic, just as I don’t know how Mann calculated the MBH99 confidence intervals or many other small climate science mysteries.
Today I’m pleased to report that I think that I can report a solution to this small mystery, which I propose below.
I recently re-read Lamb 1965. Its Table II provided a table of Central England temperature estimates in 50-year intervals. As an exercise, I manually transcribed the numbers from the table and plotted them as shown in the figure below. It obviously has some similarities to IPCC 1990 Figure 7c, though it is obviously not the same. I’ll discuss this below.
In the above figure, the red dots are 50-year annual averages from current Hadley Center data in 50 year intervals from 1700-1950, each shown in the midpoints (1725, 1775,…,1925). As one can see, for the period of overlap, Lamb 1965 values pretty much exactly match the Hadley Center 50-year CET averages. This is not totally surprising as Lamb said that he used Manley’s data, but it’s useful to confirm that the historical versions have remained essentially unchanged for these periods since 1965 (something that would obviously not be the case for Hansen’s data.)
Next here is a plot of the Lamb CET together with Hans Erren’s digitization of IPCC 1990 Figure 7c (this digitization interprets the hash marks on the IPCC 1990 diagram as degree C intervals, as is clearly indicated in the caption and commentary, though even this is contested by Connolley). On the left, I’ve denominated the series in anomaly deg C (as IPCC), while on the right I’ve denominated the series in deg C (as Lamb 1965). The zero appears to be the mean of the entire series; the dashed reference line appears to be the value circa 1900.
It is clear that the two series are closely related in the period after 1400 (up to smoothing), but, for some reason, are dissimilar in the MWP portion. From Hans Erren’s digitization, the IPCC figure definitely continues one more 50-year period with a value in 1975, even though the corresponding Lamb 1965 figure ends with the 50-year interval 1900-1950. I experimented a little with choice of end-point (considering the fact that the graphic was prepared for printing in 1990). In the figure below, the end-point
(open black dot) shows the the average CET value for the 35-year period 1950-84 – this matches very closely and I think that it’s safe to conclude that the end-point of IPCC 1990 Figure 7c was determined through some method very close to this. But the MWP period is still a mismatch.
Now look at the top panel of Lamb 1965 Figure 3 top panel, shown below. The more jagged graphic matches the values in Table II, while the rounded curve matches IPCC 1990 Figure 7c almost exactly from 900 to 1400 (with the IPCC curve receiving some additional rounding in the 1400-1700 period.)
Lamb 1965 Original Caption: Fig.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). Observed values (as standardized by MANLEV, 1958, 1961) from 1680. Values for earlier periods derived as described in the text. The ranges indicated by vertical bars are three times the standard error of the estimates.
So I think that we have a solution to the provenance of IPCC 1990 Figure 7c. It is derived from the rounded CET from Lamb 1965 Figure 3 top panel, with portion after 1400 smoothed somewhat. It has been converted to anomaly deg C (using the average of the entire period) and extended to include the average CET for the period 1950-1984.
In Figure 7, IPCC 1990 used the CET estimates as a “schematic” for the global temperature, even though, as shown above, it was an estimate for the CET, which is taken from only one location. While one can reasonably criticize the use of CET as a “schematic” for global temperature, I’m not at all convinced that IPCC TAR reliance on Graybill’s bristlecone ring width chronologies (via Mann’s Hockey Stick) as a uniquely sensitive antenna for world temperature (or the AR4 spaghetti graph) represents much, if any, improvement, as I’ve discussed on many occasions elsewhere.
There are interesting issues pertaining to what Lamb and IPCC 1990 thought about the global-ness of the MWP. Lamb 1965 discusses climate change as associated with “global” wind circulation and clearly regarded the MWP as being a small-amplitude warm period of global extent (in my opinion, IPCC AR4 has misrepresented Lamb’s views, a point that I made as a reviewer, a point that was, needless to say, ignored). Lamb regarded the LIA as being a moderate-amplitude cold event, also related to global wind patterns. There’s little doubt in my mind that Lamb regarded these events as of global extent. That doesn’t make them so. But it’s interesting that Lamb at least tries to interpret the evidence in terms of global circulation.
IPCC 1990 (see longer excerpt here) stated:
The late tenth to early thirteenth centuries (about AD950-1250) appear to have been exceptionally warm in western Europe, Iceland and Greenland (Alexandre 1987; Lamb, 1988). This period is known as the Medieval Climatic Optimum. China was, however, cold at this time (mainly in winter) but south Japan was warm (Yoshino 1978). This period of widespread warmth is notable in that there is no evidence that it was accompanied by an increase of greenhouse gases.
There are interesting exegetical issues pertaining to this, as Zhang (1993) – see abstract here – argued soon thereafter that Zhu (Chu) 1973, a prominent source in 1990, had an incorrect understanding of the medieval Chinese lunar calendar, which had caused incorrect dating of phenological events (crops, blossoms) in the MWP and, in fact, that important crops had reached further north in the MWP in China than at present. Crowley and Lowery 2000 absurdly reverted to the use of the obsolete Zhu 1973 version – matters that I may return to on another occasion.
For now, I think that we’ve solved another climate science puzzle – the provenance of IPCC 1990 Figure 7c. Maybe one day we’ll be able to determine the provenance of the MBH99 error bars, so prominently illustrated in IPCC TAR.
Now that we’ve established that IPCC 1990 Figure 7c is derived from the Lamb 1965 CET, what would happen if Lamb’s figure were brought up to date. Here’s my calculation below. In this case, I’ve shown the 1950-1999 CET average, which is about 0.12 deg C higher than the value shown in IPCC 1990 due to the warm 1990s. I’ve also shown what the 2000-2049 average would look like if the 2000-2007 average were continued for the rest of the half-century. This would be about 0.25 deg C higher than the corresponding MWP maximum in Lamb Table II (though the last half 20th century was still below the Lamb maximum). Note that IPCC AR4 stated that it is “likely that this 50-year period was the warmest Northern Hemisphere period in the last 1.3 kyr”.
H. H. LAMB, 1965, THE EARLY MEDIEVAL WARM EPOCH AND ITS SEQUEL Palaeogeography, Palaeoclimatology, Palaeoecology
Updae: Here is triptych image from Crispin Tickell (British Antarctic Survey) 1977 mentioned in a post below (thanks for the ref) as compared to the corresponding full triptych in IPCC 1990.
Original Caption: Figure 2: Temperature variations over the last million years. Courtesy of the British Antarctic Survey.