Analysis: Michael Mann Used Well Known Deceitful Statistics to Create the Hockey Stick
Michael Mann Used Well Known Deceitful Statistics to Create the Hockey Stick
h/t Commenter Sunsettommy
The deeper I dig into the Hockey Stick, the more it stinks. As you can see from the above Hockey Stick graphic, Michael Mann combined proxy and instrumental data post-1902.
The fact that the chart dog-legs at precisely the time the instrumental data is added should have raised red flags to anyone that has graduated from Kindergarten, but Michael Mann was able to get this piece of garbage passed “Peer Review” and it is used as the basis for the scientific “Consensus.” Well, digging into this practice, I discovered that people in the field know that this is a big no-no.
Possible Pitfalls of Tree Rings as a Proxy
One way in which the presentation of data can be misleading is to combine data derived from a proxy source with observed data. In Figure 6 we show data for Dimmie in Scotland. This site is near to Edinburgh which has a two-century-long temperature record. The graph has calculated data up 1975 and observed data (plotted as 5-year moving average) from then onward. This appears to show little variation in temperature until the end of the 20th century: the classic “hockey stick” shape. Figure 7 shows the same data but with the observed data for the whole period superimposed on the calculated data. As can be seen, the calculated data has much less variance than the observed data, though still with an increase toward the end.
Note, the instrumental data shows the Little Ice Age, and current temperatures are barely above the level reached in 1845 and 1855. Dimmie is near highly developed Edinburgh so the instrumental warming is likely due to the Urban Heat Island effect, not CO2.
Digging deeper, I discovered that there are all sorts of problems with using the Bristlecomb Pine as a temperature proxy.
Other things which can affect ring width are:
- The age of the tree. The rate of growth varies through the life of the tree.
- Weather. In addition to temperature, ring growth is also affected by precipitation and to a lesser extent by wind speed and sunshine.
- Previous years. If a tree has grown vigorously in one year it is likely to grow vigorously in following years and vice versa.
- Atmosphere. Carbon dioxide is necessary to growth and increased levels of carbon dioxide can lead to enhanced growth.
- Competition. Other trees nearby or other plants can rob a particular tree of nutrients or light.
- Parasites. Infestation by insects or fungi can slow the growth of the tree.
Pay special attention to bullet point #4 “Atmosphere. Carbon dioxide is necessary to growth and increased levels of carbon dioxide can lead to enhanced growth.” CO2 was basically constant pre-1902 for the life of a tree but reached new heights post-1902. CO2 almost certainly impacts the growth of tree rings, unless the laws of Botany cease to exist in the Bristlecomb forests. CO2, after all, is plant food.
That, in fact, is exactly what happened.
I’ve recently run across an article on changing water use efficiency in bristlecones, which pretty much put the nail in the coffin on any lingering ideas that 20th century bristlecone ring widths might be a temperature proxy. Tang et al. , “The dC13 of tree rings in full-bark and strip-bark bristlecone pines in the White Mountains of California”, shows a remarkable non-climatic increase of dC13 ratios of bristlecones at Sheep Mountain, the most important site in the MBH98 PC1. The dC13 values are stated to show an equally remarkable nonclimatic increase in water use efficiency at Sheep Mountain. Improved water use efficiency was the predicted mechanism for CO2 fertilization. See their Figure 3 below. Any attempt to argue that bristlecones are a temperature proxy on scientific grounds (something that has been conspicuously absent from any response by realclimate or their associates) would need to adjust for non-climatic changes in dC13 ratios and water use efficiency.
Bristlecone pines (and Gaspé cedar) have obviously featured prominently in our critique of MBH98. In our EE article, we included an extensive literature review of issues affecting the validity of these ring widths as temperature proxies, but weere then unaware of Tang et al  (which I’ve posted up here). To my knowledge, neither Mann nor any realclimate associate has attempted to defend bristlecones as temperature “proxies” other than by arguing [Wahl and Ammann] that their use improves the reconstruction RE score ( a statistical argument leading directly into questions of spurious significance which they avoid.)
Where are the defences of bristlecones as valid proxies in tree ring terms? While we’ve pointed out a number of possible mechanisms for fertilization in the 20th century, the most prominent theory is CO2 fertilization [Graybill and Idso, 1993; Lamarche at al, 1984.] The usual mechanism by which CO2 fertilization is hypothesized to lead to increased ring widths is through increased water use efficiency. Water use efficiency changes are studied in a series of remarkable articles by Xiahung Feng of Dartmouth, of which Tang, Feng and Funkhouser  is notable. “Plant water use efficiency” was defined by Farquhar et al 1989 (cited by Tang et al) as the ratio of net carbon fixed to the total water cost. The connection between CO2 fertilization and increased water use efficiency
is summarized by Tang et al. as follows:
Experimental work has strongly demonstrated the positive response of photosynthesis and plant water use efficiency to increasing CO2 concentration [ e.g. Strain and Cure 1985; Bazzaz 1990; Mooney et al 1991; Idso 1992; Korner and Arnone 1992; Norby et al 1992; Polley et a 1993; Wullscheger et al 1995)] and the negative response of stomatal conductance of plant leaves [Woodward 1987; Beerling and Woodward 1993; van de water et al 1994]. For example, by studying a number of C3 and C4 species, Polley et al 1993 showed that both plant water use efficiency and biomass increased with increasing ambient CO2 concentration. This led to the idea that CO2 fertilization may be evaluated by measuring plant water use efficiency.
It only gets worse.
There is extreme veribility in analyuzing the tree rings.
To overcome the above, for temperature reconstruction the sites to be analysed are chosen so that these other factors have limited importance. For example trees might be chosen in areas where rainfall was plentiful so that water stress does not affect growth.
Even in well chosen sites it has to be recognised that ring width is not a uniform function of temperature but is biased toward the temperatures during the growing period. This is sometimes dealt with by analysing early and late growth separately.
Tree rings have a natural tendency to drift towards implying colder temperatures as the tree ages.
For most trees the rate of growth declines with time as the tree becomes older. If no allowance was made for this it would appear the temperatures over the life of the tree were in decline.
Note how the properly adjusted tree rings imply flat temperatures between 1600 and 2000, but the raw biased unadjusted data shows cooling between 1625 and 1850. The Hockey Stick seems to replicate the raw unadjusted data, whereas the instrumental records confirm flat temperatures over that time period. Annual Central England Temperatures in 1650 were 9.25°C, and temperatures in 2016 they were 9.50°C. Temperatures were 10.0°C in 1740.
There is one huge problem, however, tree rings LEAD temperatures. Tree-Ring Proxies imply warmer temperatures BEFORE they actually happen. That clearly implies that tree rings are impacted by something other than temperatures and that the temperature tree ring relationship may be largely coincidental.
Figure 3 shows the effect of global temperatures on tree rings widths. As can be seen below there is reasonable agreementbetween the rate of growth and the temperature record. There is one interesting feature, which may be an artefact of the method of processing; the rate of growth of tree rings appears to precede the increase in temperature.
Figure 4 and Figure 5 show data for the Northern and Southern hemispheres, which show a general agreement on the pattern of rising and falling temperatures. Again the agreement is generally good except for the fact the observed temperature record shows an increase around 1940 not noticeable in the tree ring record.
Reasonable, General and good aren’t terms one would expect to find in a “settled” science with 95% confidence and a “Consensus.” Those are terms used when guessing and betting, and you are hoping for a 51% chance of winning. Also, having observations that are unexplained pretty much proves things aren’t settled.
In conclusion, it should now be obvious to everyone why Michael Mann chose to ignore instrumental measurements and instead rely upon the highly variable Tree Ring Proxies. It allowed him the freedom to pick and choose which ones he wanted to use to make his case of dramatic warming in the 20th Century.
As far as the inconvenient issue about the tree rings not matching the instrumental records. Michael Mann simply implemented “Mike’s Nature Trick…to hide the decline.”
How Congress has not investigated this is way beyond me.
Please Like, Share, Subscribe and Comment