|On Past Temperatures and Anomalous late-20th Century Warmth
||[Nov. 29th, 2009|02:44 pm]
Impacted Wisdom RSS feed
The paper by Mann and company, "On Past Temperatures and Anomalous late-20th Century Warmth," contains some interesting graphs.|
The paper purports to show that the warming in the 20th century is "unprecented" - i.e. that it's never been warmer. The Medieval Warm Period wasn't warm, by comparison. To do this, Mann et al. use information from tree-rings and ice cores as "proxies" for temperatures going back over a thousand years or more. The idea is reasonable, but problematic in practice. Tree rings, in particular, reflect the effects of moisture, temperature (only in the growing season), nutrients in the soil, accident, and (especially in the last century and a half) the effects of carbon-dioxide fertilization. Tree ring expert Keith Briffa had this to say, privately:
There is still a potential problem with non-linear responses in the very recent period of some biological proxies ( or perhaps a fertilisation through high CO2 or nitrate input) . I know there is pressure to present a nice tidy story as regards 'apparent unprecedented warming in a thousand years or more in the proxy data' but in reality the situation is not quite so simple. We don't have a lot of proxies that come right up to date and those that do (at least a significant number of tree proxies ) some unexpected changes in response that do not match the recent warming. I do not think it wise that this issue be ignored in the chapter. For the record, I do believe that the proxy data do show unusually warm conditions in recent decades. Publicly, of course, he's fully on-board with the cause. Here's a discussion of different proxy problems, also from the CRU emails. This is by the former director of CRU, Tom Wigley, and it's a good overview. Here's an excerpt:
I am not sure that this unusual warming is so clear in the summer responsive data. I believe that the recent warmth was probably matched about 1000 years ago. I do not believe that global mean annual temperatures have simply cooled progressively over thousands of years as Mike appears to and I contend that that there is strong evidence for major changes in climate over the Holocene (not Milankovich) that require explanation and that could represent part of the current or future background variability of our climate.
(c) Climate variance explained by the proxy variable--close to zero for ice core isotopes, up to 50% for tree rings, somewhere in between for most other indicators. How valuable are such partially explained records in helping explain the past?So how well do these different proxies track with temperature in this paper? I looked at the graph on page 6 of the paper:
(d) Signal-to-noise problems---a key issue is, what role has external forcing had on climate over the past 10,000 years. There is a tendency to interpret observed changes as evidence of external forcing---usually unjustifiably. Few workers in the area realize that paleo interpretation has a detection aspect, just like interpreting the past 100+ years---only much more difficult. More work is needed on this.
(e) Frequency dependence of explained variance---the classic example here is tree rings, where it is exceedingly difficult to get out a credible low frequency (50+ year time scale) message. Work in this area could reap useful rewards.
The last 120 years or so are considered to be part of the instrumental record, i.e. the time in which actual thermometer readings have been available. (There are problems with these readings, as have been discussed.) I took a closer look to see how the proxies compared to the instrumental readings:
According to this, the red instrument readings line seems to line up modestly well with the proxies.
On the next page, they graph proxies individually:
Interesting, but it seems that the last 120 years are all over the place. I decided to try a visual experiment: What if all these proxies were synchronized at some arbitrary starting point in 1880, the beginning of the thermometer record? How close would they be to each other later? The first thing I did was to blow up the graph and carefully trace the separate lines, as is visible here (the thin new yellow line in the black original):
I added the blue 1880 line, the red century lines (1900 and 2000) and the green 1950 line. Once I did the traces, I brought them together so that they started at the same point (the horizontal line):
They seem to do a Paul Revere -- the fellow who jumped on his horse and took off in all directions. The tree rings will be very substantially affected by CO2, of course -- see, for example, this work by Graybill and Idso on an analysis of tree-ring proxies and particularly one of the proxies used here:
These new observations are provocative, and permit two inferences: (1) there probably are regional temperature signals at low frequencies in some of the upper treeline data before the mid 19th century but, (2) these signals become obscure as atmospheric CO2 has increased. While these relationships deserve further study, they initially appear to explain difficulties experienced in attempts to calibrate some of these data with instrumental records.In other words, don't try to use tree-rings to prove 20th-century temperatures.
Looking at the proxies (including the tree rings) the problems with matching in recent years is obvious. So if they don't track with recent temperatures or each other, how reliable are they to be considered for temperatures 1000 years ago?
Part of the mismatch is due to wet versus dry conditions, and part is the recent high growth from the CO2 increases. There's an effect from local versus global temperatures, too. There are other factors -- but this study does not take these factors into account. And thus the Medieval Warm Period seems to disappear, to make the current period seem "unprecedented" and "anomalous" as opposed to a return to good growing seasons in the 1000-1200 timeframe.
===|==============/ Level Head