It's pretty common knowledge by now that there is one particularly embarrassing e-mail hacked from the CRU: one where a "trick" (which is, in mathematics, engineering, or science, just a way to refer to a neat solution to a problem) is used to "hide the decline" (referring to the divergence of tree ring proxy data from the measured temperatures starting in 1960).
So, is this a case of covering up some nefarious giant conspiracy? It's actually extremely easy to find out if one is a university student. I'll now go through papers through ISI Web of Knowledge [v4.7] - Web of Science as though I was a new graduate student who new to a lab that studies climate change.
Here is thus (dun dun dun!) my first live blog of any event: me reading through research articles! Think of it like reality TV, except through blogging, with fewer insults, and a lot more exciting! ;)
1) Search "climate change" in the Topic field.
Become incredibly depressed at the fact that I have no clue what anyone is saying in most papers. Realize that despite all that hard work in undergraduate studies has done little to prepare someone for actual research. (No, I actually didn't do this, but trust me, this is what every new graduate student feels like. Eventually, though most grad students do make good use of the work habits and some skills learned from undergrad.)
2) Search "Mann M*" in the Author field, "Nature" in the Publication Name field, and 1998 in the Year Published.
Since Dr. Michael E. Mann is the one being targeted, I'll look at his paper.
Well, the name of the paper is "Global-scale temperature patterns and climate forcing over the past six centuries". That already screams out the reason why direct temperature records are insufficient, unless there was a worldwide record of temperatures dating back to 1400.
Mann et al. are already discussing "multiproxy" records (i.e. data from multiple sources, including tree ring data, ice core samples, coral, etc.). Direct temperature measurements are available only from 1902 onward. (In this paper, the data ends at 1995.)
So far, much of the paper goes into describing assumptions necessary to reconstruct the temperature record before 1902. Assumptions of largely linear relationships (you'll see this in many papers), that a sparse number of proxy data sets that are widely distributed will be enough and will capture out-of-measuring range events (such as the El Nino/Southern Oscillation), and that variations in proxy data will have analogues to variations in the temperature records from 1902-1980. Due to the duration of the multiproxy data (600 years), it is safe to assume that statistical reconstruction methods will be sufficient since most temperature cycles occur over the span on the order of a decade (hence, there will be multiple repetitions of temperature cycles over 600 years).
Now, he describes what causes the variations in the multiproxy data. There are five sources that account for 30% of the variations. The first is obviously the overall global warming trend (modern day). The second is the El Nino/Southern Oscillation. The third is the North Atlantic Oscillation. The last two are multi-decade variations in El Nino and the Atlantic basin.
Now, it seems to me that this "hiding the decline" is silly. If one looks at the "hockey stick graph" (Figure 5b), the plot shows the mean of the ACTUAL MEASURED NORTHERN HEMISPHERE TEMPERATURE DATA from 1902-1995 (mostly a mess with the reconstructed, reconstructed 50-year low-pass, and error lines, but still clearly visible after the reconstruction line ends at year 1980). Surprise, surprise, but the mean of the actual recorded temperature is going up.
Also of note is the reconstructed 50-year low-pass line: low-pass doesn't mean "lowering the temperature displayed" or anything like that. It refers to a low-pass filter to get rid of the noise in the data (since noise is high frequency and will not pass through the filter). Just to get that out of the way.
What probably happened was that Dr. Mann was faced with an incomplete data set for the tree-ring proxy data after 1960, given that the extrapolated temperature from the tree rings diverged from the data obtained from other proxy sources (as well as, you know, the actual temperature). As such, Dr. Mann just filled in the blanks with the actual temperature so that he could finish the demonstration of his technique.
To me, this paper is mostly about introducing the technique of using multiple proxies to extrapolate the temperature record. The temperature reconstruction itself is almost secondary. The tree-ring data could be replaced with some other proxy source if desired, and a new reconstruction could be made using the multiple proxy technique.
Scientists have already expanded on Dr. Mann's work with more extensive temperature reconstructions using far more proxies in far more locations. Other scientists have gone on their own way with other techniques. Guess what? Despite variations in the reconstructions, the one consistent theme is the rapid increase in modern day temperature.
-------------------------------------------
That's it for now, I guess. It's pretty easy to refute the "hide the decline" claims by just looking at Figure 5b (with the mean of the ACTUAL MEASURED TEMPERATURE plotted from 1902-1995). Furthermore, the tree-ring data is only one proxy source, with other proxy sources (such as the ice core samples and coral growth) used for the temperature reconstructions. I'm still happy that I read the entire paper, though. Learning new things is never a bad thing.
Still, the paper is over a decade old (even though it is a high-impact paper). Perhaps I'll live blog yet another paper reading one day... a newer one with the latest science.
Peace and long life.
1 year ago
My professors and tutors at school never failed to point out the dangers of extrapolation; it was a thing to be avoided at all costs. Pity that the scientists involved in this nonsense did not receive, or failed to heed, similar warnings.
ReplyDeleteThe whole AGW affair, I think, boils down to this: the public is balking at following the lead of those whose predictions are based upon computer models that have demonstrated no ability to make accurate predictions (ie. the present cooling was not predicted by the current models). All the scientific explanations in the world will not convince John Q. Public when he sees such a simple failure.
Firstly, thank you for taking the time to comment!
ReplyDeleteI suppose it depends on what one means by extrapolation. Extrapolating from limited information with no knowledge of the system being analyzed is bad, but that's not what scientists are doing. (For example, if I gave you two points on a plot and asked you whether it was linear, quadratic, logarithmic, exponential, sinusoidal, etc., it would be impossible to extrapolate.)
Trying to predict any future event is inherently difficult. This is why continued studies and improved models are being developed, such that more accurate predictions can be made. This isn't the fault of the scientists. Even engineers sometimes run aground, since engineering something is trying to predict whether something will work after it is built.
One engineering example I can recall from the top of my head would be the clock speed barrier that was encountered with the Pentium 4. The voltages necessary for the (pipe) dream of a 10-GHz Pentium 4 to be achieved became too high (and yes, that was what Intel had initially believed to be the optimistic upper limit of their design). The problem that was not incorporated into the engineering models was the effect of quantum tunneling, which was more prominent with higher voltages. These days, CPU's incorporate high-k dielectrics to minimize the tunneling effect while still reducing size.
This is why continued research into climate science is so important. With further research, models will continue to improve, and more accurate and more precise predictions can be made.
What we do know for certain (from ice core samples with records dating back to 600000 years) is that the carbon dioxide concentration in our atmosphere is increasing at a rate significantly faster than the historical 100000-year cycle ever since the Industrial Revolution. We also know that carbon dioxide is a greenhouse gas. While one can argue about the actual strength of the carbon dioxide forcing, a general global warming trend is inevitable.
I would guess that when you question the accuracy of climate models, you are really questioning the precision of the climate models. (That is to say, a ten-year trend is accurate, but a yearly prediction is tougher to pin down.) In this sense, you are correct in saying that the public is likely not likely to be convinced simply due to the variability of the weather.