On The Wings Of Butterflies

My favorite class of all time was a course on chaos theory that I took my senior year in college. I think it was listed as part of the Chemistry Department’s curriculum (hey, whaddya know, it still is!), but it also involved physics and statistics (and thankfully, little math). My fondness for the class was certainly influenced by the teacher, who was uniquely adept at fostering intellectual curiosity, but also by the broad implications of chaos theory and the study of non-linear systems that piqued my imagination. I am constantly considering lessons learned from that class when pondering questions of probability, predictability and dynamic systems of all sorts, even everyday human interactions. Of course, the easiest thing to remember from that class was the theme: Sensitive To Initial Conditions or “STIC”.

Although most people today understand the basic idea behind chaos theory, they only grasp it from that colorful metaphor known as the “butterfly effect” which describes how a butterfly flapping its wings over Tokyo may cause (or prevent) a tornado occurring in Houston. What many people forget is that the story behind that metaphor, the origin of modern chaos theory, was an accident in trying to predict the weather.

Around 1960 an American mathematician and meteorologist, Edward Lorenz, developed a computer model to study weather patterns and ultimately to predict weather. What Lorenz discovered was that minute changes in one part of the globe could dramatically alter weather patterns in completely different places. The most dramatic example of this occurred when Lorenz attempted to recreate some interesting weather patterns he noticed from the computer model’s results. At the time, it was not unusual for Lorenz to set up some weather models, feed them into the computer to run overnight, and then review the results the next morning. This particular morning, however, when Lorenz tried to recreate the patterns, he instead received results indicating tsunamis, floods, hurricanes, tornadoes, blizzards and many other dramatic weather effects, none of which were in any way similar to the overnight results.

Lorenz sat down to figure out what went wrong, and eventually realized that the coordinates he was inputting were incorrect. The problem was that he could print out the coordinates to three decimal places, but the computer itself was keeping track up to six decimal places. So instead of inputting .506127 to recreate the weather scenarios from his model, Lorenz typed in the truncated .506, and chaos theory was born. That such a minuscule change could have such dramatic and wide-ranging effects on the weather was dubbed the butterfly effect.

The immediate lesson learned from Lorenz’s discovery was that predicting weather is so dependent on an infinite number of variables that, aside from our ability to even know (much less model) all of the them, the slightest change in any single one of them can completely alter any prediction, rendering it unrecognizable. The ability to model weather outcomes is completely dependent upon, and highly sensitive to, the initial conditions set for the model.

In an entirely unrelated news event, it seems that NASA has quietly revised downward the temperature data for the US, resulting in 1934 now being the hottest year on record, as opposed to 1998 as was previously stated unequivocally and vociferously. Apparently, a Y2K bug caused some big shifts in temperature data points which had previously gone unexplained. Steve McIntyre noticed the anomaly and brought to NASA’s attention, prompting the dramatic revisions. I am still not certain as to how a Y2K bug altered the readings for temperatures prior to 2000, and further revisions may still be necessary based on the findings of Anthony Watts, but suffice it to say that the revisions certainly cast some doubt on the wilder claims emanating from AGW^TM proponents. And yet, there are still those who insist that the revisions mean nothing.

Because the 1998 and 1934 numbers were so close, minor adjustments could easily change their ordering. This is what happened with the GISS numbers released this year. In that data set, 1998 was a tiny amount warmer than 1934. This change was not much ballyhooed. Nor was it a little ballyhooed. In fact, it wasn’t mentioned by anyone at all. Because it didn’t matter. When the data correction made 1998 and 1934 flip back, this change was much-ballyhooed by Steve McIntyre, even though he knew that it didn’t matter.

In fact, the claim that 1998 was the warmest on record was much ballyhooed, by people such as Al Gore in his film An Inconvenient Truth, and by the media such as recently when it was thought that 2006 had topped 1998′s temperatures:

Here’s one convenient truth for anyone worried about heating bills: 2006 was the warmest ever recorded in the United States.

[...]

“This new information represents another warning that climate change is happening around the world,” Britain’s Meteorological Office said.

The center’s preliminary data, reported Tuesday, listed the average temperature for the 48 states last year as 55 degrees. That’s 2.2 degrees warmer than average and 0.07 degrees warmer than 1998, the previous warmest year on record.

“There’s no denying that climate change is occurring, and warmer winters and warmer years are more common for that reason,” said Jay Lawrimore, monitoring chief for the U.S. climate data center, which keeps the nation’s weather records. “What we’re seeing (in 2006) is just becoming so much more common.”

Worldwide, the agency said, it was the sixth warmest year on record.

Indeed, none other than Michael Mann (no, not that one) decried 1998 as possibly the warmest year in the last 1000:

Climate scientist Michael Mann (famous for the hockey stick chart) once made the statement that the 1990′s were the warmest decade in a millennia and that “there is a 95 to 99% certainty that 1998 was the hottest year in the last one thousand years.” (By the way, Mann now denies he ever made this claim, though you can watch him say these exact words in the CBC documentary Global Warming: Doomsday Called Off).

In addition to being much ballyhooed, the claim was simply wrong about 1998 (just as it was about 2006). The significance is not in the amount of difference between the relative temperatures which, as Lambert notes, are quite small (albeit bigger now), but instead in what the corrected readings mean for the also-much-ballyhooed predictions based upon those numbers. Recall my previously unrelated anecdote, and apply STIC here now. If computer models attempting to predict future weather conditions were using the wrong data, just how accurate could they have been? Ponder that for a moment while we move along.

In addition to those questioning whether the relative order of warmest years makes much of a difference in the long run, there are also some who think it highly relevant to note that the recent change was merely for US temperatures and not for global temps.

This all had virtually no bearing on the global temperature record, in which 2005 still appears to be the hottest year on record, and Al Gore’s claim that nine of the ten warmest years in history have occurred since 1995 is still operative.

See also (HT: Insty).

That would be all well and good except for the inconvenient fact that the US historical temperatures are by far the most complete and accurate (scary, huh?) set of data around. To argue that changes in historical global temperature are not greatly affected by these downward adjustments by NASA is to misunderstand just how great the divide is between US temperature gathering and that of the rest of the world. As Warren Meyer observes:

The GISS today makes it clear that these adjustments only affect US data and do not change any of their conclusions about worldwide data. But consider this: For all of its faults, the US has the most robust historical climate network in the world. If we have these problems, what would we find in the data from, say, China? And the US and parts of Europe are the only major parts of the world that actually have 100 years of data at rural locations. No one was measuring temperature reliably in rural China or Paraguay or the Congo in 1900. That means much of the world is relying on urban temperature measurement points that have substantial biases from urban heat.

Steve McIntyre goes into greater depth about the problems with discounting the importance of U.S. historical temperature data (my emphasis):

Schmidt observed that the U.S. accounts for only 2% of the world’s land surface and that the correction of this error in the U.S. has “minimal impact on the world data”, which he illustrated by comparing the U.S. index to the global index. I’ve re-plotted this from original data on a common scale. Even without the recent changes, the U.S. history contrasts with the global history: the U.S. history has a rather minimal trend if any since the 1930s, while the ROW has a very pronounced trend since the 1930s.

[charts omitted]

These differences are attributed to “regional” differences and it is quite possible that this is a complete explanation. However, this conclusion is complicated by a number of important methodological differences between the U.S. and the ROW. In the U.S., despite the criticisms being rendered at surfacestations.org, there are many rural stations that have been in existence over a relatively long period of time; while one may cavil at how NOAA and/or GISS have carried out adjustments, they have collected metadata for many stations and made a concerted effort to adjust for such metadata. On the other hand, many of the stations in China, Indonesia, Brazil and elsewhere are in urban areas (such as Shanghai or Beijing). In some of the major indexes (CRU,NOAA), there appears to be no attempt whatever to adjust for urbanization. GISS does report an effort to adjust for urbanization in some cases, but their ability to do so depends on the existence of nearby rural stations, which are not always available. Thus, there is a real concern that the need for urban adjustment is most severe in the very areas where adjustments are either not made or not accurately made.

So does any of this mean that AGW^TM is just a big fraud? Well no, not exactly, although the latest news does take the wind out of Al Gore’s sails and devalues previous claims that we’re in an unprecedented warming period. In addition, James Hansen’s credibility has taken another hit, IMHO, since it was not an internal investigation that revealed the Y2K glitch, but a reverse-engineered analysis (because Hansen will not allow the source code and formulae used to generate his graphs to be released) that brought the problem to light. Moreover, what’s to be done about the other glaring problem with data-collection, namely the 1,221 poorly maintained weather observation bases? Similar questions are raised about the data-gathering techniques being used around the world. And although the change in U.S. temperatures has not altered the global record by much (how that is possible, one can only wonder), the veracity of the entire process certainly seems to be due for some review and adjustment. Accordingly, even if this news doesn’t sink AGW^TM it certainly casts grave doubts on the numbers backing up the theory.

In the end, those who are skeptical of the claims made by AGW^TM alarmists, and who dare to challenge their received wisdom will be cast as loutish Luddites afraid to face the plain facts so helpfully adduced by the believers. “It’s scientific consensus!” the believers shout. “Look at the bones!” they bray. Even as they studiously ignore the light cast upon the faulty premises underlying their wisdom.

But science doesn’t work that way. Like any other dispassionate, fact-oriented system, it functions on a “garbage in-garbage out” principle that is highly sensitive to the inputs. As we know, weather is also highly sensitive to initial conditions, so much so that a butterfly’s wings may make the difference between a hurricane in Seattle and a blizzard in Sydney. The systems designed to model the weather are even more sensitive, since they are largely incomplete, and thus the slightest change in the parameters can render previous results obsolete. Some will attempt to argue that the GISS temperature data is not used in the climate-prediction models, so the latest revisions really don’t mean anything, but they are being disingenuous at best.* Without the ability to accurately record the past, what chance do we really have of accurately predicting the future? But most importantly, science is not about “consensus” but instead falsified hypotheses deduced from re-creatable experiments producing the same results, over and over again. When the data underlying the experiment change, so too must the results. In this case, as Lorenz discovered on that fateful morning in 1961, changing the initial inputs may have dramatic changes. A little acknowledgment of that fact wouldn’t hurt the global warming crowd.

___________________________________________________________________________
* Here’s a hint as to why: even if the temps aren’t plugged into the models themselves, the data is compared to the modeled predictions, and then the models are tweaked to replicate the historical record as accurately as possible. If the models are being tweaked to generate an incorrect historical record, why would we expect them to produce an accurate prediction of the future?
___________________________________________________________________________

[tags] environment, global warming, climate change, weather observation stations, James Hansen, NASA, NOAA, U.S. historical temperatures, Steven McIntyre, Anthony Watts, Roger Pielke [/tags]