Monday, August 13, 2007

1998 or 1934?

Last week one of the pillars of the case for global warming appeared to crumble, in the wake of reports that NASA's global temperature data had previously been distorted by a glitch in the data, perhaps even an unresolved Y2K problem. And apparently NASA only noticed the problem after a well-known critic of climate change pointed it out to them. The result of correcting this bug is that the revised US temperature data from NASA no longer indicate that 1998 was the warmest year in the last century. 1934 now holds that record. Does this call into question the entire edifice of climate change? No, but it does point out some pitfalls in the way that the public has been warned about it.

Somewhere between the scientists who generated the previous ranking of the ten warmest years, topped by 1998, and the folks who cited this statistic to emphasize the urgency of addressing climate change, some vital information was lost. Using the old data, the average temperature for 1998 was only 0.01 degrees Celsius warmer than the figure for 1934, expressed as temperature differences--"anomalies"--versus the average US temperature from 1951-1980. Once the data were corrected by NASA, that difference flipped to 0.02 deg. in favor of 1934. But in reality, no one could say that either of these years was warmer than the other--before or after the correction--because the uncertainty in the data is apparently 0.1 deg., or 10 times the original edge that 1998 had over 1934. (If you want a technical explanation of all this, from the perspective of climate scientists who deal with these issues routinely, I refer you to realclimate.org.)

Take a look at the actual revised data. It includes a lot of year-to-year temperature variation, with many years as much as 0.5-1.0 degrees C warmer or cooler than the preceding year. However, there's still a discernible trend, even though it may have stalled a few times. While the 1930s do seem to have been at least as warm as the 1990s, the current decade is on track to be warmer than any going back to 1880. By itself, that wouldn't be significant, but in the seventy years from 1880 to 1950 there were only 9 years in which the mean temperature exceeded the baseline 1951-80 average by 0.5 deg. C or more, and a dozen years when it was at least 0.5 deg. C colder than the baseline. Since 1980 that same comparison is 13 years to none. Consecutive colder-than-baseline years, which prior to mid-century were quite common, have vanished entirely. If these patterns don't add up to a nice straight line, then they at least ought to look peculiar enough to prompt concern. That message might not be as sensational as the one about individual or top-10 hottest years, but it is a lot more robust.

Now, it ought to be obvious that the basis of climate change is rather more complicated than whether 1998 was the warmest year in over a century, or if it was warmer or cooler than 1934. And yet, by having made such a big deal out of 1998 in the first place, or by allowing the media to focus so tightly on that factoid, some folks who perhaps ought to have known better tied the perceived validity of their argument about this critical issue to an inherently weak assertion. As a result, the science of climate change looks a tiny bit shakier today, when nothing of significance has actually changed. There's a lesson here for anyone trying to explain such a complex technical subject to a general population, using mass media in which news coverage has acquired many of the elements of entertainment programming. You can't bore your audience or talk over their heads, but you also can't reduce complex arguments to such thin reeds that they snap at the tiniest shift.

No comments: