No Se Nada

Received this email in December: "Dear Kevin, On December 15, they release a statement that argues that 2005 is the warmest year ever. But there's still two weeks left in the year -- how do they know what the mean global temperature will be for this year? Besides, they don't get many stations in real-time, so in reality, their assessment is made with data that extends only until the end of October. Doesn't it seem odd that they can proclaim the outcome before it happens?"

First, the press releases (NASA here and WMO here) and news stories I tracked down were pretty careful to describe the data as preliminary. In fact, the abstract of the WMO release contained this very important caveat: "The uncertainty in the global temperature values, arising mainly from gaps in data coverage, are such that 2005 could be the warmest year or the eighth warmest year on record."

Which is very telling about how warm 2005 was.

In terms of "there's still two weeks left..." there's an easy way to answer this: 92% of the data is in*. (And for the NASA GISS work, the averaging period is Dec. 2004 – Nov. 2005, giving a full year's record.) In order for the remaining 8% to be able to alter the record to something outside the projection based on 92% of the data would require global average temperatures to suddenly swing wildly away from both the natural variability seen in the December long term average and the 2005 trend up to Dec. 1, 2005. To elucidate the former: from a century of data, we know both the average and the standard deviation of December global temperatures. Even if December 2005 deviated by more than 2 standard deviations away from the long term December mean, that last month wouldn't likely have a big influence on the entire year record. To elucidate the latter: there is also a long term relationship between the average December temperature and the average October-November temperature. A big deviation from that relationship would require a drastic weather change that affects the entire globe, which is half in winter and half in summer. You can derive the statistics to tell you just exactly how unlikely that is, but even without the stats in front of me I am pretty certain of its unlikelihood.

*I say that 92% of the data is in, ignoring another aspect of the question above: "...they don't get many stations in real-time..." Sure enough, but they get enough stations, distributed with enough geographical diversity, to be confident that the subset of the data they do have accurately represents the entire dataset. If one, as a skeptical climate scientist, were inclined to test this, it would be fairly easy to grab global datasets from years past, extract subsets from the full set, and test to see whether those subsets were consistent with the entire record.

In order to make these projections, the WMO actually grabs and synthesizes three different temperature data sets. As all three have their own methodology for collecting, processing and averaging data, synthesizing them helps ensure that processing quirks don't unduly influence the overall picture. Even so, the many thousands of temperature stations used in each dataset is an inherent check on the system, ensuring that outliers cannot weight the average (true outliers, such as a station that is giving you 40°C when everything around it is giving you 20°C, are discarded).

(If you don't like the idea of relying on averaging, think of it this way: average Vince Young's future NFL salary with ten average Americans, and the average is very close to Young's salary. But average his salary with ten million average Americans and Young's salary becomes less influential on the final number.)

The NASA GISS averaging done by Jim Hansen's group is independent of the WMO projections and comes to the same conclusion. Incidentally, Hansen's group includes this caveat:

"Our analysis differs from others by including estimated temperatures up to 1200 km from the nearest measurement station (7). The resulting spatial extrapolations and interpolations are accurate for temperature anomalies at seasonal and longer time scales at middle and high latitudes, where the spatial scale of anomalies is set by Rossby waves (7). Thus we believe that the remarkable Arctic warmth of 2005 is real, but our rank of 2005 as the warmest year depends upon that assumption."

And (7) is a footnote that refers to this paper: Hansen, J.E., and S. Lebedeff 1987. Global trends of measured surface air temperature. J. Geophys. Res. 92, 13345-13372.

I have no problem with that assumption, either.

Of course, we can now bring this discussion (see the comments there) about science and politics into the fray, as Hansen's group say on their webpage: "Recent warming coincides with rapid growth of human-made greenhouse gases. Climate models show that the rate of warming is consistent with expectations (5). The observed rapid warming thus gives urgency to discussions about how to slow greenhouse gas emissions (6)."

But my synthesis take is that despite the politics, the results are robust.