20-year surface trends close to models

Recently, I’ve been commenting to anyone who’ll listen (or not) that short-term comparisons of global temperature trends are not very meaningful and that at least the last 20 years should be analyzed. This was based on the seemingly paradoxical observation that, while trends from 2001 to present are down or flat, long term trends increased during the same period.

Finally someone (Lucia of Blackboard fame) has listened. Unfortunately, she has compared 20-year observed trends to a model subset that has an unrealistically high projected trend.

Lucia’s graphs have the model trend rising as high as 0.28 deg C per decade, whereas the commonly accepted IPCC benchmark projection is 0.2 deg C per decade for the current decade.

[Update/correction (April 20): Lucia has pointed out that the period for the IPCC 0.2C projection is not the same as the 20-year trends. In fact the trend rose through the period, if calculated directly from the model ensembles. However, there are numerous issues with using the detailed model trends for this purpose. In any event, I had already decided to perform a trend analysis based on a comparison of the post-baseline period to the baseline, as seen in the immediately subsequent post.

Meanwhile, I'm still pondering the correct values for the IPCC 20-year trend. According to Lucia's calculations, it climbed from 0.16C/decade in 2000 to 0.24C/decade at present for the A1B scenario. This is hard to square with the stated projection of about 0.2 deg C for the 2000-2008 period relative to the baseline, and points to possible noisiness in the model realization that should be smoothed out. In colloquial terms, the model noise appears to have "zigged" where the observations noise "zagged". In addition to the "averaging" approach mentioned above, other ways to handle this include using even longer trend periods or different forms of smoothing for both observations and models.

So the IPCC benchmark in the graph should be taken with a grain of salt, although it is apparently very close to the smoothed 20-year trend over the period. Or skip ahead to the next post which is a more cogent analysis in any case.]

Her graphs are getting very wide exposure indeed (they’re already up at WattsUpWithThat). So without further ado, here’s the real scoop:

20 Year Trend New

[Revised (May 8)] Using the standard 0.2 deg. per decade benchmark, Lucia’s A1B multi-model mean trends, the 20-year trends were ahead of the IPCC projection earlier this decade, but now are substantially below as of 2008. Certainly such fluctuations are to be expected with the larger interannual variability of the “real” climate compared to the models. (For comparison, look at Lucia’s original post here; it shows the model trends way above the observations; the difference is that Lucia used the “volcano” subset, which is a good 0.04 deg C higher in the critical 2006-8 period .)

Later, I’ll go into greater detail on the background and various issues involved [including derivation of the IPCC benchmark].  A more detailed analysis shows an increasing trend decade over decade (despite a relatively cool 2008). But for now, it should be noted that claims that models represent an “abject failure” or even that global warming has stopped are clearly unfounded. To be continued …

About these ads

3 responses to “20-year surface trends close to models

  1. Deep Climate–
    As I said in my post, I compared to the trend computed from runs downloaded from The Climate Explorer, using only those that include volcanic eruptions. I used this because Pinatubo erupted in 1991 and El Chicon erupted in 1982. Both affect the multi-model trend. For this reason, if you download the runs and compute the 20 year multimodel trend both exceed 2C/century currently.

    I could of course compare the 20 year trend to the “about 2C/century” which is an average over a period quite a bit longer than 20 years. But I prefer to compare 20 year trends to 20 year trends.

    For what it’s worth, when I did compare the current data to the 2C/century finding the current trend are inconsistent with the “standard” 2C/century, “people” complained that it was unfair to compare trends to the commonly used 2C/century– I should download the models and use the specific current values since those vary over time.

    [DC: 0.2C/decade is the projected IPCC temperature rise only for the earliest part of this century. Even by 2030, the scenarios diverge slightly. The total century rise is 2.65 C under A1B and 3.13 C under A2.

    I'm not sure who complained about your previous use of the "2C/century" trend, but my biggest concern with your analysis (and that of Pat Michaels) has always been the over reliance on analysis of short term trends derived from specific start and end points.]

  2. Lucia is disengenous. I have seen her evade and obfuscate in the past. When she is in the wrong, she refuses to be pinned down. JohnV got pretty disgusted with her. It is a shame since she is moderately smart.

    She is also an eefil acrylic knitter.

  3. Pingback: Recent surface temperature trends relative to baseline « Deep Climate