For nearly a week, a leaked draft of next year’s “Fifth Assessment Report” on climate change, by the United Nations’ Intergovernmental Panel on Climate Change (IPCC), has been burning up the blogosphere. Since it’s everywhere, I’ll take my liberties and join the party.

The most impressive figure shows how badly one of their most-cited series of predictions is faring. Explanation follows.

The colored shading shows the projected range of global annual average surface temperature change from 1990 to 2015 for models used in the succession of IPCC assessment reports, labeled “FAR” (First Assessment Report, 1990), SAR (1995), TAR (2001) and AR4 (Fourth, 2007). The “emissions scenarios” generally cover the range from each report during the period from 1990 to 2015, and the changes in atmospheric greenhouse gas concentration that have been observed pretty much fall within this range. The very large grey zone is irrelevant to the forecasts that were made.

The three small black rectangles each year are the observed global temperature histories in common use. For every year except the last one (2011), the black “whiskers” are an estimate of the 90% confidence range for the observed temperature. Since the three records pretty much use the same data, I wouldn’t have a lot of faith in the reality of those whiskers. Data were not fully available for 2011, so any whiskers would not be comparable to the others. Quite obviously, for more than a decade, the observations have fallen near or below the lower end of the IPCC projected range. Houston, we have a problem.

Will this chart will be altered or disappear completely in the final IPCC report due in 2013? Consider what happened the last time around, in the 2007 Fourth Assessment Report.

The original draft contained a plot of latitude and historical temperature change. While the colors are a little washed out, the large red bar from 60–70° North latitude, labeled “Siberia and East Russia” is a massive summer warming revealed in a meta-analysis of the Arctic tree line by Glen MacDonald, from UCLA’s Geography Department. It lasts for more than 6,000 years, or more than half the time since the end of the ice age. The IPCC legend says that, in the region “temperatures [were] above pre-industrial levels by 2°C [3.6° or more”.

“More”, indeed! MacDonald shows July temperatures were as much as 7°C [12.6°F] warmer than what he calls “modern”. Obviously the Arctic Ocean was nearly or completely ice-free at the end of summer for centuries or even millennia, and still the polar bear survived and the Inuit culture radiated. 

By the time the final report came out, the ten-degree latitude warming shrunk to a thin stripe that just said “North Eurasia”, and 3,000 years were chopped off:

At the time, I called MacDonald at UCLA and asked if there was any intervening finding that overturned his results, and he said no. His truth was simply inconvenient to the IPCC.

As shown by the Climategate emails, when dealing with the IPCC, people need to keep their eyes on their fries. I’ll suggest the same vigilance concerning this remarkable chart showing the failure of the one or the UN’s most-cited series of forecasts, and I’ll predict that it does not see the light of day when the final report is released next year.