IPCC AR4 is going to report that the difference between present temperature and late 19th temperatures is about 0.8 deg C, as opposed to the 0.6 deg C in TAR. This difference seemed a little higher to me than temperature changes from 2000 to 2005 (the last reported years in the respective reports) and I thought that I’d try to figure out why.The first complication is locating a digital version of the actual temperature series used in TAR. I’m sure you’ll recognize the series. This is from HadCRU2, but the earliest version that I could locate at CRU here had been updated to an end date of 2002 and was slightly different in detail – not hugely different, but annoyingly different. It’s possible that the series is archived somewhere and I failed to locate it, but I’m pretty good at finding digital data and I couldn’t find a 2000-2001 vintage version of this series at CRU or WDCP. The earliest version that I was able to find ended in 2001 – aversion located at John Daly NH here. In order to proceed, I averaged the NH and SH versions and got a better match to the IPCC TAR version. It’s pretty amazing to me that the iconic version used in IPCC TAR is so hard to find. (BTW if anyone knows of an archived version, please let me know.
Figure 1. Excerpt from IPCC TAR
I downloaded HadCRU3 global data from CRU and compared it to the earlier version in the graph below (5-year smooth). Red is the TAR version (or as close as I could get) and black is HadCRU3, which itself is updated from the AR4 version.
Notice the lowering of all past temperature estimates since TAR. So the “late 19th century” values have been lowered by nearly 0.1 deg C, with this revision accounting for about half of the increase from 0.6 to 0.8 deg C change. What is the reason for the change? I haven’t parsed the publication of HadCRU3 because of Jones’ obstinate refusal to provide station data and the resulting inability to replicate results. If it ever becomes possible to see what they did, I’ll spend more time at it. For now I merely observe the difference.
One of the most remarkable aspects of the general lowering of past values is that this included temperature data into the late 1990s – the most recent period reported by IPCC TAR. Average values for 1996-2000 were reduced by 0.05 deg C in the late version, with one individual change in the 1990s exceeding 0.1 deg C. I haven’t investigated the reasons for the change in CRU estimates in the late 1990s, but others may be interested.
Figure 2. Comparison of HadCRU2 (IPCC TAR) and HadCRU3 GLB temperature series.