While looking at the various curves and swerves on CA Eli was impressed with how well the year to year changes in temperature anomalies between the GISS ocean atmosphere record and RSS followed each other until 2007. Indeed this was a clear tip off that the RSS 2007 or the GISS 2007 values were wrong.
So he decided that it would be sensible to plot the various series against each other
No adjustments have been made to the series. They are plotted as reported. There are several interesting things to note. First, after 1980, if the anomaly goes down in one series, it goes down in the others. If it goes up in one series, it goes up in the others. If it goes sideways in one series, it goes sideways in the others. Kind of like Daniel Drew and Erie Railroad stock
This lead the bunny to some interesting conclusions. If there are errors in the satellite reconstructions, a place to look might be
1995 1985 [thanks to Alocacoce for the correction] and 2003, where the RSS and UAH records diverge. Eli would be somewhat more suspicious of the UAH record because both of the GISS anomalies vary as the RSS record. Also, the GISS Ocean Land anomaly in 1997 appears to rise more than all of the other records.
Because the MSUs are subject to long term drifts, this agreement of the yearly variations appears to be a decent first order check. We can add the Hadley Center global temperature anomaly to the comparison (warning, symbols change)
HadCRUT appears to have less annual variability than the other three reconstructions. More numerology tomorrow but this establishes that GISS, UAH and RSS are measuring the same thing.
UPDATE: Upon awakening some thoughts.
1. RSS and UAH use the same data from the same instruments. GISS and HadCRU use in different ways and combinations the same sets of measurements. Above, Eli has shown that interannual comparison is a good way to check changes in algorithms, particularly for GISS, UAH and RSS.
2. As Rabett Holmes would say, the failure of McIntyre and Motl not to graph UAH was the dog that needed to go for a walk in the night. If UAH was plotted, (see top graph)
a. Their arguments would have been weaker3. Understanding the differences in interannual variability between HadCRU and the other sets is potentially important, not only on the annual level, but also seasonal and geographically. For example, in HadCRU looking at 1996, 7 and 8, it looks like more of the El Nino was captured in 1998 than in the other records. (Eli used J-D for the year in all the others, HadCRU may be using D-N??). There is also a lot less variability in HadCRU overall.
b. It would have been clear that there was something amiss with RSS v3.0
c. This is the sort of thing you try and teach your students, there is no such thing as a negative volume. If that is your answer, you have made an arithmetic error
4. We can perhaps use the variability (to some extent) to estimate the difference in climate sensitivity at the surface and in the lower troposphere. For example if we ratio the differences in the global temperature anomalies between 1998 and 1996 for UAH and GISS Ocean Land we get .49/.27. for the GISS surface stations it is .49/.33. Similar trends appear to hold at other places where the differences between years are significant. For UAH/HadCRU we get .49/.41. The title of this thumbsucker is Food for Tamino.
5. It looks like UAH and RSS have converged with the launch of the AMSU.