C. The Temperature Anomaly was Overstated by 8.6% ​

The reported anomaly for 2019 was 0.79°C.  However, if a weighted averages were used for both land and ocean temperature then the average anomaly would be 0.73°C and not 0.79°C.   This indicates that the temperature anomaly was overstated by 8.6%.  The following is the basis for this conclusion. The quote below is from the HadCRUT4 website.

“How are the hemispheric and global anomaly series calculated?”

“Values for the hemisphere are the weighted average of all the non-missing, grid-box anomalies in each hemisphere. The weights used are the cosines of the central latitudes of each grid box. The global average for CRUTEM4 and CRUTEM4v is a weighted average of the Northern Hemisphere (NH) and Southern Hemisphere (SH). The weights are 2 for the NH and one for the SH. See Osborn and Jones (2014) for how this differs from previous versions of CRUTEM. For HadCRUT4, the global average is the unweighted average of the NH and SH values.”  Emphasis added.  [https://crudata.uea.ac.uk/cru/data/temperature/]

 As stated above, the CRUTEM4 (surface air above land) is based on a weighted average so that the Northern Hemisphere (NH) is doubled over the air temperature measurements taken in the Southern Hemisphere (SH).  The NH has a higher temperature anomaly than the SH.  For example the Northern Hemisphere can have a +1.5°C higher anomaly than the SH.  [Feulner, G et al (2013) Why is the Northern Hemisphere warmer than the Southern Hemisphere (2013) Journal of Climate, Vol 26, pgs. 7136-7150]  The CRUTEM4 showed the difference in 2019 to be about 0.5°C.  The justification for doubling the NH land temperature is that the landmass in NH is twice the size of the landmass in SH.  To obtain the average land temperatures, the arithmetic average of the NH with the SH would not be correct.  A weighing between NH and SH would be more appropriate and that is what they did.

However, it would also be appropriate to do the same thing in determining the average ocean temperature.  Based on the information from the Met Office Hanley Centre, the ocean surface temperature was obtained from the NH and the SH oceans.  They were then added together to get an arithmetic average temperature.  That is not the same procedure as followed with the land.  The SH has twice the ocean surface as the NH and simply adding the two together does not yield the average ocean temperature.  They must double the ocean surface temperature weight in the Southern Hemisphere to determine the weighted average ocean temperature.  Because the Southern Ocean does not change as much as the Northern Ocean, the arithmetic average skews the anomaly to the high side.  For example using 2019 data the SH had an ocean temperature anomaly of 0.47°C while the NH had an ocean anomaly of 0.72°C.  The arithmetic of average of the two is 0.595 which is what HadSST3 reported for the global ocean temperature anomaly.  If they gave the SH a weight of 2, as they did with the land for the NH, then the 2019 average global ocean temperature anomaly would have been 0.55°C and not 0.6°C.  This is a significant different since it appreciably affects the thermodynamic heat balances.  HadCRUT4 then arithmetically averaged the Weighted CRUTEM4 (land temp) with the Unweighted HadSSL3 (ocean temp).   Averaging a weighted temperature anomaly with an unweighted temperature anomaly is not appropriate.  

Other climate change entities appear to have done a similar weighted average land temperatures and arithmetic average ocean temperatures.  Example see data at Hansen J and Lebedeff  S (Nov 1987) Global Trends of Measured Surface Air Temperatures Journal of Geophysical Research, Vol 92, No D11, pgs. 13,345-13,372.