By Dr. Lubos Motl
The growing discrepancy between the satellites and the terrestrial measurements could be naively interpreted as bad news for science – there has to be some big mistake at least in one class of these measurements. But this conclusion isn’t really right. Their different stories about the global mean temperature probably aren’t due to a “mistake” but due to their different definition of the global mean temperature. The terrestrially measured temperature is increasing at this rate approximately 2 °C per century while the rate is less than 1.5 °C – and, according some proxies, even much smaller than that – according to the satellites. That’s not a contradiction because they mean different quantities by the “global mean temperature”.
There is no “canonical” global mean temperature. It’s an artificial quantity whose detailed value – and whose detailed change in 18 or 100 years – significantly depends on all the details about how the global mean temperature is defined and measured. I’ve mentioned that 2016 was 0.02 °C warmer than 1998 by satellites but by 0.36 °C warmer than 1998 according to the terrestrial measurements. You may pretty much say that the difference of the annual global mean temperatures in 2016 and 1998 was 0.19 °C plus minus 0.17 °C. It’s still pretty much compatible with the zero at the one-sigma level.
The GISS temperature anomaly for 2016 was 0.99 °C, about 0.13 °C warmer than the anomaly in 2015. It’s a significant increase but the increase between 1997 and 1998 was larger still, 0.15 °C. According to GISS, the temperature in 2016 was 0.36 °C warmer than in 1998. You may see that the ratio is almost exactly 0.02 °C per year or 2 °C per century. This deduced trend is almost 20 times larger than that from the satellites but even if we trusted the linear extrapolation, it would still be far from a tragedy if the temperature in 2116 were 2 °C warmer than in 2016.