Roger Andrews as promised has published his analysis of SST (Sea Surface Temperatures) datasets, based on some years of research. The essay is Making the Measurements Match the Models – Part 2: Sea Surface Temperatures and well worth a look.
Some years ago while reading to get up to speed on climate science, I was struck by a Roger Pielke Sr. comment. He said that surface temperatures are serving as a proxy for changes in heat content of the earth climate system, which is the real concern. And air temperatures are contaminated by fluctuations in water content, such that a degree difference in the humid tropics involves much more additional heat than does the same change in extremely dry polar air.
For those who want to see the math, here it is from the Engineering Toolbox.
The enthalpy of humid air at 25C with specific moisture content x = 0.0203 kg/kg (saturation), can be calculated as 76.9 (kJ/kg). . .The same calculation for moist air at 20C gives a heat capacity of 58.2, so the 5C increase requires 18.7 kj/kg for moist air vs. 5.0 kj/kg for dry air, or a ratio of 1:3.7. Similar ratios apply at all air temperatures above 0C. Subzero air, like that in the Arctic most of the year, shows little difference in heat content between dry or saturated, since cold air doesn’t hold much water vapor. See Arctic Amplication?
One implication is that polar air temperatures lacking moisture are 2-3 times more volatile, leading to the “Arctic Amplication” effect. Even so, a thorough look into weather station records around the Arctic circle undermines fears on that account. See Arctic Warming Unalarming.
The larger point made by Pielke Sr. was that a much better proxy for global warming or cooling is provided by SSTs. Measuring temperature changes in the water itself is a much better idea, giving a more exact indication of changes in heat content. There is also the point that SSTs cover 71% of the planet surface.
Andrews knows well the difficulties in assembling SST datasets, including the bucket era and the engine intake era. He addresses directly the problematic WWII measurements, suggesting they can simply be excluded as bad data without affecting the pattern. He also compares the International Comprehensive Ocean-Atmosphere Data Set (ICOADS) raw global SST series used to generate the global HadSST3 series, which is the most widely cited of the currently-published SST series.
There he finds that prior to 1940, there was systematic warming adjustments making HadSST temps higher than ICOADS. He attributes this to the long-standing belief that Night Marine Air Temperatures (NMATs) should synchronize with SSTs. That assumes that air moisture over the water should be fairly consistent from one location to another, and that marine air would be in thermal equilibrium with the water.
But apparently no studies have proven that assumption. I know of one empirical study of the ocean-air interface which shows considerable fluctuation in both the heat exchange and evaporation rates. See Empirical Evidence: Oceans Make Climate
Thanks to Roger’s work on this, we can conclude that SSTs prior to 1950 have issues, but can be encouraged that HadSST3 since then is reasonably consistent with the raw data. And in the future the ARGO record will become long enough for us to follow the trends.
The best context for understanding global temperature effects in recent years comes from the world’s sea surface temperatures (SST), for several reasons:
- The ocean covers 71% of the globe and drives average temperatures;
- SSTs have a constant water content, (unlike air temperatures), so give a better reading of heat content variations;
- Major El Ninos have been the dominant climate features these decades.