This week’s ECS estimate comes from UCLA scientist Troy Masters, who in 2013 published a paper in the peer-reviewed journal Climate Dynamics called “Observational estimate of climate sensitivity from changes in the rate of ocean heat uptake and comparison to CMIP5 models”. We’re finding a pattern in these ECS studies where the longer the title, the more it gives away the plot. In Masters’ case he, like the others, gathered up data on air temperatures, changes in atmospheric CO2 levels, changes in ocean heat content etc. and started crunching numbers. But he also decided to compare what the real world was doing to what the world inside climate models was doing. And he noticed a difference.

He focused on estimating a process called “radiative restoration strength”, which refers to how quickly the climate system sheds excess energy to space after a change in “forcing” (in the CO2 case, after a change in the amount of greenhouse gases and their energy-trapping activity in the atmosphere). Whether the Earth has a high ECS or a low ECS depends on how quickly the atmosphere can restore its radiative balance. If the climate is slow and inefficient at radiative restoration then ECS will be high because the climate has to heat up more to get the energy out to space. To estimate radiative restoration Masters decided to use observations on ocean heat content since most of the warming removed from the atmosphere ends up in the ocean.

The first step was to compute the radiative restoration strength (RRS) in the observed data and compare it to the models. Masters noted right away that the RRS in the data was twice as large as in the models, and the only two models that matched the data were also the ones with the lowest ECS. He then went on to estimate the implied ECS values in the observations and found the best estimate was 1.98 degrees C. He noted that this was low compared to the usual IPCC range and the range found in climate models, and he added:

“The latter is not surprising given that only three of the CMIP5 members examined matched or exceeded the median strength of radiative restoration as diagnosed from observations, and they generally produce a value [for RRS] that appears to be biased low.”

And an RRS biased low means an ECS biased high. It’s starting to look like a pattern.