There is much uncertainty in estimates about ocean warming and its changing heat content.
To appreciate that one only has to look at the differences between OHC (Ocean Heat Content) graphs given in the IPCC AR4 and AR5 reports. In the 20th century the sampling of ocean temperature at depth was sparse in both space and time and biased towards coastlines, shipping routes, the Northern Hemisphere and summer. Various measuring devices were used that had their own, usually time-varying, biases and knitting all this disparate data together was problematical.
In my view the only conclusion that could be drawn is that it was probable that the OHC has increased over several decades but the error on estimates of that increase are of the same magnitude as the ‘signal.’ Go much further and one risks imposing preconceptions on the data.
Sea surface temperature (SST) have shown no significant trend since 1998 and possible explanations for it are many. Once – when it was rising between the 1970s and the 1990s SST was one of the prime metrics to measure ‘global warming’ deemed important because the greater heat capacity of the oceans would mean it would absorb more heat than the flighty atmosphere. When it became obvious that surface temperatures did not show the increases some expected it was replaced by ocean warming.
The introduction of the Argo array of some 3,750 submersibles has allowed a new look at OHC changes by providing a more coherent data set and in a new paper published in Nature Climate Change Roemmich et al use three statistical methods to group Argo data in grids to extract OHC data. They conclude that for the 0 – 2,000 m layer between 2006 – 2013 the ocean heat gain was equivalent to 0.4 – 0.6 W m-2 .
They maintain that the net 0 – 500 m global average temperature increased by 0.005°C between 2006 – 2013 and the 500 m to 2,000 m by 0.002°C per year over the same period. However none of those temperature estimates has been given an associated error. Some previous papers have suggested that between 1971 – 2010 the 0 – 700 m layer increased by 0.015°C!
The accuracy to which some Argo temperature measurements are made has always worried me. Sea-Bird electronics who make the thermometers say that under laboratory conditions they have an accuracy of +/- 0.002 when treated carefully in a calibration bath. Clearly, the sea is not laboratory conditions.
Roemmich et al claim to see significant bottom-intensified, multi-decadal warming, as detailed in their figure 2a. You will notice that in 2009 bottom-warmed water rises. But without an associated error in the temperature readings Fig 2a is meaningless. The temperature difference that could separate a temperature reading in any two adjacent bins, such as the yellow and the light green bin could be as little as 0.002°C and if the error on that measurement is of that order (I suspect more) then the presence of the ‘bottom-intensified, multi-decadal warming’ is in doubt. (Click on the image to enlarge).
There is a wider point to be made. The quoting of errors and the use of error bars on graphs should be the norm. Currently there are many papers that do not follow this practice. It’s important especially for OHC measurements where the changes observed are still about the same as the errors.