Skip to content

Warming Underestimated – Does It Matter?

Dr David Whitehouse

It’s a curious phenomenon that examples of climate ‘forcing’ always seem to occur just before big environmental summits, and that the forcing only ever goes in one direction. The UN Climate Change meeting at Cancun this week is no exception. The UK’s Met Office, among others, released a series of statements and in the Met Office’s case a brochure about climate change. Their conclusion is that things are probably worse than we thought, and in their opinion, is worse than the current science is telling us. I suppose in the face of uncertainties in the science, and contradictory data, touting authoritative opinion is seen as a way to influence important meetings (the Royal Society obviously thinks the same), although it must be said that when it comes to opinion the Met Office track record for accuracy is not shining.

Just four lines of information released in the Met Office’s brochure attracted most of the attention in the media. It seems that there is a case to be made that ocean temperatures need to be adjusted. Prior to about 2002 they need to be lowered, and post 2002 they need to be raised slightly.

Those four lines were;

Changes in the way sea-surface temperatures were measured

over the last decade have introduced a small artificial cooling

of up to 0.03 °C over the last decade. This is being corrected

in a new version of the Met Office dataset.

The reference for the statement was given as; J. Kennedy, R. O. Smith and N. A. Rayner, 2011. Using AATSR data to assess the quality of in situ SST observations for climate studies, in press Remote Sensing of Environment.

It would be fair to say that the most inconvenient truth in climate science at the moment is that the world has refused to warm in the past decade. That includes the land as well as the oceans and the scientific literature is replete with research that arrives at this conclusion. It’s a topic we have discussed several times in the Observatory, most recently here. Obviously given the importance of such a finding that the ocean temperature dataset needs adjusting it is important to check, and recheck, the data on which it is based.

This is what the Met Office has done showing that recent warming may have been as much as 0.03 C per decade larger than previously thought. But does it matter, and does it justify the headlines?

Despite the unequivocal headlines no mainstream environmental journalist (in the UK at least) did anything other that repeat those four lines, and the associated comments on the Met Office’s press release. Indeed, when contacted for the scientific paper on which those four dramatic lines are based the Met Office Press Office didn’t have it and had to scramble to track it down.

Measuring Temperature

The research paper deals with different ways to measure the sea’s temperature, from ships, buoys (drifting and moored) and satellite-based observations.

They all measure different things. Temperature measurements from ships are the most variable. Some have done it by lowering a bucket (sometimes a specially designed one), which is raised, and the temperature measured and recorded with the time and ship’s position. Some ships measure the temperature of the water engine intake that comes from a different water depth and is specific to the design of the individual ship. Buoys are specially designed to take meteorological readings and sea temperatures but until recently were of a mix of designs each with their own idiosyncrasies and errors. Satellite observations (looking at the infra-red spectra of the ocean) measure something different, the temperature of a very thin slice of the ocean’s surface. When compared to the other data satellite observations have to be converted to ‘bulk temperatures’ which is a non-trivial process with scientific problems of its own.

To investigate the relationship between ships, buoys and satellites the researchers take the satellite data as the most accurate and (taking only night-time satellite observations in the first instance) then look for simultaneous ship-satellite observations as well as simultaneous buoy-satellite data between 2002 – 2007.

The satellite temperature data has an average scatter of 0.14 C. Ship data are less accurate with a scatter of 0.71 +/- 0.74 C when compared to the satellite data. Buoy data are better with a scatter of 0.29 +/- 0.26 C when compared to satellite data. None of these figures are surprising, or particularly new.

Taken together these figures suggest that, when compared to (processed) satellite data buoys tend on average to read cooler and ships warmer. According to the researchers this means that ship temperatures must be depressed and buoys raised. In addition the increasing number of measurements of sea surface temperature from ocean buoys and the decreasing proportion of measurements from ships since 1980 should be taken into consideration.

As the number of buoys has increased, the proportion of ship measurements has fallen. The researcher’s Fig 13 shows this. Click on image to enlarge.


The top graph shows satellite data (labeled ATSR) intermittently from 1991. Night-time satellite data is used post 2002, whilst daytime satellite data has been used 1992 – 96. The solid line is the ocean temperature uncorrected whilst the dashed line is the corrected curve.

The correction has been done with reference to the second graph which is the percentage contribution from buoys shown up to 2008 that has been rising steadily up to 2006, with particularly strong rises 2005-2006. This produces a sliding scale adjustment that gets steadily larger up to 2006.

Upper Limit

The researchers conclude that this difference spread across the globe and over the years is sufficient to add a warming of 0.03 C per decade to the HadCRUT surface temperature record. Despite the impression given in the media this is a small correction. It should also be noted that the 0.03 C is very much a statistical upper limit on the purported shortfall in warming. It assumes that the bias in global average sea surface temperature is on the large size of estimates and that the sea surface temperature contributes around 70 per cent of the average global surface temperature.

Superficially then, one can say the temperature in the past decade has been adjusted upward and therefore the oceans have warmed more than was realised. That view however does not take into account the variability in the data, which should not now be ignored, as that was the whole point of the exercise in the first place.

The correction is smaller than the inter-year variability and does not change the impression that there was no oceanic warming before 1997 and after 2002, after which there is if anything a slight cooling. Also note that this lack of warming occurred when the percentage of buoys rose from 40% to 80% of the data set and the cooling when the percentage of buoys remained constant at about 80%.

To my mind the new corrected data tells us nothing new and nothing that the satellite data when taken in isolation (it is after all claimed to be the best data) hasn’t already revealed.

When the errors in measurements and the scatter in the data are taken into consideration the adjustments, if confirmed and accepted, do not make much difference to way the global average annual temperatures have changed in the past twenty years and in fact confirm the non-warming of the oceans in the past decade.

So the media headlines could have just as accurately have read ‘New Met Office data confirms no warming of the oceans in past decade.’ But that would have meant abandoning journalistic acceptance of authority statements, as well as reading beyond four lines in a brochure.