November 2013 - A research paper published this week looks at different methods of using data from the HadCRUT4 global temperature dataset, which is compiled by the Met Office and the University of East Anglia's Climate Research Unit.
The paper looks at a well-known issue - namely that some parts of the globe are very sparsely observed and this creates uncertainty in the global temperature records.
There are various ways of dealing with this issue.
HadCRUT4 tackles this by using observations alongside estimates of the uncertainty in the global and regional temperature series that arises from incomplete coverage of the planet. These uncertainty estimates sit alongside estimates of other sources of uncertainty arising from factors such as changes in observing methods and urbanisation.
These measures of uncertainty can be useful for researchers since, for example, HadCRUT4 global temperature maps indicate where observations are not available, as well as how certain we are of temperatures where we do have observations.
In the new paper by Kevin Cowton and Robert G Way, two different methods of dealing with incomplete global observation are considered in relation to the data available in HadCRUT4.
Both of these focus on so-called 'extrapolation' - where sparsely observed parts of the planet are 'filled in' by using existing data. The first method does this by using nearby observations; the second method uses satellite measurements of temperatures higher in the atmosphere as a guide to temperatures near the surface.
The paper applies these methods to the HadCRUT4 data to provide a new estimate of global-average surface temperature change from 1979 to 2012.
The results suggest that, in comparison to HadCRUT4, there were cooler global average temperatures in the late 1990s and warmer global average temperatures in recent years, resulting in an overall larger warming trend globally from the late 1990s to present.
It's suggested that much of this additional global average warming relates to the relatively sparsely observed high latitudes of the Northern Hemisphere which have been warming at a faster rate than other regions.
Comparing the results of the study to trends in HadCRUT4, there is very little impact on the trend from 1979 to 2012.
The new paper suggests that the global average warming trend from 1997 to 2012 could be up to 2.5 times greater than the trend using HadCRUT4 alone. The estimated trends from the paper are also slightly larger than those from other global average temperature datasets that also use extrapolation techniques (such as NASA GISS).
Of course, before drawing any firm conclusions about what this means for recent warming trends, it's important to look at the range of evidence available from other studies - which each have their own methods, assumptions and uncertainties.
The new paper provides additional evidence that, as has long been suspected, the lack of observational coverage of the Arctic region could mean that the relatively rapid warming occurring in most recent decades has not been properly captured by global temperature records, and this may bias global warming rates from the late 1990s to 2012 towards lower values.
However, there is strong evidence that the region of the Earth that is well observed, has been warming less rapidly, and that this is partly related to ocean variability.
Furthermore there has likely been a small cooling effect from decreasing output from the sun as part of the 11-year solar cycle, and from volcanic eruptions, both of which have slightly offset the warming from greenhouse gas emissions in the last few years, even in the global mean.
Similarly, it's important to remember that global mean temperature datasets are just one component of the evidence of a changing climate and it's important to consider all the available information.
Indicators such as continued sea level rise, melting of snow and ice, and warming of the upper oceans have consistently shown that our climate continues to warm.
Last updated: 20 November 2013