Large-scale temperature trends
Jun 5, 2012
Bishop Hill

This is a guest post by Rob Wilson. It addresses some concerns I raised when I spoke in St Andrews the other week. I had discussed Climategate and the lack of trust this engendered and then went on to briefly cover other issues that made me uncomfortable. In particular, I mentioned the tendency of corrections to the temperature records to produce cooling in the nineteenth century and warming in the twentieth, and the recent lack of warming.

In light of Myles’ and other comments w.r.t. instrumental data, I thought this might be a good time to quickly try and address some of Andrew’s observations that he made in his talk at St Andrews in April.

I had hoped to write a guest post along with colleagues from the Met Office showing temperature trends along with AR4 projections, but I already see the summer running away and what spare time I have, I would rather concentrate on a series of later guest posts focussing on dendroclimatology. So below, I concentrate only on temperature trends in the HADCRU and HADSST data-sets. Thanks to Ed Hawkins and John Kennedy for providing feedback.

There were two issues that Andrew raised:

  1. That updates of large scale temperature data-sets appear to depress 19th century and raise 21st century temperature values.
  2. That over the last decade or so, there had been a flattening off in temperature trends.

At the time, I could not comment on either as I had not looked at the new data-sets in detail.

So – the following link will take you to a series of figures that compares CRUTEM3 and 4 (land temperatures) and HADSST2 and 3 (SST) for northern and southern extra-tropics (ET) and tropical (TROP) latitudinal bands. References at end.

For those of you who want to check and replicate my plots, the data can be easily accessed from the Met office website:

or perhaps an even more user friendly site is:

I have purposely not added trend lines, or smoothing functions and have just plotted the temperature anomalies (w.r.t. 1961-1990).

I am not going to describe trends in exhaustive detail, but really want to address Andrew’s two main concerns.

Older vs newer temperature data-sets

There has been little change in the NH ET input data-sets. The major changes I am aware of are some early instrumental corrections of temperature data-sets in the Greater Alpine Region. However, this is only a small number of records in the extensive NH data-set so does not impact the large scale mean series. For those interested, the Alpine data-set correction is detailed here:

Böhm R, Jones PD, Hiebl J, Frank D, Brunetti M, Maugeri M (2010) The early instrumental warm-bias: a solution for long central European temperature series 1760 – 2007. Climatic Change 101, 41-67.

For TROP and southern hemisphere ET land temperatures, the major changes are in the 19th century which reflects the addition of newly digitized station records – probably mainly from Australia. Early instrumental temperatures are always going to be less certain and there is less data.

Changes in the late 20th century appear to be minimal.

w.r.t. SST, again little difference between HADSST2 and 3 in the ET NH.

The period of greatest difference in the TROP SST data is around the post 1940’s period which are related to biases in HADSST2 w.r.t. an “uncorrected change from engine room intake measurements

(US ships) to uninsulated bucket measurements (UK ships) at the end of the Second World War.” These have been adjusted in HADSST3. For those interested, the relevant paper is:

Thompson et al. (2008) A large discontinuity in the mid-twentieth century in observed global-mean surface temperature. Nature 453, 646-649

For the southern hemisphere ET SSTs, we again see similar corrections in the 1940s as in the tropical SSTs, but interestingly, the HADSST3 are actually marginally “colder” in the recent period than HADSST2. I only highlight this to show that correction can go both ways.

As a final note, correction for homogeneity biases in temperature record is very important and if you want more information on the basic theory, a really good review paper is:

Peterson,T.C. et al, (1998). “Homogeneity adjustments of in situ atmospheric climate data: a review.” International Journal of Climatology, 18 1493-1517


The recent flattening of temperature trends

As for the recent flattening. Well this appears to vary markedly. For NH ET winter temperatures, there is clearly an “eye-ball” flattening in winter temperatures, but likewise, a continued increase in summer temperatures. Tropical land temperatures appear to show continued warming for all seasons, but tropical SST records could be argued to have flattened. SH ET land temperatures is a little mixed – perhaps a flattening in summer, but still increasing in spring and autumn.

Statistically, due to internally forced multi-decadal variability expressed in all of these records and the fact that we are “at the end of the time-series”, I think it is really very difficult to “quantify” a flattening or even a continued increase. Yes, we can fit linear trend lines to the latter end of the time-series, but with the known naturally forced decadal variability expressed in these data-series, I personally think that such exercises are really not very helpful. This issue will simply become clearer over the next 10-20 years............but should we wait until we have statistical certainty?

Final thought

So – my take home message. Let’s not generalise too much. The newer HADCRU4 and HADSST3 data-series are incrementally improved data-sets using new data and utilising corrections related to robust theory and methods. Many on this blog will disagree with this statement, but all I can urge is please read the papers below. Much effort is focussed on the uncertainties and biases in these records. I do not see a systematic change (between old and new) to cooling (warming) of early (late) large scale instrumental series – rather I see improved data-sets (HADCRU4 and HADSST3) with well documented uncertainties.

As for temperature trends, in the same way that it does not really matter if the medieval period was warmer or cooler than today, it does not really matter if a particular seasonal time series shows an increase or flattening in temperatures. What MATTERS is that we need to understand the drivers of these changes. Natural or anthropogenic (or a mix of both). CO2 cannot explain all trends since the 1850s, but likewise internal dynamics (PDO, ENSO, NAO etc) or changes in the sun or large-scale volcanic events cannot alone explain the variability in climate.


Jones, P. D., D. H. Lister, T. J. Osborn, C. Harpham, M. Salmon, and C. P. Morice (2012), Hemispheric and large-scale land surface air temperature variations: An extensive revision and an update to 2010,

J. Geophys. Res., 117, D05127, doi:10.1029/2011JD017139.

Kennedy J.J., Rayner, N.A., Smith, R.O., Saunby, M. and Parker, D.E. (2011b). Reassessing biases and other uncertainties in sea-surface temperature observations since 1850 part 1: measurement and sampling errors. J. Geophys. Res., 116, D14103, doi:10.1029/2010JD015218

Kennedy J.J., Rayner, N.A., Smith, R.O., Saunby, M. and Parker, D.E. (2011c). Reassessing biases and other uncertainties in sea-surface temperature observations since 1850 part 2: biases and homogenisation. J. Geophys. Res., 116, D14104, doi:10.1029/2010JD015220

Morice, C. P., J. J. Kennedy, N. A. Rayner, and P. D. Jones (2012), Quantifying uncertainties in global and regional temperature change using an ensemble of observational estimates: The HadCRUT4 dataset, J. Geophys. Res., doi:10.1029/2011JD017187, in press.


Article originally appeared on (
See website for complete article licensing information.