HadCRUT 4.6 temperatures compared

CRUTEM station data for October was released a few days ago. So I can now make a direct comparison of different surface averaging methodologies. The calculated October temperature anomaly for the 3 different spatial averages, as described previously are the following :

Classical (5×5 Lat,Lon Grid) 0.57C
Icosahedral (2562 node, 3D triangular Grid) 0.66C
Spherical Triangulation (3D) 0.72C

Figure 1: Comparison of the 3 averaging methods. Bottom graph shows the differences between the 3D methods and the HadCRUT4.6 classic 5×5 latitude , longitude binning

The 3D methods tend to give higher values for recent years and accentuate the peaks and troughs. The reason for this is the way polar regions are handled. Spherical triangulation essentially interpolates nearby stations right across polar regions by forming large triangular areas. Icosahedral binning however treats polar regions identically to any other region on the earth’s surface.

Figure 2. Comparison of Polar processing by on the left Spherical Triangulation and on the right Icosahedral binning.

Figure 3. The same comparison over Antarctica. Note how there is no interpolation into areas without measurements.

I discovered earlier that the Spherical Triangulation method gives almost exactly the same result as Cowtan and Way, who use kriging to extrapolate into polar regions. Both methods essentially interpolate temperatures into poorly covered regions. Icosahedral binning avoids any interpolation,  while treating polar regions in an unbiased way. It should really be compared  to the official HadCRUT.4.6 version, which is based on a regular 5 deg. lat, lon grid.

Figure 4: HadCRUT4.6 gridded results for October 2017

It is often argued that HadCRUT4 has a coverage bias in the Arctic due to using fixed sized bins, which converge in area with latitude ( as cos(lat) ). However all other groups who attempt to correct for this, interpolate results or smoothed fits into regions without any measurements. Rightly or wrongly HadCRUT4 remains the only pure measurement based result. I argue that the methodology would be improved if they moved to using Icosahedral 3D binning rather than sticking to 2D based binning. This would still retain the advantage of being a measurement only result.



This entry was posted in AGW, climate science and tagged , . Bookmark the permalink.

7 Responses to HadCRUT 4.6 temperatures compared

  1. MarkR says:

    I think the “measurement only” definition sounds nice but is arbitrary. Why does a 5 degree grid count as “measurement only”? Without referring to spatial correlation anyone could argue that (1) it’s not measurement only because measurements are only good to 20 metres, so we should average on a 20 metre grid or (2) 5 is arbitrary and 90 degrees is better.

    Extreme numbers chosen for illustration.

    • Nick Stokes says:

      I agree with that. Measurement only works at the points you measure. Sampling. Everything beyond that is interpolating, or inferring, from the sample. Triangular mesh does that systematically. Grids say that we’ll use sample values in an arbitrary local area, and if there aren’t any, we’ll use the global average (which is the effect of omission).

      IMO, to get a proper average, the best estimate should be used for every point. “Measurement only” just means that a local estimate is used some places, and a crude approx (global) elsewhere.

      • Clive Best says:

        Yes, we have to assume that there is some continuous temperature field across the surface of the earth which is then sampled at fixed points. On a daily basis a passing weather front will cause discontinuities, but we still must assume that monthly averages smooth this out so that regional climate changes slowly and continuously.

        I agree with you that a Triangular mesh is the best systematic way to infer the earth’s surface temperature. It does implicitly interpolate, but has the advantage that it handles the spatial geometry correctly, unlike kriging in 2D for example.

        The argument for binning is that you have a fixed grid which doesn’t change with time. Just how important that really is I don’t know. I suppose you could always project a changing triangular mesh onto a fixed grid. However if you want an unbiased grid which respects the earth’s geometry then the icosahedral grid is the one to go for.

    • Clive Best says:

      There is a basic assumption that underpins all ‘global’ temperature series, no matter how they are derived. Namely that all stations react in synchrony to warming (or cooling) on a local level. On land weather stations lie at different altitudes so cannot measure the same absolute temperatures even if they are only 10 km apart. However it is assumed that after seasonal averages are subtracted then the ‘anomalies’ behave the same.

      This assumption becomes less and less likely to be true the further away the stations lie from each other. Another point worth bearing in mind is that the agreement between different temperature series is because everyone uses a core set of underlying data (station and SST data). Small remaining differences are then due to methodolgies.

  2. ralfellis says:

    Just looking at Clive’s electrical generation meters, and solar is producing 2gw at 19:00 today. How is that possible…?


Leave a Reply