A bit of a nerdy subject this one but it seems that the worlds weather services just can’t agree on a common period of thirty years to calculate the ‘normals’ they use to calculate their global temperature anomalies with. NASA and the University of Berkeley prefer to use the 1951-1980 period, NOAA uses the 1971-2000 period, HadCRUT5 and the JMA use the 1981-2010 period, and the ECMWF and University of Alabama would rather use the latest 1991-2020 period. Confusing? You bet. But who’s right? The WMO have ruled on this from on high, but as usual no one takes a blind bit of notice of what they say, infact the guidance they lay down on the matter is ignored by all seven major global temperature series. Here is what the WMO does say on the matter:-
Taking into consideration issues identified in The Role of Climatological Normals in a Changing Climate (WMO, 2007) and elsewhere, the Seventeenth World Meteorological Congress (WMO, 2015) endorsed a number of changes, which are reflected in the Technical Regulations, in definitions relating to climate normals. The most significant of these changes was that the definition of a climatological standard normal changed, and it now refers to the most-recent 30-year period finishing in a year ending with 0 (1981–2010 at the time of writing), rather than to nonoverlapping 30-year periods (1901–1930, 1931–1960, 1961–1990, and in the future 1991–2020) as was the case previously. However, the period from 1961 to 1990 has been retained as a standard reference period for long-term climate change assessments.WMO Guidelines on the calculation of Climate Normals
The ECMWF have just moved to using the 1991-2020 period for their long term averages of global temperatures in ERA5 and explain why they’ve done it:-
But as we move into a new decade, the WMO recommends calculating the new climate normals as soon as possible, although it is common for organisations to adopt a transition phase, during which they produce their regular reports based on both the old and new reference periods. The WMO advocates using a historical base period (1961-1990) for assessing climate change, as well as the most recent 30-year period, in order to standardise and harmonise across institutions. However, the 1961-1990 period begins several years before the satellite era and most C3S datasets do not extend back that far. So, in order to compare across a wide range of variables, C3S will use the 1991-2020 climate normal as the main reference period for the monthly climate bulletins from January 2021 onward and for the European State of the Climate for 2021.ECMWF
New decade brings reference period change for climate data
It does make a difference
As these graphics from the ECMWF show it does make a difference to how these maps are perceived by anyone viewing them. I wonder if anyone has looked at the underlying psychological differences these changes might make to the impact of the global warming message?
It’s easy to see why both the ECMWF and University of Alabama use the 1961-1990 period their data series don’t extend that far back, but as far as the other five series go, not one of them use the recommended 1961-1990 normals to calculate their global temperature anomalies with. You could cynically conclude the reason they do this is so that one can directly compare one series with another, and that’s probably also the reason why they only produce global monthly anomalies, rather than global monthly temperatures. One good thing from them all using different periods LTA is that I can produce a graph like this with little in the way of overlapping.