Climatological Averaging Periods

[Thundery wintry showers]

The World Meteorological Organisation uses 30 year averaging periods to filter out year-on-year variability, and updates the 30 year averages every 30 years but many countries also update them each decade.

Having seen the comparisons between the UK's 1961-1990, 1971-2000 and 1981-2010 averages, I'm left wondering if maybe 30 years isn't long enough to filter out year-on-year variability? For example, the 1961-1990 period contained anomalously cool Aprils and Mays and anomalously warm Octobers, 1971-2000 had anomalously cool cloudy Junes, and 1981-2010 had anomalously cool Decembers, particularly assisted by the period both starting and ending with an exceptionally cold December.

I think when we're comparing recent weather to the "average" that we're accustomed to, it makes sense using the most recent 30-year average, but for long-term climatological analysis I'm starting to wonder if we'd be better off using a longer period like 1951-2000, or even 1901-2000 for cases where records go back that far, to filter out the year-on-year variability that evidently skews some of the 30-year averages. In view of the fact that we have less in the way of reliable records prior to 1951, I'm currently quite attracted to the idea of using 1951-2000.

The question first started coming to my mind when I saw discussions on Hadley refusing to update to 1971-2000 in their "CET diagnostics" analysis and thinking that updating the reference period each decade for a climatological analysis risks drowning out climate change, but using an "antiquated" reference period like 1961-1990 feels a bit irrelevant to the present, so neither are particularly ideal.

[Iceberg]

30 years is choosen as thats the original timeframe thought for the first generation of climatic ecosystem based succession to occur.

I.e climatic systems are based into generic types for a wet tropical forest to succeed into a dry tropical forecast is thought to take 30 years based upon only climatic factors.

30 years is of course only an average and some of the biggest longer lived species can take 1000 or so years to migrate.

This is all pretty old stuff and might be updated by now.

[Terminal Moraine]

I remember reading somewhere that 30 years was used as it was thought that such a period would encompass 90% of the temperature and rainfall variability at any given place likely to be experienced over a longer period of time.

This however relied on the climate being stable, something we now know is not the case.

Having kept records for almost 50 years myself I'd agree that a 50 year period is better at filtering out short term anomalies and diluting the effect of a single very cold/warm/dry/ wet month, in fact the longer the period of average the better it is at this.

If the climate is changing, in whatever direction, the magnitude of that change is better shown by referring to a long period average than a relatively short period recent one which is constantly updated. The latter still remains useful however as an indicator of anomalies from current conditions.

[AderynCoch]

I myself prefer to use long-term averages, as it's interesting to see how our climate has changed over the centuries. For the CET, this means using a stupendous average of 1659-2011. Although it covers a relatively small area, the sheer length of the recording period makes for some fascinating insights. For instance, a month like December 2010 is all the more remarkable when you consider only one December in 352 years of records was colder, and of course the April just gone was far warmer than any other April in the same lengthy period.

I've collated the CET monthly and daily data on a spreadsheet (I should get round to the max/min data at some point), and shaded red those months above the 1659-2011 average and blue those below. It's noticeable scrolling down the years how the preponderance of blues gives way to reds.

[Methuselah]

I see no problem with 30-year means per se: they are what we know even though they may be entirely arbitrary, but I can see no need for change. However, wouldn't adopting 10-yearly adjustments make the 'grain' in the records finer??

[VillagePlank]

I'd go for 70 years (at least for the CET)

On my blog, here, I decompose the CET signal into it's constituent sinusoids and the top 4, in terms of magnitude, and therefore variance, add up to 70 - a happy coincidence.

I'd also make sure that the average was 70 years, 70 years ago (so this year we'd use 1871..1941) and update every 70 years. Here's what it looks like, and I've added the 70 year moving average as well:

Data is from the MetO, the 1871..1941 mean is 9.24 to 3 sig-figs. Pivoted data for easy use in spreadsheets and database is attached

cet_data.txt

Edited June 8, 2011 by Sparticle

[ZONE 51]

When i think about climate patterns i always think that only great time periods can show the bigger pattern, but i do also think that 5-10 years can show small changes, these small changes occuring frequently over decades on and off depending on what other factors are to play during those periods of time, not only that, i think delayed reaction in the atmosphere can confuse things, as what has happened in the past can affect the future, also what effects from what we do having an effect on the climate and of course changing the climate patterns of what would otherwise be without our doings here on earth, that of course does not include earths natural effects like volcanoes, and the main one being the sun.

so all in all i do think long periods of time would show what we need to know, so 100years as a rounded off figure would be interesting to me, but i would and have looked at shorted periods. i do think though that its important to look into the effects of man-made factors adding to the change in the pattern of the climate. what ever we do we would effect the climate in some way i would have thought, but i do believe a great pattern is to be discovered thats looped over many many decades and hundreds of years, this pattern altered by what we do on earth. patterns altered by what we do on earth and changed by natural factors to. its a compicated process and something we need to continue to study with great importance.

Edited June 9, 2011 by ElectricSnowStorm