How long to get an average temperature/wind/precipitation etc?

Considering how many different weather setups that can occur ...how long would it take to get a true average temperature or wind or precipitation?

For example to get the extremes at each end of a histogram for each variable of wind or precipitation or temperature you need the various weather setups that create the extremes to occur across each week or day etc one in 100 year storms etc

And you need the extremes to record a true average

riffmongous

What do you mean by 'True' average? Why do you even want it? As in, what information does it impart and what use does this serve? An average can very useful but also very limited depending on your distribution

Gaoth Laidir

Thirty years is considered a good period as it pretty much does what you're saying. Obviously the once-in-lifetime events may not happen in every such period but 30 summers, 30 winters, are considered enough to give a true average.

PS. I don't like the word "normal' being used when it comes to weather. e.g. it's 2 degrees above normal. This has increasingly become understood by the general public to mean that anything that is not this "normal" value must be abnormal, when in fact it's normal to have wide variation about the seasonal 'average". Yes, statistically it is a normal distribution, however most people don't see it this way and believe that the weather is somehow "abnormal" and therefore changing if the e.g. temperature is not exactly bang on the climatological average value.

M.T. Cranium

Yep, we're stuck with the use of the term normal to mean 30-year average. Nothing is really normal about weather statistics, they are whatever they turn out to be and "normal" can only have some rather vague meaning, for example, it would be fair to say that it is normal for the winter months to be quite cloudy in Ireland, or for June to be hot and dry in the desert southwest, but when extending that to say today's normal temperature is 8 degrees, then it loses most of its meaning as there's a good chance the outcome will be significantly different from 8 degrees.

At least if we educate the users of forecasts to understand that normal means average for a recent 30-year interval, we might be a bit further ahead. The other question is how to define what 30-year interval should be the current reference point. In some countries, 1981-2010 is the widely accepted "normal" in use, whereas in at least the UK and perhaps to some lesser extent Ireland, it seems to be 1961-1990 which is already showing its age in this warming epoch. I expect the UK to flip over to 1971-2000 soon which won't be any improvement. In some of the contests that I help to organize, I give people a 30-year average as recent as possible, currently for December that would be 1989-2018 and in January it would be 1990-2019. This is converging on what will be the "new normal" soon to be in use, 1991-2020. A study of CET values showed that in that case, most of the 1991-2020 normals (averages) will be about 0.2 C higher than 1981-2010 averages.

Oneiric 3

I put this chart together to show the daily average maxima over Ireland on this day (Dec 17th) for the period between 1981-2010 (Note, the data begins in 2010 and goes back from that... as that is just the way I have my data sheet arranged.. and is more just for illustration)



The average (normal) national max temp for this day in this climate period is 8.5c, yet.. not on one occasion in this period was 8.5c recorded. A few close calls, but as the chart above shows, this 'normal' is generally made up of extremes on both ends, as has already been said above by M.T & G.L.

It really bugs me when you hear forecasters saying that 'temperatures are where they should be for the time of year', when really, there is no real 'should be' at all, because this is a highly misleading statement to make.

Gaoth Laidir

Yes, the term Long-term Average (LTA) or seasonal average should be used.