Does anyone know the "life-cycle" of the averages used for daily and yearly temperatures and rainfall by the National Weather Service?

I have lived in the same (semi-desert) area for over 40 yrs, and have seen the yearly rainfall average go from 13 inches per year to 24 inches per year and back down to 18 inches per year over that period of time.

Unfortunately, I failed to pay attention to the specific "whens" of this phenomenon...

Simple math tells me that this is NOT possible if we are using total data - only possible if we use an interval. But, I cannot find any reference to such an interval ... . Same things I have been seen with average temperatures ... .

  • $\begingroup$ No need to write your question in all caps. THAT'S SHOUTING. I edited it out $\endgroup$ – Jan Doggen Feb 23 '17 at 8:29
  • $\begingroup$ Perhaps he is channeling his National Weather Service self :-p $\endgroup$ – JeopardyTempest Feb 24 '17 at 4:23
  • $\begingroup$ (Noting you're from the Netherlands, perhaps that's not familiar to you. But traditionally that's how weather forecasts/discussions were transmitted. The Storm Prediction Center in the US just finally changed over) $\endgroup$ – JeopardyTempest Feb 24 '17 at 4:24

Climate normals are computed over 30 years. Not sure why that number of years was chosen specifically. Perhaps a recent incentive is to identify climate changes (though it actually would make warming appear less significant when looking at anomalies). But, given that those trends have only gotten interest in the past few decades, the only historical reason I can fathom is to sooner allow more sites to be used. 30 years is a fair number of data points statistically, but for daily normals, it should still have a significant error range. It would seem a 50 or mean year normals length might be more useful at this point, though indeed climate changing might be comparable to the confidence interval.

But regardless of why they chose that value, that should be the period they are reporting.

If you wanted to verify these values, I was able to find pages to get the 1961-1990 normals (use the find command within this file to determine station id), 1971-2000 normals, and of course 1981-2000 normals.
Or a more robust option would be to use NWS' great ClimateNow data portal (select your local NWS office). There you can look at annual data for the entire record length, search for specific values, or compute normals for any period you wish (select your station, then Monthly summarized data, and then the mean at the bottom is your normal over the time period you selected) (can also change to avg temp).

I know these portals can be a bit complex to some, so if you wish to mention the area, I can pull the values and edit the answer. And then you can see just how much the normals have changed :-)

| improve this answer | |
  • 1
    $\begingroup$ Average is not necessarily normal. It irritates me when people (particularly weather forecasters & reporters who should know better) don't seem to recognize that average and normal are two very different things. I live in the Sierra Nevada, where we get an average X inches of snow/rain a year. But it's pretty rare to actually get anything close to that average in any given year. We usually either get considerably less, or (like this year) considerably more. $\endgroup$ – jamesqf Feb 23 '17 at 20:00
  • $\begingroup$ Indeed a distribution is the truest representation of climactic possibilties. But the public doesn't seem to great with that amount of information. I'd just as much like to see confidence levels presented better with forecasts. But good luck to us :-p $\endgroup$ – JeopardyTempest Feb 24 '17 at 4:22
  • 1
    $\begingroup$ Thanks, Tempest!!! 30 yrs is what I thought, but had no proof... And thanks for the portals!!! $\endgroup$ – S. E. Charles Feb 24 '17 at 14:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.