Weather Recording time -Any affect on averages?

This forum is for comments about the weather spanning several days, developing phenomena eg drought, comparisons with previous years, local climatic features eg Cheshire Gap showers, and anything that does not fit any of the other forums.

Weather Recording time -Any affect on averages?

Postby Graham Easterling » 2014 Mar 28, 11:36

I record maximum/minimum temperature and rainfall at 09:00, with the rainfall and maximum temperature data cast back.

There are 2 'standard' times for recording, 09:00 and 18:00, with midnight becoming increasingly used in the AWS era.

If we take 27th of March, the maximum temperature in Penzance would have been using:-

18:00-18:00 8.1C
09:00-09:00 7.0C (0.7C lower than anything recorded over the winter and the lowest max since 6.7C on 1/4/13!)
The midnight to midnight max would have been lower than either of these figures - approx 6.5C.

I've always assumed that the period used makes no real difference to average temperature, as long as you are consistent, but is this actually the case? Have parallel recordings at 18:00 & 09:00 been undertaken?
Graham

Penzance Weather Station
Grid Ref:
SW464231 - Post Code: TR18 4TP
19m AMSL - Aspect SSE
In a SE - NW orientated valley

23 Years of Penzance Weather Records : http://penzanceweather.atspace.com/weather.html
Graham Easterling
 
Posts: 316
Joined: 2011 Nov 30, 20:29
Location: Penzance, Cornwall

Re: Weather Recording time -Any affect on averages?

Postby greg_gruner » 2014 Mar 28, 21:44

I have no data to back this up (a study is needed!) but my thinking is as follows:

The minimum temperature, other factors being equal, is around sunrise. In winter this is at around 0800 in southern England, but closer to 0900 in the north and north west. Splitting the "day" at the time of mimimum would lead to lower average mimima, since if a cold night is followed by a warm night, the 0900 minimum becomes the minimum for 2 days. In summer this would less noticeable as the minimum is reached at around 0500. I would therefore conclude that average winter minima would be lower 0900-0900 than for 1800-1800 or 0000-0000.

Regarding 1800 readings, the problem would be more in summer, when the maximum temp is often at around 1600 and tempearture is still quite high at 1800 on a warm day. If this is followed by a cool day, the highest in 1800-1800 would be the previous day's 1800 reading. This would lead to higher average maxima than 0900-0900 or 0000-0000.
Greg Gruner
Farnborough, Hampshire
greg_gruner
 
Posts: 224
Joined: 2012 Feb 04, 17:28

Re: Weather Recording time -Any affect on averages?

Postby Stephen Burt » 2014 Mar 29, 19:08

The answer is - yes, quite a big effect - comparable to or larger than calibration errors or urban heat island effects.

I devoted an entire chapter in my book to this under-appreciated major cause of inhomogeneity in climatological records. I examined 09-21/21-09 and 00-00h periods against 09-09, for maximum and minimum temperatures using 10 years data from my own site. (There's even bigger attribution issues for rainfall, of course, but mandating a move to reading manual raingauges at 00z would certainly not be popular!)

Over the year as a whole, using 09-21h max vs 09-09h reduced the annual mean by 0.09 degC, while for minima 21-09h vs 09-09h were 0.29 degC higher. The greatest differences occur in the winter months, of course - the December mean maxima were 0.33 degC lower and mean minima 0.94 degC higher than their 09-09h means.

Comparing 00-00h with 09-09h, mean maxima were 0.04 degC lower (little seasonal variation here) but mean minima 0.35 degC lower, the greatest differences here being in summer (June 0.47 degC lower). The full table appears on p. 277. I think we'd all agree that half a degree difference in averages is significant if trying to compare two sites, or one site over different periods with varying terminal hours.

In terms of frequency, the 00-00h min differed from the 09-09h min on 172 days per year on average - almost one day in two, increasing to two days in three in January.

So yes - your choice of terminal hours does make a very big difference to your averages and, particularly in winter, to your extremes too. I'm not a great fan of 09-09h other than for historical continuity, but would not advocate a change unless statistics for both terminal periods were prepared and overlapped for a long period. For what it's worth, I suspect that automation will eventually drive a move to 00-00h, though.

Stephen
-----
Stratfield Mortimer, Berkshire - central southern England
51.4°N, 1.0°W, 60 m AMSL, station grade A - AAAA47R
Records commenced here 1987 - local records available back to 1862
The Weather Observer's Handbook: www.measuringtheweather.com
Stephen Burt
 
Posts: 247
Joined: 2011 Dec 02, 19:36
Location: Stratfield Mortimer, Berkshire; a well-exposed rural site, 10 km SW of Reading

Re: Weather Recording time -Any affect on averages?

Postby Graham Easterling » 2014 Mar 29, 21:14

I devoted an entire chapter in my book to this under-appreciated major cause of inhomogeneity in climatological records


Indeed you did Stephen, and rather embarrassingly I have the book! I knew I'd read comparisons of differing observing hours somewhere, but couldn't remember where. I had hoped my question would get a response and point me in the right direction, it did, thanks!
Graham

Penzance Weather Station
Grid Ref:
SW464231 - Post Code: TR18 4TP
19m AMSL - Aspect SSE
In a SE - NW orientated valley

23 Years of Penzance Weather Records : http://penzanceweather.atspace.com/weather.html
Graham Easterling
 
Posts: 316
Joined: 2011 Nov 30, 20:29
Location: Penzance, Cornwall

Re: Weather Recording time -Any affect on averages?

Postby John Wilson » 2014 Apr 04, 15:38

For what it is worth, this phenomenon has been studied in the (relatively) distant past.
See Tinn AB 'Variations in daily maximum and minimum temperatures owing to difference in hours of setting of thermometers' QJRMS 1945; 71: 421-4.
Arnold Tinn lived at Woodthorpe, Nottingham, not far from my home. Between 1937 and 1944 he used two sets of thermometers, one using the usual 09.00 to 09.00 regimen and the other set recording the maximum between 09.00 and 18.00 and the minimum between 18.00 to 09.00 next morning. He found considerable differences between the 'orthodox' readings and the 09.00/18.00 set.
Apologies to Stephen if this study is mentioned in his book. I MUST buy a copy!

Regards to all

John
John Wilson
COL station id44018
John Wilson
 
Posts: 23
Joined: 2012 Apr 09, 21:28
Location: Arnold, Nottinghamshire, just off Mapperley Plains. 115m. amsl

Re: Weather Recording time -Any affect on averages?

Postby Dick Lovett » 2014 Apr 06, 10:52

The Met Office did an investigation of temperature extremes at 38 stations over the period 1957-70, comparing 09-09 max's and min's with those of 09-21 for max's and 21-09 for min's. Essentially, from Apr to Sept inclusive, mean differences were no more than 0.1C. However, differences gradually increased in the 'Winter' months such that Dec mean maxima were 0.4C higher when the 24 hour period was used, while Dec and Jan mean minima were both 0.7C lower.

Dick Lovett
Charlbury (Oxfordshire Cotswolds) 122m AMSL
51.87N 1.48W Station Grade C1 CC-A15 Records began April 1990
Dick Lovett
 
Posts: 16
Joined: 2011 Dec 10, 18:24
Location: Charlbury, 18km NW of Oxford. Sheltered site 122m amsl.

Re: Weather Recording time -Any affect on averages?

Postby Graham Easterling » 2014 Apr 06, 16:01

Thanks everyone for your replies.
Graham

Penzance Weather Station
Grid Ref:
SW464231 - Post Code: TR18 4TP
19m AMSL - Aspect SSE
In a SE - NW orientated valley

23 Years of Penzance Weather Records : http://penzanceweather.atspace.com/weather.html
Graham Easterling
 
Posts: 316
Joined: 2011 Nov 30, 20:29
Location: Penzance, Cornwall


Return to Everything Else Weatherwise



Who is online

Users browsing this forum: No registered users and 1 guest

cron