In article , Jeremy Handscomb
writes
Hi all,

This morning my max/min thermometer showed that last night the temperature
dropped to minus 3 - it was minus 3 at the time I do my observations too
(8am), so when I came home from work I realized that because tonight is
likely to be less cold, I will get a 'false' reading from my max/min
thermometer. It will say the coldest temp in the last 24 hours was minus 3,
but what I really want to know is tonights minimum - I have reset the
thermometer so I can know, but was wondering what the standard convention is
if the minimum (or indeed maximum) temperature coincides with the time of
observations?

Regards,

Jeremy

(Hope this post doesn't get swallowed up by all the excitement!)



Hi Jeremy,

Like many I've been with this dilemma myself a few years back. The 24-hr
convention dates back over a century when manual readings were even
difficult to do once a day in some parts of the world and had to be done
at a reasonable time too. Of course for the sake of records this
convention has stuck, nothing will change in that regard. However as you
have experienced it has great big holes in it as far as accuracy is
concerned, you have very nicely highlight this. On the 16th at obs time
you read -3.0C, on the 17th, this a.m you probably have read something
like +1 or 2 degC (like here in Coventry for instance) this was the TRUE
minimum on this day Thursday 17th Feb 2005. But the 24-hour system would
have shown it as a -3.0C for today. ABSOLUTE RUBBISH!!

In real accuracy terms it was indeed rubbish, but in met office protocol
terms it was in fact correct according to the 24 hr format. You have
come across this problem from a temperature point of view, have you
thought about rainfall?? In fact it was my rainfall recording which
really got my beef up. Many a time I was recording rain say for example
this a.m from 0400-0800, this day the 17th. But because of the 0900-0900
convention that rain actually is to be recorded on YESTERDAY'S date,
CRAZY!!
I got so disillusioned from an accuracy point of view I changed as I
know others do also, like Tudor who has responded in this thread to
another time scale. I do my readings from 0900-2100 for max temps, and
2100-0900 for minimum temps. Using this system, MOST of the time you
will have the correct min for example and also the correct max. Of
course this also is not perfect, esp in winter when after a cold spell a
warm front moves in and night time temps are rising to a max after 2100,
but on the whole it is far more accurate than the 24-hour format the WMO
use to compile their statistics.

Most times a 24-hr and a split 12-hr will show exactly the same
readings, in years to come wouldn't it be nice to look back on the 17th
Feb 2005 and see that the min was in fact 1.4C and not the 'official'
false figure of -3.0C which in fact belonged to the day before.

Don't forget you can keep both systems going for a while and see how
many times anomalies occur. My personal pref of course is to use a
split 12-hour format. Sorry if I've gone on a bit here, this was
causing me great moral anguish in times past, recording false data and
for the wrong days record too!

Just my tuppence,

Keiron Carroll
--