Thanks for the responses:

Articles by Ian Strangeway (formerly CEH), amongst others, suggest that
gauges should be flush with the ground - with a grid, as Steve suggests to
reduce splash-in. Failing that the gauges should be as aerodynamic as
possible.

The gauges here are the Young's 0.1 TBRs, and therefore are not terribly
aeordynamic. They are spaced 2m apart - so unlikely to be local effects
affecting the totals. Yes, the sensible thing would be rearrange them to
take into account any individual gauges inaccuracies... .

At the end of the day, as I've found elsewhere gauges close together can be
10-20% out quite easily: I believe the NWS recommend three gauges for each
site now...

Chris



"Jonathan Stott" wrote in message
...
Chris Kidd wrote:
It's probably been discussed before, but...

The general idea is that rain gauges should be as flush with the ground
as possible to ensure that the catch is affected as little as possible by
the wind. Just to test this I put three idenitical TBR (Youngs) gauges at
rim-heights of 40cm, 100cm and 160cm above the ground. From 12/11/07 to
29/01/08 they collected 210.4, 213.2 and 216.5 mm respectively. Um...

Speaking as someone with a science background (and not necessarily in
climatology or whatever), I would be performing the same test with the
same gauges but changing the order of heights. Repeating this over several
observations (of, say one month) and trying all combinations will minimise
the effect of any difference in calibration that the gauges have. You may
also have localised effects, so trying the experiment in many different
locations and analysing the statistics over all places will help to
minimise that effect over time.

What you can't do is to do it once and take one set of results as
verbatim. The variability of precipitation could make accurate statistical
analysis difficult though.

--
Jonathan Stott
Canterbury Weather: http://www.canterburyweather.co.uk/
Reverse my e-mail address to reply by e-mail