"Martin Rowley" wrote in message
...


... many thanks for that: would be interesting to find out how they
relate the CS to modern records - I know the Met Office are (have)
undertaken the same comparison. Using the standard WMO definition for
'bright sunshine' ( direct irradiance 120 W/m^2), it's study suggests
an rough 10% *reduction* (for the units used by the UKMO) wrt
Cambell-Stokes. I think most would agree that the C-S always did tend
(either through it's design or through difficulty of estimating
intermittent sunshine), to *over-estimate* sunshine amounts. This would
make the De Bilt record set this year even more remarkable, and in any
case, it confirms what we over this side of the North Sea have been
experiencing - a remarkable year for sunshine.

You are quite right to talk about a "rough" 10% reduction. The trouble
is that there is no way of emulating the CS recorder when you only have
a KZ (or whichever other) sensor. I'm not sure, having decided to
change the sunshine recording standard, why one would want to
emulate the CS instrument, but the Met Office certainly looked at
it seriously.

I've written elsewhere (often!) that the differences between the two
methods of measuring sunshine are not systematic, because the big flaw
of the CS recorder is its substantial over-recording when sunshine
is intermittent (there are other flaws, too, notably the requirement
for human interpretation of the trace on the sunshine card, and also
its inefficiency at recording sunshine early and late in the day). Thus,
on a day of intermittent sunshine a CS may record 50% more than
a KZ* sensor, whereas on a day of unbroken sunshine it may well
record up to 10% less. In a sunny month when a large proportion
of the sunshine comes on days of unbroken (or nearly so) sunshine,
such as March this year, it is clearly inappropriate to add 10% to
the KZ record to provide some sort of homogeneity at a long-
standing site. The same is true, to a lesser extent, over a year.

So to add 10% to the 2003 KZ sunshine records to make
comparisons with earlier years would, it could be argued, over-
estimate the duration of sunshine this year. In other words, it makes
it easier to break records. In a cloudy year, 10% would not be
enough, so it would be easier to break the "dull" records as well.

One can reasonably argue that this should be sufficient reason not
to try to homogenise to the two sorts of sunshine records, but there
will, of course, always be pressure to do so. My solution is that
the Met Office should encourage certain manned sites (say, a
dozen, at the very least) to maintain a CS recorder to provide
some sort of continuity for as long as it is possible so to do. It is
hardly an onerous task.

* KZ = The Kipp and Zonen sensor is the Met Office's preferred
new instrument.

Philip Eden