"John Dann" wrote :

Anyone know of a discussion - ideally available online - of the
preferred criteria for locating an observing site? (To be clear,
recommendations for the site itself, ie the physical geography, not
for instrument positioning).

snip

John, allow me to quote from the /Observer's Handbook/:
quote
Selection of site
To ensure that the observations are representative of the place
and sufficient comparable with those made at other stations to
permit their use in national or regional studies, the following
basic requirements are laid down for synoptic and climatological
stations.
(A) Outdoor instruments should be installed on a level piece
of ground, approximately 10m by 7m, covered with short grass
and surrounded by open fencing or palings to exluded
unauthorized persons ......
(B) There should be no steeply sloping ground in the vicinity
and the site should not be in a hollow. If these conditions are
not complied with, the readings of temperature and amount of
precipitation may show peculiarities of entirely local significance.
(C) The site should be well away from trees, buildings, walls
or other obstructions. The distance of any such obstacle (including
fencing) from the rain-gauge should not be less than twice the
height of the object from the rim of the gauge, and preferably
four times the height.
(D) The sunshine recorder, rain-gauge and anemometer must
be on sites with exposures to satisfy their requirements and they
need not be on the same site as the other instruments.
(E) As noted [elsewhere] the enclosure may not be the best
place from which to estimate the wind speed and direction;
another observation point, more exposed to the wind, may be
desirable.
(F) Very open sites which are satisfactory for most instruments
are unsuitable for rain-gauges. For such sites the rainfall catch
is reduced in other than light winds and some degree of
shelter is needed [refers to section on turf walls]
(G) If the instrument enclosure does not command a
sufficiently extensive view over the surrounding country,
alternative viewpoints should be selected for observations of
visibility.

The position used for observing cloud and visibility should be
as open as possible and command the widest possible view
of the sky and surrounding country.

In selecting a site the future should be considered as well
as the present. A good site may become a bad one because
of the growth of trees or the erection of buildings on
adjacent plots. Where the station is owned by an urban
authority it should preferably be sited on a scheduled open
space, and the layout of the remainder of the open space
should be such that the exposure will remain unimpaired
for many years.
end quote

These rules were written over half a century ago and
reflect (e.g. the open fencing) a rather different social
environment to today. But they give a good idea of what
was considered important.

Philip