Tim, firstly congratulations on even attempting the herculean task of rationalising historical weather stations and identifiers.
I'm looking only at Australia. On a quick inspection, there are numerous entries that appear to be incorrect - I say "appear" because the information may be correct in the context of the database from which it came, but in the context of a database which lists both historical and current weather
stations it is wrong. Some examples, based as I say on quick inspection:
--Some ICAO indicators are listed that do not have, and never have had, weather stations, e.g. Kayrunnera. These are possibly locations used for the purposes of navigation, or tiny country or homestead airstrips.
--Many weather stations are listed as airports that are not airports (the point of your post), e.g. Mangrove Mountain.
--For some locations where there are two weather stations, both are listed as airports and have IATA codes, but only one is an airport and has an ICAO code, e.g. Orange
--Many weather stations are listed that record only rainfall, e.g. Louth
--Some locations have erroneous data, e.g. Katoomba has two stations listed, one with an elevation of 1030m, the other 305m which could not possibly be correct.
I will need to spend more time on this, in particular looking at the databases from which the material was sourced which I hope will be evident from the table or the description at http://www.weathergraphics.com/identifiers/