Some time ago an FOI (Freedom of Information) request to the UK Met Office got the following response:
Background: Class 1–2 are regarded as excellent or very good exposures, Class 3 permits uncertainties up to about 1 °C, Class 4 permits uncertainties up to about 2 °C, and Class 5 permits uncertainties up to about 5 °C because of nearby artificial heat sources or other exposure problems.
According to analyses of the released data, approximately:
22% of stations were Classes 1–3, about 49% were Class 4, about 29% were Class 5.
I guess there must be very good reasons for not ensuring that the majority of stations are Class 1-3.
What's involved in ensuring a weather station meets class 1 requirements?
It's a theoretical optimum location for measuring whether temperature at a weather station is likely to be representative of average temperature across the surrounding area, with requirements for flatness of land, vegetation of the appropriate type and height, no shade between sunrise and sunset, no windbreaks and distance from reflective surfaces.
Has nothing to do with its suitability for use in longitudinal study, because whilst a site next to a massive concrete windbreak might be a degree or two hotter than other locations in a region, unless the concrete windbreak is a recent phenomenon, it can't be the cause of temperatures at that site being higher in the 2020s than they were in the 2000s or 1970s...