What You Measure

It is a undisputable fact - but not one widely discussed out of skeptical circles - that the number of temperature stations over the world has changed with time, going up and down.

Not only the number of stations has changed, but also their geographical distribution - maps of these changes can be found at Climate Audit. As you can see, station coverage is dense in the USA, Europe, Japan and south-east China, but sparse in Central Asia, Africa, South America and of course the oceans.

So what does it matter, it's still a global temperature record, no?

It is in the sense that we do not have anything better. We are stuck with this, like or not.

However, if we wanted to produce a global temperature record over some extended area (for example, the land owned by a farm) what we need to do is to place calibrated thermometers in a regular grid; record metadata for each station and apply standard data recording and processing techniques for all stations. Hot and cold spots are almost sure to appear, and they should be explained properly before being either ignored or mixed with all other data without any adjustement.

What I wonder is if with all station changes and moves the temperature record of this year is the same thing as the record of fifty years ago. Because we have two different sets of stations (different as location and other conditions), which in part overlap, producing the global record.

What can guarantee that the two sets are measuring exactly the same quantity?