I am also dubious about the use of airports for temperature readings (Note this is a recent development). Airports can introduce a mini-UHI effect producing an artificially higher temperature.
http://www.niwa.co.nz/news-and-publications/news/all/niwa-co...
Graphs are used to answer a question: "what is the relationship between these objects" or "how are these things changing" or "what is the trend in their rates of change".
In this case, we're less concerned with the actual temperatures at the individual stations than we are with how they're changing over time. So, there's nothing wrong with adjusting and merging them together like that.
Think of it this way: what if we were to remove the temperatures from the left side of the graph, and instead plot the temperatures as a series of differences? i.e., temperature station 1 has +5 at one point, -1 at another point, and so on.
You could plot the changes in temperature for a number of stations that way, all on the same graph, and it would make perfect sense.
That's really all they're doing.
As for measuring temperature at airports: if there is such an effect, then it will produce uniformly higher temperatures, but it will not affect the rate of temperature changes -- so, again, as long as the question is, "how is the temperature changing", taking measurements at airports is acceptable.
This is not unlike a non-programmer being skeptical of Quicksort because it's "arbitrarily dividing data".
Hypothesize. Predict. Observe. Judge the prediction. Refine the hypothesis. That is science.
As for airports, this should be obvious. Many major airports have grown over the past several decades. And have seen major increases in air traffic. And have been situated in neighborhoods that have seen massive increases in development. All of these factors contribute to the urban heat island effect (more traffic, more tarmac, etc.) These effects can cause an increase in measured temperatures over time even if overall the local climate saw no increase in temperature. This is an important effect due to the very high proportion of weather stations placed in urban areas and at airports.
There's no gentle way to say this, but: you're trying to argue in a field in which you clearly have even less understanding than I do. The "adjustments" made to the graph do not make it any less accurate for what it is representing; it is not just fiddling around with numbers or making up data. It's a very simple, and very well-understood, way to represent data.
> These effects can cause an increase in measured temperatures over time even if overall the local climate saw no increase in temperature.
How does that even make sense? If the human activities at the airport location are causing an increase in temperatures at the airport, then that by definition is an increase in the temperature of the local climate.
If the temperature is simply slightly higher than surrounding areas, but the overall trend of the temperature is not, then it can still be adjusted into a graph with temperature data from surrounding areas using the exact same methods that were used in the linked example.
How does that even make sense? If the human activities at the airport location are causing an increase in temperatures at the airport, then that by definition is an increase in the temperature of the local climate.
Climate and local weather are not the same. If I build a parking lot the local temperature may increase by several degrees C on average, yet this represents an inconsequential change to the local climate within a radius of, say, 100km. More so, it represents an inconsequential change to the climate of the entire Earth. The significance of the urban heat island effect with relation to temperature records is that the proportion of weather stations which are affected by an increased urban heat island effect (due to increased urbanization) exceeds the percentage of the Earth affected by many orders of magnitude.
Yes, urban heat island effects can be accounted for, but I think the criticism is that for the most part they have not been.