GHCN Part 6: Robe, A Tale of Bad Data

In my last post I discussed the station at Low Head and issues with the quality of the data. I also made some conjectures as to why the data is questionable. In this post I will discuss the site located in Robe, Australia. This time I won't be guessing as to what is wrong with the data.

Robe is yet another site which appears to invalidate the relationship between population growth or population density and temperature rise.

 

As before, I have transposed the data into standard deviations. It is pretty obvious the site measurements show some significant changes in measured temperature over time. A number of such shifts are evident. However, the major shift appears to happen between 1958 and 1971. As before, I will isolate those time periods.

 

 

 

I am not going to bother with a detailed statistical look at the data for reasons which will soon be obvious. The important information from this is in the pre 1958 time frame the temperature remained close to the average within minor oscillations. Post 1971 there is no appreciable change of any significance. In other words, no increase or decrease.

 

So, what happened? Let's look at the actual site itself.

 

 See the Stevenson screen holding the thermometers and other equipment? it is that white box visible behind the upper left corner of the gate.

 

And here is the view from the heavens courtesy of Google Earth. The little pointer misses it just a bit, but that ain't bad.

 

 So, let's go over what we are seeing here. First of all, the location changes since 1895 are probably pretty profound. I doubt they had Chinese restaurants and drive through banking back then. Or cars for that matter. There probably weren't any commercial AC units blowing hot air around back then either. I am betting development probably started after the war, but probably began really taking off in the late 50's. It probably reached a saturation point by the early 70's.

 

So, what do we have here? Simply put, a serious case of site contamination. While the population density is low, the development density is not. The factors of change are roads, parking lots, buildings, cars, AC units, and all the other things development brings. Remember my example from my last post of the black asphalt square? Same thing but with lots of hot exhaust added.

 

The bottom line on this is this station has been compromised to an extent where the data is just not usable. This situation exists on three certainly and probably four out of the four sites in Australia I have looked at in detail.

 

Which means, in all likelihood, I am down to just six sites.

 

What do you guess a good, detailed look at sites in the US is going to show?

 

Until next time.