Many that pay close attention to the climate debate are aware that there are different choices of instrumental data to choose from. The ones that show lower temperatures get used by skeptics and the ones that show higher temperatures get used by warmists. This is a form of cherry picking. The person choosing the data gets to pick which one they use. In this way they feel they can make the stronger case.
Considering that there is a non-trivial difference between the sets of data I feel that this problem needs a solution. Since there is no standard, I decided to make one that I will use myself and make it available to anyone that wants to use it as well. Here are the four different sources I use for my merged or blended set. They are comprised of two station sets and the two satellite sets.
CRU – The original global set. It covers from 1850-current. It is a station set.
Hadley – A different global set that also covers from 1850-current. Also a station set.
UAH – This is a satellite measurement of wavelength. It has only been around since 1979, but covers most of the Earth instead of just stations.
RSS – Also a satellite measurement of wavelength. Also around since 1979, but covers less of the Earth, it especially misses portions around the poles, especially in the Antarctic region.
The station data goes back to 1850. These sets of data are constructed by taking the high and the low temperature of the stations around the planet and averaging the temperatures until the annual average temperature anomaly is reached. It is less accurate as many factors can impact the results. One of the biggest limitations is that the temperature can only be measured where there is a thermometer hooked up. The rest of the area is essentially guessed. Many portions of the Earth do not have stations and this greatly limits the station data.
Satellite data is not direct measurement of temperature, but since each satellite is measuring each spot of the Earth the same error would apply to the whole measurement. Since they cover most of the Earth directly they measure much more of the Earth’s surface. For these reasons they are considered to be more accurate. The big limiting factor is the data only goes back to 1979. I think most people would agree that having 160 years of satellite data would be very helpful in the debate.
I am frustrated by the limitation of the available data. So I decided to use the above 4 measurements to create a single temperature record. From 1850-1978 it uses a simple average of the CRU and Hadley data. From 1979 it weights the two satellites equally with 33% and then the station average gets a 33% weight. One fortunate event was that there was not much of a difference between the 4 sets in 1979. That the methods have diverged since then is the reason why picking one set is a form of cherry picking. Warmists like the station data and skeptics like the satellite data.
The weighting options are pretty clear in the excel chart. Changing a single cell will change the weighting for all everything. I used the older version so more people will be able to open it without problem. If many people request the data to be in csv form I will also add that as a download option.
While I know that this is not a pure scientific method, it is a great engineering method for using as much of the information as possible. There is little chance it will be more inaccurate in the long run and any problems with one set will be lessened by the inclusion of the others. I will also make this available for anyone to use. I will update it each month as soon as I have all the data.
There is also a 9 year moving average that is also charted. I would rather use a standard set that includes as much information as possible and includes more than a single source. I hope that more people will start using this or one similar to it. A lot of work goes into these data sets and while none of them is perfect, hopefully the merged set will be helpful to everyone.
Finally. I am not happy with the name, if you have a better one, please let me know. I have been switching between merged and blended, but I don’t really like either of them.