2011-03-26 10:28:05Uncertainty on temperature records
John Cook


Can anyone help with this email I just received?

We want to re-create the temperature plot showing HadCRUT, GISTemp and NOAA (mainly so that the graphics is styled in line with the rest of the sheets). The data here is obviously really helpful, but we wanted to include some uncertainty estimates and these aren't included in the excel spreadsheet. Any ideas where we can get them? It would be good to get upper and lower uncertainty bands for all three sets, and then we can choose how to present them (either as a greyed out backing to the trendlines - individually or amalgamated). I'm sure this info is on the individual websites, but massive data files aren't my strong point, and I don't have access to the software I would have used in previous incarnations to get the numbers we need out of the mass of data.
Any ideas? Can you help at all?
2011-03-26 17:38:35comment
Robert Way


Depends on what they want, if they want just a standard uncertainty of a standard deviation or two that can be computed for each rather easily but if they want the uncertainties in terms of confidence intervals then I don't know where they can find that (the number of sites used per year for example)

2011-03-26 18:53:23Uncertainty on GISS
John Cook

I recall GISS showing a few uncertainty bars on their temp record, indicating recent temp has lower uncertainty than older temp (thank you, captain obvious). But I don't know of a data series showing this for every point on the time series. If anyone does, please post here.
2011-03-26 20:21:19
Ari Jokimäki


I haven't been very happy on the temperature series (or many other series) in this sense. I think uncertainty should be shown and included to the data.