Saturday, July 24, 2010

Colorado temperatures using TOBS

Back in March I first looked at the temperature data for Colorado, comparing the GIS station data with that generated with the homogenized data from the USHCN. (Incidentally their process is described on that site, explaining in part how they were able to generate data from sites that did not have full sets of information. The immediate concern that I have with that, for the purposes of the present discussion, was that as I went through the data tables for the sites in Coloradolooking at the data corrected only for time of observation (TOBS) I was struck by how few stations were providing data in the early years. Averaging has some problems, that I will likely discuss in a future post, but when there are less than half the stations reporting, then using them to estimate the values for the unknowns becomes a little more questionable.

OK, so let me go back and insert the TOBS data in the spreadsheet of the form that I used for the original post: I note, while doing so that there are 24 stations in the USHCN group, while there is only one for the GISS network. Yet of those 24 only 6 have temperatures given for 1895, when the series starts (and for which there were a full set of values for the homogenized data I had originally used – for that set the only value missing was an 1896 value for Telluride – ah, well.) There was also only one GISS station at Grand Junction.

Running the averaged values and comparing the difference between that station and the USHCN TOBS values gives an average temperature difference of 6.15 deg. While that is still high, it is less that the 6.65 deg warmer that the homogenized data suggest. As to how it played out over the years:

And there has been a clear increase in the recorded GISS temp, relative to the USHCN over the decades. The correlation has a lower correlation than there was with the homogenized data, but that data plot (in the original post) showed that the difference was getting less over the years not greater.

Now as for the average state temperature itself, and this is a bit of a surprise, since the first time I ran this, there was a clear increase in temperature over the years. Using the TOBS data (but including the GISS value) that is no longer the case.

There is that curious blip upwards after 1984, but looking at the overall not any remarkable changes. In regard to the change in the Standard Deviation with time, the shortage of data points in the earlier years skews this value beyond where it might completely valid, though the decline in recent years might be more credible.

Moving on to the effect of population, Colorado has a smaller population in larger cities, and without the homogenization to remove the effect of UHI there is a correlation to the log relationship that I have noted before (and which has previously been suggested by others).

The correlation coefficient is slightly lower, in fact, with the TOBS data, which isn’t supposed to be the way that it correlates, since the homogenization was supposed to remove those effects.

Looking at the effects of Latitude, there was no correlation with the homogenized data, and that also holds true with the TOBS data. (This, from the other states reviewed is relatively inconsistent).

Because, to a great extent, the mountains get higher going toward the west, there is a correlation with longitude:

However, when checked, the correlation is better with elevation, explaining the apparent result.

And that correlation has held true for both sets of data.

Interestingly, since my first post, there has been some interest in Colorado climate, with there being a paper out by Noah Diffenbaugh of Stanford, which apparently claims that the increase in global temperatures will be emphasized in higher states such as Colorado.

Given that there has been a CO2 increase Dr. Alan Keen refuted that prediction based on an examination of the trends in temperature data, which included the USHCN homogenized values, though after they had been used to give a value for the “Colorado” grid – a value that I will get to in future posts, when I have some more state data completed. Dr Keen provided the following plot:

Well let’s press on and get some more data.


  1. more interesting reading, thanks.
    I realize this question is outside your research but maybe another person has an answer
    I notice the 1930's warm spike in your data and have seen it recorded before in USA charts. Did that warm period exist in the rest of the world?

  2. It is one of those questions that, by the end of this series, I hope to have a better answer to. I do plan on going beyond the USA but that depends on how the data ultimately plays out.

  3. This comment has been removed by a blog administrator.

  4. This comment has been removed by a blog administrator.

  5. This comment has been removed by a blog administrator.

  6. This comment has been removed by a blog administrator.

  7. This comment has been removed by a blog administrator.