Posted by Jeff Id on November 29, 2009
A post from an anonymous blogger who requested a repost here. Since he did a decent job of demonstrating his methods (far better than CRU) it’s worth taking a look even if his name turned out to be Dr. Evil. From the savecapitalism blog, the recent discovery by the drive by media that New Zealand’s temperature data also contains substantial corrections is confirmed.
Guest post – HPX83 from the SaveCapitalism blog.
Okay, so I couldn’t help but do it. I had to try and reproduce the New Zealand “glitch”. And I did. When reading this, please remember that it is NOT scientific evidence, it is a first draft for reproducing the results found by others. I have not interpolated to get missing data, and some calculations may have some lack of precision, so do not rely on this data as “evidence” – simply as a reason to look further into the matter yourself. I am not a climatologist.
- Fetch data from GHCN datasets found here : ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v2/ (v2.mean and v2.mean_adj)
- Get all lines from v2.mean_adj where country code is 507 (New Zealand)
- Remove all lines that have data missing for one or more months (I did NOT interpolate to get missing data, I simply excluded the entire year)
- Get annual mean for each line by adding together the months and dividing by 12 (does this muck up the averages, since different months have different number of days?)
- Group data by nearest WMO Station, and use the average of all annual means for that WMO Station as data for that WMO Station / Year
- Repeat steps (2-6) for all lines in v2.mean that correspond to a line in v2.mean_adj
The result is rather interesting. I took this data and jammed it into the OpenOffice Calc program and let it plot the following for me (thick line is average of all series, with linear regression curve)
Note how the linear regression curve shows that average of all series start slightly above 12 and ends at around 13, which would be very much in line with the “consensus” agreement that temperatures have gone up 0.7 degrees since the 1840s. However – this is not what you get if you take the adjusted data.
Whoa! A whopping 2 degreees increase in temperatures instead of the 0.7. How is this done? Well, look how the adjusted values makes the average start at around 11-11.5, and ends a bit above 13. The linear regression really says it all. I’ve tried to get the same coloring for the different series in both graphs, so the “adjustments” of different series become apparent. The series 93119 shows much of the story.
Now – I’m not saying these weren’t valid adjustments – but it is VERY INTERESTING that the raw data seems to show a COMPLETELY different picture than the adjusted data. If one “adjusts” data, one could reasonably expect (or??) that about the same amount of adjustments would be done up and down. Or at least that the adjustments would be reasonably spread over time, with somewhat similar magnitude (or??). Since I am not a climatologist, maybe I’m completely wrong.
MAYBE THE MOST PROBABLE THING IS FOR THE GRAPH TO LOOK LIKE SOMEONE SAT ON THE BACK END OF IT.
Food for thought, anyways ……