Thursday, May 21, 2009

No doubts on Electra

Dave in a comment on my Electra California post, a couple of posts below, said that Electra probably had missing data causing the yearly averages to be so bad and so varied. In that post, the readers of this blog will recall, I showed that there were changes of yearly temperature average of over 10 deg F. Dave thinks it is an artifact of missing data. So, I downloaded the monthly data from and here are the plots for each month, from 1960 to 1994 (all plots say 2006, but the data only goes to 1994). As you peruse each picture notice that post 1985 there are sharp spikes in the average monthly temperature. These spikes are what causes the spikes in the annual data. Note the size of these amazing temperature jumps in the raw data. The data is crap, yet this is what we are using to calculate how much the temperature has warmed. Note that there are no missing numbers, as Dave claims

Ok, we see the jumps in the monthly temperature. So, lets check to see if the night time temperatures are missing. If so, then the cooler minimums would be gone. They wouldn't be hot. What we find when we look at the minimum and maximum temperatures for the months which have these mysterious jumps of many degrees is that it affects both max and min temperature. Here are some examples. First again, July Mean. Notice in the 2 pictures below the mean that I have July min and max and the same two weird jumps in the later years are in both the max and the min.

Same thing for November. First the mean picture followed by the min and max. Same weird jump in temp affects all datasets.

Whatever the source of these jumps in the record, it is affecting not only the mean but also the maximum and minimum temperature. That says something environmental, something next to the thermometer is screwing up the reading. It isn't missing data as Dave says.

I say again, the data is crap. It can't be used for the purpose the Historical Climate network says it is used for, i.e. to measure the subtle change in the earth's global temperature. And if you think this data can be fixed, go look at the Chinese data here


  1. Why don't you try again? Rather than dealing with co2science crap data, why don't you go to the actual, real raw daily data at the USHCN website like I suggested?

    It looks like starting in 1985 there starts to be missing data. Just like I predicted.


  2. "The data I downloaded was from CO2 and they have an early 2007 download of the raw data. I do know that the USHCN has altered the raw data since that time. I don't know how one changes raw data, but that is what they did."

    The problem is that monthly data is by definition not raw data. I see a couple of possibilities:

    1) CO2science scraped the raw daily data and used their own algorithm to make monthly or yearly averages, leading to differences with the USHCN monthly or yearly averages.

    2) USHCN changed their FILNET interpolation algorithm that is used to fill in data where raw data does not exist, leading to a change in monthly data between when CO2science scraped the monthly data and the current version.

    3) Some underhanded, deceitful US HCN employee snuck in in the middle of the night and changed the raw daily data just to miss with your mind.

    It is a pity that CO2science does not present the raw daily data from 2007, or we'd be able to figure this out.

  3. Well it is interesting that when one goes to the USHCN it doesn't show or give any indication of all those missing days, Dave. The USHCN must equally be making up the data, Dave. If according to you there are only 100 days of data in 1994 from Electra, it is quite curious to me that the USHCN carries on as normal. Here is the monthly data for 1998-1994 for Electra. Remember this is the time you say all that data is missing.

    1988 46.55 50.66 53.01 58.33 61.55 69.96 79.74 75.33 70.56 64.00 51.32 43.44

    1989 41.75 44.66 52.61 59.03 62.05 69.26 73.64 71.13 67.36 59.70 51.52 43.44

    1990 43.85 43.96 51.91 58.93 61.85 69.67 76.90 73.63 69.40 63.10 50.42 40.04

    1991 43.75 52.36 47.61 53.03 59.65 66.86 76.64 71.63 73.46 66.20 53.02 44.34

    1992 42.25 51.26 53.61 59.13 67.85 70.36 73.24 76.43 69.66 63.30 51.12 43.78

    1993 44.05 47.06 54.11 54.33 62.05 67.36 72.74 72.33 69.36 63.30 49.82 42.84

    1994 44.75 45.16 53.01 55.83 62.24 67.56 75.55 74.30 67.56 58.50 45.28 42.74

    No sign in the USHCN data that all these days are missing. Are you saying that the climatologists are simply doing bad statistics here? Are you saying that they are making up the data?

    To me, it seems that if you are correct that there are lots of missing days, it means that the data is crap and is utterly fabricated. And if you are wrong the data is crap because of huge temperature jumps. In either case, the data is crap. Have you painted yourself into a corner?

  4. I forgot to post the place I got the USHCN data from. It is from their site
    Why don't you go there. There is no hint of a problem. No notes to the effect that this data is crap, no nothing, only numbers.

  5. No, I haven't painted myself into a corner. My goal was to point out that you have very little idea about how to use this data, and I think I succeeded quite well, thank you.

    As far as the data: US HCN has put a heck of a lot of data up on the web. They haven't designed little hand-holding documentation in every single possible file, but there are plenty of pointers to documentation all over the US HCN website. They assume that if someone is doing real work, they have a basic amount of knowledge of how to do things and how to find and read documentation. The raw, daily data is there, with annotations in the comma delimited file. The monthly data uses the FILNET routines to do data infilling, as is clearly stated in a number of places, and again, it is really easy to figure out what the underlying data is like - in the interface on the monthly temperature page, there is a box you can check right next to "Mean Temperature" which is "Mean Temperature Flags". On the main monthly data page,, it says "You can download and save the data, and also plot data from individual stations in a number of ways, yielding insight into the station's record, and thereby helping to determine the suitability of a station's data for particular applications."

    eg, "we'll provide everything you have. But you may have to put in a little bit of work to figure out if a particular record is suitable for your purposes."

    One man's crap data is another man's data guano mine. Some people may want data with FILNET extrapolations. If so, it is there. Other people may not trust the FILNET stuff. If so, they can go to the raw daily data.

    It seemed easy enough to figure out for me, and I've never worked with temperature record data. Though I guess obviously it wasn't so easy for you.