Saturday, May 30, 2009

Clinton Vs Morrison More Bad data

Again we visit Dave's favorite place to get data. Dave thought this site showed that global warming was real. A real look at the data shows how unreal it is.

Today we look at Clinton Iowa and Morrison Illinois, two cities separated by 14 miles and a state line. The first picture shows the daily difference in temperature.



In 1975 the Clinton station moved 1 mile. Note that the noise introduced by the movement of the station by 1 mile is far greater than 100 years worth of global warming. See the green line inside the red around 1970. That green line represents 100 years of warming. The error in the data is 20x greater than 100 years of warming. And we are supposed to believe Global Warming is measurable! We can't even measure a similar temperature 14 miles away

Let us look a little closer at the data. I prepared a histogram of temperature differences for these two towns just 14 miles apart and within 100 feet of the same elevation. How often are the daily temperatures divergent?



You can see that half of the days have greater than a 2 deg F difference in temperature and 10% of the days have greater than 6 degrees. Let's put this in perspective

A suggested set of criteria based on the horizontal temperature gradient has been devised. A weak front is one  where  the  temperature  gradient  is  less  than  10[deg]F per    100    miles;    a    moderate    front    is    where    the temperature gradient is 10 [deg]F to 20 [deg]F per 100 miles; and a strong front is where the gradient is over 20 deg]F per 100 miles.
http://www.tpub.com/content/aerographer/14312/css/14312_117.htm

A strong front is when the temperature gradient is .2 deg F per mile, yet the temperature record says that on 10% of the days the temperature gradient between these two towns is more than twice that due to a strong cold front! Is it conceivable that these two towns experience super strong cold fronts that lie BETWEEN THE TWO TOWNS 10% of the time? That certainly seems ridiculous to believe. It is easier to believe that the thermometers are not working correctly, and if they don't, then neither does the silly global average temperature which is based upon crap data like this.

Now let's take a closer look at the 365-day running average. That is the third picture.



The interesting thing about this is that in general the temperature difference would be one way for years and then it would reverse for years. Given all the examples of this I am finding when I examine the temperature records for closely spaced towns, one would think that there are cold fronts almost constantly between closely spaced towns all over the US. Or could it be that the temperature record is just so bad that no conclusions can be made from it?

Does anyone really believe that a 1-3 deg warm front sat between Morrison and Clinton for the past 43 years? I for one am not that gullible.

4 comments:

  1. I downloaded the data and did a quick difference calculation as you did from 1948 to 2005 (when the two stations share common collection data) and I came up with a mean difference of 0.65degrees with a 95% confidence on the mean of +/- 0.35

    Now there might be a small deviation because the data set was 20900+ data points and I might have missed a "missing value" when I did a count on the number of samples.

    Now I'm not going to say that these two samples don't show alarming differences. I can't understand how they mis-measured so badly, but I can assume there was some instrumentation error.

    Thankfully the Climate Models are tested against a gridded average on a continental scale, which should help eliminate this kind of noise.

    But when these two stations only differ by about 0.65degrees on average +/- 0.35 that doesn't sound too bad considering the noise.

    I think this is why statisticians like the Central Limit Theorem.

    But again, I'm no more a statistician than I am a scientist.

    ReplyDelete
  2. I noticed a correction that was needed in the alignment of this data. The dates didn't line up so when I went through and lined them up the mean difference came out to be about 0.7degrees with a 95% confidence on the mean of +/-0.06degrees.

    (I mistakenly appended this note to the post on the two Iowa stations, sorry for the double posting)

    ReplyDelete
  3. No problem about the double posting. Yeah, almost all these stupid programs that one downloads data from don't output missing days, requiring one to go through the data set and make sure everything is aligned. That is a source of error for one and all.

    I am glad, though that you agree that this is some sort of instrumentation error, but you are wrong when you say that it isn't bad. When you average across both the positive and negative trends, you effectively average out the step change that is seen in the data. Let's say we have two rulers with which to measure a mile. You beleive at the time that each is measuring a foot. But one gives you 6500 feet per mile and the other gives you 4060 feet per mile. Both measures are attrociously wrong but averaged, they give you the actual value.

    But we must ask what is the actual value. I will guarantee you that if you take those towns and compare each with a different city, the second nearest, you would get a third pattern of errors, which wouldn't match the pattern seen above. So, then the game becomes who, if anyone is actually measuring the true temperature? The real answer is no one and the above data difference becomes the error bar for the data set. That means the error in measurement is greater than 100 years of collective signal, and statistically that means that you can't possibly know whether the world has warmed or not from this data set. And that is why as I showed a couple of posts below (search for Idso to find the chart), that editing is actually the cause of the warming trend, or at least a large large fraction of it. When editing data this attrocious, one can justify to oneself all sorts of reasons to make it hotter if you actually beleive it is hotter. Go see the post on homogeneity.

    ReplyDelete
  4. As has been pointed out many times, land temperature data is NOT the primary measure of the planet's temperature.

    This is well explained in Spencer Weart's book THE DISCOVERY OF GLOBAL WARMING as well as on the two web sites:


    http://www.skepticalscience.com/argument.php

    http://scienceblogs.com/illconsidered/2008/07/how_to_talk_to_a_sceptic.php

    Burgy

    ReplyDelete