Same Data, Different Graph


Graphs.png

How can the same data be shown on both of these graphs?  The temperature is shown on one graph and the change in temperature is on another.  Does one graph hide the Global Climate Change, and the other exaggerate it?

 

Retired chemist and physicist C.R. Dickson points out that the large error rates in global climate measuring combined with the very tiny increases shown make for very shaky conclusions.  We’re simply not measuring the temperature well enough to trust an increase of a degree or two.

Because it’s so difficult to observe man-made global warming, some experts at NASA GISS believe the accuracy of climate models requires a one hundredfold increase in order to see the small amount of warming.

“A doubling in atmospheric carbon dioxide (CO2), predicted to take place in the next 50 to 100 years, is expected to change the radiation balance at the surface by only about 2 percent. If a 2 percent change is that important, then a climate model to be useful must be accurate to something like 0.25%. Thus today’s models must be improved by about a hundredfold in accuracy, a very challenging task.”

Thermometers were invented in the early 1700’s, by the way.  Everything before then is an estimate.  There have been five ice ages in earth’s history, and the Wikipedia article indicates that outside these ice events, the earth is ice free even at higher latitudes.

It is almost as if the actual temperature of the earth has been changing for millions of years, and that with today’s measurement techniques, we can’t figure out how warm the earth is with precision.

 

Advertisements

5 thoughts on “Same Data, Different Graph

  1. JeffM

    I am going to be pedantic here so BE WARNED. The article is implicitly premised on something that people may feel intuitively but do not fully grasp intellectually because it has never been explained.

    We think of numbers as Platonic entities, absolutely perfect. But most numbers in the real world are the result of countings by fallible humans or measures from intruments that are not perfectly exact. When numbers are not perfectly exact, they SHOULD be (but seldom are) subjected to an analysis of error. The topic is usually studied in a math course called Numerical Methods or college courses in the physical sciences. but it should be introduced much earlier to a much broader audience.

    One of the things that analysis of error teaches is that a difference, whether relative or not, between numbers is MUCH less reliable than the numbers themselves. This may seem counter-intuitive so let’s take an example.

    Let’s say we have the numbers 990.0 and 1000.0, and further suppose each is exact to within plus or minus 1%. So the first number really represents a range from 980.1 to 999.9, and the second number represents a range from 990.0 to 1010.0. When we take the difference between 1000 and 990, we get 10. We may feel that with inputs exact to plus or minus 1%, we can be pretty confident in that difference of 10. Wrong. What is the range of the difference? It goes from 1010.0 – 980.1 equals plus 29.9 to 990 – 999.9 = minus 9.9, which, relative to 10, is an error of plus or minus 199%. In other words, we do not even know whether the sign of the difference is correct, let alone its magnitude.

    In short, for a difference to be reliable, the items being compared have to be super-reliable. I have no clue why this is not taught. Sorry if the tone was pedantic.

    Liked by 2 people

    Reply
    1. rt895

      Not unreliable… The way the trends are shown are deceptive. Much difference in intent. Restricted scales, offset zeros, and using a subset of the data to show the preferred trend when a larger time period shows no trend or the opposite of the preferred direction are species of fraud.

      Like

      Reply
    2. JeffM

      Dave

      I was trying to make explicit a generally little known fact about numerical analysis that was implicit in the article, namely that a comparison between two numbers of equivalent magnitude and reliability is ALWAYS much less reliable than the reliability of the numbers being compared. (Comparing numbers and taking their difference is logically the same thing.) If the numbers being compared are very reliable, then the comparison MAY be reliable.

      With respect to the specific data being discussed, I have not reviewed them personally and so cannot opine with ANY degree of intelligence on whether or not they are highly reliable.

      Now to be fair, the arguments about global warming are about changes in averages. If numbers are subject only to random errors, averages are much more reliable than individual numbers because the random errors tend to cancel out. If numbers are also subject to consistent bias, that does not affect comparisons between averages. However, data collected over time may be subject to inconsistent, non-random sources of error. If that is the case, comparing averages will not increase reliability.

      Let me give an example. A project that I once worked on required getting data on trends in population in India. Simple you say: look at the census data. The first real census was taken in 1881, when the British controlled India, with follow ups every ten years thereafter. However, the lessons learned from each census were used to improve the methods used in the next census. Consequently, the later data were generally more reliable than the earlier data. Then, in 1931, Ghandi told his followers not to cooperate in the census. The data for that census are certainly less reliable than for 1921. During 1941, World War II was raging, and the census was quick and dirty. In 1951, two different censuses were taken, one in independent India and another in independent Pakistan, each using different methods from each other and from the British in 1941. With respect to long term population trends in the Indian sub-continent, the language of certainty is completely inappropriate because the data are almost certainly filled with errors, some of which may be material.

      And then you have the issues mentioned by rt. There are ways to present even reliable data misleadingly. If you start presenting unreliable data in misleading forms, you have an utter mess.

      If you want my opinion on global warming it is this. Probably, average temperatures have risen in the last several centuries. Our capacity to measure that change numerically is weak. Our understanding whether the change is consistent with normal cyclic change is virtually nil. Our understanding of what proportion of a poorly measured and poorly understood change is due to human activity is exactly nil.

      Like

      Reply

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s