The Problem with Short-Term Analysis


I am a fan of the webcomic XKCD and tend to read it at least weekly.  The writer is a very clever and well versed in science and his degree is in physics.  Some of the most amusing scientific oddities are discussed in his What-If column.

So why am I disappointed?  He recently wrote an article about how foolish it is to think global warming is over because of the recent cold snap that has been taking place in the midwest and eastern US.  Certainly I agree that taking weather events and using them as the basis for long-term trends is folly.  No, my disappointment is with his analysis itself.  Here is the webcomic.



He is nice enough to provide the source of his data which is not an actual source of data, but a pro global warming website known as climate depot.  If you are going to source data, please use the real data source and not the intermediary.  Climate Central Depot is the kind of place that will show the temperature trend from Central Park in New York and take that seriously.

Aside from that problem, it was the scale of the analysis that really disappointed me.  Picking 1970 as the starting point seems awfully one sided and ultimately the problem with almost all global warming material.  I have long argued that trends on anything less than 1,000 year time scales are meaningless.  The problem is that finding longer term data is always tricky.  If I had only looked from 1970 forward, even I might have concluded that global warming was something to pay attention too.

However the data he found was very specific and from a big city.  It was also incorrect which shouldn’t be surprising considering the original source that he used.  I found this NOAA dataset for St. Louis, MO.  Lower on the page is also a dataset for Columbia, MO which is due west of St. Louis.

Here are the charts of the data from the NOAA site.

The Inconvenient Skeptic

The Inconvenient Skeptic


It is possible to see the differences in the climate depot data he used and the actual NOAA data for St. Louis.  It should also be no surprise that Columbia experiences more extreme cold because St. Louis will have a UHI effect, even in the winter.  It is also evident that picking 1970 was especially useful if showing a downtrend is the goal.  28 of the 38 years from 1920-1957 had zero days at or below zero.

Columbia which experiences the same climate, but less UHI effect shows little of significance.  On average over the past 120+ years there is an average of 4.5 subzero days.  In 3 of the past 5 years (including only January of 2014) they have had at least 4 such days.  Columbia is showing no signs of change.

What is most interesting is looking at the most extreme temperatures for the two locations over the entire time frame.  What it shows is the expected cyclical behavior that is typical of climate to those that pay close attention.



Both locations show periods of time where there is more instances of extreme cold.  There are also periods where there is little extreme cold.  When all things are considered, it seems likely that the next decade will have an upswing in the number of extreme cold events that reach the Midwest.  I suppose that we will have to get used to hearing this all blamed on the Polar Vortex.



I meant Climate Central, not Depot as correctly pointed out and shown in the actual comic.  Ooops.

Posted in General by inconvenientskeptic on January 29th, 2014 at 11:47 am.


This post has 7 comments

  1. I was interested in the heat extremes.

    Number of days >= 90F

    79 1954
    79 1953
    79 1936
    73 2012
    71 1934
    71 1901
    69 1933
    69 1881
    68 1983
    67 1991
    67 1980
    67 1952
    67 1937
    66 2010
    65 1988
    65 1941
    63 2011
    63 1955
    63 1944
    62 1931
    61 1914
    60 1913

    What “modern” extremes?

    I was disappointed in XKCD too.

  2. Gary Meyers Jan 30th 2014

    You say that Climate Depot is a “pro global warming” website?
    I will have to disagree with you on that point.

  3. Exactly, but whether the temperature record starts from 1970 or from the year 1000 is irrelevant really, all you’ll discover is a trend.

    A temperature trend is evidence of a trend only, it is not evidence that CO2 is the cause. This is where the alarmist argument goes off the rails.

    Blaming CO2, and therefore human activity, is merely wishful thinking.

  4. inconvenientskeptic Feb 13th 2014


    The ultimate problem is that it is impossible to separate short-term from long-term unless you look at very long-term trends.

    Anything less than 1,000 year averages mean nothing. There is so much variation in all the time scales shorter than that, that is truly makes it impossible to make any valid conclusions. 150 years truly has no meaning.

  5. lexharrison Feb 22nd 2014

    I think the issue would be best addressed by sequential hypothesis testing (SHT). Over to any statisticians who may be reading this page. Anyone else should take a look at Wikipedia to get some understanding of the techniques which seem well suited to looking at changes in data covering an extensive time series. One application of SHT step detection is highlighted in Wiki “Step detection is the process of finding abrupt changes in the mean level of a time series or signal. It is usually considered as a special kind of statistical method known as change point detection. Often, the step is small and the time series is corrupted by some kind of noise, and this makes the problem challenging because the step may be hidden by the noise. Therefore, statistical and/or signal processing algorithms are often required. When the algorithms are run online as the data is coming in, especially with the aim of producing an alert, this is an application of sequential analysis”

  6. Onlooker Feb 24th 2014


    It’s good to see that you’re posting again! I discovered your site and bought your book about 6-7 months ago and I was worried about your absence from your site since last Spring. I hadn’t looked in here for quite a while and just happened to come by to find an old post and saw your recent posting.

    Anyway, I hope you’ll be posting more often again as I’ve thoroughly enjoyed your blog (and book) and it has helped me to better understand this whole debate.

Web Design & Dev by

Mazal Simantov Digital Creativity