.

Saturday, August 9, 2014

How to read a global surface temperature chart

Sou | 6:34 PM Go to the first of 16 comments. Add a comment

Someone on Reddit linked to an old HotWhopper article that had some charts, and it got a couple of complaints:
  • Those arrows do a piss poor job explaining the data contained within them. (mythrowaway2000)
  • Some of them go from the average to specific extreme data points. Incredibly misleading even if you do understand the data in the charts. (TheTruthandtheAnswer)

The charts in question were pretty basic. I updated them recently and you can see the updated versions here.

It's easy to forget that there are probably a lot of people who've never studied subjects like science or economics or finance, or worked in an area where charts are commonplace. The above comments underline the fact that there are people who are trying to understand what is happening with climate, but are unfamiliar with charts. So I've made a short video to show what you can glean from a straightforward temperature chart. It's based on GISTemp data from NASA Goddard Institute for Space Studies.



If you click down the bottom right hand corner, you can watch it full screen or on YouTube.

16 comments:

  1. I'm not 100% sure, but I think people also were struggling with global annual average temperature anomaly and moving averages.

    ReplyDelete
    Replies
    1. I wrote something about anomalies a while back. Maybe I'll do a video on that. No promises. The trick is to try to figure out why a person doesn't understand them. What's the mental blockage. I think Nick Stokes at Moyhu might have written some articles about anomalies.

      Moving averages is something they can tackle after getting to grips with what a chart means. Tamino has a series of articles about smoothing - which is the main purpose of moving averages. There are other and often better ways to do smoothing as well. Moving averages is easy for a quick and dirty smooth :)

      Delete
  2. Neat video, Sou.
    Reading graphs is an art. One exercise for 'mythrowaway2000' and 'TheTruthandtheAnswer' would be for them to draw the graph that they think would show the pressure in a spherical balloon vs time (or vs diameter of the balloon or vs surface area of the balloon) as it is blown up until it bursts. It may make them think about relationships and, aside from the physics involved, convince them that they 'don't know graphs like they think they do'. There could even be a pressure anomaly involved if the balloon pressure is plotted relative to or above atmospheric pressure.
    Post graph, 'mythrowaway2000' and 'TheTruthandtheAnswer' could actually blow up a balloon to see what the changes in pressure actually feel like and then compare it with their graph. Or, if they have access to a manometer, ...
    Finally, interpreting graphs correctly is helped if you actually understand the science, economics, etc that is inherent in the graph. But that's often a step-too-far for some folk who infest the blogosphere with their crackpot ideas… Just saying.

    ReplyDelete
  3. Moving averages are a pretty easy concept to understand, so much so that I invented them for myself about 30 years ago.

    It was at a time I'd just gone freelance and, no longer receiving a monthly pay cheque, I was undertaking some pretty big projects that I would only invoice in stages. So although I was working pretty consistently all the time I found that my monthly incomings were fluctuating wildly, as some months I was writing just one invoice and the next I'd be writing five or six. As I kept a graph of my monthly earnings you can imagine how it looked: erratic. After scratching my head, to solve the problem I decided each month to add that month's invoice total to the three preceding months' totals and then divide by four. This created a graph that much more accurately reflected the amount of work I was doing on a monthly basis and my consequent income. It was much later that I discovered that what I'd been doing was called 'smoothing'.

    If I, a mere creative with only schoolboy maths can get my head round this, why can't anyone?

    ReplyDelete
  4. Something I have never been able to work out. They give an average annual temperature, but are they referring to the average of the maximum daily temperatures or the minimum? I assume it is the maximum daily temperature.

    ReplyDelete
    Replies
    1. What is reported by GISTemp is the mean anomaly.

      http://data.giss.nasa.gov/gistemp/

      I think it's based on the average temperature (eg max + min / 2) though it probably doesn't make a huge difference when it's spread over the entire world surface. As long as it's consistent.

      http://data.giss.nasa.gov/gistemp/FAQ.html

      http://data.giss.nasa.gov/gistemp/abs_temp.html


      If you want to look at differences in trends between max and min temps, there is more meaning at the regional level - eg whether the night time minima are rising more quickly than the daytime maxima for example, in a particular locale. Whether winter temps are rising more quickly than summer temps. This sort of information is provided by BoM.

      http://www.bom.gov.au/climate/change/index.shtml#tabs=Tracker&tracker=timeseries

      Delete
    2. Makes sense. All the more reason to focus on the anomaly value.

      If you think about it the satellite measurements beg the same question - obviously the satellite doesn't just happen to be passing over a point when the maximum daily temperature is occurring.

      Delete
    3. The satellite passes over each spot at the same local time. That of course isn't always the maximum, but it can be the usual time that you have a maximum.

      Delete
  5. Why would you not assume that it is the average of the *average* daily (or monthly) temperatures?

    ReplyDelete
    Replies
    1. Because it makes sense that they do it that way, at least for temp measurements from surface stations using max and min thermometers.

      Balloon and satellite stuff I am not commenting about, I do believe they are very complicated due to their use of the microwave proxy etc.

      Delete
    2. Please explain. You could, of course, have a yearly average of min, max, or average, but I would assume the last unless told otherwise. (Of course anomalies are used, not actual temperatures, but that doesn't change things.)

      Delete
    3. Using the max daily temperature makes more sense to me. Actually I would like them to show min and max separately to get a picture of the diurnal spread of temperatures.

      Delete
  6. Greg Laden has commented on this article, on his Facebook page, copied below. He is mainly picking up on the use of anomalies rather than an explanation of how to read a chart (any graph). I'll be using his thoughts if I get around to writing about anomalies again:

    https://www.facebook.com/laden.greg/posts/10203610449478512

    My comment on this post (it uses authentication I have a hard time with so I'm putting my comment here):

    I think that the anomaly problem is serious. Anomalies are used as though they were raw basic data in demonstrating (mainly temperature) change, but they have their own analytic purpose as well. For example, the chart used in the very nice video cold have been done with raw temperatures and nothing whatsoever would be lost, unless one is specifically interested in comparing specific years or trends of years to a particular base period. In this case, the anomaly years are chosen because they represent old times (before-ish global warming) yet goo data (which is always better later). In fact that particular time period, use frequently, represents neither very well but since it is 30 years long it is hard to argue with. But, this graph could still have been just degC and if one wants to make a comparison to some baseline you just throw the baseline on.

    A proper use of anomaly as an analytical tool would be monthly (or seasonal) change. Raw temperatures don't work well because of seasonal effects (even globally when one might think the seaons cancel each other out).Or, for example, say you want to ask the question, "Is June really extra warm in the Northern Hemisphere compared to globally") in which case you are comparing an apple to an apple plus an orange where every other month would be a different set of fruit. Here, anomalies solve the problem easily.

    Which points out a terminological thing that might help people understand:
    "When we are using anomalies, we are looking at RELATIVE temperature (or other measure)."

    So anomaly is used in the global annual average not improperly but not usefully, and somewhat misleadingly or at least, in a way that ads an unnecessary distraction. But this is something climate scientists are used to so they don't see why other people can be confused, and I think to some extent, don't acknowledge how these arbitrary 20 or 30 year anomalies are misleading.

    For example, in this graph, a useful "zero" line would be an estimate of the average global temperature from 1650-1750. That would be based on proxies, and might be off by a tenth or a couple of tenths of a degree, but the advantage of being a true pre-industrial measure might outweight that.

    Maybe I'll work on a new chart...

    [Sou: I have as much or more trouble with Facebook as Greg has with Captcha and Google blogger :D]

    ReplyDelete
    Replies
    1. Averaging raw temps is not used as all temp stations are not equal eg you can have stations located near one another, yet show different absolute temperatures eg one located on plain and another located on the top of a nearby hill.

      Delete
    2. I'm not saying that there are not reasons to use anomalies. I'm indicTing that in some cases the fixing of the data using an anomaly provides is not appropriate. The other and larger point is that the climatic baseline is usually chosen for good reasons foe a climatologist but bad reasons for making certain points.

      Delete
  7. I don't think Greg is right here. Anomalies are essential when calculating a spatial average. The reason is homogeneity. The logic underlying a spatial average is that the temperatures at unsampled points can be inferred from the sampled. This uses a spatial numerical integration formula, which is based on the idea that you can interpolate. Anomalies usually can be interpolated; they have long range correlation. Absolutes have all the variability of altitude, coastalness etc.

    NOAA tries to quote a US temperature, and this causes problems. Basically, they implicitly calculate anomalies, and then graft them back onto a fixed set of stations. Or did until recently. So, for example, USHCN and USCRN give different averages.

    ReplyDelete

Instead of commenting as "Anonymous", please comment using "Name/URL" and your name, initials or pseudonym or whatever. You can leave the "URL" box blank. This isn't mandatory. You can also sign in using your Google ID, Wordpress ID etc as indicated. NOTE: Some Wordpress users are having trouble signing in. If that's you, try signing in using Name/URL or OpenID. Details here.

Click here to read the HotWhopper comment policy.