Spiralling global temperatures

Updates to this animation are now available

Spiralling global temperatures. Click for full size animated version.

The animated spiral presents global temperature change in a visually appealing and straightforward way. The pace of change is immediately obvious, especially over the past few decades. The relationship between current global temperatures and the internationally discussed target limits are also clear without much complex interpretation needed.

Data: HadCRUT4.4 from January 1850 – March 2016, relative to the mean of 1850-1900, available here

1. Features you can see:
1877-78: strong El Nino event warms global temperatures
1880s-1910: small cooling, partially due to volcanic eruptions
1910-1940s: warming, partially due to recovery from volcanic eruptions, small increase in solar output and natural variability
1950s-1970s: fairly flat temperatures as cooling sulphate aerosols mask the greenhouse gas warming
1980-now: strong warming, with temperatures pushed higher in 1998 and 2016 due to strong El Nino events

2. Why start in 1850? Because that is when the HadCRUT4 dataset starts, as we don’t have enough temperature data before then to reliably construct global average temperature

3. Are temperatures ‘spiralling out of control’? No. Humans are largely responsible for past warming so we have control over what happens next.

4. What do the colours mean? The colours represent time. Purple for early years, through blue, green to yellow for most recent years. The colour scale used is called ‘viridis’ and the graphics were made in MATLAB.

[Thanks to Jan Fuglestvedt for the idea to make a spiral.]

153 thoughts on “Spiralling global temperatures

  1. I’m a physics major at BYU and I stumbled across a website that had posted this data and I had some questions about your data analysis. First off, how did you collect data from 1850-1900 and did you, in your opinion, collect enough to have sufficient data points? My second question is what made you decide to measure the relative temperature of the earth to what it was in 1850-1900? As a fellow scientist, I want to understand your methods to better solidify my own understanding of how to analyze data. Thank you for your time.

    1. There is not the least of ambiguity here on the source of the data. It is described as HadCRUT4.4, and you can read up the full documentation of how this is generated at http://www.metoffice.gov.uk/hadobs/hadcrut4/.

      Evidently, you want to plot positive numbers on this diagram, and we have increasing temperatures. Hence, one really had to choose the baseline at the beginning of the plotted interval for this figure.

      1. Most line and bar graphs representing global temperatures are plotted above and below a mean of relatively recent temperatures – “last 30 years”, or “since 1980”, for example. Comparing to these means, it seems to me, give a doubly-false impression. First, the means have been badly skewed – raised – by global warming. Second, there is no normal any more.

        This representation avoids those problems.

    2. Hi Bryce,

      You can read more about how the HadCRUT4 data set was put together here:

      There are more details of the land components (CRUETM4) here:

      and the ocean component (HadSST3) here:

      A lot of work goes into understanding the uncertainties in the estimates, which is, I guess, what your question is getting at.

      Best regards,


    3. As an astrophysicist myself I can tell you that all respectable Universities across the globe have large archive libraries, in which very meticulous weather observations are kept on a daily basis from the 19th century.
      where I work in particular, our archives go back to the 1770s.
      As for the 2nd question, my immediate guess is that the author (like most studies actually) sought to use as reference a period where the consequences of man-produced greenhouse gases hadn’t had an impact yet on the climate yet.
      Obviously some regions (e.g. the UK) where literally black from coal burning in the 1850s already, but that was really localized compared to the whole reservoir that is the atmosphere.

      1. Okay, but how do you ensure your datapoints e.g. for water masses and non-inhabited land masses are there?

        The thing is, if you can’t provide the same datapoints in every set you want to compare, that’s not a very scientific comparison…

        1. I think I can answer that one, although I am NOT a climate scientist. I think you’ll find the data for the sea-surface temperature (SST) dataset comes from ships logs. The British Royal Navy has been criss-crossing the globe since the 1600’s. They were in competition with other navies doing the same thing (especially the French, the Spanish and the Portuguese). Every four-hourly? daily? captain’s log for every ship starts with the same information including the lat/long, date/time, weather, wind speed/direction, cloud cover, sea temperature etc. I expect the US Navy would record the same information since the 1770’s too.

          1. Datasets from ship logs are completely unreliable as the methods they employ for taking temperature readings are extremely poor. Dipping a bucket in the water followed by hoisting it up to take a reading can shift the temperature by several degrees and depending on how conductive the material of the bucket is made out of, along with a dozen other corrupting factors, further corrupts the data.

          2. Hi Dog,
            Not true – the sea surface temperature observations are corrected for the effects you describe – read the papers discussing the construction of the datasets. Also, if we used the raw data then the global temperature change is larger – the corrected data (as used here) has lowered the trend.

    4. Bryce, I’m not a physicist, but like you I’m also wondering how data from that far ago was collected as well.

      I also would like to know what records were kept at that time period that actually collected ‘global temperatures’. I’d love to see a reply to this as well!

    5. I don’t believe the data during that perid was actually collected by the author 🙂 Sorry, I couldn’t resist.

      Considering temperature data collection between 1850 and 1900, envision an older man wearing a bathrobe and bifocals, in blowing snow, in South Dakota, in February. The measurements are taken using a thermometer sent form Washington D.C. by horse drawn carriage and steam locomotive.

      Given this, representing temperature measures with one tenth of a degree precision from this time period is frankly absurd, although that doesn’t seem to slow anyone down.

      1. Robert – remember that we have thousands of temperature measurements spread across large areas so the precision of each individual measurement is not as important. Ed.

      2. After telegraphs became available, field reports were sent by telegraph.

        Hoisting a bucket of water takes mere seconds. The adjustment for hoisting is small. Experiments were carried out in the 1880’s to compare temperatures measured by hoisting versus a thermometer directly in the water.

    6. Ok trying to wrap my head around this. But what you are saying is from 1850 the planet has warmed 35.6 degrees which would mean that the average New England summer before say 1980 was between 50 to 60 degrees fahrenheit. at 55 years of age I can tell you that is bunk. What I think you are saying is the planets temperature has increased by 35 degrees fahrenheit I have to say impossible. You Ed being a climatologist must know that we have had at least 17 ice ages and as many warming periods over the last 200,000 years. If you want to do this correctly you need to go back to the ice cores from hundreds of thousands of years ago and look at the changes in the planets temperatures. I am pretty sure humans have very little to do with any warming but what do I know, Im no scientist just a lonely IT guy…

      1. Hi tmac,
        You’ve done the conversion from Celsius to Farenheit incorrectly. The *change* in temperature we’re discussing is 2C which is equivalent to a *change* of 3.6F, not 35.6F. I think that should fit better with your experiences.

  2. Thanks for your info and infographic. I cannot get the animated gif to work. 404 — File not found.


  3. Hello. Cool graphic. I wanted to ask what exactly is being plotted. My understanding is that each point is (given month’s given year’s temp minus same month’s 1850-to-1900 average temp). Is this correct? If so, there could also be a “zero degrees of temp change” circle somewhere toward the center of this graphic? The blue and green early year lines would then zig zag across that zero circle, while more recent time periods’ lines (yellow) spiral further out from the zero circle. But the bull’s eye of the graphic doesn’t represent zero degrees of temp deviation because of the possibility that a given month’s temp could be lower than its historic average. Could you please confirm that I am understanding this correctly? Or explain where I’ve gone wrong? Apologies if I overlooked an explanation you already posted. Thanks.

    1. Hi Betsy,
      The first frame of the animation shows the zero degree circle, which the early lines do cross (although the line is not shown in the other frames). I have used the annual average between 1850-1900 as the baseline as that is what the IPCC AR5 chose, and because the data is less comprehensive in the early period.

      1. Great work Ed,
        Since you said yourself, (I have used the annual average between 1850-1900 as the baseline as that is what the IPCC AR5 chose, and because the data is less comprehensive in the early period.
        I have what I believe to be a fair scientific question and possibly of our own ignorance I could misunderstand. If it was less comprehensive and less ability to monitor Global anything back before the 1900s let alone before telegraph and so on…
        How can we be sure enough equatorial measures were made when it is known some of the harshest conditions and least evolved indigenous cultures resided in the tropical climate area, making for less communication of the higher numbers added to the global climate average during the earlier times? Are the non-included areas all warmer areas that would tip the average up and make this whole thing look more like a tight spiral rainbow? with little peaks instead of what is portrayed?
        Thank You, I hope these questions are fair and not considered biased I want to believe but I need more information to be sure.
        David Mormak

        1. Hi David,
          The way that the global datasets are constructed accounts for the different temperatures over the planet. This is done using ‘anomalies’ everywhere rather than the absolute temperatures.

          1. Could you turn your animation so the rings of various diameters are stacked and looked at from a side view? I think this would look like a global warming tornado. Thanks

          2. Could you make a series of similar charts showing CO2, CH4, etc with known values over time in a similar spiral graph? Then perhaps overlay a chart with the spiraling gases onto of the temps such that visual correlations can be easily seen?

  4. Great gif! I can’t wait to show it to my students at school (secondary school teacher in London).

    1. Could you do graphs that show temperature changes caused by each of the greenhouse gasses, the Sun, the Milankovitch cycle, and the other causes of warming? That would really seal the deal.

  5. This is a wonderful info-graphic.

    I would be exceptionally interested in seeing additional historical years of data displayed. I understand that there will be a slight decrease in the reliability of the data the farther back you go but I believe the temperature estimates that have been used to build the climate models are quite accurate and it would be a fairly straight forward effort to display that data visually.

    It would be great to see 300,000 (or more) years of data plotted in this manner rather than being limited to only the past 66 years.

    Thanks again for all your hard work providing such an excellent visualization that can be easily consumed.

    1. I’d also like to see individual spiral graphs that interpret other contributing data from the same historical period to see the effect of additional factors individually and then superimposed over one another, to expand upon the new view as a key to data accessibility and understanding from the masses.

      Upon an initial view, I had a personal predisposition to look more closely at the the era of industrial revolution, watching for that fossil fuel CO2 spike in usage and the possible global effect. My background is more in history and literature, so I was looking to draw a correlation between what I had been taught previously and how this new data might revolutionize that outlook. I wonder what graphs of other measured social and environmental variables, put into the same interpretive software might add to our understanding.

      Global human population estimates, global species extinction rates, internal known solar system events, evidence of possible external system events, etc.,; whatever elements the brightest minds can think to add to measuring the elements that help to create the stew of life as portrayed in similar individual graphs for their own individual unique development and then overlay end to see how they may have contributed to the whole.

      1. Sorry, I don’t see a way to edit. There is an outlier in the late 1800’s, do you have any insight? Also, after reading the site more closely, a graph on the growth of domesticated crops versus the loss of natural environmental zones/foliage would be interesting, I would think. Thanks.

  6. This illustration would be ideal for use in lectures that I give on cruise and expedition ships in the Arctic. It always amazes me that some people are still convinced that climate change is a myth. I have just finished writing a book on the Atlantic Arctic, which looks at the Inuit of Eastern Canada and Greenland and the Sami of Arctic Norway. It covers both their histories and the effects of climate change on their cultures. Currently I am trying to source illustrations for this work (which is for educational purposes).
    Is it possible that I could use this wonderful illustration as a final page? (or maybe even on the back cover?) It would also be very useful when giving lectures in both University colleges and schools where I guest lecture.

  7. How does one capture only the GIF for use on a web page?

    (Why not attach your address to the GIF then people could make their way back?)

    It’s an extremely impressive piece of graphing. I think it’s going to have a lot of impact.

    1. Hi Bob,
      The image above was originally linked to the full GIF, but our webservers were struggling to cope. There are other copies of the GIF available, e.g. here.

  8. Hi Ed. This is a beautiful graphic and I fully support the scientific consensus behind what it shows. I have a minor concern over how it could possibly be slightly misleading, however.

    The way I interpret the plot, the area within the spiral corresponds to some measure of the magnitude of the year-round average temperature. If the temperature is scaled linearly along the radius of the plot, wouldn’t the interior area then grow with the square of the temperature, which might be misleading?

    1. Hi Jack – interesting point which a few others have made as well. I think different people will interpret the figure in different ways. If you are watching how close the spiral gets to the 1.5C line then I think a linear scale makes most sense. Others may see it differently. When I have some time, I will remake the figure with a square root scaled radial axis to see how much difference it makes!

      1. I agree with your point on the linear scale . Polar plots always set off warning bells in my head when circular area has any kind of significance to the interpretation of the plot, though. Would love to see what the square root scaled radial axis plot looks like.

      2. Hi,

        That would give a specific significance to the zero value when that value is quite arbitrary. You could use absolute values in which case the difference between a linear and square-root plot would be rather small.

      3. I very much agree with the point raised that the main mechanism which the viewer sees as at work is that a year’s worth of data (one full circle) is come to be represented by that circle’s area/size, which will over-state a difference in comparison to other years. I am not so much worried about the actual effect of that, but rather by the degree to which this makes the graphics vulnerable to argumentative attacks from certain sides. However, psychophysics to the rescue, until the square-root scaled plot is ready, one can defend by the fact that the human mind tends to underestimate area relationships: https://makingmaps.net/2007/08/28/perceptual-scaling-of-map-symbols/ and http://courses.cs.washington.edu/courses/cse512/15sp/lectures/CSE512-Perception.pdf . However, the apparent bible on the topic (https://www.edwardtufte.com/tufte/books_vdqi ) still seems to strongly advocate to not correct for that human bias…

  9. Hi, this is very interesting. Could you please plot two more animations – one with the data plus one standard deviation, and one with the data minus one standard deviation?

    1. I should clarify that I meant the variance output by the algorithm that estimates the global mean, not the variance of the input data.

  10. Hi Ed,

    What a powerful visualization. I would love to understand what tools you used to create it. Could you please comment?

    Thank you!

    1. Hi Joe,
      Good question – I made each frame in MATLAB and then combined into a large GIF. The colours of the lines are from the ‘viridis’ colour scale which is colour-blind friendly and perceptually uniform.

      1. Thanks for being colour-blind friendly. Graph reading can be really difficult for us tritanopes.

  11. Good visual. I imagine it though in 3-D. The 2-D information would stay the same but let it rise in the z axis with time. It woudl look like a tornado.

  12. Because man’s contribution to atmospheric carbon dioxide was so low in the 1800s (really pre-WWII), this chart is significant evidence that other factors than just man-made CO2 are in play.

    1. Of course – many factors have always been known to be important. Volcanoes, aerosols, solar activity, land-use change etc all play a role, as well as other greenhouse gases, but CO2 is the dominant factor.

    2. That is correct. Here is a good animated slide show that shows how the contributions from all the different factors add up to give us our observed trend. There are many factors, but CO2 has been the dominant contributor to the trend since around 1900. Interestingly, this also shows it would be even warmer if not for the cooling effect of man-made aerosol pollution. http://www.bloomberg.com/graphics/2015-whats-warming-the-world/

  13. So, there is a temperature increase. It is largely man made, this time, what about previous times? What happened previous times? Did people go to the moon or Mars and build vast cities, were people happier? And, when did species go extinct at a pace unmatched by hunting due to warming? Were there floods and droughts displacing legions of people in the data (because that is in the news)?

    1. On much longer timescales, changes in Earth’s orbit are the main culprit for climatic changes, e.g. the semi-regular ice ages over the past ~million years. Variations in volcanic eruption frequency and solar output are also sometimes important.

      1. OK, interesting. But my point is that in reality, chemical weapons are being used in Syria. In reality, people in China have chest pains and die from air pollution. It is a nice graphic. Millions of people are living in the poorest of conditions in refugee camps while cities like Doha are growing to become new pollution hubs. Ultra fine dust is entering the bloodstream of millions in Europe. Poland is like a burning open coal mine. The whole food chain is pervaded by aerial and ground water pollution and by profit driven junk food production.

  14. Is that why GISS starts in 1880, 1878 had inconvenient data. Looks like Ed accidentally let a little cat out of the bag, how can 1878 March be as hot as 1989 March with global warming?

  15. Hi, great graphic, but I miss the legend for the colors used. Violet in the Center -> blue -> green -> yellow. What does that exactly mean?
    Thanks for your answer!

  16. At least we have a new entry for https://en.wikipedia.org/wiki/Misleading_graph ?

    I believe it was Tufte who said that if you see circles on a chart, be immediately suspicious of biased presentation?

    “…By their very nature, circular gauges use a great deal of space to say relatively little, and they fail spectacularly when intended for comparison…”

    To be specific, such a graph as above INHERENTLY visually emphasizes later temperature increases which seem to be actually measured by their radius from the center (this is also misleading, mentioned later), yet appear visually either as circumference or area which increase by the 2(pi)r or (pi)r^2 respectively. (Setting aside the inherent inability of the graph to show any value less than positive.) The zero-value ‘circle’ is undesignated, so even a no-change result *appears* to be an increase as it’s likely perceived by the viewer to be ‘distance from center point’ not ‘distance from arbitrary circle around the center date’.
    Finally, the presentation more or less deliberately hides zero or small values in the “mess” while extreme increases are highlighted by the dark space around them, (deliberately?) taking advantage of probably the Müller-Lyer, Jastrow, or Hering illusion effects, not sure which.

    It’s a pretty animation, but egregiously misleading. Which is great if you’re preaching to the choir, I guess.

    1. Steve, is your snarky dismissal of Ed’s graph derived from a purely academic rejection of the messenger’s tools or do you genuinely dismiss the mountains of data confirming the message: The earth is getting warmer? I am curious how you dismiss the dozens of unadorned line graphs whose interpretation cannot be argued by any rational person (see link).


      And please don’t use the natural variability argument. The attached article clearly shows that human impact far exceeds anything natural.


      1. While Steve’s comment is perhaps not nicely worded it has validity.

        Consider: Take the negative value of all the anomalies and create the same animation. Would it leave the viewer with the same impression? I’d say no, so the presentation is biased. That is sad since Ed Hawkins is a proponent of avoiding biases in scientific presentations (see: “ending the rainbow”).

        Possible solution 1: Animation of two side by side circles. One for Negative, one for Positive anomaly.

        Possible solution 2: A circle plot with months from 0 to 180 degrees for positive values, and 180 to 360 for negative values (e.g. Jan would 0 to 15 degrees for positive, or 180 to 195 for negative). Radius would still represent change in anomaly.

  17. This is an interesting method for presenting the global time series. I especially like how you get to focus on the evolution of the time series one year at a time.

    Not to pile unto the many suggestions you have already received, but it would be interesting if you varied the line width each year to reflect uncertainty in the data set.

  18. Great graphic. Would be wonderful to see a corresponding graphic showing fossil fuel consumption…

  19. It would also be poignant to post an overlay of other atmospheric deposition products.
    The last 30 years have been measured. Magnesium, Sulfur, Sodium, Chlorine, Potassium, Nitrogen. Together they represent a comprehensive view of human impacts on the troposphere.

    1. Exactly. Why are climate scientists not talking publicly about the geoengineering programs going on?

  20. “we have control over what happens next”

    Do we? We have control over how much greenhouse gases we emit from now on, but there’s a lot already in the atmosphere and I don’t think we have a way of removing them other than waiting decades.

    Is there a way of cleaning up the mess previous generations have made?

Leave a Reply

Your email address will not be published. Required fields are marked *