Thursday, July 16, 2015

Watching the global thermometer - year to date GISTemp with June 2015

NOTE: GISS has announced corrections and thanked Nick Stokes (see comments below). Refer to the updated version of this article.


Every month since March, I've posted a chart of the progressive year-to-date global average surface temperature, from GISS. This is the update with June included. I'll repeat the explanation with each update and add what seem to be things to watch.

Note: I made a mistake with the June YTD in the original, which I've now fixed. (Sou 9:00 am)

Update: I've added a chart of annual temps for GISTemp, comparing the version put out in June with the version put out in July. That's so that you can see what difference the shift to ERSST v4 makes.
Sou 12:18 pm Thursday 16 July 2015


Worth noting


  • June was an average of 0.76°C above the 1951-1980 mean and is equal hottest June on record, with June 1998.
  • The only other month this year that's been hottest on record is February with an anomaly of 0.89°C.
  • The highest anomaly this year is now March at 0.91°C, which makes it the second hottest March after March 2010, at 0.92°C.
  • 2015 is still hottest on record so far, equal with 2010. With the adoption of ERSST v4, some of the temperatures are higher. So are those of some other years, particularly in 2010, temperatures have been upped quite a bit. But all have changed.
  • May is still relatively cool, unlike in some other data sets. By cool I don't mean cold. It was 0.73C above the 1951-1980 mean. It's just that most people thought it would be among the hottest of Mays.
  • The lowest anomaly was in April this year, at 0.71°C above the 1951-1980 mean. 
  • The progressive year to date average up to and including June is 0.8°C above the 1951-1980 mean. In June 2010 it was also 0.8°C above.




Explaining the chart


The chart is a progressive year to date average for all years from 1995 to the present. (Thanks to JCH for the tip.) What that means is for January each year, it just shows the anomaly for January. For February it shows the average of January and February for each year. For March, its the average of the monthly anomaly from January to March.

If you look at December, each year shows the annual average temperature for the full year. For November, each year has the average for the year up to November, not including December. (As before, I've made it extra large because of all the fine detail.)

Data Source: NASA GISS

2015 is now equal with 2010. The years to watch are 2014, 2010 and 2005. I've plotted them with slightly thicker lines so they stand out more easily.

The coldest year of the lot was 1996, which still ended up more than 0.3°C above the 1950 to 1981 average.  The next time someone tries to tell you that "it hasn't warmed since 1996" then show them this chart :)


GISTemp changes with ERSST v4


Now that GISTemp is using ERSST v4, the temperature data has changed some over all years. Below is a chart showing the annual average for the June version and this latest July version, which uses ERSST v4. I've also plotted the difference between the two versions. It's really piddly in the whole scheme of things as you can see.

Data Source: NASA GISS

Points to note:
  • The 2015 temperature for both series is year to date average to May (not including June), so they can be compared.
  • The "difference" is the June version minus the July version. Negative means the July version is higher than the June version.
  • As you can see, mostly it's almost zilch. The biggest difference is 1943, which is 0.09 so still less than 0.1C difference. Most years it's +/- 0.01C to 0.03C. 
  • The July version is higher than the June version at both ends. There's barely any difference in the middle of last century.
  • There's that questionable peak around 1940 that someone will undoubtedly be able to sort out one of these days. (Probably because something different happened on war vessels to how the temps were usually taken.)

Related updates




I'll add a link to the table shortly. Here is the link to the table for the top Year to Date chart.

30 comments:

  1. Because GISTemp has shifted to using ERSST v4, all the numbers have changed. I'll do a comparison plot so you can see the difference - but I may not get to it for a little while yet.

    ReplyDelete
  2. Ah, good. You now have 2015 YTD correctly as .80, whereas you had it at .78 a few minutes ago. Bring on the comparison between the old GISTEMP and new. Can't wait to see that :-)

    I've been putting this off for a long time now, but I'm a freakin' professional programmer. It's about high time I downloaded something like R and started playing with it...

    ReplyDelete
    Replies
    1. I use R for various things all the time. If you are a professional programmer it should drive you crazy for the first few weeks but the graphics alone are worth the aggravation. Be sure to install the ggplot2 package, of the the best of the plotting packages. And get a good R dedicated text editor or IDE. Best of luck

      Delete
    2. I use R for various things all the time. If you are a professional programmer it should drive you crazy for the first few weeks but the graphics alone are worth the aggravation. Be sure to install the ggplot2 package, of the the best of the plotting packages. And get a good R dedicated text editor or IDE. Best of luck

      Delete
    3. I use "R Studio" for editor and IDE. It is pretty good.

      Delete
    4. Thanks for the tips, folks. Looks like ggplot2 will work with R-Studio too. Downloading this stuff now...

      Delete
    5. There are several R IDEs. Besides R-Studio, Revolution R has a free IDE here http://www.revolutionanalytics.com/revolution-r-open (which is compatible with R-Studio as well, btw). There are a number of others. See https://en.wikipedia.org/wiki/R_%28programming_language%29#Interfaces All can use any R package found on CRAN including ggplot2.

      Some, like Revolution R Enterprise, do have their own additional functions to increase capabilities. But the normal R user doesn't need them normally unless they are into big data.

      Delete
  3. Yeah, I messed up the June number - but it was changed fairly soon after I posted the article. (Might have needed a browser refresh.)

    I've now added the annual comparison of pre- and post-ERSST, showing the two line charts plus a bar chart of the difference - on the one chart above.

    ReplyDelete
  4. Sou, I think you need to update the first plot some more. March 2015 is now given as 0.91, for example.

    ReplyDelete
    Replies
    1. It's the latest (July) data, Andy. The March 2015 number is the average of January, February and March - not the value for March. The purpose is to track how 2015 is going compared to prior years - at a glance.

      You're not the first person confused by what I've done :)

      Delete
    2. Aha! Sorry.

      Is it possible for you to highlight 1998? It's a little hard for my poor old eyes to distinguish.

      Delete
    3. Andy, I was caught twice by the running mean in the graph when we were speaking about monthly means in the comments.

      It makes sense to plot the values that way though, because it shows at a glance how each year has trended (and how this year is trending) toward the annual anomaly.

      Delete
    4. Done, Andy. 1998 is now a thick dashed purple line.

      Delete
    5. Thanks, Sou.

      It's amazing how many years beat 1998 now, nine by my count. But I thought warming stopped then? Yet it is now barely in the top ten, very likely to be relegated to the second division of warm years by 2015.

      I demand a recount, an audit! (Coming soon to blog near you.)

      Delete
  5. Another change that has slipped below the radar (as far as I can tell anyway) is NOAA replaced GHCN-M version 3.2.2 with version 3.3.0 in June 2015. It changed the ranking of at least one of the years I believe.

    ReplyDelete
  6. Sou,
    GISS posted a discussion of the change here. The difference plot shows a pretty steady recent difference of 0.06°C; v4 higher. But a puzzle is that April and May were quite exceptional, graph here. I think they may have made some mistake in the update. Those two months look like old values (April was unchanged). If they have, then 2015 will be back in front.

    ReplyDelete
    Replies
    1. A lot of people were thinking that May was too low, compared to other datasets, so you could be right there Nick. Have you asked anyone at GISS about it?

      Here's a spreadsheet for anyone interested. It shows the monthly anomalies as at June and the monthly anomalies as at July, and the difference - July minus June. I've only included years from 1990 onwards.

      In recent years the adjustment was relatively high (around 0.06 on average), but not for April or May this year. May should probably have gone up as much as it did using ERSST v3b. With v4 some of us would have expected it to go up quite a bit more, with sea surface temps showing the hottest May on record. And April should possibly have been higher too. It didn't change.

      Delete
    2. Sou,
      No, I haven't contacted GISS. I've found something a bit similar in TempLS, though with more irregularity earlier, which I'm trying to work out.

      The low value in May with GISS 3b, when as you say SST went up, and NOAA LO went way up, may be part of the story. But it is the March-April transition that stands out.

      Delete
    3. I did contact them, and now they've fixed it and thanked you, Nick :)

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

      (h/t David Sanger)

      Delete
    4. Just wait for the reaction from our little Conspiracy Theorists! :-)

      Delete
    5. They are a bit slow on the uptake, as usual :D

      Delete
    6. Thanks, Sou. I think the acknowledgement counts for you too. It does make quite a difference.

      Delete
    7. It was nothing to do with me, Nick. I have it on good authority it was GISS people reading your article at Moyhu :)

      Delete
  7. Sou, a very minor crow to pluck with you: shouldn't the difference be inverted in the graph? I believe you have graphed old - new, when the other way around would be more intuitive.

    ReplyDelete
    Replies
    1. I can do that if you want me to. Maybe next time.

      Delete
  8. In the new version, 2014 is no longer the hottest year: we're back to 2010 now.

    ReplyDelete
    Replies
    1. Not according to GISTemp (see above). 2014 is still the hottest year so far, but probably not for long.

      Delete
  9. GISS just announced updated figures since the found an error. June now warmest in their record, +0.80

    ReplyDelete
    Replies
    1. Thanks David. Rather than correct this article, I've written a new one.

      Delete

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. Details here.

Click here to read the HotWhopper comment policy.