V4C August Temperature and the size of V4 adjustments

Only GHCN-V4 is currently available for August. V3 has not been updated since 17th August so I wondered if it had perhaps been discontinued, but it turns out they have had  IT problems. My calculation of the latest V4C combined with HadSST3 gives a global temperature anomaly of 0.82C, which is an insignificant increase of 0.01C from July. The annual results are shown in Figure 1. in red compared to those of V4U in green.  What I find strange is that the discrepancy with the unadjusted data (V4U) after 2015 is large and growing. The annual data for 2019 is the average of the first 8 months.

Figure 1. Annual global temperatures for unadjusted and adjusted GHCN-V4 data combined with HadSST3

After 2015 the adjustments made to the recorded temperatures in V4U have increased dramatically. There is currently (2019) a huge difference of over 0.2C between V4C and V4U, which is unprecedented in recent times. We can see this in more detail by looking at the monthly temperatures.

A comparison of the last few years of monthly data from V4U(uncorrected) and V4C(adjusted)

These are changes in global temperatures so the real adjustments to land temperatures  are far larger – well exceeding 0.5C! To put these change into context let’s simply look at long term trends. Shown below are all monthly corrections made to V4 station data back to 1850, based on a standard 1961-1990 baseline. The zero adjustment baseline line is shown in black.

Monthly comparisons between V4C and V4U from 1880.


Before 1960 land temperatures have mostly been adjusted downwards while after 1990 they have been slightly adjusted upwards. Overall you can see that this process has added about 0.06C to global warming. However what is particularly alarming are the huge corrections that have made since 2016 which are comparable to or even larger than those made during  the 2nd world war years.

I have no idea as to what caused this sudden surge in temperature adjustments around 2015. Does anyone else ?

Finally here is a “slow” animation of the August temperature distribution calculated using spherical triangulation in 3D.

About Clive Best

PhD High Energy Physics Worked at CERN, Rutherford Lab, JET, JRC, OSVision
This entry was posted in AGW, climate science, NOAA and tagged . Bookmark the permalink.

15 Responses to V4C August Temperature and the size of V4 adjustments

  1. The Paris COP 21 conference in late 2015 and Trump’s election victory in late 2016 changed data doctoring strategy?

  2. Jim says:

    Stepping back to the solar system point of view and a layman’s perspective – the earths orbit is not exactly circular as are the other planets. All combine to modify the distance from the earth to the sun. Therefore the earth receives slightly different amounts of energy over the centuries as Jupiter’s orbit effects the earth’s orbit. Jupiter’s year is 12 of the earth’s year so there is at least a 12 year cycle withing the yearly climate cycles. I also understand that the sun is not a constant homogeneous machine. The sun’s output seems to vary with the amount of “sun spot” activity. Decades ago I heard an expert say that the sun was on a 11 year cycle. More recently I heard it was on a 7 year cycle. Half a millennium ago the Vikings settled in Labrador Canada and then left as the neighborhood cooled. Climates have varied over the centuries on longer cycles than we can discern over a century of data collection. Fourier analysis might hint at longer frequency cycles. One could extrapolate known weather to times when those conditions occurred in the past, but there would be a large about of uncertainty in the results. The sun has not been the same size and output for the past billion years. The sun is evolving as nuclear processes move forward. Comets drag matter from out side our system which could account for adding fuel to the sun over the millenniums. In the same sense, the sun will at some point dramatically change. The sun will collapse to a neutron star or other thing with a different energy output. Mixing politics with natural processes should be left to FB. If mankind doesn’t look after his planet – you won’t have to worry about small men in powerful positions. In Hawaii they banned suntan lotion and then there was a volcano eruption that killed more coral than any man made pollution! Man is not in control of nature so we had better get a handle on changing it for man to continue his existence in this vast universe. If we don’t get it right – a few a-bombs will set off a crescendo. I believe that God has other plans in mind!

    • Clive Best says:

      Jupiter’s orbit around the sun changes the earth-sun distance every 12 years. I calculate the maximum effect at about 0.8*10^6 km. The earth’s elliptical orbit changes the annual distance from the sun by 5*10^6. Perihelion occurs in early January or summer in the southern hemisphere.

      Insolation goes like 1/R^2 so the earth’s elliptical orbital has roughly 40 times more effect than Jupiter’s orbit. However Jupiter affects both hemisphere’s equally and over longer timescales. I am not aware that anyone has shown a 12 year cycle in climate. There are instead claims of a 60y cycle in the temperature record.

      We don’t have to worry about the sun’s evolution for a billion years or so !

  3. Fred Byren says:

    This is extradordinary. It conrtradicts satellite measurements . They do not show any warming during this period in fact a little cooling if anything. The next few years should clarify matters. The Met Office people should supply us with a plausible explanation. Surely, there hasnot been an decrease in Urban heating effect. That seems to me to be the only explanation.The satellite measurements are considered to be accurate. I hope the explanation is not skulduggery.

  4. Fred Byren says:

    Sorry about the typos. I’ve just had a cataraCT removed and am waiting for new spectacles

  5. What might be happening is this …

    Temperature data has been adjusted in 2016 and years since. These adjustments involved decreasing the historical data by a constant amount.

    The crucial point is that data for the baseline period for the deriving the anomalies, 1961-1990, was also adjusted downwards. With the 1961-90 averages being decreased by the same amount as all other data prior to 2016, those older anomalies don’t change. (Simple analogy: 10 – 6 = 4 but reduce but the 10 and the 6 by one to get 9 and 5 and we get 9 – 5 = 4)

    This downward adjustment of 1961-1990 data has lowered the average, which means that the anomalies for recent temperature data that hasn’t been adjusted are now HIGHER.

    Applying adjustments to more and more temperature data during 2018 and 2019 would account for the increasing global average temperature anomaly.

    • Clive Best says:

      You are quite right. If you adjust the data within the baseline period then it will have a dramatic effect on all temperatures for 150 years. This is the problem with using anomalies.

      This is exactly what has happened with the new HADSST4 ocean temperatures. They changed the correction for bucket and ERI measurements such that temperatures between 1961-1990 reduced slightly. This caused recent ‘anomalies’ based on buoys to apparently increase in exactly the way you show.

      I can probably check whether V3C have different normals to V3U.

      However the temperature difference increases after 2015 so this must be localised in those years.

      I suspect both effects are occurring.

  6. paulski0 says:

    There are a few oddities here. I don’t know if NOAA/GHCN publish a global land average with unadjusted data (there is one shown in the GHCNv4 paper but it’s smoothed and obviously out of date) but Berkeley Earth do. Comparing the Berkeley Earth final adjusted global land average against the raw average reveals negligible difference. Comparing a GISS land-only global average (from here), with GISS mostly being GHCNv4, with the Berkeley raw average does indicate an apparent recent structural “high stand” starting at 2012. However, the magnitude relative to the early 2000s is only about 0.05C, and that’s in a land-only average. That difference would be barely visible in a land+ocean global average such as you’re showing.

    And on that note, the time series for your V4U average looks strange given it’s land+ocean and the ocean data is identical. In HadSST3 2018 is more than 0.05C warmer than 2010 and about 0.1C warmer than 2005, but your V4U-HadSST3 average has 2018 more than 0.05C colder than 2010 and at the same level as 2005. Given the much greater fraction of ocean than land this suggests a representative land-only average of your analysis would have 2018 maybe around as cold as 2008, which is wildly implausible. And is not remotely similar to what the Berkeley Earth raw average produces.

    This leaves two primary possibilities that I can see: 1) There’s some very inaccurate raw data being supplied to GHCNv4 for recent years, that Berkeley Earth somehow aren’t also seeing… but the GHCNv4 corrections appear to deal with them appropriately, 2) There’s something not quite right with how you’re doing this analysis.

    I wonder if you can do a land-only version to avoid any confusion with the ocean data?

  7. Olof R says:

    Hi Clive,

    Nick Stokes makes a similar dataset, TempLSmesh, using GHCN v4 and ERSST5, both in adjusted and unadjusted versions.

    Data can be found here:


    Press the TempLS button and you will find 4 different datasets using adjusted or unadjusted GHCNv4 (8 versions totally), with data for the recent 6 years

    The differences between adjusted and unadjusted are very small, and nothing like those you show above. Here’s TempLSmesh:

    I don’t think the choice of SST dataset interacts with the effect of GHCN adjustments. It must be something else that is wrong..

    • Clive Best says:

      Nick Stokes and I normally get almost exactly the same results as we are doing nearly the same thing. I insist on all stations having a minimum of 10y of data within the baseline period which is maybe one difference. Remember also that I am comparing annual values rather than monthly values.

  8. Clive Best says:

    I found this in the updated V3 status.txt file.


    GHCNM version 3 is no longer operational (TAVG, TMAX amd TMIN). Users should use
    GHCNM version 4 for TAVG, found here: ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v4/

    In addition Version 4 TMAX and TMIN is under development and will be available in
    the future.”

  9. ljhooge says:

    3rd last paragraph. Should the .06 be .6? Wondering.

    • Clive Best says:

      The adjustments refer to the right hand axis. So I am being conservative in saying that the adjustments made to V4U in the early periods to reduce temperatures increase apparent warming by about 0.06C. We could simply say 0.1C instead. However, it is the adjustments made in the last few years that are surprising.

Leave a Reply