The Diverging Surface Thermometer and Satellite Temperature Records Again

Joint post with Roger Andrews

In my recent post titled The Diverging Surface Thermometer and Satellite Temperature Records, Roger posted 4 charts in the comments that I felt were both interesting and important. For those not up to speed with the importance of comparing surface thermometer with satellite data it boils down to understanding the rate of global warming and whether or not the lower troposphere is still warming as all the climate models predict should be happening. The satellite data (UAH and RSS) shows little to no warming since 1997 – the famous pause – while all the surface thermometer records now show continued warming (e.g. GISS LOTI and HadCRUT4). One of the data sets must be faulty and Roger’s charts cast some further light on this issue.

Here we deal with the charts in the reverse order from Roger’s comment. Figure 1 compares the UAH satellite (over sea only) and Hadley sea surface temperature (HadSST3) records. HadSST3 makes up 71% of HadCRUT4, the combined land-ocean record commonly used to define the Earth’s ‘surface temperature’. The grey trace shows the difference between these two data sets and shows a near flat line. There is literally no difference between surface temperatures and temperatures measured by satellites over the oceans. This suggests that both these data sets are reliable.

Figure 1 Satellite over-sea temperature (UAH) compared with sea surface temperature (HadSST) data. The difference between the two is the grey trace at bottom which shows the gradient of satellite and surface data over oceans are the same.

This points to the faulty data lying in the land based data. Why should the satellites work reliably over the oceans and not over land? We can’t think of a reason. But there is a good reason for suspecting the land based thermometers since these measure temperatures in a totally different way to the SSTs. Roger has long argued that SST and air temperature data over land should not be combined into a single index since they are measuring different things.

If we do the same comparison over land we see there is a difference (Figure 2) with the land based Crutem4 index showing about 0.3˚C more warming than the UAH over-land data since 1980. One needs to recall that the difference between the gross satellite and surface data is tiny, of the order +0.15˚C since 1980. Since the land based data only accounts for 29% of the total, this difference between the land based data sets may account for about 0.09˚C of the gross difference, i.e. most of it.

Figure 2 Comparing the satellite data over land with the land surface thermometer data shows a small difference with the surface thermometers running about 0.3˚C warmer. Enough to explain most of the difference between the global satellite and surface indexes.

We can take this a step further. Comparing the UAH over-land and over-sea we find that the air over land does indeed appear to be warming marginally more rapidly than the oceans. I would feel inclined to put that down to non GHG related human activity such as urban sprawl, deforestation and irrigation.

Figure 3 Comparing satellite data over sea and over land suggests that the land may be warming slightly faster.

If we compare the surface records we see that the land is warming much faster than the oceans, of the order +0.5˚C since 1980. Does this mean that CO2 is a more potent GHG over land? We don’t think so. The different behaviour can be explained by either adjustments made to land surface thermometer records or by land surface thermometers being more sensitive to growing population and land surface changes than satellites. The thermometers are after all normally located close to human habitation.

Figure 4 Comparing the surface thermometer data over land and over sea suggests that the air over land is warming much faster than the oceans.

Finally, in the comments Luis pointed out that HadCRUT3 was largely in agreement with UAH since 1997. So I have checked this out.

Figure 5 Comparison of HadCRUT3, 4 and UAH since 1997. HadCRUT4 re-writes the record books. Note the offset between HadCRUT and UAH is down to different datum / base periods used.

It is indeed the case that since 1997, UAH has been on a flat, slowly declining trend (-0.13˚C per century). HadCRUT3 was on a flat, slowly rising trend (+0.18˚C per century). HadCRUT4 is on a much more steeply rising trend of +0.59˚C per century. With a stroke of the brush, the pause was written out of history.

Concluding Comment

Satellite and surface thermometer data agree over the oceans. They used to agree better over land until HadCRUT4 supplanted HadCRUT3, ending the pause and causing land surface thermometers to diverge from the satellite data sets.

This entry was posted in Climate change and tagged , , , , , . Bookmark the permalink.

27 Responses to The Diverging Surface Thermometer and Satellite Temperature Records Again

  1. Euan Mearns says:

    The bigger picture comparing Had3 and Had4. From this it can be seen that the main divergence occurs in the post 1945 cooling episode and in the post-2000 flat spot. There is no material difference in the gradient since 1850.

  2. Yvan Dutil says:

    Temperature growth faster over land because the thermal inertia of land is much lower than water, also there is less evaporation.

    • Javier says:


      The different behaviour can be explained by either adjustments made to land surface thermometer records or by land surface thermometers being more sensitive to growing population and land surface changes than satellites.

      On a warming planet global land temperatures should rise faster than global sea surface temperatures, while on a cooling planet the opposite is expected, global land temperatures should lower faster than global SSTs. Humans can contribute to that, but they are not needed to explain the difference. If we are not able to quantify the natural and human contributions, that difference only tells us that the planet is warming.

    • Euan Mearns says:

      Thanks Yvan, no one is disputing that the atmosphere is warming. The debate is about rate, cause and the most reliable means of measuring.

      One of the things I’m trying to bring to the fore here is that the differences are tiny between different means of measuring.

      • Javier says:

        Euan, If you want to discuss rate of warming, then you should look at rate of warming.

        For example this UK Met Office graph through BBC:

        Or this graph from the European Environment Agency:

        On the first one you can see very clearly that land temperatures (red curve) have a more extreme rate of change since the start of the records in 1880. This alone explains why we are seeing more warming on land.

        On the second one you can see that differences in records don’t matter much because sooner or later the warming rate has to become very similar for all the databases.

        Based on rates of warming (see first graph), it is very difficult to make the argument that the planet is warming significantly faster in the last 120 years as the peak rate of warming appears to be the same, about 0.4°C/decade. The warming is coming from the peak rate of cooling going down from -0.4°C, to -0.2°C, to 0°C. Interesting, isn’t it?

        Regarding the causes, I would say that CO2 makes a good candidate for explaining at least a good part of the reduction in the rate of cooling, but then it would look as it has very little effect on the rate of warming. The hypothesis would be that something else is warming the planet and CO2 is reducing the cooling. That is not very supportive of CO2 based catastrophism, as increased CO2 seems to be reaching its limits in its cooling quenching capacity, and whatever is warming the planet must follow its natural oscillation, be it centennial or millennial.

        Regarding the most reliable means of measuring temperature changes, it does not matter in the long run, but in the short run it matters a lot for political reasons. On theoretical grounds, satellites are a lot better in measuring globally, much cheaper that stuffing remote weather stations (there is a reason for the huge drop in stations in the 90’s), and less subject to artifacts and variability, so it is clear that they are better for giving a global picture, while thermometers are much better at recording local changes.

  3. Elvis says:

    I’m wondering about the theoretical foundation to the expected correlations. Is it generally accepted that thermometer readings of near surface temperatures really should track the satellite measurements of the bottom 5km of the atmosphere? They are after all physically separate with the atmosphere undergoing convection and translation (wind). Also, is the microwave spectrum emitted by land the same as that emitted by water?

    These question seem fundamental to understanding the records.

    • duwayne says:

      Elvis, those are good questions. As to the first question, it is my understanding that greenhouse gas theory and the modelling thereof predicts that temperatures in the lower atmosphere will rise faster, about 1.2 times faster, than the surface as the earth warms due to increased atmospheric CO2. This makes the disparity between the reported satellite temperatures and the surface temperatures which show the reverse even more puzzling.

    • Euan Mearns says:

      Elvis, good questions. I can’t answer. I’ll see if I can get Clive Best to call by and provide his greater insight. Following from what Yvan said, warming over the oceans is less because there the Green House effect is dominated by water vapour. Extending that argument, one may expect the CO2 greenhouse effect to be greatest over deserts.

      • Javier says:

        Actually, CO2 greenhouse effect should be greatest over polar regions where the air is the driest.

        “Specific humidity is at a maximum in the Tropics and declines as one moves poleward reaching minimum values in each polar region. This implies that the air in the most “arid” desert regions of the planet (30 degrees N or S latitude) actually contains more water vapor than temperate regions much farther north, and certainly more than the polar regions.”

        Deserts have a lower relative humidity but higher specific humidity than polar regions. Greenhouse effect only cares about the mass of water vapor, not its relative value to temperature dependent capacity.

      • Elvis says:

        I’m uncertain whether you comment on deserts was tongue in cheek or not. CO2 is well mixed (concentration much the same everywhere), so it will have the same greenhouse effect everywhere.

        • Euan Mearns says:

          It wasn’t tongue in cheek. My understanding of Yvan’s comment was that over the oceans, the GHE was buffered to water vapour. dCO2 of 0.014% has little relative effect when the air contains 1 to 5% H2O.

          And I’m afraid I don’t undersand Javier’s comment about specific humidity. My knowledge and understanding has boundaries. I thought the air over deserts was extremely dry, hence at night they cool down really fast. And I find it hard to swallow that grey leaden skies i’m looking at right now have less water vapour than the Sahara.

          And the GHE is not the same everywhere since it depends on incident solar radiation and nature of the secondary irradiating surface and on the temperature gradient through the air at the CO2 emission height. If it is getting warmer as you go up, i.e. in the stratosphere, then CO2 actually works as a conductor and not an insulator.

          But lets not lose sight of what this post is about – the various ways of measuring lower troposphere temperature all agree with remarkable consistency, the differences are tiny, a few tenths of a degree, but the diverging trend between satellites and thermometers is significant and hugely important for political reasons.

          • Javier says:


            Specific humidity is the amount of H2O vapor per unit of air mass, so it is a measure of total amount of water vapor, while relative humidity is the percentage of H2O vapor respect the maximum amount of H2O vapor that the air can hold, so it is a measure of relative saturation.

            Over the deserts relative humidity can be as low as 20% due to hot air being able to hold a lot of water vapor, and yet the total amount of water vapor per unit of air mass is still significantly higher than over polar regions, where with very little water vapor the relative humidity can be as high as 80%.

            The air over polar regions is the driest on Earth in absolute terms. You can check it very easily in any text book or over the internet. As the greenhouse effect only cares about the number of molecules, CO2 molecules get the most bung for the buck over polar regions as more IR radiation is available for them.

            My hunch is that CO2 effect is not as we have been told and is maximal over polar regions and acts mainly reducing the equatorial-polar thermal gradient that is the engine that cools the Earth. So we are not warming unusually, we are just not cooling as efficiently. I think there is evidence that this could be the case.

          • Elvis says:

            Euan, yes indeed, the differences are small. Getting to the bottom of the divergence is important.

            Following on what Javier was saying, here’s a view of Total Precipitable Water vapor in May 2009, which is akin to specific humidity.

            Also to bear in mind is that CO2 is well mixed and will be 400 parts per million up to the space station (there’s a lot fewer millions up there though). Water vapor on the other hand is predominantly in the low atmosphere, 95% of it below 6km. Beyond the tropopause (12km) is pretty dry.

          • Euan Mearns says:

            @ Javier, the first part of your explanation is stretching my comprehension. And this is where I need 4D climate models I can trust, which are sadly lacking. But on this I have an opinion:

            My hunch is that CO2 effect is not as we have been told and is maximal over polar regions and acts mainly reducing the equatorial-polar thermal gradient that is the engine that cools the Earth.

            My opinion is that in Antarctica, in winter there is no sun and so no GHE. In summer, the surface lies in the stratosphere and so CO2 acts as a conductor feeding heat away from the surface. But since the surface is covered in ice, there is no secondary IR to feed into space.

            Not sure what happens over the Arctic.

          • Javier says:


            I am far from being knowledgeable in these things, but AFAIK greenhouse effect acts mainly on IR going from the surface upwards, and so it cares little for the sun, specially in the polar regions, where the incidence angle is low and they are energy deficient (i.e. they radiate to space a lot more energy that they receive from space).

            Most of the energy radiated to space at polar regions is not of polar origin. See for example:

            It is energy transferred to polar regions from tropical areas through atmosphere (Hadley cells -> Ferrer cells -> polar cells) and oceanic currents (warm superficial waters to the poles and cold deep waters away from the poles through the Meridional Overturning Circulation) in the form of heat. Thus most of the IR that is radiated to space from polar regions comes from far away and goes up day and night, summer and winter. The entire Earth is venting to space through the polar regions and CO2 has a party there reducing that venting without almost any competition from water vapor due to the extreme dryness of the polar air.

            That is in my opinion the basis of the Arctic amplification, the increased warming over the Arctic. Regarding Antarctica, the increased radiation to space does not take place mainly over the center of the continent, but over the relatively warmer Southern Ocean that surrounds it, that is fed warm water from the Pacific and Indian oceans and were waters cool down and lose heat.

    • Elvis says:

      When I first thought about this I thought, well of course the troposphere should track the surface – after all, the basic explanation of the GHE is that extra CO2 raises the altitude at which radiation to space occurs and, assuming a fixed lapse rate, we can follow that back down through many km of atmosphere and get an increased surface temp. The logic of the explanation is that the troposphere heats up in step with the surface. However, this basic explanation is doubtless a simplification; wind and convection make it much more complicated.

      Added to that, the TLT is for the lower 5km and the troposphere is perhaps 12km and contains about 1.5 times as much stuff as the lower 5km (half of the atmosphere is in the lower 5.6km, half of the rest in the next 5.6km, half of the rest … etc). And there is the contamination of the T2 microwave band by surface radiation (which does indeed vary depending upon what the surface is, water, ice and soil all having different characteristics) and of T2 and 3 by stratospheric radiation, all of which need to be removed to create TLT. And that is not mentioning the cross calibration and orbital issues of the dozen (or more?) satellites involved over the years. After all that, it would not surprise me to learn that TLT (as measured) does not track the thermometer record so well, even if theoretically is should (and I still don’t know if it should :-).

  4. Willem Post says:

    If an extra terrestrial person compared a series of European “satellite-like” maps since 1850, and he knew nothing about CO2, he would see huge changes in deforestation, industrial farming, metropolitan sprawl, and transport interconnections, and conclude the European climate would, all other things being equal, show an increase in temperature.

    Then, someone comes along and tells him about fossil fuels, etc., and the ET’s response might be, oh, I thought the energy had come from nuclear.

  5. Alan Poirier says:

    Nicely done. Personally, my own research leads me to believe GISS temperatures, which form the basis of most land/sea data sets, have been subject to so many “administrative” changes over the past few years that they bear little resemblance to reality. The divergence in trends between the satellites and thermometers is now approaching ridiculous levels, as a number of bloggers have already pointed out. I refer your readers to the excellent work Dr. Ole Humlum has done in this respect over at Of course, the premier debunker of GISS and NOAA remains Steve S Goddard, aka Tony Heller.

    • Roberto says:

      ‘Of course, the premier debunker of GISS and NOAA remains Steve S Goddard, aka Tony Heller.’

      Where?… I’ve visited T Heller’s blog and after reading for a while I’ve not seem a single datum, only useless childish posts.

      • A C Osborn says:

        None so blind as those who do not wish to see.

        • Elvis says:

          I suppose you are referring to Heller. You and Roberto are right. His blog does seem to be a mixture of cherry picking and crying “fraud” without cause as often as he can. Yet he seems to have some qualifications so is not totally without intellect. Remarkable.

  6. People may be interested in this old post by Tamino, which seems relevant to this topic. There are also other interesting posts, about the so-called pause, on that site.

  7. Luís says:

    Hi again Euan, I’ll be more careful with comments onwards, folk actually read them.

    I just want to note another thing, the trend since 1997 in version 5 of the UAH dataset is actually positive:

    Version 6 is still in beta and missing some of the series (poles, etc). It might too early to take conclusions on it.

    In any case, the step from HadCRUT3 to HadCRUT4 was clearly a move away from the satellite data close to the peer thermometer datasets. This means modifications are almost exclusive to the past 10 years or so. This definitely needs some deeper study, are such precise modifications to the dataset legitimate or nor? I hope you can use your peer network to get to the bottom of this.


  8. clivebest says:

    There is another subtle difference between Hadcrut3 and Hadcrut4 which is not well known. When CRU calculated CRUTEM3 they summed both hemispheres and took the average (NH+SH)/2

    Initially they did the same with CRUTEM4 but some time soon after they switched to (2NH+SH)/3. This caused a significant increase to the land average. So it really depends how CRUTEM4 and HADSST are merged.

    CRUTEM4 also added over 600 stations in Arctic regions. Yet they added none in Antarctica and dropped many in South America. They chose to populate the small (lat,lon) grids near the North pole but decrease those near the South pole. I wonder why?

  9. Pingback: Recent Energy And Environmental News – October 5th 2015 | PA Pundits - International

Comments are closed.