The iconic plot of Global Mean Surface Temperature (GMST), such as figure 1 above which is taken from the article, Indicators of Global Climate Change 2022, form the focus and principle argument for action to combat Climate Change. This plot is an average over the whole global while in reality the temperature evolution is very different from region to region around the world.
In this iconic plot, a level of temperature rise of 1.5°C relative to a “pre-industrial” period, is frequently quoted to suggesting the point where serious, irreversible, damage will occur for the environment. The website, Indicators of Global Climate Change for Policy Makers, explores in detail when the temperature values will exceed the 1.5°C limit. Figure 2 shows a plot from this website.
Figure 2. Plot from IGCC with decadal average GMST data curve (light grey) compared to their predicted human-induced global warming (orange). The annotations, “Natural?”, for the dip in the data around 1910 and the bump around 1940 have been added to the original image.
The current page studies the uncertainties on temperature measurements and the variations of GMST data around the world in order to see if the focus on a single plot makes any scientific sense. The 1940s bump in figure 2 is very prominent in the data of some regions but not others. A secondary question is if the fuss about a precise 1.5°C level has any significant meaning.
Differences between land, sea, north, south
Figure 3 shows the world divided between northern and southern hemispheres and between land and sea. For the northern hemisphere it can be seen that the temperature increase from the 1850-1900 reference period to 2022 is 1.9°C over land while it is only 1.0°C over the sea. For the southern hemisphere the increase is 1.3°C over land and 0.9°C over sea.
Figure 3. Plot of evolution of temperature for Northern and Southern Hemispheres showing temperature anomaly separately over land regions and sea regions of the world. (HadCRUT5 data, 1850-1900 reference period).
Temperature measurements have uncertainties
Any measurement of a physical quantity has an experimental error and the proper treatment of these errors should be an essential part of any analysis of experimental data. The data in the iconic GMST plot comes from numerous individual measurements around the world. A major input into the global mean temperatures comes from land based weather stations, the number of which varies with year in the range 10000-20000 stations. In early years, the instruments used were thermometers each of which could be read with an experimental accuracy of 1-2°C. In a full year this make 4-7 million recordings. If these recordings were uncorrelated and unbiased the uncertainty on the combined annual average would be less than 0.001°C. In fact the measurements are not uncorrelated and unbiased and so the real uncertainty is much bigger than this tiny value. The groups providing the data used for the GMST sometimes provide data uncertainties and sometimes do not. The HadCRUT group provides uncertainties which can be converted to one standard deviation error bars. These are plotted on the GMST data below and are in the range 0.05-0.09°C for the years 1850-1950 and in the range 0.01-0.02°C for the years 1970-2022.
The GMST plots are presented as “Temperature Anomalies” where the data is compared relative to a reference period. Sometimes the reference period is 1850-1900 but sometimes other periods are used for various reasons. Figure 4 displays the GMST data with error bars and with different reference periods showing the change. The conclusion must be that quoting any temperature rise relative to pre-industrial time with a precision less than 0.05°C is meaningless.
Figure 4. GMST data plot with error bars from HadCRUT. The data is displayed with two different reference periods 1850-1880 and 1850-1880.
Temperature evolution is not monotonic
Figure 5. Fits to iconic Global Mean Surface Temperature described in Clintel’s IPCC Report 2023.
The iconic temperature evolution plot clearly does not show a monotonic increase in temperature as would be expected if the change in temperature was only controlled by a monotonic rise in atmospheric CO2 levels. Figure 5 shows fits to different time periods indicating falling temperatures in the periods 1850-1910, 1945-1976 and increasing temperatures in the periods 1910-1945, 1976-2022.
Temperature rise are largest in the Arctic
The slopes of the temperature changes in different time periods differ greatly around the globe. For the plots in the rest of this page the data is mainly taken the ERA5 global analysis reanalysis which covers both land and sea regions. Linear fits to this data are made allowing different slopes for two time periods 1945-1976 and 1977-2022. The example plots in figure 6 show linear fits for three latitude ranges. These plots indicate different trends: for the high latitudes, 3°C temperature rise 1977-2022 but lower slope for 1945-1976; for mid-latitudes, 2°C rise for 1977-2022 but 0.5°C fall for 1945-1976 and for low-latitudes, 0.6°C rise for 1977-2022 and lower slope for 1945-1977.
Figure 6. Example temperature evolution plots for three latitude ranges: high-latitudes 75-90°N; mid-latitudes 30-45°N and low-latitudes 0-15°N.
Figure 7 summarizes the temperature trends as a function of latitude for the two time periods. In the southern hemisphere the slope are much smaller than in the northern hemisphere with negative slopes in the some regions.
Figure 7. Temperature evolution slopes vs. latitude averaged over all longitudes. Plotted for two time periods 1945-1976 and 1977-2022.
Figures 8 and 9 show the slopes of the temperature trend data in 15° X 15° bins of latitude x longitude. The first plot show the temperature slopes for 1977-2022 and the second for 1945-1976. Again the very significant difference in warming between the two time periods is seen with the warming being largest in the longitude regions north of Siberia.
Figure 8. Temperature trends for bins of 15°x15° of latitude and longitude, for 1976-2022. Red regions have temperature increases >7°C/century and white increases <1°C/century.
Figure 9. Temperature trends for bins of 15°x15° of latitude and longitude, for 1945-1976. Orange regions have temperature increases >5°C/century and blue regions have decreases >3°C/century.
The high warming in the Siberian arctic region in 1977-2022 is well known and described in the paper: Rantanen et al., The Arctic has warmed nearly four times faster than the globe since 1979. Figure 10 shows the high warming region analysed in this paper.
Figure 10. Annual mean temperature trends for the period 1979–2021, derived from the average of the observational datasets from the paper Rantanan et al.
The page on the website: Plots for Arctic Land Regions, show temperature evolution plots for various arctic land regions from 1900-2022 taken from Berkeley Earth. These particular plots, together with figure 5, show that the drops in temperature in the period 1945-1976 in some regions of figure 9 are preceded by increases in the period 1910-1945, so the driving feature is the bump around 1940 seen in figure 2. In the Arctic Land Region plots the 1940 bump appears with very different amplitudes in difference places. The origin of this bump which is surely “natural” and not caused by a monotonic increase of CO2 levels needs to be explained.
Temperature rise in the Arctic is mainly in the winter
Seasonal variations are another big factor in temperature trends. Figure 11 shows examples of summer/winter variations at Arctic and Antarctic latitudes. For the Arctic most of the warming takes place in the northern winter and for the Antarctic the largest warming is also in the southern winter, but the warming is much smaller than the Arctic. Figure 12 illustrates the summer/winter variations as a function of latitude.
Figure 11. Examples of seasonal variations of temperature evolution for latitude ranges: Arctic 75-90°N and Antarctic 0-15°N, both integrated over all longitudes. Blue curve is Jan, Feb, Nov, Dec (NH winter) and red curve are May, Jun, Jul, Aug (SH winter).
Figure 12. Seasonal variations of temperature evolution slopes vs. latitude averaged over all longitudes. The two curves are for the month ranges indicated and for the time periods 1977-2022.
Figure 13. Histograms of temperature trend slopes in bins of 15° X 15° in latitude x longitude as in figures 8, 9. Left-hand plots are for fits to the period 1977-2022 and right-hand to the period 1946-1976. The top plots are fits to the full year and the bottom plots are to the summer (May, Jun, Jul, Aug for NH, Nov, Dec, Jan, Feb for SH) and for the winter (reverse month attribution).
Figure 13 illustrates the vast range of temperature trends around the globe by making histograms of the slopes displayed in figures 8 and figure 9. The histograms are for the two periods and for the full year, summer and winter as indicated. Again it can be seen that the temperature increases are much larger for the 1977-2022 period than the 1945-1976 period and that the slopes for some latitude, longitude bins in winter have temperature increases up to 20°C/century. Some regions for the period 1955-1976 have temperature decreases up to 5°C/century.
Figure 14. Temperature trends in Northern Hemisphere and Southern Hemisphere temperate zones and the Arctic and the Antarctic.
Figure 14 emphasises the major differences in warming trends for the Northern and Southern Hemispheres where for the temperate zones, where the north warms 2.5 faster than the south during the periods 1976-2022 and the Arctic which warms 4.6 times faster than the Antarctic in the same period.
Conclusions
The data clearly shows that on average the surface temperature around the world is increasing. This increase is very clear in the period 1977-2022 but less clear for earlier periods with some periods where the temperature actually decreased. The regions where the temperature increases are large in the Arctic show a very much bigger rise in winter compared to summer.
The iconic plot of global mean surface temperature hides a multitude of temperature variations which are much larger than the anomalies in this particular plot. It is unconvincing that comparisons, such as figure 2, showing “human-induced warming” with the temperature trend data prove that the causes of the temperature changes are completely understood. Many questions must be answered:
- Why does the Northern Hemisphere warm so much more than the Southern Hemisphere?
- What causes the dip in the data around 1910 and the bump around 1940?
- What is special about the Siberian Arctic?
- To what extent do climate models explain the regional effects in general.
Finally, the emphasis on the limit of “1.5°C rise pre-industrial” is obviously used as a propaganda stunt and has nothing to do with scientific interest. As well as the large regional variations, the value depends completely on the definition of the pre-industrial reference period during a time where the temperature uncertainties are large and have significant temperature variations with year.
This post is copied from the website of the author, John Carr. This website, climate-and-hope.net , contains much data on Electricity Generation and Technologies and subjects related to Climate Change.
John is an ex-colleague of Clive at CERN.
You also need to take into account the actual data availability which, on the oceans in the southern hemisphere, can vary quite a lot, especially before the satellite era..
https://imgur.com/gallery/Ulah5KV
The use of anomalies to derive global temperatures hides huge spatial and temporal differences in coverage, and in methodology with time.
Perhaps the only consistent long term temperature measurements is CET (Central England Temperatures). This shows only a modest rise in temperature averages and annual extremes over more than 370 years.
Thanks for the interesting post John.
Thanks for the graph!
The uptick in the late 70’s is close to the time frame I started driving. Back then the limited number of roads seemed to have been paved with concrete. The 4-lane highway near my after-school service station job is now 8 to 12 lanes wide. The surfaces are paved with asphalt these days.
Q: “Why does the Northern Hemisphere warm so much more than the Southern Hemisphere?”
A: Because of the larger proportion of land. Land warms faster than ocean.
Q: “To what extent do climate models explain the regional effects in general.”
A: Pretty well. See Figure SPM.5 Panel A which shows observed and simulated warming (normalized to 1 degree global warming): https://www.ipcc.ch/report/ar6/wg1/figures/summary-for-policymakers/figure-spm-5/
Thanks for the replies.
I am searching for quantitative answers.
Q1.The southern oceans seem to warm more slowly than the northern oceans. Is that just because they have a bigger surface area?
Q2.Thanks for the picture. I find it very difficult to conclude anything from these tiny images. Do you have a link to the data? I would like to plot data and models on the same plots to really see the agreement.
How about the atlas at KNMI climate explorer? You can compare the CMIP5 set with observations? See https://climexp.knmi.nl/plot_atlas_form.py. I compared the temperature difference of 1950-1980 and 1990-2020 and they look pretty similar to me.
Note that ideally, in addition to plotting the multi-model mean, you’d also look at multiple results from one climate model based on starting with different initializations to get a feel for how different an individual instantiation of the climate can look relative to the model mean – natural internal variability is a thing, and while it is difficult to estimate real world internal variability, we can at least investigate model-world internal variability as a proxy.
Thanks a lot. I have tried The Climate Explorer, it is interesting. I will see if from this I can find out how to make the comparisons I want.
Clive
It has been a while for me. A few years ago, you and I agreed on the 67-ear cycle e shown in the graph below that comes from the H4 data. I have the same cycle showing in the H5 data.
I prefer the older H4 data. I still analyze the H5 data, but it is tainted.
https://1drv.ms/i/s!AkPliAI0REKhhMslBUImiwrlLh1kzw?e=dbU0PH
The MDO is a 67-year cycle and the DeVries is 209 years.
I also do an analysis where I assume that CO2 alone drives the bus. Here is what I get.
https://1drv.ms/i/s!AkPliAI0REKhhMsmoSmKnbPbuK0TaQ?e=awWTow
I have the figure below to give you a clue as to how poor the climate models are.
https://1drv.ms/i/s!AkPliAI0REKhhMsnqsXBUSHG4xt7zQ?e=NzBkig
This figure gives me a problem. CMIP5 predates CMIP6. I want to think that with additional time a later version of the models would be improved and have better agreement with the data. That is not the case.
Here are my projections.
https://1drv.ms/i/s!AkPliAI0REKhhMsnqsXBUSHG4xt7zQ?e=NzBkig
This figure gives me a problem. CMIP5 predates CMIP6. I want to think that with additional time a later version of the models would be improved and have better agreement with the data. That is not the case.
Here are my projections.
https://1drv.ms/i/s!AkPliAI0REKhhMsoJvYXo-167RSMWA?e=g4ftML
Global cooling started around 2015 when the 209-year and 67-year cycles were both near their peak. You will see a trend line in the earlier figure that indicates this. The forthcoming El Nino might disrupt this, but the decline will reveal itself again after the El Nino.
I am like other skeptics. I don’t think it is possible to construct an accurate climate model unless natural variability is included.
Charles, I am a bit new to all this and am looking for details.
Clive has kindly let me share my observations in the data on his blog.
Do you have a write-up which explains all the pictures you have sent? They all look very interesting and useful but with my own level of understanding difficult to digest.
John,
I appreciate your interest.
The best way for me to answer you is to give you some history.
I retired from the Naval Nuclear program 13 years ago. I am now 76.
From that work, I became familiar with the FFT. I can remember the first desktop FFT analyzer we used, the Nicolet 444.
I did not do the signal analysis. It was my job to look at the results and try to come up with design remedies for peaks with significant signal-to-noise and reduce them to background levels. The FFTs of accelerometers told us what our equipment was producing.
I became involved in analyzing the climate due to my interest in the subject at the time and I wanted to get myself involved in something that would preserve and possibly enhance my skills.
I became aware that Mccracken identified the solar cycles in a paper. The table below shows what I started with.
https://1drv.ms/i/s!AkPliAI0REKhhMsvt7VBGEXW6pYoSQ?e=nu1boG
I tried a few of these cycles and found they would fit the data. I sent that analysis to JoNova, and she got me in touch with her husband Dr. David Evans.
Dr. Evans had come up with his own analysis technique called the Optimal Fourier Transform. He had a spreadsheet that contained the OFT and that is what I used to identify the signals that were involved in the datasets. I used the output from his analysis and used that output as inputs in a program I had that used the sum of the squares error to obtain a more precise fit to the data.
For that first figure, there are 107 cycles. However, with as few as 13 cycles I can get a good fit to the H4 data.
I wanted to know whether the climate was driven by CO2 or natural cycles.
In Dr. Evans’s spreadsheet, I found the answer. To me, this figure destroys the idea that CO2 drives the climate. The figure below is equivalent to Dr. Evans’.
https://1drv.ms/i/s!AkPliAI0REKhgZMDMRK-IJl5JCkP-g?e=MeE2Bx
Note that the cO2 line is flat from 0 until around 1800. If CO2 is flat there can’t be a temperature change. How then can the Medieval Warm Period (MWP) be
Charles,
Thanks for the reply, I understand how you have analysed the data now.
John
That was when temperature data was simple before “homogenisation” and “blending” to fit the narrative.
Perhaps you refer to this 😉
https://clivebest.com/blog/?p=2295
I rather like ocean heat content as the measure of global warming.
https://climate.nasa.gov/vital-signs/ocean-warming/
I can certainly believe that the ocean heat content is a better measure of global warming than surface temperature.
I find zetajoules a bit difficult to understand. Are there plots of temperature profiles with depth at a few points in the deep oceans?
“This chart shows annual estimates for the first 2,000 meters of ocean depth.”
Sadly more scientific fraud from the “activist-research” crowd.
Using the deep ocean as a calorimeter makes a lot of sense …. if you have the data. NASA are misleading the public as usual. The reason they post in zetajoules which means absolutely nothing to most people, apart from sounding enormous, is because if they showed the temperature changes they claim to be measuring it would raise questions.
They claim to have data for top 2000m going back to 1960. They had a handful of temperature profiles for the entire planet at that time.
Also the temperature change in deep ocean is in tens of millikelvin IIRC. That has no scientific value as a measure of global OHC.
Even post y2k, when they have a fleet of diving Argo floats it is questionable whether geographic coverage is sufficiently uniform to accurately obtain a global OHC to the accuracy they claim.
Welcome to the muddy world of unscientific data mangling and political agendas. 😉
In fact I worked on an experiment which made measurements of temperature at a depth of 2000m in the Mediterranean Sea. I know it only changes by a tiny amount.
Even if the data is not geographically uniform, I think it would be honest and clear to present it. I have looked a bit for data compilations but not found any yet.
Ultimately the role of climate reports such as the IPCC reports is to provide information to policy makers, to assist them in policy making.
The article lists 14 different measures of temperature change, which make sense to those in the trade, but not necessarily others.
Clearly you are not happy with the basic temperature anomaly/time graph. What alternative would you suggest to show policy makers, bearing in mind that most of them have PPE or law degrees.?
I have been wondering who can actually read and understand the IPCC SPM reports. Certainly not myself with a physics PhD and a lifetime or research experience. I have guessed that the real policy makers all have aides who write them a summary of the summary.
“Someone writes them a summary of the summary” sounds a bit like the design control abstracts and fact books I got suckered into writing for a few projects back before the internet. It would have been nice to have read some of Paul’s papers back in the day-
The politics of policy analysis: theoretical insights on real world problems | Paul Cairney: Politics & Public Policy (wordpress.com)
Full article: The politics of policy analysis: theoretical insights on real world problems (tandfonline.com)
An interesting procedural detail of the workings of INTERGOVERNMENTAL Panel on Climate Change, that not many people realise, is that they write the Summary FIRST.
Yes, really. All the national teams get together and agree on every line, phrase and term of the SPM, in every language, then they go away and spend 3 months writing “scientific” basic for the summary.
IIRC the SPM of AR4 came out in Feb 2007 to much (organised) fanfare in the global media. The full report was published in June.
It is important to realise that this is a top down political process by govt. bureaucrats, not a bottom up filtering of scientific knowledge.
That maybe why it reads more like a powerpoint presentation than anything a scientist may recognise.
The SPMs and the Synthesis reports are surprisingly messy and difficult to absorb. I find myself relying on other people to summarize the summaries for me and then I go back to check.
It is all a revelation.
The author has identified the known 1976 climate shift from the data. It was discovered in the early 1990s. It is all over the literature, and despite being identified in the Pacific Ocean, it affected the entire world producing a global change in atmospheric circulation.
Marcus, S.L., de Viron, O. and Dickey, J.O., 2011. Abrupt atmospheric torque changes and their role in the 1976–1977 climate regime shift. Journal of Geophysical Research: Atmospheres, 116(D3).
Less known is the climate shift that took place in 1997 resulting in the famous pause and initiating Arctic amplification.
Chavez, F.P., Ryan, J., Lluch-Cota, S.E. and N?iquen C, M., 2003. From anchovies to sardines and back: multidecadal change in the Pacific Ocean. Science, 299(5604), pp.217-221.
These shifts take place every 25-35 years and result in multidecadal climate regimes that alter climate trends in almost every variable. They are not understood and little studied because models do not produce them, yet they are the more defining events for the climate we experience in our lifetime.
There aren’t very many climate variables you can study and not find them affected by these shifts. The 1997 shift is very clearly visible in lower stratosphere temperature trend, supposedly a fingerprint of the human effect on climate, except the 1997 shift says it is not.
That’s why for the past decade we stopped hearing about the cooling of the stratosphere. Things that don’t support the party line are not discussed.
Javier, I am certainly not claiming original discoveries in the data.
Just pointing out features of the data I have recently learned about and which may not be widely known because of the focus of the “iconic plot”.
Thanks for the links to these paper, I will read then.
Hi John, I see you are an ex-CERN college of Clive, that explains the excellent content despite you lack of familiarity with climate arguments.
One thing I find fundamentally stupid in all this is the idea of “land and sea average temperatures”. From a physics point of view, there is no such thing as an average temperature across different media. Temperatures are not extensive quantities and thus cannot be added or averaged. If the idea is that temperature is a proxy for thermal energy then it must be restricted to only SST, at it’s crudest simplification.
7 years ago I compared rate of change of SST and BEST land surface temps. Land shows twice the rate of change, suggesting that damp rock has about half the SHC of sea water.
The climate pseudo-scientists seem happy with this aberration because it increases the rate of change in SST. However, since the whole argument is about “radiative forcing” of GHG and they are implying energy balance changes are causing the warming , it clearly should physically meaningful.
I wrote an article on this which Judith Curry put on Climate Etc. but it did not get much useful discussion.
https://climategrog.wordpress.com/2016/02/09/are-land-sea-averages-meaningful-2/
Average land temperatures are of interest in the practical sense that is where we live and grow our food but “adding” them to SST is meaningless from either angle.
The “iconic graph” is , IMO, a scientific fraud.
Greg,
As you recognize I am an ex-colleague of Clive but not an expert on global warming data and arguments. I have followed the AGW story for a few decades but without looking myself at the data in any detail.
About 6 months ago I began to understand disturbing details of the temperature data and so made the study that Clive kindly allowed me to post on his blog. This gives me a good forum for feedback from which I can advance my knowledge. Thank for you feedback I will look at it in detail.
It certainly is the case that the “iconic graph” is being used to deceive people and so I think you are correct it is “scientific fraud”.
The temperature discussion in the media this summer is certainly even worse with the Financial Times trying to cause a panic about pavement temperatures of 82°C in Arizona. It is clearly easy to deceive the general public and even people who appear to have some scientific knowledge are caught in the deceit.
Yes, I pointed out in 2014 that we should be taking more notice of TLS since it does not have all the noise and oscillations of the tropospheric record.
https://climategrog.wordpress.com/uah_tls_365d/
A recent update of the graph shows post 1997 linear trend is 0.87deg C / century cooling , a fraction of what they were getting by pretending the previous slope was proof of AGW, whilst ignoring the flagrant cause was volcanic. (Ozone destruction by sulphuric acid aerosols and post eruption cleansing of stratosphere. )
Thanks for this, it is very interesting.
I’ve been blogging and researching this for a few years now. And a presentation to NASA JPL last year. My plots are a little more graphic on the topic. https://www.abeqas.com/earths-warming-and-cooling-surfaces/
And I’m closing on a paper with others which addresses much above. Curious how none seem to cognify even though web site gets over 40,000 visits on some posts. Maybe I should publish a paper. Given last solar forcing paper though, it is clear that none of the poser class will read.
MW, It’s tides, not sunspots.
Thanks for pointing to the intriguing charts. It’s a pity they are not numbered for easy reference, but the top left chart on the second image shows a prominent warming area to the north west of Novaya Zemlya. Could this be influenced by the dumping of “spent” nuclear reactors and materials ? According to a post by “Swedish Bill” (link below) a conservative estimate is that over 100 nuclear reactors have been dumped on the sea bed… here is the link:
https://judithcurry.com/2023/07/02/whats-causing-the-extremely-warm-temperatures-in-the-north-atlantic/
“In this iconic plot, a level of temperature rise of 1.5°C relative to a “pre-industrial” period, is frequently quoted to suggesting the point where serious, irreversible, damage will occur for the environment.”
You may read this on the Guardian but it is NOT what 1.5 limits is.
Firstly the 2degC idea was pulled from the air by Phil Jones et al based on the idea that by the 90s we’d seen 1deg rise since pre-industrial. They figured we’d probably be OK with another 1deg. After that, they reckonned it not possible to tell at which point there would be irreversible change.
IPCC adopted 2deg as the “target” limit for Paris Climate Accord.
As the hiatus in global warming stretched to 16 years and got a lot of coverage by Monckton and GWPF the alarmists got alarmed that it may take too long to get 2degC and everyone may decide it was not as urgent as we have been told , so they decided to up the ante. IPCC added the extra “it would be nice to stay below 1.5deg if we can”.
So it’s 2deg , not 1.5deg and it is not “will occur” , it’s “may occur” but we don’t really if it’s 2deg, 3deg or more.
The Guardian operation on the Goebels principal that if you repeat a lie often enough people will start to believe it.
It seems to be working !
Thanks for these clarifications.