Check: CO2 emissions fingerprint on temperatures

From InfoCheckers

By Tim88 -v.1.2 - Overview article: Check: Effects of CO2 emissions on nature -


It's hard to decide where the truth is between "climate deniers" who hold that CO2 emissions have practically no effect on climate, and "climate alarmists" who foresee the end of the world as all ice will melt in the near future. Here we make thankful use of a mostly rural temperature data set of the Northern hemisphere that served to estimate an average Northern hemisphere temperature history. That temperature estimate has been compared with simulated effects of varying solar irradiation and historical CO2 concentrations. By performing an updated, revised simulation we obtain a reasonably good fit in which about 0.5 °C temperature increase since 1880 can be ascribed to CO2 emissions. Assuming that there are no other causes, the effect of CO2 emissions appears clearly and quite convincingly (please judge for yourself!). Extrapolated, an estimated 1°C temperature increase can be expected from a doubling of CO2. This result is consistent with recent findings of "moderate", non-alarmist climate scientists. The spreadsheet can be downloaded from the bottom of this page.


Here we attempt to find evidence for an effect of CO2 emissions on temperatures. In an earlier analysis we found that very likely current atmospheric CO2 concentrations are increasing due to man-made CO2 emissions. Now we want to compare the theoretically possible temperature increase due to increasing CO2 with the real temperature record over the last 100 years. Most people -including until recently this writer- rely on the officially presented historical temperature graphs such as GISTEMP (presented by NASA) and GHCN by NOAA. However, it has been argued rather convincingly that those data sets are unreliable (see also here and here). Not only historical land temperatures have issues, also ocean temperature records are inaccurate.


Estimations of historical temperatures going back to before the use of thermometers are much debated, with as most notorious example the "hockey stick" reconstruction by Mann, which yielded an almost straight temperature curve for the past thousand years, effectively leading to denial of the Medieval Warm Period. That in turn led to accusations related to the climate gate scandal and a defamation court case that was thrown out by the judge in 2019 as Mr. Mann refused to hand his data to the judge.

Based on the available evidence we may assume that the Medieval Warming Period was about as warm as temperatures at the start of the 21st century.


Historically changing climates give context to a possibly significant influence of CO2 emissions in our time. It would be great if we could create a small data set based on a sample of high quality rural temperature stations. Regretfully, such information is hard to obtain.


In 2017 John Christy and Richard McNider published a study based on satellite temperature data of the lower troposphere. They found a transient climate response of 1.10 ± 0.26 °C - which is about half of the average of commonly used climate models. It should be noted that also satellite temperature data have issues.

A good effort using again another approach was made by Willy Soon and co-authors in 2015, and their data is used for our own analysis here below. In their paper "Re-evaluating the role of solar variability on Northern Hemisphere temperature trends since the 19th century", they presented a weighed average of purified temperature records of the USA, China, Ireland and the arctic circle. Instead of trying to correct doubtful station records they weeded out such stations form the records in order to rely on mostly good, rural data. Their historical Northern hemisphere land temperature estimation yielded about 1/4 °C less temperature increase since 1880 than the GHCN reconstruction. From their land data record they made an estimated average Northern hemisphere temperature record by accounting for the reduced temperature variation of the oceans. It should be noted that while Willy Soon has been accused (apparently wrongly) of being a mouthpiece for oil industry, his co-authors Connolly and Connolly are beyond any doubt environmentalists and independent researchers, not acting for big oil.

Soon and co-authors estimated the possible effect of varying solar irradiation as well as the possible effect of increasing CO2 concentration on temperatures. The authors chose the same solar irradiation reconstruction as James Hansen in 1981, and it makes sense to agree with him for the same reason ("The improved fit provided by Hoyt's solar variability represents a posteriori selection, since other hypothesized solar variations that we examined [..] degrade the fit" - Hansen 1981). Nevertheless, there is a certain risk of confirmation bias as also discussed in the paper of Soon.

After they found the best solar fit (freely adjusting the unknown amplification), they came to the conclusion that "it seems that most of the temperature trends since at least 1881 can be explained in terms of solar variability, with atmospheric greenhouse gas concentrations providing at most a minor contribution."

However, there is a notable mismatch after the year 2000. And since their publication, five years have passed during which global warming as well as warming of the Northern hemisphere strongly increased, even though solar irradiation decreased to about the level of 2009 (according to Virgo and IRMB satellites).

Further, their fitting method (with correlation r2=0.496) is debatable. Arguably, they should have tried to find the best combined fit of CO2 and solar influence, but instead they followed an overly simple fitting method that left little room for adding a reasonable fit of the effect of CO2. Their simulations also approximate the effects of CO2 concentration and solar irradiation on temperature as linear, which adds in theory small but unnecessary estimation errors.

Taking those factors together, a better and more plausible fit may be obtained that acknowledges a much more significant role of CO2 in recent warming. The CO2-induced average warming since the year 1880 may easily be as much as 0.5 °C. That is demonstrated by comparing the following two graphs:

Updated simulation of global temperature variation due to variations of solar irradiation and CO2 emissions. Temperature [°C] as function of time; the last 5 years are approximate.
Grey: proxy of yearly global temperature anomaly based on rural measurements. Blue: 3 yr temp. average. Red: simulated temperature variation due to Total Solar Irradiation and CO2.
r2=0.499. With many thanks to Soon, Connolly and Connolly for the original spreadsheet.

In the figure here above a logarithmic effect of atmospheric CO2 on temperature was modeled, combined with a T4 effect of total solar irradiation (TSI). The last five years were completed with approximate values as some data are not yet made available. Here the simulated overall warming until 2019 due to TSI was 0.31 °C while the simulated warming due to CO2 was 0.50 °C (until 2014: TSI 0.42°C, CO2 0.45 °C). This was sought for, but the result appears to be not far from optimal (see Discussion for the deviation in the second half the 20th century).

Since 1880 the CO2 concentration in air has increased from about 290 ppm to 413 ppm. The here above pictured simulation corresponds with a temperature increase of 0.95°C resulting from a doubling of the atmospheric CO2 concentration. Consequently, with this re-analysis we obtain a similar result as Christy (note however that this analysis is only for the Northern hemisphere), in significant disagreement with both Soon and the average of climate models.

In comparison, here below is a fit with solar irradiation alone (not optimized):

Simulated effect on temperatures due to variation of solar irradiation alone. r2=0.482

Telling is the increasing deviation during the last 20 years; increasing the amplification of the solar effect does not make much difference for that discrepancy. With the addition of a strong assumed effect from atmospheric CO2 as shown in the top graph, one obtains a much better match for the trend of temperature increase during the period 1980-2000 as well as the attained temperature and the recent temperature trend.


A clear deviation occurs in the period 1960-1990, in which the simulated temperatures are too high. The simulated curves were aligned as such on purpose, because that period corresponds to the so-called period of global dimming. Possibly most or all of the discrepancy can be explained by strongly increased airborne particles from air pollution, which was more and more reduced starting from around 1990. Note that the Northern hemisphere land temperature data (which served for the total Northern hemisphere temperature estimation) was likely most affected by that pollution. Also the big eruption of mount Agung in 1963 may have played a role.

The fit with possible TSI is quite good if we take the global dimming period in account. Further, the simulated effect of atmospheric CO2 is consistent with (debated) predictions based on radiation physics. Thus it looks plausible that a significant warming occurred in the Northern hemisphere due to greenhouse gas emissions.

Note that CO2 is only the main greenhouse gas, other emitted gasses and aerosols should also play a role but we did not account for their influence which largely was in tune with CO2.


We have good reason to believe that significant man-made warming occurred due to greenhouse gas emissions (primarily CO2). Our analysis suggests that greenhouse gas emissions since 1880 may have caused an average temperature increase of around 0.5°C in the Northern hemisphere.

Download: modified spreadsheet, originally from Soon, Connolly and Connolly File:Soon et al 2015 data-updated-Tim88.ods
You can play with fitting yourself: modify the amplification factors of cells U2 and W2. Readjust the offset with cells U3 and W3.

Discussion via the Forum