Population increase impacts the climate, using the sensitive Arctic as an example

Ol M. Johnnessen , , Elen V. Shlin

a Nansen Scientific Society, Bergen, Norway

b Nansen International Environmental and Remote Sensing Centre, St. Petersburg, Russia

c Institute of Earth Sciences, St. Petersburg State University, Russia

Keywords: Population Climate Sea ice Arctic

ABSTRACT The global population during the last 100 years has increased from 2 to 7.7 billion, causing an increase in green- house gases in the atmosphere. In order to see how population increase is directly related to physical variables of the climate, this Perspective article places observations and scenarios of climate change into context and puts forth a statistical modeling study on how the sensitive Arctic climate responds to the increasing population. The relationships between population, Arctic sea-ice extent (SIE), and surface air temperature (SAT) are very strong, with the increasing population explaining 96% of the decreasing SIE and about 80% of the increasing SAT in the Arctic. Our projection for the SIE using the population as a “proxy predictor ”for a projected population of 10 billion people on the Earth in 2100, yields a SIE of 9.30 and 8.21 million km 2 for a linear and squared rela- tionship, respectively, indicating no “tipping point ”for the annual ice extent in this century. This adds another dimension to climate understanding for the public at large using population as a proxy variable, instead of the more abstract CO 2 parameter. This also indicates that it is important to attempt to limit the ongoing increase in population, which is the main cause of the greenhouse gas emissions, in addition to reducing per capita emissions by an exponential increase in implementing renewable energy, a formidable challenge in this century.

1. Introduction

The world population is the major source of the increasing emis- sions of CO2and other greenhouse gases (GHGs) to the atmosphere, impacting global climate, health, living conditions, economies, nature, and ecosystems. The correlation between CO 2 and population in the pe- riod 1963—2019 is nearly perfect, close to unity ( Figs. 1 and 2 (a)). For the present climate and projections in the future, global climate models such as those in phase 6 of the Coupled Model Intercomparison Project (CMIP6) are used with different CO2scenarios, called Shared Socioe- conomic Pathways (SSP) ( IPCC, 2021 ). The SSPs are based on business as usual, including population growth, gross domestic product, fertility, poverty, education, and other factors ( O’Neill et al., 2017 ; Gidden et al., 2019 ). However, there have been no studies addressing the direct re- lationship between physical climate variables and population, although there have been many studies mentioning the importance of popula- tion growth’s impact on climate ( Lutz, 2017 ; Scovronick et al., 2017 ; Bongaarts and O’Neill, 2018; Vollset et al., 2020 ). Studies of the direct relationship between physical variables and the population will give an- other dimension to the importance of the population impact on climate, which is easier to understand for the public at large rather than an ab- stract CO2number.

In order to investigate this relationship, we have selected the sensi- tive Arctic climate as an example, which has seen a dramatic decline in the sea-ice extent (SIE) and an increase in the surface air temperature (SAT) ( Johannessen et al., 2004 , 2016 , 2020 ; Davy and Outten, 2020 ; SIMIP Community, 2020 ). In this Perspective article, we establish the re- lationship between the SIE and SAT from 70°N to the Pole and the world population increase by correlating SIE and SAT directly with the popula- tion as a “proxy variable ”for present and future projections. Since there is a near perfect relationship between population and CO2( Fig. 2 (a)), and since CO2is the main driver for climate change in the Arctic, pop- ulation is also a reasonable proxy variable for climate change.

2. Data

2.1. Projected population increase

There will be an increase in the population in the future, projected to reach a maximum in 2064 of 9.7 billion people, with an uncertainty interval (UI) between 8.8 and 10.9 billion. Thereafter, it may decline to 8.8 billion (UI: 6.8—11.8) in 2100 ( Vollset et al., 2020 ), while the United Nations’ latest revision forecasted 10.9 billion (UI: 9.4—12.6) by 2100 ( UNDP, 2019 ). An alternative forecast for 2100 is 13.6 billion (UI: 10.7—17.7) ( Vollset et al., 2020 ). From the above, it can be seen that there are significant differences between these forecasts for 2100.Therefore, in our study we use different population projections in the future for 2100 up to 13 billion as the input to the historical relationshipregression equations between SIE and SAT and the population increase for the period 1963—2019, when these variables started to change dramatically ( Fig. 1 ).

Fig. 1. Arctic SIE, Arctic SAT, GHGs, CO 2 , and population for the period 1850—2019. Annual SIE, annual SAT, GHG, CO 2 , and global population. Note: Inverted scale for GHGs, CO 2 , population, and SAT.

2.2. SIE, SAT, population, CO 2 , and GHGs from 1850 to 2019

In the Arctic region, the SAT and the sea-ice cover are highly sensitive to climate change, mainly due to the increasing GHGs. For example, it was shown that 90% of the annual SIE decrease could be explained by the increasing CO2in the atmosphere in the period 1961—2007 ( Johannessen, 2008 ). In order to provide an overview on how the SIE, SAT, population, CO2, and the total GHGs have varied since 1850,these variables are plotted in Fig. 1 .

The annual relationship between population ( Worldmeter, 2020 ),SIE ( Walsh et al., 2019 ) and SAT from 70°—90° north ( Rohde and Hausfather, 2020 ), CO2and total GHGs ( Butler and Montzka, 2021 ) ( Fig. 1 )shows a high degree of correspondence between these variables after 1963. The population in 1850 was 1.2 billion people, growing slowly to 3 billion in the 1960s, and thereafter increasing dramatically to 7.7 billion by the end of 2019. The annual SIE in the period 1850—1920 was quite stable at about 13.5 million km2. However, in response to the Early Warming/Cooling (EWC) period ( Bengtsson et al., 2004 ;Johannessen et al., 2004 ), shown in the SAT curve between 1917 and 1940, it decreased to 12.5 million km2in 1945 and thereafter increased again during the subsequent cooling period to 13.9 million km2in 1963,before decreasing dramatically to 10.5 million km2in 2019 ( Fig. 1 ). The SAT curve in the EWC period increased from ? 14.7°C in 1917 to ? 12.0°C in 1940 and subsequently decreased again to ? 14.6°C in 1963. Thereafter, the SAT increased to about ? 11°C in 2019, which was about two times larger than the average annual increase in the Northern Hemisphere ( Johannessen et al., 2016 ). There was also a pronounced cooling from 1860 to 1864 of about 3°C before a warming of 3.5°C from 1864 to 1880, and thereafter a similar cooling up to 1889 without any response in the SIE, so far not explained in the literature but perhaps caused by the quality of data during this period. This cooling—warming—cooling period and the EWC period were not influenced substantially by the CO2,which only increased slowly during these two periods. CO2and the total GHGs in 1850 were 285 ppm and 288 ppm, respectively, increasing in 1963 to 319 ppm and 342 ppm, before increasing dramatically to 410 ppm and 500 ppm by the end of 2019, when the CO2was about 80% of the total GHGs, while in 1963 it was 90%. The 410 ppm in 2019 in the atmosphere was caused by a total emission of about 40 GtCO2yr?1, of which more than 50% was absorbed by the ocean and land ( IPCC, 2021 ).Since all the variables started to decrease/increase dramatically in the 1960s after the EWC period, as well as the fact that the quality of the data is better after the 1960s, we selected the period 1963—2019 for further analysis.

3. Results

3.1. Statistical relationships between SIE, SAT, and population

The relationship between SIE and population is shown in Fig. 2 (b),where the coefficients of determination (R2) between them using both linear and second-order polynomial regression equations were 0.95 and 0.96, respectively, with standard deviations (SDs) of 0.21 and 0.19, indicating that about 96% of the SIE decline can be explained by the increasing population, at least during this transient period. The correlation between SAT and population, both linear and squared ( Fig. 2 (c)), again shows a highR2(0.77 and 0.84, respectively, with SDs of 0.57 and 0.48),indicating that the population could explain about 80% of the warming in the Arctic, while the natural variability caused by the North Atlantic Oscillation and Pacific Decadal Oscillation (e.g., Johannessen et al.,2016 ) could be partly responsible for most of the rest (see Fig. 1 ).

Fig. 2. (a) Correlation between annual CO 2 and population. Linear (red) and squared (purple) relationship for the period 1963—2019, including the R 2 and SD. (b) Correlation between SIE and population. The panel shows linear and squared relationship including R 2 and SD for annual values in the period 1963—2019. (c) Correlation between SAT north of 70°N and population. The panel shows linear and squared relationship including R 2 and SD for annual values in the period 1963—2019. (d) Statistical models (both linear and squared) of CO 2 ,SIE and SAT north of 70°N as functions of population, annually for 4—13 billion.

3.2. Projections of SIE and SAT in the Arctic

Next, we project what could happen in the Arctic in the future when using the SIE and SAT relationships with population (both linear and squared) as a “proxy predictor ”. The reason for using both the linear and squared relationship with population is that we want to see the spread using these two statistical models. Since the population estimates in 2100 are uncertain, we use the population estimates up to 13 billion for 2100 according to the different estimates of the population increase( Vollset et al., 2020 ).

For a projected population growth to 10 billion in 2100, our statistical analyses indicate the Arctic Ocean SIE will still be about 9.30 and 8.21 million km2, compared to 10.5 million km2in 2019, based on the linear and squared statistical models, respectively, using the population as a “proxy predictor ” ( Fig. 2 (d) curves marked SIE-P lin and SIE-P sq),with a spread of about 1 million km2, which is much less than the 6 million km2spread when compared with the projections using the SSP5-8.5 and the SSP1-2.6 scenarios in the CMIP6 climate models ( Davy and Outten, 2020 ). Even with a population increase to 13 billion people in 2100,which according to a recent study ( Vollset et al., 2020 ) is not realistic,the SIE will be about 7.31 and 4.15 million km2( Fig. 2 (d)), based on the linear and squared relationship to the population “proxy predictor ”,respectively, suggesting strongly that the Arctic will not become ice free in the future on an annual basis, i.e., no “tipping point ” will be reached.

It should be mentioned that our projection for CO2using the relationship between CO2and population ( Fig. 2 (a)), both linear and squared,for 10 billion at 2100 is 450 ppm and 482 ppm, respectively ( Fig. 2 (d),curves marked CO2-P lin and CO2-P sq) are probably as realistic as some of the SSP scenarios. Furthermore, our scenarios are higher than what is needed to attain the Paris Agreement target for a global average temperature increase of less than 2°C, which will require the SSP1-2.6 scenario ( IPCC, 2021 , table SPM.1), under which the CO2concentration is projected to be 425 ppm in 2100. Therefore, our dynamical transient scenarios are conservative when compared with the Paris Agreement.

But what about the summer ice? It was already projected in 2004 that 80% of the summer (September) ice could melt in response to a doubling of CO2( Johannessen et al., 2004 ), recently updated by many subsequent studies (e.g., Notz and Stroeve, 2016 ; Davy and Ouetten, 2020 ;Johannessen et al., 2020 ; SIMIP Community, 2020 ). Furthermore, a statistical analysis between the September SIE and the ln(CO2/CO2r),which is the empirical law for longwave radiation back to space from the surface ( Myhr et al., 1998 ), for the period 1901—2010, where r is the reference level of CO2in 1901, indicates that all the summer ice would melt if the CO2level in the atmosphere reaches 502 ppm ( Johannessen et al.,2020 , their Fig. 4.36) —a finding recently updated by our group using the shorter satellite period between 1979 and 2018, which gave a result of 566 ppm for zero September SIE. These two values are much higher than under the 2°C Paris Agreement target with the SSP1-2.6 scenario,which projects values of 450 ppm in 2060 and 425 ppm in 2100( IPCC, 2021 , table SPM.1), as mentioned above. Based on this, we put forward the hypothesis that the summer ice will not melt completely;no tipping point will be reached, which indicates the ice cover will remain for all seasons in the future, under the assumption that the Paris Agreement target can be reached or nearly reached.

As was done for the SIE, we also calculated the SAT for a population increase of 10 billion, using the population as both a linear and squared proxy predictor ( Fig. 2 (d), curves marked SAT-P lin and SATP sq). The results showed that the SAT for a population of 10 billion in 2100 is ? 9.5°C and ? 5.8°C respectively, compared to ? 11°C in 2019,with a spread of 3.7°C compared to a spread in 2100 of 9.0°C using the SSP5-8.5 and SSP1-2.6 scenarios in the CMIP6 climate models ( Davy and Outten, 2020 ).

4. Summary and conclusion

It is well known that the human population, through the GHG emissions we generate, is impacting the global climate, health, economies,and society at large. In order to see how the population is impacting physical variables of the climate, we have for the first time carried out a study in which the relationships between the SIE and SAT in the Arctic have been directly correlated with the population increase during the period 1963—2019, as summarized in Fig. 2 . Based on the very high correlations between these variables, we have made projections for the SIE and SAT in the Arctic using the projected population increase as a proxy predictor. Our projection for a population of 10 billion in 2100 indicates that the annual ice cover will be between 9.3 and 8.2 million km2, based on a linear and squared statistical model, respectively,reduced from 10.5 million km2in 2019. This means that the SIE will decrease by between 0.52 million km2and 1.00 million km2per billion of population increase from the present day (2019) to the projected 10 billion in 2100, based on a linear and squared relationship, respectively.Even for an unrealistic increase of the population to 13 billion in 2100,the remaining SIE is projected to be between 7.3 and 4.1 million km2,based on the linear and squared population predictor, respectively, indicating no “tipping point ” for the annual ice cover in the Arctic in the future.

We have also made some reflections about the summer Arctic sea ice and suggested a new hypothesis that the summer ice will not melt if the Paris Agreement can be reached. This is the topic for a more detailed study currently underway.

The Arctic SAT is projected to increase from ? 9.5 to ? 5.8°C, based on the linear and squared models, respectively, for a 10 billion population in 2100 —an increase from about ? 11°C in 2019. This implies a warming above 70°N in the Arctic of 1.5°—5.2°C from 2019 to 2100.

In this study, we have directly shown, using the Arctic as an example, that the human population is impacting the climate. Therefore, it is very important to attempt to intensify the work involved in limiting the population growth in the future in addition to reducing GHGs. It is also very important to break the relationship/correlation between the population and CO2in order to attain the Paris Agreement, a formidable challenge for the world’s population, which will also require an exponential development of renewable energy and a modification of people’s living standards, at least in the industrial countries.

Data and materials availability

All data in the main text are freely available.

Competing interests

Authors declare no competing interests.

Acknowledgments

We acknowledge Stephen A. Montzka of NOAA for the GHG data,Richard Davy of the Nansen Environmental and Remote Sensing Center for discussions, Martin Miles of NORCE Norwegian Research Centre for editing, and funding support from the Nansen Scientific Society.