CO2 Concentrations in February and March, 2020
March 10, 2020
Affiliation: Missouri University of Science and Technology (emeritus)
The COVID-19 pandemic caused China, currently the world's largest producer of CO2 emissions, to shut down much of its economy in January and February, 2020. This article provides indications that this was at least partially the cause of a brief reduction in atmospheric CO2 concentrations in February and March, 2020. It also gives evidence that a similar reduction in atmospheric CO2 concentrations in February and March, 2008 might have been caused by the economic meltdown of that period.
I wish to thank the United States of America National Oceanic and Atmospheric Administration (NOAA) for its wonderful data on average daily atmospheric concentrations of carbon dioxide at Mauna Loa, without which, this article (clearly) could not have been written.
Northern Hemisphere winter is typically a time of increase in atmospheric CO2. Trees enter a dormant state in which they absorb little carbon dioxide, removing an important carbon sink. Thus, when we see a decrease in atmospheric CO2 during this period, it behooves us to search for a cause. In February and March of both 2008 and 2020, we saw just such a decrease in atmospheric CO2. These winters marked the beginning of what were perhaps the two greatest economic upheavals in recent times. In late 2007, as a result of large scale speculation and corruption, global markets started a precipitous decline which continued for over one year. In January, 2020 as a result of the COVID-19 pandemic, China, the world's largest producer of CO2 emissions, shut down much of its industry and global markets soon reacted to the pandemic with another precipitous fall. Could the economic upheaval and the reduction in atmospheric CO2 be related? Perhaps. In this article, we attempt to show that natural variation is unlikely to be responsible for the observed decrease in atmospheric CO2 concentrations. To my knowledge, no other cause has been suggested for these unexpected decreases in atmospheric CO2 except economic meltdown.
The starting point in this research was NOAA's daily averages for CO2 concentrations in the atmosphere as measured at Mauna Loa. Missing data, which were thought to have skewed previous attempts were filled in through rough interpolation. (See below for details.) Then 30 day running averages were computed, starting with Dec. 1 through Dec. 30 and ending with April 1 through 30, for the 33 Northern Hemisphere winters: 1987-88 through 2019-20. Then from each 30 day average ending in February 28 through March 31, the previous 30 day average was subtracted. (Eg: The average for the 30 days December 31 through January 29 was subtracted from the 30 day average of January 30 through February 28.) The minimum was then taken over these 32 subtractions (33 in leap years) and assigned as xyear to that winter.
30-day averages were chosen in order to winnow out very short-term variations, but yet leave a window for a significant number of readings within the winter.
These 33 data points, x1988 through x2020, were analyzed and found to have the mean, μ=0.31, and the standard deviation, σ=0.35, Not surprisingly, the lowest distance from the mean was -2.56σ for the winter 2007-08 and the second lowest was -1.36σ for the winter 2019-20. No other winter had a distance from the mean less than -σ.
Assuming a normal distribution, the probability of any two random data points lying below
-σ from the mean and at least one of them lying below -2σ from the mean is about 0.007. Thus, natural variation seems highly unlikely to be responsible for these unexpected February and March falls in atmospheric CO2 concentrations.
Discussion of Attached Graph:
30 day running averages for six selected winters were then graphed: 1998-99, 2004-05, 2007-08, 2016-17, 2018-19 and 2019-20. (See accompanying graph below.)
i=2007-08: xi=-0.60; (xi-μ)=-2.56σ: the largest distance from the mean. The graph shows a definite depression in February and March as atmospheric CO2 concentrations decreased. A slow recovery began in late March.
i=2019-20: xi=-0.17; (xi-μ)=-1.36σ: the second largest negative distance from the mean. The graph also shows a depression in February and March as atmospheric CO2 concentrations decreased, although not so pronounced as with 2007-08. The recovery in late March and April was much more pronounced than in 2007-08.
i=1998-99: xi=0.02; (xi-μ)=-0.80σ: the third largest negative distance from the mean. The depression in February and March is very noticeable in the graph, but much shorter in duration than in 2007-08 or 2019-20. To my knowledge, there were no great economic downturns that could have caused this.
i=2004-05: xi=1.03; (xi-μ)=2.06σ: the largest positive distance from the mean. This winter appears on the graph as very close to straight upward sloping line.
i=2016-17: xi=0.31; (xi-μ)=0.00σ: an “average” winter according to the above criteria.
i=2018-19: xi=0.14; (xi-μ)=-0.47σ: an anomalous winter. The depression in February and March is very noticeable in the graph, but too short in duration to show up according to the above criteria. Another mild depression occurred in late March and April which came too late to show up according to above criteria.
Conjecture 1: The economic downturns in the winters of 2007-08 and 2019-20 caused an initial decrease in CO2 emissions, which showed up within a month or two as a decrease in atmospheric CO2 concentrations. However, polluters quickly found ways to makeup the deficit by quickly reopening polluting industries or using the crises to circumvent environmental protection measures.
Conjecture 2: The economic downturns in the winters of 2007-08 and 2019-20 caused an initial decrease in CO2 emissions, which showed up within a month or two as a decrease in atmospheric CO2 concentrations. However, within another month Earth's atmosphere made up the deficit by sucking CO2 back out of carbon sinks such as the oceans.
These two conjectures are not mutually exclusive.
Looking to the Future:
In any case, the reductions in atmospheric CO2 concentrations were short-lived, and the Earth's atmosphere soon made up for the deficit and more.
Clearly, more research is required to reach a conclusion. Nevertheless, an answer to these questions might yield a clue as to how these physical processes work and how one might best reduce atmospheric CO2 concentrations in the future.
Having found that natural variation is very unlikely to be responsible for the observed February-March decreases in atmospheric concentrations of CO2 in the winters of 2007-08 and 2019-20, and for lack of any other reasonable explanation, it seems that economic upheaval would be the most likely cause.
In general, missing data are interpolated by averaging all valid readings within the four days preceding the missing datum and averaging all valid readings within the four days following the missing datum. The two averages are then averaged to replace the missing datum.
If necessary, five, six or even seven days are used to get at least one valid reading both before and after the missing datum.
Generally, readings in November and May are only used if no appropriate valid readings can be found in early December or late April and then only one valid reading is used.
NOAA Earth System Research Laboratories Recent Daily Average Mauna Loa CO2
This article is posted at tomsager.org/COVID-CO2-article.html.
The spread sheet used to make these calculations is posted at tomsager.org/OpenOfficeDocs/COVID-CO2-DATA.ods.
The graph attached to this article (before photo manipulation) is posted at tomsager.org/OpenOfficeDocs/COVID-CO2-GRAPH.odt.