Did Global Lockdowns Impact Atmospheric CO2 Concentration?

I recently learned that my local district council has signed-up to the 15-minute scheme that is rolling out across the UK.

Residents are worried, and quite rightly so, for a semi rural community like ours relies heavily on private transport, there being few bus services to speak of: no car, no cake!

Any sensible transport planner would undertake surveys both before and after introduction of a new policy so they can quantify the benefits. Not so our council.

Schemes they are going to design themselves will go ahead whether residents like it or not and whether they actually work or not. There’s a fair chance the knock-on effect will be increased travel and emissions as residents find ways around the road blocks and closures.

The Big Scheme

But there has been a rather big emissions reduction scheme in operation; a global scheme, in fact, and in place of flower beds and bollards placed across residential roads our respective governments have gone and locked us all away for weeks on end.

Businesses have gone bust, shops have closed for good and vehicles have rusted. In the height of incarceration some wag joked that they’re now getting three months to the gallon. Quite. So did this big scheme have a big impact on atmospheric CO2? Please place your bets now and let’s go find out…

Scripps CO2 Program

If you want decent data on atmospheric CO2 your best bet is the Scripps CO2 program. There’s a fascinating history to this institute, which lies right at the heart of the global warming story.
I’m not going to chew on that bone today for I want to get ahead and pull down monthly and weekly in situ reference data from the Mauna Loa Observatory (MLO). Yes, I know that the MLO is stuck on top of a volcano that also belches CO2, and agree this is well weird.

The idea was to get up as high as possible, away from as much vegetation as possible, and as far away from any continent as possible, for all these things affect readings enormously.

Spot The Ball

Although I can run both the monthly and weekly data series back to 1958 this doesn’t give us a decent visual for our eyeballs to dance over. I have thus chosen the period 2015 to 2022 to give us a better flavour, marking in the period March 2020 to April 2021 to remind us of the most intense lockdown period.

Here’s the slide for the monthly mean series followed by the slide for the weekly series. Note the strong seasonal pattern:

Like you I could not detect any noticeable impact of the global pandemic and all the extraordinary restrictions this entailed. Were it not for those two dashed lines I’d have no clue as to when the world closed down!

The Big Spanner

This is one of those moments when we need to reach for the big spanner of Autoregressive Integrated Moving Average (ARIMA) intervention modelling.

We’re looking at time series data so we better use a proper-job time series technique to see if we can spot anything our eyes have missed by dialling-up a fancy method for outlier detection. Believe it or not there are no less than 7 outlier types defined within the ARIMA module of my stats package and here they all are:

Being a reckless sort I decided to flip them all on to see what came out in the wash.

Monthly wash

The monthly series yielded a rather pleasing ARIMA(1,1,0)(0,1,1) structure with the following primary outputs (please follow this Wiki link to see what this gibberish means):

his tells us that monthly CO2 follows a strong seasonal pattern (MA seasonal Lag 1, p<0.001), with a modest degree of ‘memory’ (AR Lag 1, p<0.001); that is to say this month’s value for CO2 will depend on last month’s. This is all well and good but I want you to ignore everything apart from the lower table, for this tells us if anything unusual has happened at any point.

We observe that the model detected a slight drop in atmospheric CO2 to the tune of -0.957ppm (p<0.001) back in April 2017 and another slight drop to the tune of -1.066ppm (p=0.001) back in March 2022.

Before alarmists get too excited at this point we should note the March 2022 drop is earmarked as a transient with a decay factor of 0.915. What this means is that there was an immediate drop that dissipated back to normal.

If you take a magnifying glass to the monthly chart you can see this blip for yourself – it looks like somebody took a small bite out of the spring 2022 series.

This bite is not unique, and if you study the monthly chart carefully you’ll see an odd kink each March or thereabouts going all the way back to 2015 and before.

I am hoping folk have figured for themselves what is causing this, for it is a wonderful realisation… those regular nibbles in the CO2 record are the plants of the northern hemisphere coming back to life and fixing CO2 like crazy as new growth gets underway.

For some reason the plants of the northern hemisphere were pretty hungry during spring of 2022, and this may well be the impact of global greening (i.e. there are more hungry plants wanting dinner).

No doubt there are alarmists who will try to pin this seasonal nibble on the impact of global lockdowns but global lockdowns over an extended period are not going to cause a single transient dip a year or so later.

Weekly Wash

Here’s the same approach used on the weekly time series data, which yielded a sophisticated model structure of ARIMA(2,1,2):

This tells us that atmospheric CO2 has a 2-week ‘memory’ (AR lag 2, p<0.001) over which it is subject to random shocks (MA Lag 2, p<0.001). Those wacko four digit numbers in the outliers table refer to the week number since w/e 29 March 1958.

It would indeed be lovely for the stats package to print dates instead of cryptic week numbers but that is how the cookie crumbles! I shall therefore announce that the first pandemically-flavoured observation in March 2020 (w/e 7 March) possesses a week number of 3161, with global lockdowns ending sometime around week 3216.

If we are to detect an impact of global shutdown we must therefore be seeking negative coefficients for outlier weeks occurring after week 3161.

Clearly there ain’t any! The closest we get is a transient lowering of the mean weekly atmospheric CO2 by -2.350ppm (p<0.001) during week 3112 (w/e 23 March 2019). Now, lockdowns weren’t even born back then but I can tell you what was being born and that was the flowering spring of 2019.

Ain’t that just wonderful? If you squint at the weekly slide you’ll see this sharp drop, which is mirrored in notable spring drops for 2020 and 2022. We might call this the ‘spring signal’. Setting the spring signal aside my eyeballs cannot detect any impact of global lockdown, and neither can the ARIMA procedure.

Wood For Trees

OK, so we have been unable to detect the impact of lockdown on weekly and monthly values for atmospheric CO2, but what about longer timescales? Are we missing the wood because of the trees?
At this point we can either resort to an analysis of annual means or do something sophisticated and rather elegant using seasonal decomposition. As Wiki states this is an important statistical technique, being one that decomposes a periodic data series into a seasonal component, a combined trend and cycle component, and an error component.

In plain English this technique separates the seasonal ‘wobble’ from a wobbly series so we can look at both the wobbly bit and the underlying trend bit. Here’s the wobbly bit for the period Jan 2015 – Dec 2022 followed by the seasonally adjusted data (i.e. wobbly bit removed), with global shutdown marked by two dashed lines:

Can we spot anything yet? I say not! Atmospheric CO2 carried on rising just as if total global shutdown of business, schooling, social life, travel and pretty much everything we do as a civilisation meant absolutely nothing.

Now that’s pretty darn compelling evidence of the utter futility of Net Zero, especially when we stop to consider geologists measured the impact of global lockdown on movement of the Earth’s crust!

Grab The Spanner!

At this stage we’ve got a lovely run of seasonally adjusted monthly time series data, so what we can now do is conjure an indicator variable to define the period of global shutdown, as indicated by those dashed grey lines in the plots, and stuff this into our ARIMA modelling as an independent (predictor) variable.

The resulting best-fit ARIMA(0,1,3) model yields the following:

We observe that our indicator for the lockdown period of March 2020 – April 2021 is utterly insignificant as a predictor of seasonally adjusted atmospheric CO2 (p=0.846). Again there is zero evidence to support Net Zero but, of course, that won’t stop globalists doing exactly what they want regardless of what the science actually says.

More, Bigger, Better

So… we’ve not detected any impact of global shutdown in a weekly or monthly time series, and we’ve not detected any impact using an indicator variable to mark out a sustained period of 12 months for seasonally adjusted data.

But should we be thinking bigger? I’m thinking in terms of a sustained impact of global lockdown that kicks off in March 2020 and runs right through to the end of the series in December 2022.

Perhaps this ain’t some short, sharp shock but a more genteel affair draped over many months. The BIG problem is that atmospheric CO2 will be greatly dependent on many factors over this extended time frame I can’t possibly hope to account for.

Chief among these will be land surface and sea surface temperature. Fortunately for us these are brought together in one handy super yummy data series that goes by the natty name of HadCRUT.5.0.1.0.

In two shakes of a locust’s leg I had the monthly time series for the HadCRUT5 northern hemisphere temperature anomaly bolted onto the Mauna Loa in situ reference data ready for action. The resulting best-fit ARIMA(0,1,1)(0,1,1) model for the prediction of atmospheric CO2 (raw data rather than seasonally adjusted) using an indicator variable to mark a sustained effect of global shutdown yields the following:

There’s a big surprise here and that is the HadCRUT5 northern hemisphere combined land and sea surface temperature anomaly failed to perform as a statistically significant predictor of atmospheric CO2 at the global reference laboratory at Mauna Loa (p=0.435).

This indicates that not only is Net Zero a nonsensical political farce but there’s also no empirical evidence of a link between current levels of CO2 and surface temperatures. I didn’t expect that! What I did suspect was that my indicator for a sustained effect of global shutdown would be a flop, and I was right (p=0.375). So there you go, computer says no!

Less, Smaller, Worse

Just to prove I’m not playing tricks we are going to go down the alternative route of looking at annual mean atmospheric CO2. The time series since 2000 is pretty boring:

And my eyeballs still can’t detect any change to the incessant rise in CO2 despite the mother of all global shutdowns! Sure puts my local council’s grand scheme of a few annoying traffic bollards into perspective, I can tell you.

Dessert

I guess I better end this article with the results of an ARIMA(1,1,1) model for the prediction of mean annual atmospheric CO2 that will cheer up despondent alarmists. Try this for size:

Tada! Now that we’ve smoothed the data out to the creamy texture of the annual mean we suddenly find HadCRUT5 northern hemisphere anomaly bursting back into statistical significance as a predictor of atmospheric CO2 as measured over at Mauna Loa (p=0.001).

This is interesting because that relationship is simply not there down at the monthly level of data resolution.

Alarmists should find this pudding more palatable but there’s no getting away from the fact that a global shutdown indicator marking out 2020 failed to reach significance as a predictor of atmospheric CO2 as measured at the global reference laboratory at Mauna Loa (p=0.977).

In plain English this means we locked down the world like never before and it had no detectable effect on CO2 concentration.

Net Zero must be referring to common sense.

Wriggling Free

I think it fair to say we’ve covered a few half-decent classical ideas on how to assess the immediate impact of global lockdown on atmospheric CO2 only to find there ain’t any detectable evidence.
However, it’s very easy for clued-up alarmists to wriggle out of this by playing the residency card. They’ll argue that since residency is up around 200 years then the concentration of CO2 we are measuring today is the product of 200 years worth of human emissions of which a decline in just one year will be an undetectable drop in the ocean.

If residency is a genuine thing, and is up at 200 years, then this will indeed be the case and we won’t be able to detect anything as paltry as a year of global shutdown. This harsh fact cuts both ways for it means Net Zero (or any means of slashing ‘fossil fuel’ use) is not going to have the slightest impact on the climate for a very long time.

Thus, all those desperate souls throwing soup on paintings, those folk scared sh*tless about the future, and those folk seething with anger are basically screaming at the wind.

Nothing they can do, say, think or believe will make any difference in their lifetime or that of their children even if all ‘fossil fuel’ was banned globally at midnight tonight.

The only solid option these souls have open to them is to learn to live with the weather like our ancestors did.

All this, of course, assumes that The Science™ we’re all supposed to follow is 100 percent correct, yet it is abundantly clear that it is fatally flawed and requires serious revision.

See more here substack.com

Bold emphasis added

Header image: BBC

Please Donate Below To Support Our Ongoing Work To Defend The Scientific Method

PRINCIPIA SCIENTIFIC INTERNATIONAL, legally registered in the UK as a company incorporated for charitable purposes. Head Office: 27 Old Gloucester Street, London WC1N 3AX. 

Trackback from your site.

Comments (1)

  • Avatar

    Peter F Gill

    |

    The lack of effect on the Keeling curve was predicted by Murry Salby and myself.

    Reply

Leave a comment

Save my name, email, and website in this browser for the next time I comment.
Share via