CO2 calibration impact on outdoor measures

[Guilherme]

Hi all,

I was reading this post on the global CO2 map (We Launched the Largest Real-Time Global CO₂ Map), and found it fascinating. I particularly liked the graph showing how the average global CO2 ppm increased over the years.

Now, while I understand the blog described use cases of detecting CO2 spikes, and how one can correlate that to particular emissions, I am also a bit concerned about a potential limitation on measuring baseline CO2 levels.

I thought the latter would be relevant for (a) measure the global average trend of outdoor CO2 levels; (b) identify places / cities that have baseline CO2 concentration higher than others; and (c) fine tune calibration of indoor monitors by understanding what is the actual CO2 outdoor level around me (instead of hardcoded 400ppm).

However, as the sensor drifts over time and calibration is needed (by default every week), I believe it might not be possible to evaluate any of the aforementioned items, right? Because either the sensor will get an offset to force its baseline to artificially become 400ppm again, or we disable the calibration, but then no longer know if the observed baseline drift of CO2 level is caused by environmental conditions or rather sensor electronics. And yet another factor to account for is the altitude above sea level where the monitor is located.

I wanted to share these raw thoughts on the forum to confirm whether these are indeed limitations, or if I may have misunderstood or overlooked something here. I am looking to learn more from the community and also to understand how I can best make use of the global CO2 map. Any thoughts would be very appreciated, thanks!

[Achim_AirGradient]

Yes, as pointed out, the low cost CO2 sensors are not suitable for absolute atmospheric measurements in the single digit ppm range that would be required to track the average increase of atmospheric CO2 levels.

However they work well in showing local emission spikes and also certain patterns, e.g. day night etc.

Now if we think that actually the progressive increase in atmospheric CO2 levels is actually the result of localised emissions averaged in the atmosphere measuring and reducing these localized effects can actually have an impact on slowing down the atmospheric increase.

[Guilherme]

Hi @Achim_AirGradient , thanks for confirming. I was now further exploring the CO2 global map, and identified indeed a few cases that are showing higher CO2 values at the moment, but their daily time series seems to confirm these to be rather spikes, e.g.:

However, there is one particular case that intrigued me - even if I click to visualize the last 2 weeks samples, all values are >= 450 ppm (always on yellow or orange color coding), as below:

So in the case above we can likely conclude there are spikes (orange bars) in compared to a given baseline, but it is difficult to interpret whether the yellow bars are actually a consistent baseline value or rather caused by other unknown factors.

And yet another interesting pattern is to be found here:

There seems to be isolated samples with quite low CO2 values (green bars), and one of these being even at 331 ppm. But then I don’t know if the green bars are the actual baseline values (can we trust 331?), or rather artifacts (though I’m not yet familiar with potential sources of artifacts in CO2 sensors). And therefore I find it difficult to conclude on whether the orange bars are sources of CO2 emissions, quite prolonged over the entire day, or rather the actual (drifted?) baseline value for CO2.

Speaking of drift, and aiming to try to make these values more interpretable - would it be possible to send to the server the timestamps in which the CO2 sensors were calibrated? And then show then as vertical lines in the time series, so one could potentially understand what could have changed over time there. Also, is it possible to retrieve from the CO2 sensor (and display in those time series too) what was the offset obtained from the calibration and being applied to the measures?

[Achim_AirGradient]

As far as I know we don’t know when the sensor module auto calibrates but I like the idea to have a notification.

In general for the outdoor CO2, we might implement in future a fixed time that the calibration happens so that all sensors have a specific calibration date.

All in all, I would say we are still more in an explorative and data collection phase here as there are hardly any large outdoor monitor networks with low cost CO2 sensors and to some extent the best calibration model still needs to be developed.

[MallocArray]

Out of the box you may not know when the sensor calibrates, but you can disable the automatic calibration and then have the firmware send the calibration command every week to match the default behavior, but then you are in control of it and can write back to the API when it was initiated. If the interval is configured differently, you still have full control of the notification when that happens