Good performance of low-cost carbon dioxide sensor based on intercomparisons with the standard eddy-covariance system

Rannik, Üllar; Mammarella, Ivan; Vesala, Timo; Väkimies, Pirkko; Heiskari-Tuohiniemi, Hilkka; Korkiakoski, Mika

doi:10.5194/egusphere-2026-144

Preprints

https://doi.org/10.5194/egusphere-2026-144

Preprints

05 Mar 2026

| 05 Mar 2026

Status: this preprint is open for discussion and under review for Atmospheric Measurement Techniques (AMT).

Good performance of low-cost carbon dioxide sensor based on intercomparisons with the standard eddy-covariance system

Üllar Rannik, Ivan Mammarella, Timo Vesala, Pirkko Väkimies, Hilkka Heiskari-Tuohiniemi, and Mika Korkiakoski

Abstract. Flux measurements have started to play an important role outside academia in assessing carbon sinks of different ecosystems and land-use types. If natural carbon solutions are to be deployed and monetized in carbon markets, more low-powered and low-cost flux systems should be deployed. There is a growing need for low-cost sensors that still fulfil the requirements for scientific applications. We present a case study where Vaisala company and the University of Helsinki joined their industrial and academic competencies to create an inexpensive yet precise fast-response carbon dioxide (CO₂) and water vapour (H₂O) sensor. A working prototype was developed and field-tested against a scientific reference eddy covariance (EC) setup. Special attention was paid to response time, lowered sampling frequency, and auto-calibration related to the temperature. The results at the end of the project were very promising. The enclosed-path EC prototype had a CO₂ response time of 0.18 sec and a noise level of 1 ppm at a 5 Hz sampling rate. The internal auto-calibration procedure was continuously improved such that CO₂ signal drifting was avoided and the instrument was capable of measuring CO₂ fluxes with high correlation relative to the reference EC setup (R² = 0.98).

Received: 11 Jan 2026 – Discussion started: 05 Mar 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Preprint (PDF, 3329 KB)

Supplement (614 KB)

Download & links

Üllar Rannik, Ivan Mammarella, Timo Vesala, Pirkko Väkimies, Hilkka Heiskari-Tuohiniemi, and Mika Korkiakoski

Status: open (until 09 Apr 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2026-144', Anonymous Referee #4, 26 Mar 2026 reply
This study presents a fast-response, low-cost carbon dioxide sensor for flux measurements. It is based on approximately five months of field observations and includes comparisons with a standard eddy covariance system. The sensor’s performance is thoroughly evaluated, and it appears to offer a cost-effective alternative for field CO2 flux measurements. However, a few major issues should be addressed before the manuscript can be considered for publication in AMT.
Major Comments:
Clarification of technical details: As this is primarily a technical paper intended for users of CO₂ flux measurements, I think some methodological aspects need to be explained more clearly in the manuscript itself. In particular, it is not sufficiently clear what improvements were introduced with the software update and how these changes affected the sensor performance. A more explicit description would be very helpful. Similarly, the autocalibration procedure should be described in more detail. At present, it is difficult to understand how this procedure is implemented in practice and how it affects the long-term behaviour of the sensor. Related to this, it would also be useful if the authors could comment on the temporal evolution of sensor drift. For example, does the drift increase approximately linearly with time, or does it follow a different pattern?

Response time and wind conditions: The approach used to estimate the sensor response time would benefit from some further clarification. As I understand it, the estimate is derived from data collected under unstable atmospheric conditions and within a relatively narrow wind speed range of 2.5–3.5 m s⁻¹.This raises the question of whether the derived response time is partly influenced by the prevailing wind or turbulence conditions. If so, the authors should discuss to what extent the estimated response time may depend on wind speed or atmospheric state more generally. Would the same response time be obtained under calmer conditions, or under stronger turbulence? Some discussion of the robustness of this estimate would strengthen the paper.

Uncertainty in weak flux measurements: I also feel that the manuscript should discuss more carefully the uncertainty associated with very weak CO₂ fluxes. For example, in Figure 6 the H₂O fluxes show a relatively smooth decline from July to September. The August H₂O fluxes are weaker than in July, but they are still clearly visible and remain higher than those in September. In contrast, the CO₂ fluxes over the same period remain close to zero. This makes me wonder how reliably the sensor can quantify very small CO₂ fluxes. A similar concern arises from Figure 7. If one focuses on the August data only, excluding the period from 21 July 2022 to the end of July 2022, the agreement between prototype and reference appears rather weak for small CO₂ fluxes, especially in the range of about –5 to 5 µmol m⁻² s⁻¹. This suggests that the uncertainty may be substantial in the near-zero range. I would therefore encourage the authors to discuss more explicitly the reliability and uncertainty of the sensor under weak-flux conditions.

Other Comments:
In Figure 4, sensible heat flux is used as an indicator of atmospheric stability. I wonder whether momentum flux or friction velocity (u*) might be more appropriate for this purpose. At least, the choice of sensible heat flux should be justified more clearly.
It would be useful to include some practical recommendations for long-term operation and maintenance. Based on the results shown, the system appears to require relatively frequent updates, and some guidance for future users would add value to the paper.
The prototype sensor operates at 5 Hz, whereas the reference eddy-covariance system typically operates at 10 Hz. The manuscript should discuss how this difference in sampling frequency may affect the flux calculation and the comparison between the two systems. In particular, could the lower sampling rate lead to a bias or loss of information?
Technical Corrections:
Line 16: I would suggest avoiding the word “promising”, which sounds slightly informal in this context.
Line 137: Please clarify which lag time was used for the CO₂ flux calculation.
Line 147: Delete the extra word “for”.
Line 213: Please clarify whether this refers to r or R², and use the appropriate notation consistently.
Figure 5: It would help readability to use slightly more distinct colours or symbols for the different periods.

Reply
Citation: https://doi.org/10.5194/egusphere-2026-144-RC1
RC2: 'Comment on egusphere-2026-144', Anonymous Referee #2, 27 Mar 2026 reply

This manuscript describes the development and evaluation of a low-cost CO₂/H₂O sensor for eddy covariance applications, which is particularly relevant given the growing demand for scalable carbon flux measurements. The results of this sensor are in agreement with the chosen reference system. Before publication, some clarification should be required as follows:
1. Impact of software updates on results: The paper describes some updates during the field campaign, particularly a major change on July 21. More clarification should be given to the software update and autocalibration and how that affected performance. Would this be a reliable method for a more long term use or how does drift affect over long periods of time?
2. Differences in the prototype and reference system: As stated in the paper, the prototype and reference systems are installed at different heights and separated horizontally. Additionally, the prototype operates at 5 Hz, compared to 10 Hz for the reference system. Clarification should be given in the manuscript of how these changes impact the data between the prototype and the reference. For example, how does the separation affect flux differences?
3. Line 16 says, "The results at the end of the paper are very promising..." and then proceeds to give the results. This sentence is not needed. The results should prove the promise, it does not need to be stated.
4. Line 151: "indicated the presence of noise at high frequencies", consider specifying the frequency range
5. The figures are in a blue/green color palette. Consider switching to a blue/red palette instead for readability.

Reply

Citation: https://doi.org/10.5194/egusphere-2026-144-RC2
RC3: 'Comment on egusphere-2026-144', Anonymous Referee #1, 27 Mar 2026 reply

This work designed a low-cost CO2 sensor to measure CO2 flux. The system has advantages in fast response and low noise. The good performance of the low-cost sensor was validated by simultaneous long-term field measurement of CO2 flux using a scientific reference eddy covariance setup. This study fits well within the scope of AMT and have importance implications in greenhouse gas measurements and the evaluation of their climate effects. The following concerns should be addressed before it can be accepted for publication.
Specific comments:
Lines 83-85: The fund information should go to the section “Acknowledgements”.
Lines 111-115: The software updates have a significant impact on the performance of the sensor. The authors should elaborate on what the key update is.
Lines 120-124: The CO2 mixing ratios before and after 21 July should be given to have a better understanding of how the software update shapes the measurement of CO2.
Figure 6: The current x-axis is difficult to read. I would suggest changing the x-axis to day 1 to 31.
Figure 8: The difference in CO2 flux in September is also relatively large. What is the reason?

Reply

Citation: https://doi.org/10.5194/egusphere-2026-144-RC3

Üllar Rannik, Ivan Mammarella, Timo Vesala, Pirkko Väkimies, Hilkka Heiskari-Tuohiniemi, and Mika Korkiakoski

Supplement

https://doi.org/10.5194/egusphere-2026-144-supplement

Üllar Rannik, Ivan Mammarella, Timo Vesala, Pirkko Väkimies, Hilkka Heiskari-Tuohiniemi, and Mika Korkiakoski

Viewed

Total article views: 249 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
145	88	16	249	41	15	26

HTML: 145
PDF: 88
XML: 16
Total: 249
Supplement: 41
BibTeX: 15
EndNote: 26

Views and downloads (calculated since 05 Mar 2026)

Month	HTML	PDF	XML	Total
Mar 2026	145	88	16	249

Cumulative views and downloads (calculated since 05 Mar 2026)

Month	HTML	PDF	XML	Total
Mar 2026	145	88	16	249

Viewed (geographical distribution)

Total article views: 248 (including HTML, PDF, and XML) Thereof 248 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 28 Mar 2026

Short summary

We present a case study where Vaisala company and the University of Helsinki joined their competencies to create an inexpensive yet precise fast-response carbon dioxide and water vapour sensor. A working prototype was developed and field-tested against a scientific eddy covariance setup. Attention was paid to internal auto-calibration procedure such that signal drifting was avoided and the instrument was capable of measuring fluxes with high correlation relative to the reference setup.


Total:	0
HTML:	0
PDF:	0
XML:	0