the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Field Evaluation of Low-cost Electrochemical Air Quality Gas Sensors at Extreme Temperature and Relative Humidity Conditions
Roubina Papaconstantinou
Marios Demosthenous
Spyros Bezantakos
Neoclis Hadjigeorgiou
Marinos Costi
Melina Stylianou
Elli Symeou
Chrysanthos Savvides
George Biskos
Abstract. Modern electrochemical gas sensors hold great potential for improving practices in Air Quality (AQ) monitoring as their low cost, ease of operation and compact design can enable dense observational networks and mobile measurements. Despite that, however, numerous studies have shown that the performance of these sensors depends on a number of factors (e.g., environmental conditions, sensor quality, maintenance and calibration, etc.), thereby adding significant uncertainties in the reported measurements and large discrepancies from those recorded by reference-grade instruments. In this work we investigate the performance of electrochemical sensors, provided by two manufacturers (namely Alphasense and Winsen), for measuring the concentrations of CO, NO2, O3 and SO2. To achieve that we carried out collocated yearlong measurements with reference-grade instruments at a traffic AQ monitoring station in Nicosia, Cyprus, where temperatures range from almost 0 °C in the winter, to almost 45 °C in the summer. The CO sensors exhibit the best performance among all we tested, having minimal mean relative error (MRE) compared to reference instruments (ca. 5%), although a significant difference in their response was observed before and after the summer period. At the other end of the spectrum, the SO2 sensors reported concentration values that were at least one order of magnitude higher than the respective reference measurements (with MREs being more than 1000 % for Alphasense and almost 400 % for Winsen throughout the entire measuring period), which can be justified by the fact that the concentrations of SO2 at our measuring site were below their Limit of Detection. In general, variabilities in the environmental conditions (i.e., temperature and relative humidity) appear to affect significantly the performance of the sensors. When compared with reference instruments, the CO and NO2 electrochemical sensors provide measurements that exhibit increasing errors and decreasing correlations as temperature increases (from below 10 to above 30 °C) and RH decreases (from > 75 to below 30 %). Interestingly, the performance of the sensors was affected irreversibly during the hot summer period, exhibiting different response before and after that, and resulting to a signal deterioration that was more than twice as that reported by the manufacturers. With the exception of the Alphasense NO2 sensor, all LCSs exhibited measurement uncertainties that were much higher, even at the beginning of our measurements, compared to those required for qualifying the sensors for indicative air quality measurements according to standard protocols. Overall, our results show that the response of all LCSs is strongly affected by the environmental conditions, warranting further investigations on how they are manufactured, calibrated and employed in the field.
- Preprint
(2512 KB) -
Supplement
(1017 KB) - BibTeX
- EndNote
Roubina Papaconstantinou et al.
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2022-1474', Anonymous Referee #1, 09 Mar 2023
General Comments:
In this work, the authors conducted a field collocation study of low-cost gas sensors (CO, NO2, O3, & SO2) from two different manufacturers. Using the manufacturer provided calibrations, the authors compared the performances of these sensors to that of reference instrumentation for a period of ~14 months over a wide range of temperature and relative humidity conditions. They found that the performance of the O3 and SO2 sensors was poor and did not continue further analysis with them. With the NO2 and CO sensors they demonstrated a clear change in sensor performance after exposure to high temperature and low relative humidity conditions during the summer. When compared to EU standards for indicative measurements most of the sensors only passed two of the three criteria, while the Alphasense NO2 sensor passed all three.
Specific Comments:
What is the reasoning for using two different sensor manufacturers rather than evaluating duplicates from a single manufacturer?
Although the cross-sensitivities are stated in Table S1 and line 258 states that Alphasense provides a correction for this, it is unclear in Eq 1 and lines 157-159 which variables correspond to the NO2 and the O3 LCSs. Although the Winsen specification sheet does not mention cross-sensitivity I imagine they would have the same cross-sensitivity issues as any other EC sensor. Please discuss why further NO2 and O3 corrections attempts were not made using the available reference instrument data.
Line 177: Please include more information on the chosen temperature and RH regimes and why they were chosen.
What is the scientific reasoning for subtracting 400ppb from the Winsen CO sensors? Is this just a manufacturer issue? Line 213 explaining this offset could be moved to Section 2.3 line 173.
Line 227: It is mentioned that LCSs overestimated NO2 concentrations, which could be due to their ozone cross-sensitivity as this was not corrected for, this point is only brought up later in line 279. Perhaps start this discussion earlier in line 227.
Table 1: What do you mean by “useful measurements”?
Line 221: While the mean CO values between the LCSs and references instruments are within error the Winsen CO has low R2 and both have low slopes in Figure 4a & 4e and Table 2 so I’m not sure I agree that there is “good agreement” between these LCSs and the reference instrumentation, maybe relative to the other LCSs tested. The U.S. EPA Air Sensor Performance Targets and Testing Protocols establish statistical targets for LCSs (currently Ozone and PM have been published) which includes a bias performance metric with slope and intercept calculations. Performance targets for NO2, CO and SO2 are being deliberated. It would be useful to include these metrics in Table 1.
Duvall, R., A. Clements, G. Hagler, A. Kamal, Vasu Kilaru, L. Goodman, S. Frederick, K. Johnson Barkjohn, I. VonWald, D. Greene, AND T. Dye. Performance Testing Protocols, Metrics, and Target Values for Ozone Air Sensors: Use in Ambient, Outdoor, Fixed Site, Non-Regulatory Supplemental and Informational Monitoring Applications. U.S. EPA Office of Research and Development, Washington, DC, EPA/600/R-20/279, 2021.
https://doi.org/10.1016/j.atmosenv.2020.118099
Line 247: Please state the temperature and RH conditions for winter and summer periods respectively and how you define summer and winter (date ranges).
Technical Corrections:
Line 377: Please correct to “making it the only sensor that can”
SI Line 37: Please correct to “the further we move below”
Table S1: In the top row an error pops up in the PDF version “[Error! Reference source not found.]” Was a picture supposed to placed here?
Please include more specific information on the Peltier et al. 2020 reference.
Citation: https://doi.org/10.5194/egusphere-2022-1474-RC1 -
RC2: 'Comment on egusphere-2022-1474', Anonymous Referee #2, 13 Mar 2023
This article shows a very interesting finding, where the authors evaluate the performance of electrochemical AQ sensors (two manufacturers; Alphasense and Winsen) during a period (14 months approx) comprising extreme environmental conditions (high temperature in summer). The results show how temperature and relative humidity can worsen sensor measurements and even degrade sensor performance permanently. Still, in my opinion, there are several questions that need to be addressed before it is considered for acceptance:
- You mention that you use the reference thermometers at the station to compute the different correction factors. Do you think that there can be a difference between the temperature recorded by the reference thermometers and the real temperature at which the sensor box is exposed ? Can it impact the correction factors calculations or the final concentrations? Are you including temperature sensors in the sensor box ?
- Table S1 shows some sort of compilation errors “Error! Reference source not found”.
- Many recent works use a machine learning-based in-situ calibration, comparing the LCS measurements with the reference instrument. Do you think that some sort of sensor recalibration where the sensor calibration model is trained with data with large temperatures may reduce the worsening of the sensor ?
- As noticed in Table I, different sensors have different numbers of useful samples (N), the difference can be up to 2000. Is it due to missing data, sensor issues ?
- In Figure 3, it can be seen how the O3 estimation for the Alphasense sensor completely overestimates the reference values. Does it make sense? As far as I understand this period corresponds to beginning of the testing and the temperature is not large is this case.
- Have you evaluated/shown the possible joint effect of temperature and relative humidity ? It would be interesting to show a result by diving the data into, low T+low RH, low T + high RH, etc.
Citation: https://doi.org/10.5194/egusphere-2022-1474-RC2
Roubina Papaconstantinou et al.
Roubina Papaconstantinou et al.
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
206 | 68 | 9 | 283 | 28 | 4 | 5 |
- HTML: 206
- PDF: 68
- XML: 9
- Total: 283
- Supplement: 28
- BibTeX: 4
- EndNote: 5
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1