the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Vertical profiles and surface distributions of trace gases (CO, O3, NO, NO2) in the Arctic wintertime boundary layer using low-cost sensors during ALPACA-2022
Abstract. Electrochemical gas sensors (EGSs) have been used to measure the surface distributions and vertical profiles of trace gases in the wintertime Arctic Boundary Layer during the Alaskan Layered Pollution and Chemical Analysis (ALPACA) field experiment in Fairbanks, Alaska in January–February 2022. The MICRO sensors for MEasurements of GASes (MICROMEGAS) instrument set up with CO, NO, NO2 and O3 EGSs was operated on the ground at an outdoor reference site downtown Fairbanks for calibration, onboard a vehicle moving through the city and its surroundings and onboard a tethered balloon, the Helikite, at a site at the edge of the city. To calibrate the measurements, a set of machine learning (ML) calibration methods were tested. For each method, learning and prediction were performed with coincident MICROMEGAS and reference analyser measurements at the downtown site. For CO, the calibration parameters provided by the manufacturer led to the best agreement between the EGS and the reference analyser and no ML method was needed for calibration. The correlation coefficient R is 0.82 and the slope of the linear regression between MICROMEGAS and reference data is 1.12. The mean bias is not significant but the Root Mean Square Error (290 ppbv) is rather large because of CO concentrations reaching several ppmv downtown Fairbanks. For NO, NO2 and O3, the best agreements for the prediction datasets were obtained with an artificial neural network, the Multi-Layer Perceptron. For these 3 gases, the correlation coefficients are higher than 0.95 and the slopes of linear regressions with the reference data are in the range 0.93–1.04. The mean biases which are 1±3 ppbv, 0±4 ppbv and 3±12 ppbv for NO2, O3 and NO respectively are not significant. Measurements from the car round of January 21 are presented to highlight the ability of MICROMEGAS to quantify the surface variability of the target trace gases in Fairbanks and the surrounding hills. MICROMEGAS flew 11 times from the ground up to a maximum of 350 m a.g.l. onboard the Helikite at the site at the edge of the city. The statistics performed over the Helikite MICROMEGAS dataset show that the median vertical gas profiles are characterised by almost constant mixing ratios. The median values over the vertical are 140, 8, 4 and 32 ppbv for CO, NO, NO2 and O3. Extreme values are detected with low O3 and high NO2 and NO concentrations between 100 and 150 m a.g.l. O3 minimum levels (5th percentile) of 5 ppbv coincident with NO2 maximum levels (95th percentile) of 40 ppbv occur around 200 m a.g.l. The peaks aloft are linked to pollution plumes originating from Fairbanks power plants such as documented with the flight of February 20.
- Preprint
(14723 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-2421', Laurent Spinelle, 26 Sep 2024
Dear authors, congratulations for the work carried out and presented in this manuscript. A lot of details are given in the document and the comparative approach of the different calibration methods using the Taylor diagram is very well explained and used.
I nevertheless have few comments or suggestions listed below:
Line 10, “The correlation coefficient R”: only a question at this stage of the review, is it really the R or the R2?
Line 18: What does a.g.l. mean?
Line 20: do you have some data to illustrate the "almost constant"?
Line 57: What is the range of temperature?
“Figure 3” (Barret et al., 2024, p. 8) Figure 3: If the period without data is not usefull, I would advice the author to use some cut in the time series, or multiple graphs in order to maximise the visibility instead of leaving a third of the space empty.
Line 227, “the addition of voltages from the NO sensor in equ. 2.5.2”: What is equ.2.5.2, do you mean equation (1) at the top of the page ?
Line 237, “correlation coefficient R.”: In relation to the comment in the abstract, it is maybe a good idea to write down the formula.
Figure 5: it is not easy to see the difference in color of the dot in the legend (printed or pdf), I would advice to increse the size of the symbol.
Figure 6(b): could you give some info about the 2 line (I guess red = linear regression, blu = unity)
Figure 7: I think some words are missing, "Same caption as Figure 5 but for NO2 (MLP100 calibration function)."
Figure 9: Same comment as for Figure 7.
Citation: https://doi.org/10.5194/egusphere-2024-2421-RC1 -
RC2: 'Comment on egusphere-2024-2421', Anonymous Referee #2, 16 Nov 2024
This study investigates surface distributions and vertical profiles of trace gases in Arctic wintertime using low-cost sensors. The MICROMEGAS instrument, equipped with EGSs, was deployed at a downtown calibration site, onboard a vehicle, and on a tethered balloon at the city’s edge. Regarding sensor calibration, results indicate that manufacturer-provided parameters were sufficient for CO, while a Multi-Layer Perceptron neural network yielded the best results for NO, NO₂, and O₃. MICROMEGAS demonstrated effectiveness in capturing surface and vertical variability in gas concentrations. Overall, the manuscript presents interesting findings. However, I have several major comments:
General comments:
- Based on the manuscript flow, consider modifying the title to: “Surface Distributions and Vertical Profiles of Trace Gases (CO, O₃, NO, NO₂) in the Arctic Wintertime Boundary Layer Using Low-Cost Sensors During ALPACA-2022”.
- The study emphasizes the novelty of deploying EGS in cold environments, but the impacts of temperature and relative humidity on sensor performance are not discussed. Please address this in the manuscript.
- Line 95 mentions SO₂ measurement, but no data is presented. Either provide detailed discussion of SO₂ data or remove its mention.
- Cross-interference: In lines 225–230, what contributions do NO and NO₂ concentrations have on O₃ sensor calibration under cold conditions? Would these impacts vary with temperature changes?
- Lines 145–150: Discuss whether vehicle emissions would influence sensor detection. Was the detection of the sensors based on gas diffusion, or was there an air pump? If a pump was used, what was its flow rate? If based on diffusion, would sensor response vary at different vehicle speeds? For example, if the vehicle speed is too fast, the sensor response may not be fast enough to capture the rapid change of ambient concentration. Please discuss.
- Lines 152–155: What were the vertical profiles of temperature, RH, and pressure during balloon ascents? How did rapid changes in these parameters affect sensor performance?
- Lines 152-155: In the balloon measurements, have the authors compared EGS data with high-accuracy instruments (e.g., for CO and O₃) to validate the sensor’s vertical profiling capabilities?
- Figure 3: Why not randomly separate the training and testing datasets? This approach might be more robust and objective.
- Lines 202–205: Add a brief description of the HGBT and RF parameterization methods used.
- Line 259: Include a brief explanation of how to interpret the Taylor diagram.
- Lines 259–271: Until the end of the manuscript, I realized that Figure 4b presents the information of the raw data, and the raw data for CO was used in this study without any correction. This information should be clearly stated here.
- Lines 292–296: Could the relatively worse performance of HGBT and RF models result from the data selection process? Consider discussing whether a 10-fold cross-validation or random data selection could improve predictions.
- Line 296: What is meant by the term “raw calibration method”? Does it refer to using raw data directly or applying a linear correction to the raw data?
- Line 305: The manuscript lacks information on Ox calibration. Please include this.
- Line 327: Define what is meant by “systematic bias.”
- Figure 15: Was the time resolution of sensor data consistent with that of the reference instrument? If not, could observed spikes in sensor data be due to its higher time resolution?
- Section 4: Discuss the limitations of this study and please provide suggestions for future research on low-cost sensors in cold environments.
Technical comments:
- Line 57: Extremely, not extreme.
- Line 142: Correct "ouside" to "outside."
- Line 208: Figure 7 is referenced before Figure 6. Correct the order.
- Line 227: Replace "Equ." with "Eqn. 2.5.2."
- Figure 15: Trace gas, not trace gaz, in the caption. For better visualization, consider using a different symbol instead of dark circles.
Citation: https://doi.org/10.5194/egusphere-2024-2421-RC2
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
254 | 75 | 103 | 432 | 12 | 9 |
- HTML: 254
- PDF: 75
- XML: 103
- Total: 432
- BibTeX: 12
- EndNote: 9
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1