the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 05 Aug 2026
Assessing the benefit of sample drying for AE33 aethalometers: evidence from long‑term field observations
Abstract. Aethalometers measuring equivalent black carbon (eBC) are widely used in both research and regulatory air quality monitoring networks. However, uncertainties regarding the influence of moisture in eBC measurements persist. While many networks recommend or require sample drying, there is limited long‑term field evidence quantifying the benefits of humidity control for aethalometers operated under real‑world conditions. Here we present a multi‑year field intercomparison of aethalometers (AE33, Aerosol Magee Scientific) operated with and without sample drying at an urban background site in Sydney, Australia. Three extended tests (December 2015 – August 2016, November 2016 – June 2017, January 2018 – March 2019) were conducted using four instruments in total. Instruments were co‑located within the same climate‑controlled enclosure and configured identically, differing only in sample humidity conditioning (membrane dryer). Across all tests, wavelengths and averaging periods, strong agreement was observed between dried and non‑dried instruments, with correlation coefficients (R²) consistently exceeding 0.98. For the 880 nm channel, regression slopes for non-dried vs dried sampling ranged from 0.97 – 1.07 across the three tests. Similar results were obtained for 1‑, 2‑, 5‑, and 10‑minute averages, although variability increased at shorter averaging intervals. Stratification by ambient relative humidity, dew‑point temperature, and dew‑point deficit showed no systematic degradation in agreement between hourly averages at high humidity, and moisture‑related effects were small compared with test‑to‑test variability. These results indicate that, for hourly eBC measurements in a temperate, high‑humidity coastal environment, the absence of sample drying introduces only small differences in measured concentrations when compared with humidity‑controlled sampling. While short‑term variability and instrument‑specific effects remain important considerations, these results suggest that long‑term, network‑scale eBC measurements from AE33 aethalometers can remain highly comparable even without sample drying. This study provides quantitative evidence to inform guidance on humidity control for aethalometers in monitoring networks.
- Preprint
(2395 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-4169', Anonymous Referee #1, 10 Sep 2026
-
AC1: 'Reply to RC1', Matthew Riley, 25 Oct 2026
We thank you for the attention to our manuscript and the careful comments that you have provided through the review process. We have accepted in full almost all of your comments and suggestions. Similar to comments from reviewer RC2 you have identified some issues with our description of the dried sampling system and the need to better characterise the observed humidity conditions, including the instrument response at or below 40% RH. Unfortunately, we did not log the performance of the drier and hence the hourly RH of the dried sample. But we have added some additional analysis to assist with interpretation of results in lieu of that data. We include analysis for each test segmenting data into bands of below and above 40% ambient RH. Notably this shows no difference in the results between ambient "dry" conditions and other "wetter" conditions.
We have addressed all comments/suggestions below. We thank you again for reviewing the manuscript and your supporting comments.
Regards,
The Authors
Detailed response
The manuscript does not define the RH range observed for the dry or the ambient sample. That would be important to define so that they could be compared to other measurements. Would be good to show also the RH for the dried sample as a background data. Was it measured or has it been tested if the drying capacity was enough (e.g., did it stay below 40 % as recommended by GAW)? How big fraction of the ambient sample was below RH < 40 %, which could be considered dry aerosol by GAW and is not expected to affect the measurements so much. [Comment/Response] Unfortunately we did not continuously log the sample RH after drying. We have added text that describes this and the maintenance checks using the LED indicators that suggested that the dryer was always performing sufficiently well. However, we have now included a brief analysis of the samples where ambient RH was <40% and hence drying should have little effect on the sample. The number of hours this occurred is reported and the results of the comparison between the analysis for periods where RH was below 40% and all others. We have added this text to section 3.3: This is further illustrated when comparing results by test when segregated to two bands: hours where ambient relative humidity was below 40% and all other hours. For each test the number of hours where ambient relative humidity was <40%, and hence the dryer was not altering sample humidity, was Test 1 405 hours, 6.5% of samples; Test 2 235 hours, 4.8% of samples; and Test 3 875 hours, 9.2% of samples. The difference in slope estimates across all channels and all tests was not significant when applying the Wald test with Holm-adjusted p-values.
The manuscript focuses on equivalent black carbon measurements, which are conducted on wavelength 880 nm, which is of course useful in the viewpoint of air quality monitoring. Comparison at other wavelengths is also shown but not discussed so much. This could be done with absorption Ångström exponent (AAE), which is a very commonly derived parameter from AE33 data and used in the “aethalometer model” also mentioned in the manuscript. Therefore, by extending the analysis on AAE, the scientific impact of the manuscript would be improved. [Response] We elected not to include AAE analysis originally focussing instead on the raw channel outputs. However, with your suggestion we have now included in Table 4 results for AAE (470-880 nm) and in section 3.2.
Abstract: Please mention what were the RH ranges for dry and ambient samples to define what is meant with dried and ambient samples. [Response] We have added text to the abstract defining dry as <40% relative humidity.
L179: What was the shed temperature on average and how much it varied? I think stating this shortly would be good background information for the reader to have an idea of the differences in outdoor and indoor temperatures (as it affects the sample RH). [Response] We have added another figure (new Figure 3) which shows the distribution of the ambient temperature and moisture and the dew point deficit. We have also added daily average shed temperature and humidity to Figure 2 and added text to section 2.1 "Temperature control set points for the shed was in the range 22–25 °C" and to section 3.1 “Shed temperatures ranged between 20.6 and 35.6 °C, with 89.6% of shed temperature observations between 23–27 °C ”
L195-197: It would be good to mention here the materials of the filters and if they can they be affected humidity in a different manner? [Response] We have updated text in section 2.2 “Yus-Díez et al. (2021) reported that differences in aethalometer measurements between 8020 and 8060, which are both PTFE-coated glass-fibre tapes, were within the instrument uncertainty”.
L233-234: Indeed, it is unfortunate that the instruments were not compared. Having even a short comparison period before each deployment at the station either with ambient or dried sample would have make the results more reliable. But I assume that having a regular maintenance and testing for the instruments gives good enough basis here. Performing the neutral density tests should raise an issue, if the instrument would measure too high or low concentrations. I suggested mentioning the unit-to-unit variability (Cuesta-Mosquera et al., 2021) also here to remind the reader about expected agreement between the instruments. [Comment/Response] Yes, it is unfortunate that we did not do this. We are currently constructing a further field comparison that will commence and end with side-by-side instrument comparisons (and within the test period). We thank you for the observation on the ND tests and after again checking maintenance records have added the following text “During the neutral density filter tests no significant variances between instruments were observed”. We did not add further discussion of results from Cuesta-Mosquera et al. (2021) here, leaving this to the final sections.
Figure 2: Please mention that the T, RH, and dew point are ambient outdoor measurements to avoid misunderstandings. RH from indoor could be included in the RH panel to show the variation between indoor and outdoor air. RH is more practical parameter and easier to understand than DPD. If I understood correctly, the indoor RH was measured with the Vaisala sensor (or it could be estimated based on the outdoor T and RH, and indoor T). [Response] We have updated the figure to include both the ambient and internal (shed) T and RH. The indoor T and RH were measured by a Vaisala sensor identical to the outdoor sensor.
L262-263: I do not understand why run the tests for ambient outdoor RH as that does not represent the measurement conditions, especially not in winter if the sampled air warms up to shed temperature and the sample RH decreases. This is also questioned in the manuscript on L389-394 and in the discussion. I would recommend running the tests for indoor (or sample) RH, which is typically monitored at stations or by aerosol instruments themselves and therefore the results would be easier to interpret. [Comment] Unfortunately we do not have continuously logged sample RH for the dried sample. We are unsure if there would be any benefit in comparing internal (shed) humidity as there is little to no direct moisture interaction between shed air and sample air, unlike temperature. This is why we analyse the DPD, as an effort to assess if sample train induced temperature changes may influence the results. We are preparing a further comparison where we will attempt to improve on this study be logging additional sample parameters (T, RH, DP) across multiple identical nafion drying systems in the field.
L356-359 Isn’t this expected, as those are the months when the indoor ambient sample RH is the highest because there is no “passive drying” due to the indoor and outdoor temperatures being fairly similar. [Response] Yes, this is correct for the most part. We have added text During the warm season internal shed conditions are more likely to be similar to ambient conditions than during the cooler months.
Citation: https://doi.org/10.5194/egusphere-2026-4169-AC1
-
AC1: 'Reply to RC1', Matthew Riley, 25 Oct 2026
-
RC2: 'Comment on egusphere-2026-4169', Anonymous Referee #2, 28 Sep 2026
The article provides a field comparison of aethalometer AE33 response with and without sample drying. The subject is important for monitoring networks which extensively use filter photometers for measurement of eBC.
The study was performed 3 times at one location with different instrument pairs and different types of filters. The results show that on average the instrument response is not connected to the sample humidity. There is a long timeseries and a good statistical evaluation of the results. The article is nicely written.
The main drawback of this study is that the performance of the Nafion drier is not reported. Because of that it is difficult to access the relevance of this study. The performance of the drier should be monitored regularly because it changes with the contamination of the Nafion membrane.
Another limit of this study is that measurements were conducted at a single location. For an urban site it is expected to measure fresh, local emissions. The presence of particle coating in aged aerosol could influence the results because the coating can increase the mass absorption cross-section of black carbon. Humidity can then interact with the properties and thickness of the coating.
L57: Parameter C also depends on particle size (Supplement of Romshoo et al., 2022: https://doi.org/10.5194/amt-15-6965-2022)
Table 1.: When comparing 23 instruments it is better to show the standard deviation of slopes than the range, where a bed performance of single instrument can affect the reported result. Cuesta-Mosquera et al. (2021) showed, that after maintenance the variation between the instrument response was 1%.
L207 and L213: The correct tube inner diameter is 6.35 mm.
L211: Please provide exact description of the Nafion drier: length, tube diameter. How was it operated: vacuum or dry sheath air?
Table 4. Does the slope represent the ratio between dried and non-dried BC or the oposite?
Table 6. Please explain n.
L495: A type of aerosol (fresh vs. aged) can also affect the results.
L535: “Sample drying is probably less important and provides limited benefit for hourly reporting typical of regulatory monitoring networks”. This might only apply to the urban sites with non-coated black carbon particles.
Citation: https://doi.org/10.5194/egusphere-2026-4169-RC2 -
AC2: 'Reply on RC2', Matthew Riley, 25 Oct 2026
Thank you for reviewing our manuscript. Your supportive comments have been helpful. We have accepted almost all of your comments and suggestions. Similar to reviewer RC1 you identified shortcomings regarding our lack of data on the performance of the drying system through measurements of T/RH of the dried sample. We have included some additional analysis that we believe assists somewhat in overcoming that lack of data.
We again thank you for your helpful comments and suggestions. Our detailed responses are below.
Regards,
The Authors
Detailed comments
The main drawback of this study is that the performance of the Nafion drier is not reported. Because of that it is difficult to access the relevance of this study. The performance of the drier should be monitored regularly because it changes with the contamination of the Nafion membrane. [Comment] We agree with this assessment, noting that due to data logging considerations we did not have good data to assess the performance of the drier. Ideally, we would have logged all data from the drier. While not logging data directly from the drier, inspection of the drier alarm LEDs occurred routinely, and no alarms were observed. We have added text to section 2.3 “The dryer LED status indicator was checked weekly and throughout the study was always reported as green“. Although we did not assess the drier performance this is compensated somewhat by the length of the comparison and the use of multiple instruments. The long test period allowed sampling to occur under a wide range of ambient humidity conditions. This is illustrated in a newly added figure (new Figure 3) which shows the distribution of temperature and moisture measurements during each of the tests. We also add in section 3.3. an assessment of observations where ambient RH <40% and hence dryer impacts should be low. In response to you and RC1 we added brief analysis of the samples where ambient RH was <40% and hence drying should have little affect on the sample. The number of hours this occurred is reported and also the results of the comparison between the analysis for periods where RH was below 40% and all others. We have added this text to section 3.3: This is further illustrated when comparing results by test when segregated to two bands: hours where ambient relative humidity was below 40% and all other hours. For each test the number of hours where ambient relative humidity was <40%, and hence the dryer was not altering sample humidity, was Test 1 405 hours, 6.5% of samples; Test 2 235 hours, 4.8% of samples; and Test 3 875 hours, 9.2% of samples. The difference in slope estimates across all channels and all tests was not significant when applying the Wald test with Holm-adjusted p-values.
Another limit of this study is that measurements were conducted at a single location. For an urban site it is expected to measure fresh, local emissions. The presence of particle coating in aged aerosol could influence the results because the coating can increase the mass absorption cross-section of black carbon. Humidity can then interact with the properties and thickness of the coating. [Comment] We agree and further testing at differing sites is one of our key conclusions/recommendations.
L57: Parameter C also depends on particle size (Supplement of Romshoo et al., 2022: https://doi.org/10.5194/amt-15-6965-2022). [Response] We have added that reference and the relationship of C to particle size
Table 1.: When comparing 23 instruments it is better to show the standard deviation of slopes than the range, where a bed performance of single instrument can affect the reported result. Cuesta-Mosquera et al. (2021) showed, that after maintenance the variation between the instrument response was 1%. [Response] Although Cuesta-Mosquera et al. (2021) do not report the formal variance in their results, we have calculated mean and standard deviation of the results from their instrument tests on ambient air for both before and after maintenance and include these as “The mean of slopes across all 23 instruments before/after maintenance is calculated as 0.99/0.96, with standard deviations of 0.12 and 0.11 respectively”.
L207 and L213: The correct tube inner diameter is 6.35 mm. [Response] Thank you for identifying this. We have corrected the text.
L211: Please provide exact description of the Nafion drier: length, tube diameter. How was it operated: vacuum or dry sheath air? [Response] The drier used is the Magee M5610 which uses vacuum sheath air. We have added a more thorough description of the dryer in section 2.2
Table 4. Does the slope represent the ratio between dried and non-dried BC or the oposite? [Response] Slopes are dry/ambient. We have added this text to section 2.3 “We report slopes (dry/ambient) ….”
Table 6. Please explain n. [Response] n is the number of hourly observations in the comparison. We have added this to the table caption.
L495: A type of aerosol (fresh vs. aged) can also affect the results. [Response] We have added “or with different mixes of aged or fresh BC particles”
L535: “Sample drying is probably less important and provides limited benefit for hourly reporting typical of regulatory monitoring networks”. This might only apply to the urban sites with non-coated black carbon particles. [Comment] This may be the case and our conclusion that “Further multi-instrument comparisons across a broader range of environments are warranted” reiterates the need for more work testing under different environmental/airshed/source conditions.
Citation: https://doi.org/10.5194/egusphere-2026-4169-AC2
-
AC2: 'Reply on RC2', Matthew Riley, 25 Oct 2026
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 258 | 117 | 41 | 416 | 30 | 33 |
- HTML: 258
- PDF: 117
- XML: 41
- Total: 416
- BibTeX: 30
- EndNote: 33
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review report for ”Assessing the benefit of sample drying for AE33 aethalometers: evidence from long‑term field observations”
In their study, Riley et al. investigate whether drying the sampled air is critical in Aethalometer (model AE33) measurements. As AE33 instruments are widely used in monitoring black carbon (BC) concentration and as BC is an emerging air quality parameter, the number of active AE33 measurements will likely increase even further. Aiming to keep the measurements comparable, drying the sample is recommended by several research infrastructures. However, setting up a dryer is not always possible or the drying capacity is not sufficient to keep the relative humidity (RH) under the recommended 40 %. This study raises an important point by studying whether we can still get comparable results for black carbon even though the aerosol sample has not been dried. I suggest accepting the manuscript after changes, which I have suggested below.
General comments
The manuscript does not define the RH range observed for the dry or the ambient sample. That would be important to define so that they could be compared to other measurements. Would be good to show also the RH for the dried sample as a background data. Was it measured or has it been tested if the drying capacity was enough (e.g., did it stay below 40 % as recommended by GAW)? How big fraction of the ambient sample was below RH < 40 %, which could be considered dry aerosol by GAW and is not expected to affect the measurements so much.
The manuscript focuses on equivalent black carbon measurements, which are conducted on wavelength 880 nm, which is of course useful in the viewpoint of air quality monitoring. Comparison at other wavelengths is also shown but not discussed so much. This could be done with absorption Ångström exponent (AAE), which is a very commonly derived parameter from AE33 data and used in the “aethalometer model” also mentioned in the manuscript. Therefore, by extending the analysis on AAE, the scientific impact of the manuscript would be improved.
Detailed comments
Abstract: Please mention what were the RH ranges for dry and ambient samples to define what is meant with dried and ambient samples.
L179: What was the shed temperature on average and how much it varied? I think stating this shortly would be good background information for the reader to have an idea of the differences in outdoor and indoor temperatures (as it affects the sample RH).
L195-197: It would be good to mention here the materials of the filters and if they can they be affected humidity in a different manner?
L233-234: Indeed, it is unfortunate that the instruments were not compared. Having even a short comparison period before each deployment at the station either with ambient or dried sample would have make the results more reliable. But I assume that having a regular maintenance and testing for the instruments gives good enough basis here. Performing the neutral density tests should raise an issue, if the instrument would measure too high or low concentrations. I suggested mentioning the unit-to-unit variability (Cuesta-Mosquera et al., 2021) also here to remind the reader about expected agreement between the instruments.
Figure 2: Please mention that the T, RH, and dew point are ambient outdoor measurements to avoid misunderstandings. RH from indoor could be included in the RH panel to show the variation between indoor and outdoor air. RH is more practical parameter and easier to understand than DPD. If I understood correctly, the indoor RH was measured with the Vaisala sensor (or it could be estimated based on the outdoor T and RH, and indoor T).
L262-263: I do not understand why run the tests for ambient outdoor RH as that does not represent the measurement conditions, especially not in winter if the sampled air warms up to shed temperature and the sample RH decreases. This is also questioned in the manuscript on L389-394 and in the discussion. I would recommend running the tests for indoor (or sample) RH, which is typically monitored at stations or by aerosol instruments themselves and therefore the results would be easier to interpret.
L356-359 Isn’t this expected, as those are the months when the indoor ambient sample RH is the highest because there is no “passive drying” due to the indoor and outdoor temperatures being fairly similar.