the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Feb 2023
Cloud response to co-condensation of water and organic vapors over the boreal forest
Abstract. Accounting for the condensation of organic vapors along with water vapor (co-condensation) has been shown in adiabatic cloud parcel model (CPM) simulations to enhance the number of aerosol particles that activate to form cloud droplets. The boreal forest is an important source of biogenic organic vapors, but the role of these vapors in co-condensation has not been systematically investigated. In this work, the environmental conditions under which strong co-condensation -driven cloud droplet number enhancements would be expected over the boreal biome are identified. Recent measurement technology, specifically the Filter Inlet for Gases and AEROsols (FIGAERO) coupled to an iodide-adduct Chemical Ionization Mass Spectrometer (I-CIMS), is utilized to construct a volatility distribution of the boreal atmospheric organics. Then, a suite of CPM simulations initialized with a comprehensive set of concurrent aerosol observations collected in the boreal forest of Finland during Spring 2014 is performed. The degree to which co-condensation impacts droplet formation in the model is shown to be dependent on the initialization of the updraft velocity, aerosol size distribution, organic vapor concentration and the volatility distribution. The predicted median enhancement in cloud droplet number concentration (CDNC) due to accounting for the co-condensation of water and organics is 20 % (interquartile range 29–14 %). This corresponds to activating particles 12–16 nm smaller in dry diameter, that would otherwise remain as interstitial aerosol. The highest CDNC enhancements (ΔCDNC) are predicted in the presence of a nascent ultrafine aerosol mode with a geometric mean diameter of ~40 nm and no clear Hoppel minimum, indicative of pristine environments with a source of ultrafine particles (e.g., via new particle formation processes). Such aerosol size distributions are observed 30–40 % of the time in the studied boreal forest environment in spring and fall when new particle formation frequency is the highest (six years of statistics). Five years of UK Earth System Model (UKESM1) simulations are further used to evaluate the frequencies to which such distributions are experienced by an Earth System Model over the whole boreal biome. The frequencies are substantially lower than those observed at the boreal forest measurement site (< 6 % of the time) and the positive values, peaking in spring, are modeled only over Fennoscandia and western parts of Siberia. For the aerosol size distribution regime simulated by UKESM1, offline simulations with the adiabatic parcel model reveal the ΔCDNC to be sensitive to the concentrations of semi-volatile and some intermediate-volatility organic compounds (SVOCs and IVOCs). The magnitudes of ΔCDNC remain less affected by the more volatile vapors such as formic acid and extremely low and low volatility organic compounds (ELVOCs and LVOCs) in the CPM simulations. The reasons for this are that most volatile organic vapors condense inefficiently due to their high volatility below cloud base and the concentrations of LVOCs and ELVOCs are too low to gain significant concentrations of soluble mass to reduce critical supersaturations needed for droplet activation. Suppression of the critical supersaturation caused by organic condensation is the main driver of the modeled ΔCDNC. The results highlight the potential significance of co-condensation in pristine boreal environments close to sources of fresh ultrafine particles. For accurate predictions of co-condensation effects on CDNC, the representation of the aerosol size distribution is of essence. Further studies targeted at finding observational evidence and constraints for co-condensation in the field are encouraged.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(3468 KB)
-
Supplement
(2381 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3468 KB) - Metadata XML
-
Supplement
(2381 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-164', Anonymous Referee #1, 29 Mar 2023
The manuscript describes modelling experiments regarding co-condensation of organic vapour and water in a rising air parcel. The main improvement compared to previous work is better quantification of the more volatile vapours using the FIGAERO-I-CIMS observations. The topic of the paper is interesting, as co-condensation could potentially have significant impact on cloud droplet activation but this effect is currently ignored in majority of models. The paper is generally well written and detailed.
Comments:
- In the introduction the authors should briefly explain how co-condensation differs from separately simulating condensation of water (taking into account hygroscopicity of the aerosol) and organic vapours (e.g. original VBS that doesn’t account for water in the particle), as it is presently commonly done in models.
- line 86: “the cooling of the rising air triggers also their condensation” – this gives the impression that the condensation of the higher volatility vapours is purely due to lower temperature, while actually the main driver seems to be the aerosol water content. Please restate.
- The authors should carefully check all their formulas. For example: Formula (5), line 184 : “ρa is the density of the particle dry mass” –shouldn’t this be air density, as nij is probably per cubic meter? Formula (14): The condensation of organic species seems to depend only on water density and molar mass, of which the density seems to cancel out?
- It is unclear from the description of the CPM what assumptions have been made about the hygroscopicity of the condensing organics. What value of hygroscopicity parameter for the SOA would these assumptions lead to? How does it compare with other modelling studies or observations?
- How does UKESM simulate biogenic SOA and its precursors?
- The authors convincingly show that new particle formation is an important factor for creating the suitable particle size distributions for co-condensation to affect the CDNC. However, boundary layer new particle formation is not included in UKESM simulations and the simulated size distributions are shown to be quite different from the ones observed in Hyytiälä. I am not sure if there is any value in showing these results in this already long paper.
- Lines 493-501: Topping et al. (2013) found very significant reduction in condensation of their highest volatility bin. Taking into account that more volatile species are less soluble, the reduction should be even larger for the higher volatility bins of this study. Given that most of the condensed mass seems to come from bins where log10 C* >= 3, how large are the authors expecting the ideality related overestimations to be?
- In sections 3.3 and 3.4 the frequencies of the criteria for significant impact were analyzed one by one. What about the frequency of all the criteria fulfilled at the same time?
- Lines 708-710: “While the SMEAR II PNSD data are retrieved at ground level, utilization of the UKESM1 modal parameters (only soluble modes considered) from CB is chosen, because these PNSD log-normal parameters would actually meet the cloud droplet activation scheme in the model.” - I find this selection a bit odd. Wouldn’t it make more sense to look at the near-surface PNSD that would have a chance to grow with co-condensation in updrafts before cloud activation? How large is the difference between near-surface and BC PNSDs in UKESM?
Citation: https://doi.org/10.5194/egusphere-2023-164-RC1 -
RC2: 'Comment on egusphere-2023-164', Anonymous Referee #2, 19 Apr 2023
General comment
This study investigates how cloud droplet number concentrations (CDNC) are altered through co-condensation of organic vapours along with water vapour during cloud events in boreal biomes. It uses an adiabatic cloud parcel model (ICPM), which is initialized at 90 % RH to simulate cloud formation in upraising air at updrafts of 0.1, 0.3, and 1 m/s. The simulation of the biogenic organic aerosol is based on volatility basis sets derived from measurement data of gas phase and condensed phase composition performed in the boreal forest of Finland during Spring 2014. A median enhancement in CDNC of 20 % was predicted, and the most suitable conditions for a strong CDNC enhancement through co-condensation were found to occur in the presence of a nascent ultrafine aerosol mode, conditions occurring quite frequently in fall and spring. To come to this conclusion, the authors performed 97 ICPM simulations representing the atmospheric conditions: temperature profile, aerosol composition, and particle number size distribution (PNSD) fitted as an Aitken and an accumulation mode. Each simulation was initialized at the height and temperature where 90 % RH is reached and then run at the three pre-defined updrafts. The PNSD and aerosol chemical composition (gas and condensed phase) measured near ground level were taken to be identical to those at 90 % RH. Like this, PNSD, volatility distributions and temperature were different for every simulation. This variation in conditions makes the explanation of the model initialization and the interpretation of the results quite complex, and at times confusing. As the topic of this study is highly relevant and this study is the first one to simulate co-condensation in such detail, the method description especially the adaptation of the VBS to the starting temperature and the ICPM initialization need to be improved. Altogether, this is a very thorough study that deserves publication in ACP.
ICPM initialization
Information about the ICPM initialization is spread over the whole manuscript and can sometimes be inferred only indirectly. This hampers the reading and understanding of the manuscript.
- On line 423, it is stated that an internal mixture of organics and ammonium sulfate (AS) is simulated, but the amount of AS in each simulation is not mentioned nor is the role of AS in co-condensation discussed.
- For the CJ distributions, it is stated on line 333 that “After the temperature adjustments, the volatility distributions are binned to ranges between logC* = [-8, 3] spaced by one decade in C*”. Comparing the VBS displayed in CJ with the ones labelled as CJ in Fig. 2a, similarities are difficult to detect. Figure 2 in CJ displays in 11 panels different VBS. All of them show constant low organic mass in the bins below logC* = -1 and increasing or even steeply increasing organic mass in the bins from logC* = 0 to 3. Topping et al. (2013) also displays a VBS with strongly increasing organic mass in logC* > 0 in their Fig. 2. This is opposite to the VBS distribution labelled as CJ in Fig. 2a. Why is this? To correctly adjust the VBS from CJ to the lower ICPM initialization temperature, information of the volatility bins with logC* between 4 and 6 at 298°C would be required, but such data is not given in CJ. So, how could the temperature adjustment be performed at all?
- Hunter et al. (2017) display in their Fig. 1c VBS derived from different measurement techniques including a gas-phase and a condensed-phase CIMS. The VBS derived from CIMS (g) data is very different from the one derived from PTR-MS for the same volatility range. As the choice of VBS is crucial for co-condensation, the derivation of this VBS needs to be explained better.
- For the F distribution, an adjustment to the initialization temperature is mentioned on line 370. How exactly was this adjustment carried out? How were the ΔHvap for the adjustment determined?
- Was a full equilibration carried out at the initialization temperature and RH? If yes, kinetic effects would only become important for the increase of RH from 90 % to 100 % and equilibration could be overestimated. The opposite would be the case if no equilibration were carried out at model equilibration. Can you comment on this?
- On line 162, it is stated that PNSD were constructed at model initialization. Then, on line 236, it is stated that ”The PNSD for the ICPM initialization are obtained from the Differential Mobility Particle Sizer (DMPS) measurements from SMEAR II performed within the forest canopy.” Do these measurements reflect the actual temperature at the forest canopy and dry conditions? Were they adjusted to reflect 90 % RH?
Discussion of the results
- One general weakness in the discussion of the results is that the role of temperature and RH for the condensation of organic species is not kept properly apart. The effect of the lower temperature is e.g. shown in Fig. 4, panels g-i, but in the discussion of co-condensation depending on the season, the influence of temperature on ΔCDNC is not mentioned. If a full temperature adjustment had been carried out at model initialization, a temperature dependence as shown in Fig. 4 should not be present. The question therefore remains to what degree the higher co-condensation in spring and autumn compared with summer is caused by the lower temperatures during these seasons. In a revised manuscript version, the respective role of temperature and co-condensation should be discussed more explicitly.
- On lines 493–501, the potential influence of non-ideality on gas-particle partitioning of organics is discussed. Despite recognising the shortcomings due to the assumption of ideality, quite detailed predictions of the co-condensation effect in boreal regions are made without taking the issue of solution non-ideality up again. A discussion of this aspect should be part of a revised discussion. The O:C could be estimated from CIMS and AMS data given in Fig. 2f. Such an estimate is strongly recommended to improve the manuscript.
- Uncertainties in the PNSD should also be critically assessed as these were highlighted to be crucial to determine ΔCDNC. Bimodal fits (Aitken and accumulation modes) have been performed to describe the measured PNSD from the SMEAR II campaign although the monthly averages seem to be rather unimodal for some months. It would be instructive if the bimodal fits were shown at least for some typical examples (including a seemingly unimodal PNSD) in the SI.
- Aerosol properties are discussed in detail to estimate the effect of co-condensation over the boreal biome. While this discussion is very detailed, a discussion of the meteorological conditions in the boreal biome is missing: how often are there clouds at all? What updrafts prevail?
Specific comments
The nomenclature is not always consistent or clear throughout the manuscript:
- Sometimes sc (lines 608–612; Fig. 3b) is used and sometimes s* (e.g. line 90). s* is defined as critical supersaturation but no definition is given for sc. Is both critical supersaturation? If this is the case, the nomenclature should be made consistent.
- What is the definition of r* (e.g. line 549)? Usually, it is defined as the critical radius, but for this, the values given in the text seem too low. Rather, it seems to be the smallest radius that becomes activated during cloud droplet activation. In Table 2, r*noCC and r*CC are defined, but a definition of r* is lacking.
- What is the definition of SVOC and IVOC in terms of volatility bins? Does the definition depend on the vapour pressure at the initialization temperature or at the vapour pressure at 300 K? The exact definition is e.g. important for the understanding of Fig. 3a.
Line 43: “suppression of the critical supersaturation”: this sounds a bit too strong. It is rather just a “lowering”.
Lines 88–89: As examples of direct experimental studies of co-condensation, Wang et al. (2020) and Gunthe et al. (2021) could be mentioned here. Wang et al. (2020) described a haze event in Beijing, which was driven by co-condensation of nitrate. Gunthe et al. (2021) describes a haze event in Delhi driven by co-condensation of HCl.
Lines 108–109: formulation is unclear: what is meant by “the asymptote of the curve”?
Lines 182–185: In Eq. 5, the liquid water mixing ratio is calculated from the increase in particle radii due to condensed water mass. From the increase in radius, the additional volume can be calculated and converted to water mixing ratio. It is not clear why the particle dry mass should appear in the denominator. Rather, it should be the air density to convert to mixing ratio.
Lines 283–284: “The PNSD and aerosol chemical composition measured near ground level are assumed to be identical to those at 90 % humidity”: Co-condensation between dry conditions and 90 % RH will change the size distribution. Is this neglected? See also general comment.
Lines 194–203: in the calculation of the Köhler equation, ideality is assumed. This is mentioned but not fully discussed. See general comment.
Line 316: “of the three”: are the three fitting parameters meant here?
Line 345: the sentence structure should be checked.
Lines 346, 352–360, and 401–415: here, the oxygenation of the organic compounds is discussed. The ACSM together with the CIMS information could be used to infer hydrophilicity and estimate the role of non-ideality in gas-particle partitioning. Eq. 18 requires information of oxidation state, thus the basis for such an estimate would be given. See also general comment.
Lines 369–376: How is the temperature adjustment of the volatility bins to the initial temperature done? Are the substances kept in the same bin and just assigned a lower vapour pressure or are the substances reassigned to the new bin with the correct C* at the temperature? This should be explained better. Could you comment here the role of fragmentation or dimerization of organic substances during the measurements?
Line 418 and Fig. 1a: the red crosses seem orange on my screen.
Lines 418–423 and Table 1: what is the total mass of organics (gas and condensed phase; sum of all volatility bins) and ammoniums sulfate? This information should be given e.g. as additional rows in Table 1. Table 1 should be referenced in the text here.
Line 479: “gas-phase concentration” should be specified.
Line 481: CB should be defined more exactly in terms of RH. In Fig. S2a, “below CB” seems to be taken as RH = 99.98 % RH. Was this the assumption throughout the manuscript? If yes, it should be stated in the main text.
Lines 490–501: Here, the assessment of non-ideality by Topping et al. (2013) is discussed, and it is mentioned that the simulated co-condensation should be assessed with caution. Specifically it is stated “it is likely that solubility decreases towards the higher volatility bins.” Nevertheless, in the elaboration of the relevant criteria for co-condensation, this caution was not kept up. The discussion of the role of non-ideality should be taken up again here. See also general comment.
Line 504–509: up to 100 % of organic mass in volatility bin logC* = 7 has been found to condense below CB. This is a lot! At what temperature was the bin initialized (is it bin logC* = 7 at 300 K or at the initialization temperature)? If it is logC* = 7 at the initialization temperature, the condensation seems too large compared with Topping et al. (2013) when below CB is defined as 99.98 % RH. In Topping et al., less than 90 % of the bin logC* = 3 was condensed at 99.9999 % RH.
Lines 514–516: “The condensation efficiency of the highest volatility bin shows a high ICPM initialization temperature dependence. If the model initialization takes place at 270 K, up to 100 % of the organic vapor in the bin condenses, while if the ascent starts at 290 K, only 40 % of the mass concentration is transferred to the condensed phase below CB (Figs. 4g-i)”: These two sentences seem to imply that the substances are assigned to bins at 300 K and this assignment is kept at lower temperatures. This would be opposite to the description of initialization in the method section.
Line 549: r* has not been defined.
Line 558: are “critical radii” r*?
Line 561: susceptibility to what?
Line 571: what does not happen? The steep slope? Formulation should be improved.
Lines 579–581: “This result highlights that significant quantities of co-condensable organic vapors are distributed in the higher volatility bins and these concentrations should not be neglected in further co-condensation studies.” This statement is questionable in view of Topping et al. who found ΔCDNC up to 55 % using volatility bins just up to logC* = 3 (as stated on line 577 of this manuscript). The role of the higher volatility bins should be elaborated further in view of the initialization temperature.
Line 589: the median mode diameters of Aitken and accumulation modes should be mentioned here.
Lines 595–597: “This result underlines that environments rich in particles from a local source would be more susceptible to high ΔCDNC due to co-condensation while regions with aged and cloud processed size distributions are affected less (ΔCDNC < 20% in our simulations; Fig. 5a).” This is a very general statement and might be questioned, considering that organic emissions from local sources are generally also less oxidized, which again would reduce co-condensation if non-ideality is considered.
Line 616: what is meant by "suppress the solute effect"? Bulk depletion by surface partitioning? “Suppression” is not mentioned in the cited study. It should be explained better what is meant and not just referred to Sorjamaa et al.
Line 624: what is meant by "our results" here? Surface tension is not treated in the present study.
Line 645–647: this sentence is unclear. The connection between supersaturation, gas phase concentration of organics, and particle number should be explained better.
Lines 654–657: Figure 3d shows a considerable effect due to co-condensation for updrafts of 1 m/s. Nevertheless, this updraft is not discussed at all in the next section. It should be mentioned how frequent supersaturations above 0.5 m/s are (or more generally the distributions of updrafts leading to cloud cover in Hyytiälä). If they are frequent, a discussion of higher updrafts should be added (including figures, e.g. as part of the SI).
Line 697: do you mean “correlates with a high new particle formation frequency”?
Lines 698–699, Fig. 7c and Sect. 3.5: all measured monthly PNSDs are very similar. None has a clear Hoppel minimum. In contrast, all simulated size distributions show a clear Hoppel minimum. The manuscript states that the co-condensation effect on droplet activation is very sensitive to the size distribution. How useful are then the UKESM1 simulations when they model the size distributions so differently from the measurements?
Line 774: consider the word choice: "lowering" might be more adequate than "suppression".
Figure 1c: what time of day has the temperature been measured?
Fig. 2b and c: is this the partitioning at dry conditions or at model initialization? This information should be added to the figure caption.
Fig. 2d is not explained in the text. Does the displayed partitioning refer to a specific bin?
Line 907: “in shown in”: should this be: “is shown by”?
Line 918: “supersaturation -0.1%, RH=90%”: how does this fit together?
Fig. 4d: the caption to this panel is missing.
Line 939: the figure caption and the legend within the figure do not give the same percentiles.
Figure 6a: the font of the legend is too small.
Supplementary information:
Figure S.2 The updraft should be explicitly stated. I guess it is 0.3 m/s?
Line 26: red lines seem orange on my screen.
Line 35: there is no red shaded size range.
Figure S.3: there is a lot of information in one plot. Consider to make additional plots with D2 against CDNC and D1 against CDNC. Like this, one could see better whether there is a correlation between these parameters.
Line 53: “The markers from every 97 simulations are color-coded with the initial concentration of organic vapor in both simulation sets.”: The legend within the panels states that it is the ratio between the two values. Please clarify which information is correct.
Line 67: “(86% of the time in panel a, 78% of the time in panel b and 90% of the time for panel c).” The numbers within this bracket are confusing, as they suggest that most of the time the surface area exceeds 100 mm2cm-3. Instead, the given percentages give the frequencies for which the calculated surface areas are lower.
Figure S.6: In this figure, frequencies from 10 to 40 % seem abundant. Is this true or just due to the choice of color scale? If it were true, it is not discussed like this in the text.
References
Gunthe, S. S., Liu, P., Panda, U., Raj, S. S., Sharma, A., Darbyshire, E., Reyes-Villegas, E., Allan, J., Chen, Y., Wang, X., Song, S., Pohlker, M. L., Shi, L., Wang, Y., Kommula, S. M., Liu, T., Ravikrishna, R., McFiggans, G., Mickley, L. J., Martin, S. T., Poschl, U., Andreae, M. O., and Coe, H.: Enhanced aerosol particle growth sustained by high continental chlorine emission in India, Nature Geoscience, https://doi.org/10.1038/s41561-020-00677-x, 2021.
Wang, Y., Chen, Y., Wu, Z., Shang, D., Bian, Y., Du, Z., Schmitt, S. H., Su, R., Gkatzelis, G. I., Schlag, P., Hohaus, T., Voliotis, A., Lu, K., Zeng, L., Zhao, C., Alfarra, M. R., McFiggans, G., Wiedensohler, A., Kiendler-Scharr, A., Zhang, Y., and Hu, M.: Mutual promotion between aerosol particle liquid water and particulate nitrate enhancement leads to severe nitrate-dominated particulate matter pollution and low visibility, Atmos. Chem. Phys., 20, 2161-2175, https://doi.org/10.5194/acp-20-2161-2020, 2020.
Citation: https://doi.org/10.5194/egusphere-2023-164-RC2 -
AC1: 'Author response comment on egusphere-2023-164', Liine Heikkinen, 10 Jan 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2023-164/egusphere-2023-164-AC1-supplement.pdf
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-164', Anonymous Referee #1, 29 Mar 2023
The manuscript describes modelling experiments regarding co-condensation of organic vapour and water in a rising air parcel. The main improvement compared to previous work is better quantification of the more volatile vapours using the FIGAERO-I-CIMS observations. The topic of the paper is interesting, as co-condensation could potentially have significant impact on cloud droplet activation but this effect is currently ignored in majority of models. The paper is generally well written and detailed.
Comments:
- In the introduction the authors should briefly explain how co-condensation differs from separately simulating condensation of water (taking into account hygroscopicity of the aerosol) and organic vapours (e.g. original VBS that doesn’t account for water in the particle), as it is presently commonly done in models.
- line 86: “the cooling of the rising air triggers also their condensation” – this gives the impression that the condensation of the higher volatility vapours is purely due to lower temperature, while actually the main driver seems to be the aerosol water content. Please restate.
- The authors should carefully check all their formulas. For example: Formula (5), line 184 : “ρa is the density of the particle dry mass” –shouldn’t this be air density, as nij is probably per cubic meter? Formula (14): The condensation of organic species seems to depend only on water density and molar mass, of which the density seems to cancel out?
- It is unclear from the description of the CPM what assumptions have been made about the hygroscopicity of the condensing organics. What value of hygroscopicity parameter for the SOA would these assumptions lead to? How does it compare with other modelling studies or observations?
- How does UKESM simulate biogenic SOA and its precursors?
- The authors convincingly show that new particle formation is an important factor for creating the suitable particle size distributions for co-condensation to affect the CDNC. However, boundary layer new particle formation is not included in UKESM simulations and the simulated size distributions are shown to be quite different from the ones observed in Hyytiälä. I am not sure if there is any value in showing these results in this already long paper.
- Lines 493-501: Topping et al. (2013) found very significant reduction in condensation of their highest volatility bin. Taking into account that more volatile species are less soluble, the reduction should be even larger for the higher volatility bins of this study. Given that most of the condensed mass seems to come from bins where log10 C* >= 3, how large are the authors expecting the ideality related overestimations to be?
- In sections 3.3 and 3.4 the frequencies of the criteria for significant impact were analyzed one by one. What about the frequency of all the criteria fulfilled at the same time?
- Lines 708-710: “While the SMEAR II PNSD data are retrieved at ground level, utilization of the UKESM1 modal parameters (only soluble modes considered) from CB is chosen, because these PNSD log-normal parameters would actually meet the cloud droplet activation scheme in the model.” - I find this selection a bit odd. Wouldn’t it make more sense to look at the near-surface PNSD that would have a chance to grow with co-condensation in updrafts before cloud activation? How large is the difference between near-surface and BC PNSDs in UKESM?
Citation: https://doi.org/10.5194/egusphere-2023-164-RC1 -
RC2: 'Comment on egusphere-2023-164', Anonymous Referee #2, 19 Apr 2023
General comment
This study investigates how cloud droplet number concentrations (CDNC) are altered through co-condensation of organic vapours along with water vapour during cloud events in boreal biomes. It uses an adiabatic cloud parcel model (ICPM), which is initialized at 90 % RH to simulate cloud formation in upraising air at updrafts of 0.1, 0.3, and 1 m/s. The simulation of the biogenic organic aerosol is based on volatility basis sets derived from measurement data of gas phase and condensed phase composition performed in the boreal forest of Finland during Spring 2014. A median enhancement in CDNC of 20 % was predicted, and the most suitable conditions for a strong CDNC enhancement through co-condensation were found to occur in the presence of a nascent ultrafine aerosol mode, conditions occurring quite frequently in fall and spring. To come to this conclusion, the authors performed 97 ICPM simulations representing the atmospheric conditions: temperature profile, aerosol composition, and particle number size distribution (PNSD) fitted as an Aitken and an accumulation mode. Each simulation was initialized at the height and temperature where 90 % RH is reached and then run at the three pre-defined updrafts. The PNSD and aerosol chemical composition (gas and condensed phase) measured near ground level were taken to be identical to those at 90 % RH. Like this, PNSD, volatility distributions and temperature were different for every simulation. This variation in conditions makes the explanation of the model initialization and the interpretation of the results quite complex, and at times confusing. As the topic of this study is highly relevant and this study is the first one to simulate co-condensation in such detail, the method description especially the adaptation of the VBS to the starting temperature and the ICPM initialization need to be improved. Altogether, this is a very thorough study that deserves publication in ACP.
ICPM initialization
Information about the ICPM initialization is spread over the whole manuscript and can sometimes be inferred only indirectly. This hampers the reading and understanding of the manuscript.
- On line 423, it is stated that an internal mixture of organics and ammonium sulfate (AS) is simulated, but the amount of AS in each simulation is not mentioned nor is the role of AS in co-condensation discussed.
- For the CJ distributions, it is stated on line 333 that “After the temperature adjustments, the volatility distributions are binned to ranges between logC* = [-8, 3] spaced by one decade in C*”. Comparing the VBS displayed in CJ with the ones labelled as CJ in Fig. 2a, similarities are difficult to detect. Figure 2 in CJ displays in 11 panels different VBS. All of them show constant low organic mass in the bins below logC* = -1 and increasing or even steeply increasing organic mass in the bins from logC* = 0 to 3. Topping et al. (2013) also displays a VBS with strongly increasing organic mass in logC* > 0 in their Fig. 2. This is opposite to the VBS distribution labelled as CJ in Fig. 2a. Why is this? To correctly adjust the VBS from CJ to the lower ICPM initialization temperature, information of the volatility bins with logC* between 4 and 6 at 298°C would be required, but such data is not given in CJ. So, how could the temperature adjustment be performed at all?
- Hunter et al. (2017) display in their Fig. 1c VBS derived from different measurement techniques including a gas-phase and a condensed-phase CIMS. The VBS derived from CIMS (g) data is very different from the one derived from PTR-MS for the same volatility range. As the choice of VBS is crucial for co-condensation, the derivation of this VBS needs to be explained better.
- For the F distribution, an adjustment to the initialization temperature is mentioned on line 370. How exactly was this adjustment carried out? How were the ΔHvap for the adjustment determined?
- Was a full equilibration carried out at the initialization temperature and RH? If yes, kinetic effects would only become important for the increase of RH from 90 % to 100 % and equilibration could be overestimated. The opposite would be the case if no equilibration were carried out at model equilibration. Can you comment on this?
- On line 162, it is stated that PNSD were constructed at model initialization. Then, on line 236, it is stated that ”The PNSD for the ICPM initialization are obtained from the Differential Mobility Particle Sizer (DMPS) measurements from SMEAR II performed within the forest canopy.” Do these measurements reflect the actual temperature at the forest canopy and dry conditions? Were they adjusted to reflect 90 % RH?
Discussion of the results
- One general weakness in the discussion of the results is that the role of temperature and RH for the condensation of organic species is not kept properly apart. The effect of the lower temperature is e.g. shown in Fig. 4, panels g-i, but in the discussion of co-condensation depending on the season, the influence of temperature on ΔCDNC is not mentioned. If a full temperature adjustment had been carried out at model initialization, a temperature dependence as shown in Fig. 4 should not be present. The question therefore remains to what degree the higher co-condensation in spring and autumn compared with summer is caused by the lower temperatures during these seasons. In a revised manuscript version, the respective role of temperature and co-condensation should be discussed more explicitly.
- On lines 493–501, the potential influence of non-ideality on gas-particle partitioning of organics is discussed. Despite recognising the shortcomings due to the assumption of ideality, quite detailed predictions of the co-condensation effect in boreal regions are made without taking the issue of solution non-ideality up again. A discussion of this aspect should be part of a revised discussion. The O:C could be estimated from CIMS and AMS data given in Fig. 2f. Such an estimate is strongly recommended to improve the manuscript.
- Uncertainties in the PNSD should also be critically assessed as these were highlighted to be crucial to determine ΔCDNC. Bimodal fits (Aitken and accumulation modes) have been performed to describe the measured PNSD from the SMEAR II campaign although the monthly averages seem to be rather unimodal for some months. It would be instructive if the bimodal fits were shown at least for some typical examples (including a seemingly unimodal PNSD) in the SI.
- Aerosol properties are discussed in detail to estimate the effect of co-condensation over the boreal biome. While this discussion is very detailed, a discussion of the meteorological conditions in the boreal biome is missing: how often are there clouds at all? What updrafts prevail?
Specific comments
The nomenclature is not always consistent or clear throughout the manuscript:
- Sometimes sc (lines 608–612; Fig. 3b) is used and sometimes s* (e.g. line 90). s* is defined as critical supersaturation but no definition is given for sc. Is both critical supersaturation? If this is the case, the nomenclature should be made consistent.
- What is the definition of r* (e.g. line 549)? Usually, it is defined as the critical radius, but for this, the values given in the text seem too low. Rather, it seems to be the smallest radius that becomes activated during cloud droplet activation. In Table 2, r*noCC and r*CC are defined, but a definition of r* is lacking.
- What is the definition of SVOC and IVOC in terms of volatility bins? Does the definition depend on the vapour pressure at the initialization temperature or at the vapour pressure at 300 K? The exact definition is e.g. important for the understanding of Fig. 3a.
Line 43: “suppression of the critical supersaturation”: this sounds a bit too strong. It is rather just a “lowering”.
Lines 88–89: As examples of direct experimental studies of co-condensation, Wang et al. (2020) and Gunthe et al. (2021) could be mentioned here. Wang et al. (2020) described a haze event in Beijing, which was driven by co-condensation of nitrate. Gunthe et al. (2021) describes a haze event in Delhi driven by co-condensation of HCl.
Lines 108–109: formulation is unclear: what is meant by “the asymptote of the curve”?
Lines 182–185: In Eq. 5, the liquid water mixing ratio is calculated from the increase in particle radii due to condensed water mass. From the increase in radius, the additional volume can be calculated and converted to water mixing ratio. It is not clear why the particle dry mass should appear in the denominator. Rather, it should be the air density to convert to mixing ratio.
Lines 283–284: “The PNSD and aerosol chemical composition measured near ground level are assumed to be identical to those at 90 % humidity”: Co-condensation between dry conditions and 90 % RH will change the size distribution. Is this neglected? See also general comment.
Lines 194–203: in the calculation of the Köhler equation, ideality is assumed. This is mentioned but not fully discussed. See general comment.
Line 316: “of the three”: are the three fitting parameters meant here?
Line 345: the sentence structure should be checked.
Lines 346, 352–360, and 401–415: here, the oxygenation of the organic compounds is discussed. The ACSM together with the CIMS information could be used to infer hydrophilicity and estimate the role of non-ideality in gas-particle partitioning. Eq. 18 requires information of oxidation state, thus the basis for such an estimate would be given. See also general comment.
Lines 369–376: How is the temperature adjustment of the volatility bins to the initial temperature done? Are the substances kept in the same bin and just assigned a lower vapour pressure or are the substances reassigned to the new bin with the correct C* at the temperature? This should be explained better. Could you comment here the role of fragmentation or dimerization of organic substances during the measurements?
Line 418 and Fig. 1a: the red crosses seem orange on my screen.
Lines 418–423 and Table 1: what is the total mass of organics (gas and condensed phase; sum of all volatility bins) and ammoniums sulfate? This information should be given e.g. as additional rows in Table 1. Table 1 should be referenced in the text here.
Line 479: “gas-phase concentration” should be specified.
Line 481: CB should be defined more exactly in terms of RH. In Fig. S2a, “below CB” seems to be taken as RH = 99.98 % RH. Was this the assumption throughout the manuscript? If yes, it should be stated in the main text.
Lines 490–501: Here, the assessment of non-ideality by Topping et al. (2013) is discussed, and it is mentioned that the simulated co-condensation should be assessed with caution. Specifically it is stated “it is likely that solubility decreases towards the higher volatility bins.” Nevertheless, in the elaboration of the relevant criteria for co-condensation, this caution was not kept up. The discussion of the role of non-ideality should be taken up again here. See also general comment.
Line 504–509: up to 100 % of organic mass in volatility bin logC* = 7 has been found to condense below CB. This is a lot! At what temperature was the bin initialized (is it bin logC* = 7 at 300 K or at the initialization temperature)? If it is logC* = 7 at the initialization temperature, the condensation seems too large compared with Topping et al. (2013) when below CB is defined as 99.98 % RH. In Topping et al., less than 90 % of the bin logC* = 3 was condensed at 99.9999 % RH.
Lines 514–516: “The condensation efficiency of the highest volatility bin shows a high ICPM initialization temperature dependence. If the model initialization takes place at 270 K, up to 100 % of the organic vapor in the bin condenses, while if the ascent starts at 290 K, only 40 % of the mass concentration is transferred to the condensed phase below CB (Figs. 4g-i)”: These two sentences seem to imply that the substances are assigned to bins at 300 K and this assignment is kept at lower temperatures. This would be opposite to the description of initialization in the method section.
Line 549: r* has not been defined.
Line 558: are “critical radii” r*?
Line 561: susceptibility to what?
Line 571: what does not happen? The steep slope? Formulation should be improved.
Lines 579–581: “This result highlights that significant quantities of co-condensable organic vapors are distributed in the higher volatility bins and these concentrations should not be neglected in further co-condensation studies.” This statement is questionable in view of Topping et al. who found ΔCDNC up to 55 % using volatility bins just up to logC* = 3 (as stated on line 577 of this manuscript). The role of the higher volatility bins should be elaborated further in view of the initialization temperature.
Line 589: the median mode diameters of Aitken and accumulation modes should be mentioned here.
Lines 595–597: “This result underlines that environments rich in particles from a local source would be more susceptible to high ΔCDNC due to co-condensation while regions with aged and cloud processed size distributions are affected less (ΔCDNC < 20% in our simulations; Fig. 5a).” This is a very general statement and might be questioned, considering that organic emissions from local sources are generally also less oxidized, which again would reduce co-condensation if non-ideality is considered.
Line 616: what is meant by "suppress the solute effect"? Bulk depletion by surface partitioning? “Suppression” is not mentioned in the cited study. It should be explained better what is meant and not just referred to Sorjamaa et al.
Line 624: what is meant by "our results" here? Surface tension is not treated in the present study.
Line 645–647: this sentence is unclear. The connection between supersaturation, gas phase concentration of organics, and particle number should be explained better.
Lines 654–657: Figure 3d shows a considerable effect due to co-condensation for updrafts of 1 m/s. Nevertheless, this updraft is not discussed at all in the next section. It should be mentioned how frequent supersaturations above 0.5 m/s are (or more generally the distributions of updrafts leading to cloud cover in Hyytiälä). If they are frequent, a discussion of higher updrafts should be added (including figures, e.g. as part of the SI).
Line 697: do you mean “correlates with a high new particle formation frequency”?
Lines 698–699, Fig. 7c and Sect. 3.5: all measured monthly PNSDs are very similar. None has a clear Hoppel minimum. In contrast, all simulated size distributions show a clear Hoppel minimum. The manuscript states that the co-condensation effect on droplet activation is very sensitive to the size distribution. How useful are then the UKESM1 simulations when they model the size distributions so differently from the measurements?
Line 774: consider the word choice: "lowering" might be more adequate than "suppression".
Figure 1c: what time of day has the temperature been measured?
Fig. 2b and c: is this the partitioning at dry conditions or at model initialization? This information should be added to the figure caption.
Fig. 2d is not explained in the text. Does the displayed partitioning refer to a specific bin?
Line 907: “in shown in”: should this be: “is shown by”?
Line 918: “supersaturation -0.1%, RH=90%”: how does this fit together?
Fig. 4d: the caption to this panel is missing.
Line 939: the figure caption and the legend within the figure do not give the same percentiles.
Figure 6a: the font of the legend is too small.
Supplementary information:
Figure S.2 The updraft should be explicitly stated. I guess it is 0.3 m/s?
Line 26: red lines seem orange on my screen.
Line 35: there is no red shaded size range.
Figure S.3: there is a lot of information in one plot. Consider to make additional plots with D2 against CDNC and D1 against CDNC. Like this, one could see better whether there is a correlation between these parameters.
Line 53: “The markers from every 97 simulations are color-coded with the initial concentration of organic vapor in both simulation sets.”: The legend within the panels states that it is the ratio between the two values. Please clarify which information is correct.
Line 67: “(86% of the time in panel a, 78% of the time in panel b and 90% of the time for panel c).” The numbers within this bracket are confusing, as they suggest that most of the time the surface area exceeds 100 mm2cm-3. Instead, the given percentages give the frequencies for which the calculated surface areas are lower.
Figure S.6: In this figure, frequencies from 10 to 40 % seem abundant. Is this true or just due to the choice of color scale? If it were true, it is not discussed like this in the text.
References
Gunthe, S. S., Liu, P., Panda, U., Raj, S. S., Sharma, A., Darbyshire, E., Reyes-Villegas, E., Allan, J., Chen, Y., Wang, X., Song, S., Pohlker, M. L., Shi, L., Wang, Y., Kommula, S. M., Liu, T., Ravikrishna, R., McFiggans, G., Mickley, L. J., Martin, S. T., Poschl, U., Andreae, M. O., and Coe, H.: Enhanced aerosol particle growth sustained by high continental chlorine emission in India, Nature Geoscience, https://doi.org/10.1038/s41561-020-00677-x, 2021.
Wang, Y., Chen, Y., Wu, Z., Shang, D., Bian, Y., Du, Z., Schmitt, S. H., Su, R., Gkatzelis, G. I., Schlag, P., Hohaus, T., Voliotis, A., Lu, K., Zeng, L., Zhao, C., Alfarra, M. R., McFiggans, G., Wiedensohler, A., Kiendler-Scharr, A., Zhang, Y., and Hu, M.: Mutual promotion between aerosol particle liquid water and particulate nitrate enhancement leads to severe nitrate-dominated particulate matter pollution and low visibility, Atmos. Chem. Phys., 20, 2161-2175, https://doi.org/10.5194/acp-20-2161-2020, 2020.
Citation: https://doi.org/10.5194/egusphere-2023-164-RC2 -
AC1: 'Author response comment on egusphere-2023-164', Liine Heikkinen, 10 Jan 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2023-164/egusphere-2023-164-AC1-supplement.pdf
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 577 | 261 | 25 | 863 | 76 | 23 | 24 |
- HTML: 577
- PDF: 261
- XML: 25
- Total: 863
- Supplement: 76
- BibTeX: 23
- EndNote: 24
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Liine Heikkinen
Daniel G. Partridge
Wei Huang
Sara Blichner
Rahul Ranjan
Emanuele Tovazzi
Tuukka Petäjä
Claudia Mohr
Ilona Riipinen
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3468 KB) - Metadata XML
-
Supplement
(2381 KB) - BibTeX
- EndNote
- Final revised paper