Using a region-specific ice-nucleating particle parameterization improves the representation of Arctic clouds in a global climate model

Gjelsvik, Astrid Bragstad; David, Robert Oscar; Carlsen, Tim; Hellmuth, Franziska; Hofer, Stefan; McGraw, Zachary; Sodemann, Harald; Storelvmo, Trude

doi:https://doi.org/10.5194/egusphere-2024-1879

Preprints

https://doi.org/10.5194/egusphere-2024-1879

Preprints

09 Jul 2024

| 09 Jul 2024

Using a region-specific ice-nucleating particle parameterization improves the representation of Arctic clouds in a global climate model

Astrid Bragstad Gjelsvik, Robert Oscar David, Tim Carlsen, Franziska Hellmuth, Stefan Hofer, Zachary McGraw, Harald Sodemann, and Trude Storelvmo

Abstract. Projections of global climate change and Arctic amplification are sensitive to the representation of low-level cloud phase in climate models. Ice-nucleating particles (INPs) are necessary for primary cloud ice formation at temperatures above approximately -38 °C, and thus significantly affect cloud phase and cloud radiative effect. Due to their complex and insufficiently understood variability, INPs constitute an important modelling challenge, especially in remote regions with few observations, such as the Arctic. In this study, INP observations were carried out at Andenes, Norway in March 2021. These observations were used as a basis for an Arctic-specific and purely temperature-dependent INP parameterization, and implemented into the Norwegian Earth System Model. This implementation results in an annual average increase in cloud liquid water path (CLWP) of 70 % for the Arctic, and improves the representation of cloud phase compared to satellite observations. The change in CLWP in boreal autumn and winter is found to likely be the dominant contributor to the annual average increase in net surface cloud radiative effect of 2 W m^-2. This large surface flux increase brings the simulation into better agreement with Arctic ground-based measurements. Despite that the model cannot respond fully to the INP parameterization change due to fixed sea surface temperatures, Arctic surface air temperature increases with 0.7 °C in boreal autumn. These findings indicate that INPs could have a significant impact on Arctic climate, and that a region-specific INP parameterization can be a useful tool to improve cloud representation in the Arctic region.

Received: 23 Jun 2024 – Discussion started: 09 Jul 2024

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 preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Astrid Bragstad Gjelsvik, Robert Oscar David, Tim Carlsen, Franziska Hellmuth, Stefan Hofer, Zachary McGraw, Harald Sodemann, and Trude Storelvmo

Status: closed

CC1:
'Comment on egusphere-2024-1879', Lin Lin, 08 Aug 2024

Hi Gjelsvik et al.,
Thank you for the interesting work! I enjoy reading your research! I have a question about replacing the Meyers scheme with your observed INP as a function of temperature. If i understand your methodology correctly, your observed INP number as a function of temperature is in the immersion freezing mode. However, Meyers scheme describes the deposition freezing and condensation ice nucleation. Could you please explain a bit more why the measured INP concentrations that are relevant for the immersion mode can be used to replace the M92 scheme?

My 2nd question is if it is appropriate to quantify INP concentration as a function of temperature in the immersion freezing mode? For visualization purpose, INP number concentration can be plotted as a function of temperature, like figure 2 in your paper, figure 7 in Geerts et al., (2022) (https://journals.ametsoc.org/view/journals/bams/103/5/BAMS-D-21-0044.1.xml). But can we quantify INP number concentrations as a function of only temperature in the immersion freezing mode, without considering other factors such as water activity or active surface area density?

my 3rd question is how different is your measured INP number concentration from that measured during the COMBLE campaign? One year apart in the same season yield similar results?

Thank you very much!

Citation: https://doi.org/10.5194/egusphere-2024-1879-CC1
- AC1: 'Reply on CC1', Astrid Bragstad Gjelsvik, 06 Sep 2024
  
  Hi Lin Lin,
  First of all, we appreciate very much that you have taken the time to read and ask about our study! We will try and address your questions listed below as best as possible here, but definitely also make sure to clarify these aspects when revising the article - so thank you very much for helping us improve it.
  Question 1: If I understand your methodology correctly, your observed INP number as a function of temperature is in the immersion freezing mode. However, Meyers scheme (M92) describes the deposition freezing and condensation ice nucleation. Could you please explain a bit more why the measured INP concentrations that are relevant for the immersion mode can be used to replace the M92 scheme?
  Regarding your first question, this is a slightly unfortunate complication due to CAM5’s parameterization choices, which is an outdated treatment of ice nucleation. It is true that the Meyers’ scheme (M92) describes deposition nucleation and condensation-freezing in mixed-phase clouds, and it is also the dominant ice nucleation pathway in CAM5. We replace this parameterization with a new function (A21) based on measurements of INPs in immersion freezing mode, which is now being accepted as the dominant nucleation pathway in mixed-phase clouds (see e.g. Ansmann et al., 2009, de Boer et al., 2011, Westbrook and Illingworth, 2011).
  
  We clarify a few additional details of our ice nucleation modifications in CAM5. It should be stressed that we do not make any changes to the deposition nucleation scheme in cirrus clouds, but only change the M92 scheme active for temperatures between -37°C and 0°C . Additionally, in the updated version of this manuscript we exclude the immersion freezing parameterization of Bigg wherever we replace M92, so that the A21 parameterization is the only active immersion freezing parameterisation in the Arctic. This makes very little change to the cloud ice number concentrations or the overall picture we see, as the Bigg parameterization contributes little to activated ice nuclei, something which is also stated by previous studies by e.g. English et al. (2014). To address the unclarities you have pointed out, we plan to update lines 148-156 accordingly:
  “In CAM5, the different heterogeneous ice nucleation pathways in mixed-phase clouds are parameterized independently, namely, contact freezing (Young, 1974), immersion freezing (Bigg, 1953, hereafter: "B53") and deposition and condensation freezing (Meyers et al., 1992, hereafter: "M92"). Here, we update the M92 parameterization. This parameterization is active in the temperature range -37°C to 0°C, and is responsible for more than 90 % of ice nuclei formed in CAM5 mixed-phase clouds (English et al., 2014). Since the measured INP concentrations are relevant for the immersion mode, replacing the M92 with our measurements entails excluding deposition and condensation freezing in Arctic mixed-phase clouds. This exclusion is justified by observational studies that found deposition and condensation freezing to be negligible for mixed-phase clouds (Ansmann et al., 2009; Boer et al., 2011; Westbrook and Illingworth, 2011). Where we update the M92 parameterization using our INP measurements in immersion freezing mode, we exclude the B53 immersion freezing parameterization.”
  Question 2: Is it appropriate to quantify INP concentration as a function of temperature in the immersion freezing mode? For visualization purpose, INP number concentration can be plotted as a function of temperature, like figure 2 in your paper, figure 7 in Geerts et al., (2022) (https://journals.ametsoc.org/view/journals/bams/103/5/BAMS-D-21-0044.1.xml). But can we quantify INP number concentrations as a function of only temperature in the immersion freezing mode, without considering other factors such as water activity or active surface area density?
  Regarding your second question, it is true that we make a simplification when we predict immersion freezing INP concentration based only on temperature. The reason we do not include aerosol surface area density as a predictor in the parameterisation is because we find quite low correlation with aerosol surface area density (see Fig. A1e in the appendix). This finding is consistent with INP measurements in other Arctic sites, e.g. in Ny-Ålesund (Li et al., 2022). Therefore we believe that including aerosol surface area density will not improve the predictive ability of our INP parametrization. The aerosol species originally used to predict INPs in CAM6-Nor (the atmospheric model we use) are dust and soot, but other sources such as marine organic aerosols or other bioaerosols are likely important Arctic INPs (see for example Creamean et al. (2022), Carlsen and David (2022), Sze et al. (2023) or Freitas et al. (2023)). Differences in water activity can indeed result from different aerosol chemical composition and in theory affect ice nucleation. However, this is difficult to account for with our measurement technique, as the aerosols are all impinged in water before their ice-nucleating ability is probed, which may dissolve and dilute substances that would otherwise affect the freezing ability of an aerosol. However, one of our motivations for using a simple temperature-based parameterisation is that we do not yet have a full understanding of how INP concentrations can be predicted in the Arctic. Therefore by keeping it simple, we avoid introducing any additional sources of error when predicting Arctic INPs, until a full understanding on the composition-dependent ice-nucleating ability of Arctic INPs is in place.
  Question 3: How different is your measured INP number concentration from that measured during the COMBLE campaign? One year apart in the same season yield similar results?
  Regarding your third question, if we compare Fig. 7 in Geerts et al. (2022, https://doi.org/10.1175/BAMS-D-21-0044.1) and Fig. 2 in our study we see that our 2021 measurements in Andenes one year after the COMBLE campaign show somewhat higher maximum concentrations (around one order of magnitude at some temperatures) than the COMBLE campaign, and similar minimum concentrations. It should be noted that there are certain important differences between the data. First of all, the COMBLE data exclusively represents cold air outbreaks, i.e. flow reaching Andenes from the Arctic, while our measurements also include air masses reaching our measurement site with a more southerly and southwesterly flow (see Fig. A2 of the air mass back trajectories in the appendix). As these air masses are going into the Arctic, we believe that they are also important to capture in our case to understand Arctic INP concentrations. Secondly, our measurements are only from March 2021, while the COMBLE measurements are from the period December 2019 to May 2020. Nevertheless, we find that our measurements are quite similar to other measurements around the Arctic (see Fig. 2). Additionally, our measurement technique does not allow us to estimate INP concentrations higher than approximately 0.1/L, or INP concentrations at temperatures lower than around -25°C. This should be kept in mind when comparing the two datasets. To make this comparison easier for the reader, we are in the process of incorporating the COMBLE data points into Fig. 2 in the manuscript.
  Thank you again for your questions and please let us know if you have any others.
  Best regards,
  
  Astrid Bragstad Gjelsvik on behalf of the author team
  
  Papers on liquid as a prerequisite for ice in mixed-phase clouds:
  Ansmann, A., M. Tesche, P. Seifert, D. Althausen, R. Engelmann, J. Fruntke, U. Wandinger, I. Mattis, and D. Müller (2009), Evolution of the ice phase in tropical altocumulus: SAMUM lidar observations over Cape Verde, J. Geophys. Res., 114, D17208, doi:10.1029/2008JD011659.
  de Boer, G., H. Morrison, M. D. Shupe, and R. Hildner (2011), Evidence of liquid dependent ice nucleation in high-latitude stratiform clouds from surface remote sensors, Geophys. Res. Lett., 38, L01803, doi:10.1029/2010GL046016.
  Westbrook, C. D., and A. J. Illingworth (2011), Evidence that ice forms primarily in supercooled liquid clouds at temperatures > −27°C, Geophys. Res. Lett., 38, L14808, doi:10.1029/2011GL048021.
  Paper from English et al. on ice nucleation parameterizations in CAM5:
  English, J. M., Kay, J. E., Gettelman, A., Liu, X., Wang, Y., Zhang, Y., & Chepfer, H. (2014). Contributions of Clouds, Surface Albedos, and Mixed-Phase Ice Nucleation Schemes to Arctic Radiation Biases in CAM5. Journal of Climate, 27(13), 5174-5197. https://doi.org/10.1175/JCLI-D-13-00608.1
  Paper on INP measurements in Ny-Ålesund:
  Li, G., Wieder, J., Pasquier, J. T., Henneberger, J., and Kanji, Z. A.: Predicting atmospheric background number concentration of ice-nucleating particles in the Arctic, Atmos. Chem. Phys., 22, 14441–14454, https://doi.org/10.5194/acp-22-14441-2022, 2022.
  Papers on INPs, seasonality and connections to bioaerosols:
  Creamean, J.M., Barry, K., Hill, T.C.J. et al. Annual cycle observations of aerosols capable of ice formation in central Arctic clouds. Nat Commun 13, 3537 (2022). https://doi.org/10.1038/s41467-022-31182-x
  Carlsen, T., & David, R. O. (2022). Spaceborne evidence that ice-nucleating particles influence high-latitude cloud phase. Geophysical Research Letters, 49, e2022GL098041. https://doi.org/10.1029/2022GL098041
  Sze, K. C. H., Wex, H., Hartmann, M., Skov, H., Massling, A., Villanueva, D., and Stratmann, F.: Ice-nucleating particles in northern Greenland: annual cycles, biological contribution and parameterizations, Atmos. Chem. Phys., 23, 4741–4761, https://doi.org/10.5194/acp-23-4741-2023, 2023.
  Pereira Freitas, G., Adachi, K., Conen, F., Heslin-Rees, D., Krejci, R., Tobo, Y., Yttri, K. E., & Zieger, P. (2023). Regionally sourced bioaerosols drive high-temperature ice nucleating particles in the Arctic. Nature communications, 14(1), 5997. https://doi.org/10.1038/s41467-023-41696-7
  
  Citation: https://doi.org/10.5194/egusphere-2024-1879-AC1
RC1:
'Comment on egusphere-2024-1879', Anonymous Referee #1, 11 Aug 2024

Publisher’s note: this comment is a copy of RC3 and its content was therefore removed.

Citation: https://doi.org/10.5194/egusphere-2024-1879-RC1
- RC2:
  'Reply on RC1', Anonymous Referee #1, 13 Aug 2024
  
  Publisher’s note: this comment is a copy of RC3 and its content was therefore removed.
  
  Citation: https://doi.org/10.5194/egusphere-2024-1879-RC2
  - RC3: 'Reply on RC2', Anonymous Referee #1, 15 Aug 2024
    
    The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1879/egusphere-2024-1879-RC3-supplement.pdf
    
    Citation: https://doi.org/10.5194/egusphere-2024-1879-RC3
RC4: 'Comment on egusphere-2024-1879', Anonymous Referee #2, 16 Sep 2024

Review of “Using a region-specific ice-nucleating particle parameterization improves the representation of Arctic clouds in a global climate model” by Gjelsvik and co-authors.
In this study the authors use measurements of ambient ice nucleating particle (INP) concentrations from within the Arctic to assess the sensitivity of the NorESM Earth System Model to the representation of immersion mode heterogeneous ice nucleation. Ambient aerosols greater than 0.5 um in diameter were collected over a period of 15 days in March 2021 and processed in an instrument that measures the number of INP activated as a function of temperature and the volume of sampled air. Using these data an INP parameterization was derived for the representation of immersion-mode ice nucleation in supercooled cloud liquid droplets. The authors incorporated the parameterization into a cloud microphysics module in NorESM to replace primary ice production in the Arctic from the default mechanism(s). A comparison between simulations using the default (named M92) and the new setup (named A21) was used to assess whether the new setup improves the model’s ability to reproduce observed quantities of the Arctic atmosphere. The new A21 setup improves the resulting bulk cloud phase partitioning between liquid and water, impacts the amount of cloud ice and cloud liquid water loadings, and reduces the bias apparent in radiative fluxes at the surface. The study is not entirely novel but does help extend our knowledge on the role of INPs in the climate, and specifically the Arctic region which is less understand than others. It also demonstrates, as have other studies, the sensitivity of simulated high-latitude clouds to the representation of ice production. However, I believe there are several major issues within the methodology and analysis that need addressing before the manuscript is in a position to be accepted for publication.
Major comments.
The main result of the study is that reducing the simulated INP concentrations results in more liquid water and less ice, which better matches remotely observed cloud-phase partitioning. One caveat to this result is that it could be compensating for other poorly represented processes in the cloud microphysics scheme. Are the authors confident that this is not the case? For instance, Gettelman et al. (2023) have demonstrated that Arctic LWP and IWP in CESM2 is sensitive to several different cloud microphysical processes. Do other models using the M92 parameterization (or similarly high primary ice production rates) suffer from the same issue? It may not be possible to answer these questions, but this caveat should thoroughly addressed in the manuscript.
“Improved representation of Arctic clouds”. The authors have only selected a few observable quantities to assess the representation of Arctic clouds. There are others that could be considered to fully justify this statement – including LWP and IWP, cloud fraction, and precipitation. Widespread LWP increases of 30 gm-2 apparent in Figure 5 will surely be seen in the observations / model biases. Similarly, the cloud fraction changes in Figure B1 may be substantial enough to be seen in observations / model biases. Also, why was the TOA shortwave CRE not compared to observations (as for longwave in Figure 11)? I couldn’t find anywhere in the text that explained their omission.
Are the INP measurements adequately representative of the whole Arctic region across the annual cycle? The authors compare their INP parameterization with others from the Arctic region and show consistency, yet a new study by Raif et al. (2024) has shown that INP concentrations may not always be so low. Additionally, there is good evidence that dust from low latitudes is present in the Arctic atmosphere throughout the year, which may strongly influence the spatial variability of INP availability (Shi et al., 2022). In terms of locally produced INP, there will be important sources that are dependent on the seasonal cycle of sea ice (sea spray aerosol), snow coverage (glacial dust sources), and temperature/sunlight (biological sources). I recommend the authors include these caveats to their discussion/conclusions on the representativeness of their parameterization. Perhaps expand lines 390-395 or 210-212.
The INP measurements were made using air sampled via an instrument that does not include aerosols smaller than around 500nm in diameter. Does omitting sub 500nm-sized particles bias the INP concentration measurements? Do the authors have a sense of how many more INP would be measured if the whole size distribution was accounted for?
The purely temperature dependent (L6) parameterization presented here is put forward as a new and simplified representation of Arctic INP, yet without any analysis on the origins of the underlying aerosol its use in future studies is limited. The better representative parameterizations are linked to a quantity or property of the aerosol size distribution (e.g., particles larger than 0.5 um or specific components like dust surface area) that are capable of being dependent on variable properties of the atmosphere / aerosol population. With these we are then able to assess changes in INP distributions for future scenarios etc. Do the authors have information on the composition of the aerosol that is acting as an INP? Is this consistent with other studies? Could the authors convert the INP spectra into a quantity related to the aerosol size distribution such as the active site density ns? This would allow a much more thorough comparison with other INP measurements.
Minor comments
L46. Dusts from lower latitudes are also readily transported to the Arctic. See Shi et al. (2022) and references therein.
L66. “To our knowledge…” I think this is likely correct, but there are a few studies that have used Arctic-based INP parameterizations in global models, which should be acknowledged for having a similar methodology. Kawai et al. (2023) incorporated an observations-based ice nucleation parameterization of Arctic dust into CAM5 and Shi et al. (2022) modelled INP concentrations in the E3SMv1 model using a parameterization based on Icelandic dust samples.
L144 and L229. Is total aerosol surface area for particles greater than 0.5 um diameter?
L146. Is this a sensible number?
L146. What impact will turning off ice detrainment have and why was it turned off?
L151. M92 is for primary ice production but the empirical parameterization is for INP. How do you relate the INP to the primary ice production? Also, do you still have Bigg immersion mode freezing and Young contact freezing? Sometimes the Bigg parameterization is used for rain drops rather than cloud droplets – if this is the case perhaps state this to avoid confusion.
L210. I suggest the authors include low latitude dust sources too.
Figure 2. As this is the first time the INP data is being shown please include uncertainty estimates to the scatter points.
L261. Do the changes in the cloud fraction profile (less/more cloud at different altitudes) have implications for the change in longwave radiation at the surface?
L274. What happens to total water path (TWP) and precipitation? Or is it all WBF? I would be surprised if the differences were entirely attributed to this. It would be useful to have all hydrometeor classes accounted for so we can understand where the liquid has come from / gone. Also, is there a reason TWP is never discussed?
L345 and others. I’m not sure the use of the word drastic (radical; extreme) is an appropriate choice. Please consider changing all instances throughout the manuscript.
References.
Gettelman et al., 2023. https://doi.org/10.5194/gmd-16-1735-2023
Kawai et al., 2023. https://doi.org/10.1029/2022GL102470
Raif et al., 2024. https://doi.org/10.5194/egusphere-2024-1502
Shi et al., 2022. https://doi.org/10.5194/acp-22-2909-2022

Citation: https://doi.org/10.5194/egusphere-2024-1879-RC4

Status: closed

CC1:
'Comment on egusphere-2024-1879', Lin Lin, 08 Aug 2024

Hi Gjelsvik et al.,
Thank you for the interesting work! I enjoy reading your research! I have a question about replacing the Meyers scheme with your observed INP as a function of temperature. If i understand your methodology correctly, your observed INP number as a function of temperature is in the immersion freezing mode. However, Meyers scheme describes the deposition freezing and condensation ice nucleation. Could you please explain a bit more why the measured INP concentrations that are relevant for the immersion mode can be used to replace the M92 scheme?

My 2nd question is if it is appropriate to quantify INP concentration as a function of temperature in the immersion freezing mode? For visualization purpose, INP number concentration can be plotted as a function of temperature, like figure 2 in your paper, figure 7 in Geerts et al., (2022) (https://journals.ametsoc.org/view/journals/bams/103/5/BAMS-D-21-0044.1.xml). But can we quantify INP number concentrations as a function of only temperature in the immersion freezing mode, without considering other factors such as water activity or active surface area density?

my 3rd question is how different is your measured INP number concentration from that measured during the COMBLE campaign? One year apart in the same season yield similar results?

Thank you very much!

Citation: https://doi.org/10.5194/egusphere-2024-1879-CC1
- AC1: 'Reply on CC1', Astrid Bragstad Gjelsvik, 06 Sep 2024
  
  Hi Lin Lin,
  First of all, we appreciate very much that you have taken the time to read and ask about our study! We will try and address your questions listed below as best as possible here, but definitely also make sure to clarify these aspects when revising the article - so thank you very much for helping us improve it.
  Question 1: If I understand your methodology correctly, your observed INP number as a function of temperature is in the immersion freezing mode. However, Meyers scheme (M92) describes the deposition freezing and condensation ice nucleation. Could you please explain a bit more why the measured INP concentrations that are relevant for the immersion mode can be used to replace the M92 scheme?
  Regarding your first question, this is a slightly unfortunate complication due to CAM5’s parameterization choices, which is an outdated treatment of ice nucleation. It is true that the Meyers’ scheme (M92) describes deposition nucleation and condensation-freezing in mixed-phase clouds, and it is also the dominant ice nucleation pathway in CAM5. We replace this parameterization with a new function (A21) based on measurements of INPs in immersion freezing mode, which is now being accepted as the dominant nucleation pathway in mixed-phase clouds (see e.g. Ansmann et al., 2009, de Boer et al., 2011, Westbrook and Illingworth, 2011).
  
  We clarify a few additional details of our ice nucleation modifications in CAM5. It should be stressed that we do not make any changes to the deposition nucleation scheme in cirrus clouds, but only change the M92 scheme active for temperatures between -37°C and 0°C . Additionally, in the updated version of this manuscript we exclude the immersion freezing parameterization of Bigg wherever we replace M92, so that the A21 parameterization is the only active immersion freezing parameterisation in the Arctic. This makes very little change to the cloud ice number concentrations or the overall picture we see, as the Bigg parameterization contributes little to activated ice nuclei, something which is also stated by previous studies by e.g. English et al. (2014). To address the unclarities you have pointed out, we plan to update lines 148-156 accordingly:
  “In CAM5, the different heterogeneous ice nucleation pathways in mixed-phase clouds are parameterized independently, namely, contact freezing (Young, 1974), immersion freezing (Bigg, 1953, hereafter: "B53") and deposition and condensation freezing (Meyers et al., 1992, hereafter: "M92"). Here, we update the M92 parameterization. This parameterization is active in the temperature range -37°C to 0°C, and is responsible for more than 90 % of ice nuclei formed in CAM5 mixed-phase clouds (English et al., 2014). Since the measured INP concentrations are relevant for the immersion mode, replacing the M92 with our measurements entails excluding deposition and condensation freezing in Arctic mixed-phase clouds. This exclusion is justified by observational studies that found deposition and condensation freezing to be negligible for mixed-phase clouds (Ansmann et al., 2009; Boer et al., 2011; Westbrook and Illingworth, 2011). Where we update the M92 parameterization using our INP measurements in immersion freezing mode, we exclude the B53 immersion freezing parameterization.”
  Question 2: Is it appropriate to quantify INP concentration as a function of temperature in the immersion freezing mode? For visualization purpose, INP number concentration can be plotted as a function of temperature, like figure 2 in your paper, figure 7 in Geerts et al., (2022) (https://journals.ametsoc.org/view/journals/bams/103/5/BAMS-D-21-0044.1.xml). But can we quantify INP number concentrations as a function of only temperature in the immersion freezing mode, without considering other factors such as water activity or active surface area density?
  Regarding your second question, it is true that we make a simplification when we predict immersion freezing INP concentration based only on temperature. The reason we do not include aerosol surface area density as a predictor in the parameterisation is because we find quite low correlation with aerosol surface area density (see Fig. A1e in the appendix). This finding is consistent with INP measurements in other Arctic sites, e.g. in Ny-Ålesund (Li et al., 2022). Therefore we believe that including aerosol surface area density will not improve the predictive ability of our INP parametrization. The aerosol species originally used to predict INPs in CAM6-Nor (the atmospheric model we use) are dust and soot, but other sources such as marine organic aerosols or other bioaerosols are likely important Arctic INPs (see for example Creamean et al. (2022), Carlsen and David (2022), Sze et al. (2023) or Freitas et al. (2023)). Differences in water activity can indeed result from different aerosol chemical composition and in theory affect ice nucleation. However, this is difficult to account for with our measurement technique, as the aerosols are all impinged in water before their ice-nucleating ability is probed, which may dissolve and dilute substances that would otherwise affect the freezing ability of an aerosol. However, one of our motivations for using a simple temperature-based parameterisation is that we do not yet have a full understanding of how INP concentrations can be predicted in the Arctic. Therefore by keeping it simple, we avoid introducing any additional sources of error when predicting Arctic INPs, until a full understanding on the composition-dependent ice-nucleating ability of Arctic INPs is in place.
  Question 3: How different is your measured INP number concentration from that measured during the COMBLE campaign? One year apart in the same season yield similar results?
  Regarding your third question, if we compare Fig. 7 in Geerts et al. (2022, https://doi.org/10.1175/BAMS-D-21-0044.1) and Fig. 2 in our study we see that our 2021 measurements in Andenes one year after the COMBLE campaign show somewhat higher maximum concentrations (around one order of magnitude at some temperatures) than the COMBLE campaign, and similar minimum concentrations. It should be noted that there are certain important differences between the data. First of all, the COMBLE data exclusively represents cold air outbreaks, i.e. flow reaching Andenes from the Arctic, while our measurements also include air masses reaching our measurement site with a more southerly and southwesterly flow (see Fig. A2 of the air mass back trajectories in the appendix). As these air masses are going into the Arctic, we believe that they are also important to capture in our case to understand Arctic INP concentrations. Secondly, our measurements are only from March 2021, while the COMBLE measurements are from the period December 2019 to May 2020. Nevertheless, we find that our measurements are quite similar to other measurements around the Arctic (see Fig. 2). Additionally, our measurement technique does not allow us to estimate INP concentrations higher than approximately 0.1/L, or INP concentrations at temperatures lower than around -25°C. This should be kept in mind when comparing the two datasets. To make this comparison easier for the reader, we are in the process of incorporating the COMBLE data points into Fig. 2 in the manuscript.
  Thank you again for your questions and please let us know if you have any others.
  Best regards,
  
  Astrid Bragstad Gjelsvik on behalf of the author team
  
  Papers on liquid as a prerequisite for ice in mixed-phase clouds:
  Ansmann, A., M. Tesche, P. Seifert, D. Althausen, R. Engelmann, J. Fruntke, U. Wandinger, I. Mattis, and D. Müller (2009), Evolution of the ice phase in tropical altocumulus: SAMUM lidar observations over Cape Verde, J. Geophys. Res., 114, D17208, doi:10.1029/2008JD011659.
  de Boer, G., H. Morrison, M. D. Shupe, and R. Hildner (2011), Evidence of liquid dependent ice nucleation in high-latitude stratiform clouds from surface remote sensors, Geophys. Res. Lett., 38, L01803, doi:10.1029/2010GL046016.
  Westbrook, C. D., and A. J. Illingworth (2011), Evidence that ice forms primarily in supercooled liquid clouds at temperatures > −27°C, Geophys. Res. Lett., 38, L14808, doi:10.1029/2011GL048021.
  Paper from English et al. on ice nucleation parameterizations in CAM5:
  English, J. M., Kay, J. E., Gettelman, A., Liu, X., Wang, Y., Zhang, Y., & Chepfer, H. (2014). Contributions of Clouds, Surface Albedos, and Mixed-Phase Ice Nucleation Schemes to Arctic Radiation Biases in CAM5. Journal of Climate, 27(13), 5174-5197. https://doi.org/10.1175/JCLI-D-13-00608.1
  Paper on INP measurements in Ny-Ålesund:
  Li, G., Wieder, J., Pasquier, J. T., Henneberger, J., and Kanji, Z. A.: Predicting atmospheric background number concentration of ice-nucleating particles in the Arctic, Atmos. Chem. Phys., 22, 14441–14454, https://doi.org/10.5194/acp-22-14441-2022, 2022.
  Papers on INPs, seasonality and connections to bioaerosols:
  Creamean, J.M., Barry, K., Hill, T.C.J. et al. Annual cycle observations of aerosols capable of ice formation in central Arctic clouds. Nat Commun 13, 3537 (2022). https://doi.org/10.1038/s41467-022-31182-x
  Carlsen, T., & David, R. O. (2022). Spaceborne evidence that ice-nucleating particles influence high-latitude cloud phase. Geophysical Research Letters, 49, e2022GL098041. https://doi.org/10.1029/2022GL098041
  Sze, K. C. H., Wex, H., Hartmann, M., Skov, H., Massling, A., Villanueva, D., and Stratmann, F.: Ice-nucleating particles in northern Greenland: annual cycles, biological contribution and parameterizations, Atmos. Chem. Phys., 23, 4741–4761, https://doi.org/10.5194/acp-23-4741-2023, 2023.
  Pereira Freitas, G., Adachi, K., Conen, F., Heslin-Rees, D., Krejci, R., Tobo, Y., Yttri, K. E., & Zieger, P. (2023). Regionally sourced bioaerosols drive high-temperature ice nucleating particles in the Arctic. Nature communications, 14(1), 5997. https://doi.org/10.1038/s41467-023-41696-7
  
  Citation: https://doi.org/10.5194/egusphere-2024-1879-AC1
RC1:
'Comment on egusphere-2024-1879', Anonymous Referee #1, 11 Aug 2024

Publisher’s note: this comment is a copy of RC3 and its content was therefore removed.

Citation: https://doi.org/10.5194/egusphere-2024-1879-RC1
- RC2:
  'Reply on RC1', Anonymous Referee #1, 13 Aug 2024
  
  Publisher’s note: this comment is a copy of RC3 and its content was therefore removed.
  
  Citation: https://doi.org/10.5194/egusphere-2024-1879-RC2
  - RC3: 'Reply on RC2', Anonymous Referee #1, 15 Aug 2024
    
    The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1879/egusphere-2024-1879-RC3-supplement.pdf
    
    Citation: https://doi.org/10.5194/egusphere-2024-1879-RC3
RC4: 'Comment on egusphere-2024-1879', Anonymous Referee #2, 16 Sep 2024

Review of “Using a region-specific ice-nucleating particle parameterization improves the representation of Arctic clouds in a global climate model” by Gjelsvik and co-authors.
In this study the authors use measurements of ambient ice nucleating particle (INP) concentrations from within the Arctic to assess the sensitivity of the NorESM Earth System Model to the representation of immersion mode heterogeneous ice nucleation. Ambient aerosols greater than 0.5 um in diameter were collected over a period of 15 days in March 2021 and processed in an instrument that measures the number of INP activated as a function of temperature and the volume of sampled air. Using these data an INP parameterization was derived for the representation of immersion-mode ice nucleation in supercooled cloud liquid droplets. The authors incorporated the parameterization into a cloud microphysics module in NorESM to replace primary ice production in the Arctic from the default mechanism(s). A comparison between simulations using the default (named M92) and the new setup (named A21) was used to assess whether the new setup improves the model’s ability to reproduce observed quantities of the Arctic atmosphere. The new A21 setup improves the resulting bulk cloud phase partitioning between liquid and water, impacts the amount of cloud ice and cloud liquid water loadings, and reduces the bias apparent in radiative fluxes at the surface. The study is not entirely novel but does help extend our knowledge on the role of INPs in the climate, and specifically the Arctic region which is less understand than others. It also demonstrates, as have other studies, the sensitivity of simulated high-latitude clouds to the representation of ice production. However, I believe there are several major issues within the methodology and analysis that need addressing before the manuscript is in a position to be accepted for publication.
Major comments.
The main result of the study is that reducing the simulated INP concentrations results in more liquid water and less ice, which better matches remotely observed cloud-phase partitioning. One caveat to this result is that it could be compensating for other poorly represented processes in the cloud microphysics scheme. Are the authors confident that this is not the case? For instance, Gettelman et al. (2023) have demonstrated that Arctic LWP and IWP in CESM2 is sensitive to several different cloud microphysical processes. Do other models using the M92 parameterization (or similarly high primary ice production rates) suffer from the same issue? It may not be possible to answer these questions, but this caveat should thoroughly addressed in the manuscript.
“Improved representation of Arctic clouds”. The authors have only selected a few observable quantities to assess the representation of Arctic clouds. There are others that could be considered to fully justify this statement – including LWP and IWP, cloud fraction, and precipitation. Widespread LWP increases of 30 gm-2 apparent in Figure 5 will surely be seen in the observations / model biases. Similarly, the cloud fraction changes in Figure B1 may be substantial enough to be seen in observations / model biases. Also, why was the TOA shortwave CRE not compared to observations (as for longwave in Figure 11)? I couldn’t find anywhere in the text that explained their omission.
Are the INP measurements adequately representative of the whole Arctic region across the annual cycle? The authors compare their INP parameterization with others from the Arctic region and show consistency, yet a new study by Raif et al. (2024) has shown that INP concentrations may not always be so low. Additionally, there is good evidence that dust from low latitudes is present in the Arctic atmosphere throughout the year, which may strongly influence the spatial variability of INP availability (Shi et al., 2022). In terms of locally produced INP, there will be important sources that are dependent on the seasonal cycle of sea ice (sea spray aerosol), snow coverage (glacial dust sources), and temperature/sunlight (biological sources). I recommend the authors include these caveats to their discussion/conclusions on the representativeness of their parameterization. Perhaps expand lines 390-395 or 210-212.
The INP measurements were made using air sampled via an instrument that does not include aerosols smaller than around 500nm in diameter. Does omitting sub 500nm-sized particles bias the INP concentration measurements? Do the authors have a sense of how many more INP would be measured if the whole size distribution was accounted for?
The purely temperature dependent (L6) parameterization presented here is put forward as a new and simplified representation of Arctic INP, yet without any analysis on the origins of the underlying aerosol its use in future studies is limited. The better representative parameterizations are linked to a quantity or property of the aerosol size distribution (e.g., particles larger than 0.5 um or specific components like dust surface area) that are capable of being dependent on variable properties of the atmosphere / aerosol population. With these we are then able to assess changes in INP distributions for future scenarios etc. Do the authors have information on the composition of the aerosol that is acting as an INP? Is this consistent with other studies? Could the authors convert the INP spectra into a quantity related to the aerosol size distribution such as the active site density ns? This would allow a much more thorough comparison with other INP measurements.
Minor comments
L46. Dusts from lower latitudes are also readily transported to the Arctic. See Shi et al. (2022) and references therein.
L66. “To our knowledge…” I think this is likely correct, but there are a few studies that have used Arctic-based INP parameterizations in global models, which should be acknowledged for having a similar methodology. Kawai et al. (2023) incorporated an observations-based ice nucleation parameterization of Arctic dust into CAM5 and Shi et al. (2022) modelled INP concentrations in the E3SMv1 model using a parameterization based on Icelandic dust samples.
L144 and L229. Is total aerosol surface area for particles greater than 0.5 um diameter?
L146. Is this a sensible number?
L146. What impact will turning off ice detrainment have and why was it turned off?
L151. M92 is for primary ice production but the empirical parameterization is for INP. How do you relate the INP to the primary ice production? Also, do you still have Bigg immersion mode freezing and Young contact freezing? Sometimes the Bigg parameterization is used for rain drops rather than cloud droplets – if this is the case perhaps state this to avoid confusion.
L210. I suggest the authors include low latitude dust sources too.
Figure 2. As this is the first time the INP data is being shown please include uncertainty estimates to the scatter points.
L261. Do the changes in the cloud fraction profile (less/more cloud at different altitudes) have implications for the change in longwave radiation at the surface?
L274. What happens to total water path (TWP) and precipitation? Or is it all WBF? I would be surprised if the differences were entirely attributed to this. It would be useful to have all hydrometeor classes accounted for so we can understand where the liquid has come from / gone. Also, is there a reason TWP is never discussed?
L345 and others. I’m not sure the use of the word drastic (radical; extreme) is an appropriate choice. Please consider changing all instances throughout the manuscript.
References.
Gettelman et al., 2023. https://doi.org/10.5194/gmd-16-1735-2023
Kawai et al., 2023. https://doi.org/10.1029/2022GL102470
Raif et al., 2024. https://doi.org/10.5194/egusphere-2024-1502
Shi et al., 2022. https://doi.org/10.5194/acp-22-2909-2022

Citation: https://doi.org/10.5194/egusphere-2024-1879-RC4

Astrid Bragstad Gjelsvik, Robert Oscar David, Tim Carlsen, Franziska Hellmuth, Stefan Hofer, Zachary McGraw, Harald Sodemann, and Trude Storelvmo

Model code and software

INP-Andenes-2021-NorESM2 Astrid Bragstad Gjelsvik, Robert Oscar David, Tim Carlsen, Franziska Hellmuth, Zachary McGraw, Stefan Hofer, and Trude Storelvmo https://github.com/astridbg/INP-Andenes-2021-NorESM2

Astrid Bragstad Gjelsvik, Robert Oscar David, Tim Carlsen, Franziska Hellmuth, Stefan Hofer, Zachary McGraw, Harald Sodemann, and Trude Storelvmo

Viewed

Total article views: 894 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
497	118	279	894	13	14

HTML: 497
PDF: 118
XML: 279
Total: 894
BibTeX: 13
EndNote: 14

Views and downloads (calculated since 09 Jul 2024)

Month	HTML	PDF	XML	Total
Jul 2024	129	35	12	176
Aug 2024	176	47	14	237
Sep 2024	76	13	42	131
Oct 2024	49	9	93	151
Nov 2024	48	10	92	150
Dec 2024	19	4	26	49

Cumulative views and downloads (calculated since 09 Jul 2024)

Month	HTML	PDF	XML	Total
Jul 2024	129	35	12	176
Aug 2024	176	47	14	237
Sep 2024	76	13	42	131
Oct 2024	49	9	93	151
Nov 2024	48	10	92	150
Dec 2024	19	4	26	49

Viewed (geographical distribution)

Total article views: 877 (including HTML, PDF, and XML) Thereof 877 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 13 Dec 2024

Short summary

Ice formation in clouds has a substantial impact on radiation and precipitation, and must be realistically simulated in order to understand present and future Arctic climate. Rare aerosols known as ice-nucleating particles can play an important role for cloud ice formation, but their representation in global climate models is not well suited for the Arctic. In this study, the simulation of cloud phase is improved when the representation of these particles are constrained by Arctic observations.


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