the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 02 Feb 2024
AeroMix v1.0.1: a Python package for modeling aerosol optical properties and mixing states
Abstract. Assessing aerosol mixing states, which mainly depend on aerosol chemical compositions is indispensable to estimate aerosol direct and indirect effects. While the limitations in the measurements of aerosol chemical composition and mixing states persist globally, the Optical Properties of Aerosols and Clouds (OPAC) model has been widely used to construct optically equivalent aerosol chemical compositions from measured aerosol optical properties using Mie inversion. However, the representation of real atmospheric aerosol mixing scenarios in OPAC has perennially been challenged by the exclusive assumption of external mixing. A Python successor to the aerosol module of the OPAC model is developed, named 'AeroMix,' with novel capabilities to 1) model externally and core-shell mixed aerosols, 2) simulate optical properties of aerosol mixtures constituted by any number of aerosol components, 3) and define aerosol composition and relative humidity in up to 6 vertical layers. Designed as a versatile open-source aerosol optical model framework, AeroMix is tailored for sophisticated inversion algorithms aimed at modeling aerosol mixing states and also their physical and chemical properties. AeroMix's performance is demonstrated by modeling the probable aerosol mixing states over Kanpur (urban), India, and the Bay of Bengal (marine). The modeled mixing states are consistent with independent measurements using single-particle soot photometer (SP2) and transmission electron microscopy (TEM), substantiating the potential capability of AeroMix to model complex aerosol mixing scenarios involving multiple internally mixed components in diverse environments. This work contributes a valuable tool for modeling aerosol mixing states to assess their impact on cloud nucleating properties and radiation budget.
- Preprint
(1061 KB) - Metadata XML
-
Supplement
(538 KB) - BibTeX
- EndNote
Status: open (until 05 May 2024)
-
RC1: 'Comment on egusphere-2024-62', Anonymous Referee #1, 29 Mar 2024
reply
AeroMix v1.0.1: a Python package for modeling aerosol optical properties
and mixing states by Raj et al.
Summary
Raj et al. describe a Python based model ‘AeroMix’ which computes aerosol optical properties and mixing state using the Mie inversion technique. The predecessor model, Optical Properties of Aerosols and Clouds (OPAC) assumes external mixing states only, meaning it’s unable to accurately depict complex “real-world” aerosol. AeroMix is capable of modeling external and core-shell mixed populations with six vertical atmospheric layers. AeroMix is validated by comparing with OPAC simulations, along with mixing state measurements with a single-particle soot photometer and transmission electron microscopy. The probable mixing state calculated using AeroMix over Kanpur, and the Bay of Bengal is presented. The authors include detailed documentation of the open-source code on Github and Zenodo.
General comments
The study fits into the scope of GMD. Minor organization of the methods section is needed as described below.
Verdict: Accepted with minor revisions
Introduction
1. When discussing motivation, the mixing state of aged aerosol (far from source) is important for CCN (Farmer, Cappa, and Kreidenweis 2015). McFiggans et al. 2006 is another source discussing CCN from close to source and aged populations.
Model Overview
2. For the readers’ convenience, add a table for chemical species modeled (IS, WS, BC, SS, MD). Table columns should include the name, shorthand name (e.g. IS), species considered/represented (IS includes soil dust, fly ash, and non-hygroscopic organic matter from biomass burning), and modes present (nucleation (nm), accumulation (am), coarse (cm)).
3. It is unclear if the optical properties for coated MD particles are calculated using Mie theory or TMM. Clarify if TMM only applies to particles comprised of 100% MD.
4. The exponential function for the vertical profile of aerosol concentration is mentioned in Eq. 5; however, only the cubic function is utilized in the modeling section considering it provides the better fit based on R2 and RMSE values. The authors do not necessarily need to include the exponential function since it is not used to model vertical profiles in this paper.
Modeling of aerosol mixing state with AeroMix
5. This section was jumpy and difficult to follow. I suggest further dividing into subsections:
3.1 Validation with OPAC
This subsection would consist of paragraphs 1 (“The primary objective of AeroMix …”, and 3 (“Past endeavours to deduce intricate …”).
3.2 Model setup
This subsection would include paragraph 2 (AeroMix performance was further assessed …”), 4 (“The analysis encompasses eight aerosol components …”), 6 (“The combined mass concentration of SSam and …”), 5 (“Previous studies have assumed either the entire mass …”), and 10 (“In this study, the aerosol mixing state …”).
3.3 Iteratively evaluating probable existence of mixing states
Consists of paragraphs 8 (“The probable existance of the components in the atmosphere …”) and 9 (“The aerosol mixture with the spectral AODs and SSAs matching well …”).
3.4 Vertical distributions of aerosols
Consists of paragraph 7 (“The vertical distribution of aerosols in the mixed layer …” through “attempting a generalized quantification of the sensitivity of vertical profile assumptions on AOD lacks meaningful interpretation”).
This can be arranged as the authors see fit, but at current the lack of structure makes reading challenging.
Summary and future scope
6. Little mention of future work/applications. Maybe mention potential model development projects and associated improvement to probable mixing state predictions.
Minor comments
Line 43: Healy et al. 2014 and Ye et al. 2018 describe in situ measurements of mixing state between external and internal.
Line 58: Unclear to reader why “batch mode” would have limited functionalities. Either clarify or reword.
Line 97-98: Unclear why density of BC is mentioned here.
Line 137 – 139: This sentence is confusing, specifically “and particles composed of multiple chemical species (internally mixed) as an external mixture.”
Equation (7): I believe index of summation should be layer, not n.
Line 163: Consider adding one sentence on the Mie inversion technique.
Figure 2: Consider adding average wind directions for winter, pre-monsoon and post-monsoon. This would aid in the comparison of modeled mixing states discussion (Sections 4.3 and 4.4). Label IGP region.
Line 243: Add reasoning for RMSE threshold after sentence: “Only those spectra with RMSE minimum 0.03 are considered as the best fit (Fig. 1). [Add here].” The last sentence of this paragraph “The RMSE threshold of 0.03 is chosen to ensure …” should be moved to this location.
Figure 3 and 4: Unnecessary to include colors for externally mixed and core-shell mixed in caption.
Technical corrections
Line 11: Add comma after “aerosol chemical compositions,”
Line 13: Remove “the” so the sentence reads “While limitations in the measurements of aerosol …”
Line 13-15: Sentence “While the limitations in the measurements …” is a run on. Consider breaking into two sentences.
Line 18: Move position of “3)” – “components, and 3) define aerosol composition”
Line 98: Add comma “The optical properties of each component, except for the MD components, are calculated”
Line 209: Run-on sentence, rewrite as “…CSR values of core-shell mixed aerosols in the model. This offers a more flexible approach compared to relying on measured…”
Line 297: Change to “… mixed states varied from 30% to 59% in premonsoon and winter, respectively.”
References
Farmer, Delphine K, Christopher D Cappa, and Sonia M Kreidenweis. 2015. 'Atmospheric processes and their controlling influence on cloud condensation nuclei activity', Chemical Reviews, 115: 4199-217.
Healy, R. M., N. Riemer, J. C. Wenger, M. Murphy, M. West, L. Poulain, A. Wiedensohler, I. P. O'Connor, E. McGillicuddy, J. R. Sodeau, and G. J. Evans. 2014. 'Single particle diversity and mixing state measurements', Atmos. Chem. Phys., 14: 6289-99.
McFiggans, Gordon, Paulo Artaxo, Urs Baltensperger, Hugh Coe, Maria C Facchini, Graham Feingold, Sandro Fuzzi, Martin Gysel, Ari Laaksonen, and Ulrike Lohmann. 2006. 'The effect of physical and chemical aerosol properties on warm cloud droplet activation', Atmospheric Chemistry and Physics, 6: 2593-649.
Ye, Qing, Peishi Gu, Hugh Z Li, Ellis S Robinson, Eric Lipsky, Christos Kaltsonoudis, Alex KY Lee, Joshua S Apte, Allen L Robinson, and Ryan C Sullivan. 2018. 'Spatial variability of sources and mixing state of atmospheric particles in a metropolitan area', Environmental science & technology, 52: 6807-15.
Citation: https://doi.org/10.5194/egusphere-2024-62-RC1 -
AC1: 'Reply on RC1', Puna Ram Sinha, 18 Apr 2024
reply
We appreciate and thank the reviewer for their valuable and detailed comments. All comments have been taken into account, as outlined in the attached response notes. The necessary corrections have been incorporated into the revised manuscript that is being submitted.
-
AC1: 'Reply on RC1', Puna Ram Sinha, 18 Apr 2024
reply
-
RC2: 'Comment on egusphere-2024-62', Anonymous Referee #2, 04 Apr 2024
reply
This manuscript is generally well-composed.
Dividing the content between lines 80 to 120 into more succinct paragraphs could significantly improve readability.
Citation: https://doi.org/10.5194/egusphere-2024-62-RC2 -
AC2: 'Reply on RC2', Puna Ram Sinha, 18 Apr 2024
reply
We appreciate and thank the reviewer for their valuable and detailed comments. All comments have been taken into account, as outlined in the attached response notes. The necessary corrections have been incorporated into the revised manuscript that is being submitted.
-
AC2: 'Reply on RC2', Puna Ram Sinha, 18 Apr 2024
reply
-
RC3: 'Comment on egusphere-2024-62', Simon O'Meara, 26 Apr 2024
reply
Referee Comment on AeroMix v1.0.1: a Python package for modeling aerosol optical properties and mixing states by Raj et al.
Summary
Raj et al. address the assumption of external mixing in the OPAC model by describing a model (AeroMix) that can treat aerosol as both externally and internally mixed. The benefit of such a model is the accurate estimation of light scattering and cloud transformation, such that radiation balances can be accurately estimated.
Other factors improve upon OPAC too, making AeroMix a good candidate as a tool for the next-generation of algorithms to more accurately constrain aerosol mixing state.
The model is verified (checking that the chosen equations are solved correctly) for externally mixed aerosol against OPAC and evaluated (assessing whether the model is a good representation of the real system) against field measurements.
Overall the paper is written and presented very well, though I agree with the other referees about rearranging section 3 to make a smoother read.
Verdict: Accept with minor revisions (detailed below)
General Comments
In the introduction (e.g. line 53), the limit on number of aerosol components that are available in OPAC is presented as a key limitation on its ability to infer mixing state. However, on line 103, the authors explain how AeroMix relies on the OPAC aerosol database for its measurements. My understanding is that AeroMix is therefore limited to the number of components available in the OPAC aerosol database. Therefore, whilst both OPAC and AeroMix rely on the OPAC aerosol database, are they not identically limited to the number of aerosol components? (By the way, I note there is an argument that users could append aerosol components to the OPAC aerosol database, and therefore extend the number of aerosol components present in AeroMix, but is this not equally true for the OPAC model? If not, then please explain.)
It is clear from Fig. 1 that AeroMix, as a stand-alone model predicts the optical properties of aerosol populations, and I can see from line 73 and line 98 that (as in OPAC) Mie theory is used to do this, and this is represented by the ‘AeroMix’ item in Figure 1. However, I am confused by the Mie inversion part of AeroMix: in line 163 several references are given for the ‘Mie inversion’ technique to find the aerosol mixing state, however none of these references use the phrase ‘Mie inversion’ for their techniques. Is this phrase newly coined in this paper? If so, it should be properly justified, including an explanation of why ‘inversion’ is the correct word, rather than iteration. I say this because it appears from Fig 1 and the description in line 241-260 that an iterative approach is used to find the mixing state that gives best fit to observations. Therefore, I question whether this is an inversion approach, or an iterative approach. Would ‘Mie iteration‘ be a better phrase?
Line 241 indicates that the iteration in AeroMix finds the probable existence of components, but Fig 1 and other part of the main text (e.g. line 253) suggest the iteration finds the probable mixing state. What does the iteration solve? Is it both the components present and their mixing state? If so, please make clearer.
The authors acknowledge that the iterative technique used to estimate probable mixing states is not unique (line 254). I would like to see more discussion, or reference (where this issue has been detailed before), around the range of mixing states that can be inferred from the iteration technique for a given set of inputs, i.e., a discussion around the probability that a given inferred mixing state is accurate. If I understand that AeroMix returns just one probable mixing state for a given set of inputs, then this seems to me to be a fundamental weakness, and future work should prioritise quantifying the uncertainty around the returned mixing state, and/or returning multiple mixing states so that users can quantify the range of probable solutions. I see that this issue is dealt with in lines 404-407, which explains that other algorithms could imbed AeroMix and therefore quantify the probable accuracy of returned mixing states. I also see that in sections 4.4 and 5, that the authors express the limitations of AeroMix, stating that only results useful for qualitative interpretation are returned. This is great, because it’s key that the authors describe the model’s limitations. However, I think this limitation should be mentioned in the model overview, so that readers are quickly aware of it. Furthermore, it would be useful to have some more description around what the ‘inherent constraints’ mentioned in line 401 are – this would greatly help authors of future minimization and machine learning algorithms to correctly utilise AeroMix.
Please detail what happens along the ‘No’ route of Fig. 1 – the authors explain in line 241 that iterative changes are made, but how is the amount and direction of change estimated?
Minor Comments
Line 35 – please provide some example references of studies that have used OPAC to estimate probable mixing state.
Line 57 – more detail is needed about why AEROgui is inferior to AeroMix, so that readers can distinguish between the two. For example, what limits in functionality does AEROgui have that AeroMix overcomes?
Line 137 – I don’t understand the sentence starting on this line. Specifically, how can an internally mixed aerosol be treated as an external mixture? This sounds like the components comprising the internally mixed aerosol are separated into separate particles so that an external mixture is formed.
Technical Corrections
None
References
None
Citation: https://doi.org/10.5194/egusphere-2024-62-RC3
Model code and software
AeroMix Sam P. Raj and Puna Ram Sinha https://doi.org/10.5281/zenodo.10552078
Interactive computing environment
Codes and model output used to generate the figures Sam P. Raj and Puna Ram Sinha https://doi.org/10.5281/zenodo.10552113
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 403 | 84 | 26 | 513 | 38 | 10 | 9 |
- HTML: 403
- PDF: 84
- XML: 26
- Total: 513
- Supplement: 38
- BibTeX: 10
- EndNote: 9
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1