the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Aug 2024
Numerical investigations on the modelling of ultrafine particles in SSH-aerosol-v1.3a: size resolution and redistribution
Abstract. As the health impact of ultrafine particles is getting better understood, modelling the size distribution and the number concentration with chemistry transport models becomes an increasingly important matter. The number concentrations is strongly affected by processes linked to aerosol dynamics: coagulation, condensation and gas/particle phase partitioning, nucleation. Coagulation is usually solved using an Eulerian approach, using a fixed diameter size discretization. In opposition, condensation/evaporation is rather solved using a Lagrangian approach, requiring redistribution of particles on the fixed grid size. Here, a new analytic formulation is presented to compute efficiently coagulation partition coefficients, allowing to dynamically adjust the discretization of the coagulation operator to the Lagrangian size mesh evolution, and therefore solve all the processes linked to aerosol dynamics with a Lagrangian approach, avoiding the redistribution on the fixed size grid. This new approach has the advantage of reducing the numerical diffusion introduced by condensation. The significance of these effects on number concentrations is assessed over Greater Paris with the chemistry transport model Polyphemus/Polair3D coupled to the aerosol model SSH-aerosol, using different size resolution of the particle distribution.
- Preprint
(1170 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (extended)
-
RC1: 'Comment on gmd-2024-150', Anonymous Referee #1, 31 Oct 2024
reply
Jacquot and Sartelet. “Numerical investigations on the modelling of ultrafine particles in SSH-aerosol-v1.3a: size resolution and redistribution”
General comments:
This paper presents the development of analytical equations and a numerical algorithm for Lagrangian calculation of the coagulation term in aerosol distribution dynamics, specifically for calculating the partitioning of coagulated aerosol (number and mass) between discrete size bins in a manner consistent with the Lagrangian formulation of other aerosol dynamics processes. The approach was developed with the purpose of reducing numerical dispersion during aerosol dynamics simulation caused by redistribution due to changing between Lagrangian and Eulerian formulations. In this study, the new approach was applied within a Eulerian chemical transport model to a real-world case study, with three different resolution particle size distribution schemes. Impacts on accuracy (compared with measurements) and numerical dispersion were investigated. The main results suggest that the impact of the new formulation/algorithm on numerical diffusion is small compared with the impact of size resolution itself (and compared with the overall error in representing measurements). Overall, I find the work to be interesting and publication of the new more-internally-consistent algorithm is likely to be useful to CTM model developers. However, the impact on the field may be small due to the findings (of small effects); hence this may be better as a technical note. Additionally, some of the explanation needs improved clarity and some corrections are required, as discussed below.
Specific Comments:
The importance and impact of this work may be limited for a few reasons, making me question whether this would be better as a technical note than a full paper.
- The findings indicate that numerical dispersion is less sensitive to the new coagulation formulation/algorithm than to the resolution of the aerosol size discretization. The differences are also much smaller than the simulation errors (determined by comparison to measurements).
- The authors state that the new formulation is basically repeating a derivation of Debry and Sportisee (2007) but correcting a mistake from that paper. This seems like a focus for a technical correction rather than for a novel contribution.
- The authors further conclude that the new formulation/algorithm may be most useful for low size resolution simulations (9 bins), due to the computational costs of the new Lagrangian formulation/algorithm. Given this, it is not clear why one should invest (computationally) in the new formulation versus just investing in a higher resolution size distribution. A discussion of the relative computational costs should be provided.
Tables 1 and 2 show comparison of model results to measurements for the mixed Eulerian/Lagrangian algorithm (1), not the fully Lagrangian algorithm (2). The text indicates this is because statistics “are very similar”. Because the fully Lagrangian algorithm is the focus of this paper, the statistics resulting from using that algorithm should be shown (and perhaps those for the mixed algorithm put in an appendix for comparison)
Additional checking of the derivations is needed (by someone whose work involves similar derivations). I did not have the time to go through the equations in detail to fully understand them, but I did compare them to equations in the existing literature and was left with some questions. First, Appendix A lists the general aerosol dynamics equations for evolution of the number and mass distributions; a citation for this classical formulation is listed as Gelbard et al. (1980) [L91]. Although the number density equations seem generally consistent with other sources (Seinfeld and Pandis, 2006 textbook), I did not find these formulations in the Gelbard reference, so the citation seems to need correction. More importantly, it looks like there may be a typo or error in the mass distribution equation (A4) as the units don’t seem to work out. Specifically, the last term in the equation (Is*n) appears to be missing some kind of multiplicative density term (mass of species s per volume of species s?). Without that, there is no mass, so it is unclear how that term contributes to a mass density.
Additional explanation of some terms is also needed. For example, what is the meaning of the blackboard bold 1 symbol in Equations 3 and 4? Perhaps this is a well-known symbol in this subfield, but I didn’t recognize it. If this work is to reach an audience that isn’t excessively narrow, non-standard mathematical symbols should be defined.
Table 2 caption. The units of mass concentration appear to be mislabeled here as #.cm-3, which are units for a number concentration, not a mass concentration. The units need to be corrected (or clarified).
Figure 2. The figure labelling should be improved. In addition to labelling the sites based on whether they measure number or mass concentration, sites should also be labelled using the network names provided in the text (LHVP, SIRTA AIRPARIF). Additionally, labeling the location of Paris and the meaning of the polygon’s boundaries would be helpful.
Technical corrections:
Figure 3. The exponent (4) is missing from the color scales and the units of number concentration are also missing (except for the exponent).
L20 typo: should be “focused”
L42 “formerly equivalent to ...’ I don’t understand this. Should it say "formally"?
L86. Missing word “be” before “a partitioning”
L107 Typo: “substraction”
L186. Should refer to Figure 6, not Figure 3.
Citation: https://doi.org/10.5194/gmd-2024-150-RC1
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 146 | 37 | 10 | 193 | 11 | 7 |
- HTML: 146
- PDF: 37
- XML: 10
- Total: 193
- BibTeX: 11
- EndNote: 7
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1