the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Jul 2026
Modeling mineral dust aerosols in the Arctic with the regional model WRF-Chem 4.6.1
Abstract. Important sources of mineral dust aerosols in the high latitudes, especially in the Arctic, are neglected in most atmospheric models. Here, we create a new description of mineral dust aerosol emissions including high-latitude dust sources, and implement it in a polar version of the WRF-Chem 4.6.1 model. We also improve the representation of mineral dust aerosol deposition processes. A 1-year-long quasi-hemispheric simulation for 2012 is evaluated against observations of surface concentrations and deposition fluxes at the Arctic surface. The updated model predicts 34 Tg yr-1 of emissions north of 60° N, and our evaluation shows dramatic improvements in model performance compared to the base model, to a state-of-the-art climate model, and to the CAMS reanalysis. We find that high latitude dust sources have significant impacts on several climate-relevant metrics in the Arctic, such as aerosol optical depth, ice nucleating particle concentrations, and especially mineral dust aerosol deposition to snow and ice, where high latitude sources are the main contributor over the Arctic sea ice pack. These results demonstrate the importance of taking high latitude dust emissions into account, provide a description of these sources reusable in other models, and open the way for future work to better estimate the climate impacts of high latitude dust sources.
- Preprint
(4201 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- CC1: 'Comment on egusphere-2026-4112', Daniel Bellamy, 22 Jul 2026
-
CC2: 'Comment on egusphere-2026-4112', Alexander Ukhov, 26 Jul 2026
1. Wet scavenging
The manuscript states that GOCART does not include grid-scale wet scavenging. However, standard WRF-Chem provides the large-scale wet-deposition option wetscav_onoff = -1 in module_wetdep_ls.F. Please clarify how the new formulation differs from this existing option and, if possible, briefly compare their performance.
2. Sedimentation mass conservation
Section 4.2 introduces an anti-diffusive sedimentation scheme, but the manuscript does not explicitly demonstrate aerosol mass conservation.
Please provide a mass-budget test confirming that atmospheric aerosol loss through sedimentation is balanced by surface deposition.Citation: https://doi.org/10.5194/egusphere-2026-4112-CC2 -
RC1: 'Comment on egusphere-2026-4112', Anonymous Referee #1, 30 Sep 2026
The study presents a new description of mineral dust aerosols, including high-latitude sources, implemented in a polar fork of WRF-Chem 4.6.1 ("WRF-Chem-Polar"), together with revised wet-scavenging and sedimentation schemes for the GOCART aerosol model. The evaluation rests on a one-year quasi-hemispheric simulation from 2012-01-01 to 2013-01-01 at 100 km × 100 km over a quasi-hemispheric domain, with 70 vertical levels from the surface to 50 hPa and the lowest level top at 20 m AGL, compared against climatological surface concentration and deposition observations, against the CAMS reanalysis and the CESM2.2 CAM-Chem simulation used to drive the model.
The paper is well-motivated, the model development is transparent, and the code and data are made openly available. The work is a useful contribution that addresses real gaps in Arctic dust modeling. I would recommend acceptance subject to moderate revisions after addressing my comments below:- Section 4.2 introduces an anti-diffusive sedimentation scheme. The statement that this implementation "almost completely cancels the effect of numerical diffusion in vertical advection" is explicitly attributed to Mailler et al. (2021) and was obtained in that framework; applying the limiter here to a downward, monotone settling operator is a narrower result than the text implies. The earlier statement that the diffusive upwind settling scheme is among the "shortcomings are corrected below" reads as a general fix to numerical diffusion in the model, which the results do not support. Related to this, the effect of the new scheme is never isolated: all three simulations include the new wet-scavenging and new sedimentation options and differ only in emissions and land use, so there is no upwind-settling control to compare against. Since the option is namelist-controlled (aer_settling_opt) a single additional run with it disabled, compared to WRF-Chem-NEW in dust load, surface and surface concentration would be sufficient to demonstrate what the scheme actually changes and to justify the claim being made.
- The simple GOCART scheme in WRF model originally already includes a dust wet removal parameterization scheme. Please explain the differences between the existed scheme and the new schemes described in section 4.1.
- The paper compares a 2012 simulation to long-term climatological observations spanning 1983–2025 at different sites, and explicitly acknowledges that this introduces an unquantified interannual signal. This is defensible given the scarcity of Arctic dust data, but the headline statistics (e.g. log-R = 0.79, log-RMSE = 0.70 for WRF-Chem-NEW versus log-R = 0.45 / log-RMSE = 1.73 for WRF-Chem-OLD and log-R = 0.43 / log-RMSE = 2.8 for CESM2.2 CAM-Chem) rest on a small number of stations and on modeled rather than observed meteorology. Please report the number of comparison points and a per-station breakdown, so readers can judge whether the "dramatic improvements" are driven by one or two high-latitude sites.
- WRF-Chem-OLD has zero emissions north of 60°N, so any positive Arctic concentration is trivially an improvement. The more informative contrast is WRF-Chem-NEW versus WRF-Chem-NEW_NOHLD, and versus observations in the well-observed 50–60°N band. The residual low-latitude underestimate (log-RMSE = 0.72) is attributed to missing agricultural sources, but that attribution is asserted rather than demonstrated. Some supporting analysis (e.g. the spatial/temporal signature expected from agricultural dust) would strengthen it.
Citation: https://doi.org/10.5194/egusphere-2026-4112-RC1 -
RC2: 'Comment on egusphere-2026-4112', Anonymous Referee #2, 06 Oct 2026
This manuscript describes improved and new implementations related to dust emissions and transport at high-latitudes in the regional WRF-Chem 4.6.1 model. High-latitude dust sources have recently been recognized as relevant for the Arctic climate systems and it is an important update of the model. While the manuscript clearly describes changes to the model and reads well, it fails to show the relevance and validity of single parameterizations. In particular process parameterizations that have not been presented elsewhere should be explained and tested more thoroughly. Moreover, sensitivity analysis of specific changes would strongly help to show the strengths and shortcomings of the model.
In my opinion, the manuscript also needs further explanations and discussion on these topics in particular:Freezing parameterization
Indeed, soil freezing processes are likely relevant for dust emissions at high latitudes. Several processes are involved however, and both limited dust emissions due to freezing and increased dust emissions due to thaw-freeze cycles or enhanced aggregation and saltation bombardment are possible. The parameterization presented here appears to be arbitrary and needs further explanation and discussions that cover following questions: 1) what observations or experiments is this based on and how were thresholds established? 2) What is the (quantitative) effect of this parameterization on (high-latitude) dust emission estimates? 3) Did it improve results and where, e.g. is it generally applicable or what are the limitations?Indeed, the number of observations to test such a parameterization is highly limited. Nonetheless, there must be some reasoning for this parameterization that should be presented. A better description and quantification of this process will help users of the model to assess whether they should include this parameterization. Furthermore, a discussion on the potential relevance of freezing processes can be a motivation for future observations and experiments.
Model evaluation
The current model evaluation is highly limited, and it is not clear what changes contribute to model improvement. The only comparisons to observations are based on modelled 2012 average surface concentrations and deposition to a climatology. Indeed, observational datasets are limited, but there are data for 2012 in the referenced datasets. A comparison to time series, or even monthly or seasonal averages would be highly valuable to show whether the model captures any dust episodes or observed seasonal cycles, especially since high-latitude dust sources are known to show a strong variability. Did you consider remote sensing products or vertical profiles?General applicability
The authors present the dust emission module as globally applicable, yet evaluation focuses almost entirely on the Arctic. The figure of total dust emission (Fig 3) shows large changes in dust emissions apart from high-latitude sources. And the revised scavenging scheme can have impacts globally. The authors should therefore show that these updates improve, or at least not degrade, model performance outside high-latitude dust source regions.Specific comments
Line 29; Several studies have already discussed or shown impacts of local dust on surface albedo in polar regions.
Line 39: Where do you evaluate impacts of high latitude dust in the Arctic region?
Figure 1: It looks like there’s a large impact beyond high-latitude sources in your model domain and globally. Can you quantify changes to total dust emissions for different regions?
Table 1/eq 5; The size classes do not correspond to LeGrand et al 2021 and it is not entirely clear to me how dust emissions are split over all size bins.
Equation 9: On what data is this based and how were thresholds defined?
Line 183-185: Why not show dust transport based on MOSAIC and GOCART aerosols? That could confirm that differences only occur close to dust sources.
Table 2: Where did coefficients for ice come from and why were these added? These are not included by Luo et a. (2020) as far as I understand.
Line 338-342: Why do you only evaluate the contributions of the new dust emission model, not the new scavenging and sedimentation routines?
Line 350-353; You could add information on the temporal variation of dust emissions.
Figure 4: Why only include stations >50 degrees north if there are several more in your domain?
Also, rather than only showing average climatology, you could add information on the range of observed annual averages at each observation site. For sites that include 2012 you could even add a direct comparison. The representation of the seasonal cycle would be quite important.Line 383: What other models do you mean?
Figure 4: Please explain what the lines in the scatter plot are.
Figure 5: Are high-latitude dust sources important for these sites?
Citation: https://doi.org/10.5194/egusphere-2026-4112-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 235 | 121 | 42 | 398 | 34 | 35 |
- HTML: 235
- PDF: 121
- XML: 42
- Total: 398
- BibTeX: 34
- EndNote: 35
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
L358: Bellamy et al. (2026) simulates dust events in Canada, not Alaska