| 20 Aug 2026
Optimized below-cloud scavenging scheme and its prominent performance of nitrogen wet deposition in polluted regions
Abstract. Simulations from Model Inter-Comparison Study for Asia (MICS-Asia) phase III generally underestimate regional nitrogen wet deposition, mainly because aerosol below-cloud scavenging processes are oversimplified or ignored in models. To address the problem, most existing studies adopt empirical corrections rather than optimizing physical mechanisms. This study develops an improved physically-based below-cloud wet scavenging parameterization scheme, which is implemented into the Nested Air Quality Prediction Model System (NAQPMS) for comparative simulations over East Asia in 2014. On the basis of conventional collision mechanisms, this scheme further incorporates three additional key mechanisms, including thermophoresis, diffusiophoresis and electrostatic interaction. The results show that the new scheme not only well captures the observed aerosols below-cloud scavenging coefficients, but also enables refined characterization of separate in-cloud and below-cloud wet deposition processes. Pronounced simulation improvements are found in heavily polluted regions including Beijing-Tianjin-Hebei region, Northeast China and Yangtze River Delta, while slight overestimation appears at several sites in Korea and Japan. Furthermore, the new scheme effectively elevates the contribution of below-cloud scavenging, especially for reduced nitrogen, achieving good agreement with observations in Beijing. Accurate simulations in both the in-cloud and below-cloud processes facilitate the reliable assessments of long-range transport and near-surface deposition fluxes of pollutants. This study demonstrates the necessity of fully resolving physically based wet scavenging mechanisms for high-precision nitrogen deposition modeling, which further advances our understanding of regional pollutant transport and the global nitrogen cycle.
Main Comments:
The authors present an analysis of the effects of updating the representation of below-cloud scavenging in a coupled meteorology-air quality model. The update includes the addition of several process representations for aerosol-raindrop interactions that are not currently included in existing modeling systems. The analysis includes a comparison of model simulations using the updated representation to model simulations with existing representations and to observations from routine and specialized monitors. The results suggest marked improvement in model performance. The authors also analyze and discuss how the new model formulation can impact inferences about modeled source contributions to downwind locations across a range of scales, making the results relevant to a large audience. The manuscript is well organized and well written and most of my specific comments listed below are minor and/or editorial in nature.
As elaborated further in some of my detailed comments below, the two main aspects of the manuscript in which I encourage revisions are a) to add a detailed description of the new method in the main manuscript rather than only showing it in the SI and to also add a discussion of future work that may be needed to further constrain the new method, and b) to add a description of the methods used to derive observation-based estimates of below-cloud scavenging and any associated uncertainties and future work needed to improve those estimates.
Specific Comments:
Line 30: Some of the nitrogen deposition networks like EANET have existed for more than just recent years, suggest rephrasing.
Line 40: suggest replacing “discharge” with “emission” because of the more widespread use of the term “emission inventories” compared to “discharge inventories” in air quality modeling
Line 48: change “second” to “secondary”
Line 74: change “month” to “months”
Line 110: please elaborate on the findings by Bae et al. (201) and Xu et al. (2019a) that BWSCs in this size range were underestimated by 1-2 orders of magnitude. Which chemical and physical environments and seasons were analyzed in these studies? What were the uncertainty bounds of observation-constrained BWSC estimates used in the comparison against modeled BWSCs?
Lines 120 – 125: The model updates summarized in these few lines are fundamental to the entire analysis presented in this manuscript and I therefore strongly encourage the authors to include the detailed description of these updates included in S1 to the main manuscript. I also suggest adding some discussion on whether the authors consider the current implementation of the three new processes to be state-of-the-science or whether important gaps remain in our understanding of these processes (and possibly other processes affecting BWSCs) that would benefit from further theoretical and/or laboratory studies.
Line 128: add “a” before “one-year”
Line 129: change “spectra was processed” to “spectra were processed”
Lines 127 – 145: I found this section to be a bit confusing. I think the intent is basically to say “we tried deriving raindrop size distribution equations for both mixed cloud and convective cloud precipitation but were not able to achieve a good fit for the latter so the calculation of raindrop size distribution in GE-14 BSWC was based solely on the fit derived for mixed cloud precipitation”. I suggest editing this section for clarity and also suggest discussing the limitations of this approach when the GE-14 BSWC scheme is applied to convective conditions and what kind of future work might help further advance the scheme.
Line 152: Please clarify and state explicitly which of the four schemes in Table 1 “diagnostic convective cloud scheme” refers to - GE-14?
Lines 222: Please specify exactly which validation metrics (presumably from Table 2) confirm that GE14-BWSC and GE14-CAMx outperform the other two schemes
Line 225: Similar to the comment as above, please be more specific which metrics were used to determine that GE14-BWSC is better than GE14-CAMx. Also suggest editing “GE14-BWSC is better than that of the GE14-CAMx” to “GE14-BWSC is better than GE14-CAMx”
Lines 240 – 241, Figures 3a-e: The coloring of the red, white, and blue circles depicting the normalized biases in Figures 53a-d is extremely hard to distinguish from the underlying contour plot showing the modeled annual TIN concentrations in precipitation. I suggest depicting these normalized biases on separate maps with just a white background. In addition, the colored dots showing observed annual mean TIN concentrations in Figure 3e presumably use the same color scheme as the color bar shown next to panels b and d but this isn’t clear from the layout of the panels and color bars.
Line 252: are there references or other supporting evidence suggesting that the model may overestimate convective transport in this region? Which aspects of the NAQPMS could be updated to improve such overestimations?
Line 265, Figure 3f: Do the modeled error bars in this figure represent the spatial variability across all grid cells (or observation sites) in each region for each model? If so, why do the observations not include any error bars given that there are multiple measurement sites in each region, too? The text should also discuss the fact that the differences between models are much smaller than the spread of the error bars.
Line 281 and Figure 5: please clarify whether these contributions are to rainwater concentrations or wet deposition fluxes.
Lines 284 – 285: “Both of these … underestimate …” . Please explicitly reference the tables and/or figures as well as the region(s) this statement is based on.
Line 287: “The Ge14-BWSC scheme … delivers the closest consistency …” Please explicitly reference the tables and/or figures as well as the region(s) this statement is based on.
Lines 288 – 289: Please briefly summarize the method used by these previous studies to estimate the observed below cloud vs. in cloud contributions.
Line 294: What are the uncertainties associated with the method used to estimate the observed below cloud vs. in cloud contributions? For example, is the uncertainty of the 65.3% observation-based estimate for Nox small enough to discern whether the Ge14-BWSC estimate of 65.5% or the Ge14-CAMx estimate of 51.7% is in better agreement with the observation-based estimate?
Line 311: remove “but” before “this only”
Line 330: Please clarify what is depicted by the vectors in panels i – j. Winds at 10m, the first model layer, some other layer? Please also provide a legend relating the length of the vectors to wind speed.
Line 339: Capitalize “the” at the beginning of the sentence
Line 342: [winter] “wet deposition decreases there” compared to what – summer? other regions?
Line 351: Figure S9 is missing a color scale bar.
Line 353: “This study proposes a new physically sound below-cloud scavenging scheme”: The fact the summary begins with this statement reinforces my earlier comment of strongly suggesting that the description of this new scheme should be moved from Sec. S1 to the main manuscript. It is a core part of this study.
Lines 353 – 354: “effective fixes the systematic biases” Please be more specific when and where improved biases were found.
Line 359: suggest changing “emphasized by” to “implemented in”
Line 360: suggest briefly mentioning any uncertainties associated with observation-based estimates of below-cloud scavenging coefficients and future research needs to further constrain such estimates.
Lines 363 – 364: suggest rephrasing “avoiding the situation that the below-cloud deposition relies” to “instead of below-cloud deposition relying”
Lines 366 – 367: change “updated mechanisms reasonably produce” to “updated mechanism reasonably produces”
Line 371: change “mechanisms reshape” to “mechanism reshapes”
Lines 381 – 382: While I realize that the preprint is targeted for ACP rather than GMD, the updates to the NAQPMS code play a central role in the analysis presented here. As such, the original and updated model code should be made publicly available rather than relying on email correspondence with the author.
SI line 39: change “differs from thermophoresis” to “Different from thermophoresis”
SI line 172, Figure S9: See comment above about adding a color bar.