Particle dry deposition algorithms in CMAQ version 5.3: characterization of critical parameters and land use dependence using DepoBoxTool version 1.0
Abstract. This study investigates particle dry deposition by characterizing critical parameters and land-use dependence in a 0-D box model as well as quantifying the resulting impact of dry deposition parameterizations on regional-scale 3-D model predictions. A publicly available box model configured with several land-use dependent dry deposition schemes is developed to evaluate predictions of several model approaches with available measurements. The 0-D box model results suggest that current dry deposition schemes in 3-D regional models underestimate particle dry deposition velocities, but this varies with size distribution properties and land-use categories. We propose two revised schemes to improve dry deposition performance in air quality models and test them in the Community Multiscale Air Quality (CMAQ) model. The first scheme improves the previous CMAQ scheme by preserving the original dry deposition impaction calculation but turning off redundant integration across particle size for each aerosol mode. The second scheme adds a dependence on leaf area index (LAI) to better estimate uptake to vegetative surfaces while using a settling velocity that is integrated across particle size for the Stokes number calculation. CMAQ model performance was evaluated for a month in July 2011 for the conterminous U.S. based on available observations of ambient sulfate (SO4) aerosol concentrations from multiple routine particulate matter monitoring networks. Incorporation of the first scheme has a larger impact on coarse particles than fine particles, systematically reducing monthly domain-wide average particle dry deposition velocities (Vd) by approximately 96% and 35%, respectively, and increasing monthly average SO4 concentrations by 395% and 21%. After incorporating LAI into the boundary layer resistance (Rb), the second scheme creates more spatial diversity of Vd and changes SO4 concentrations (coarse = −76% to +336%; fine = −7% to +18%) with land-use categories. These modifications are incorporated into the current publicly available version of CMAQ (v5.3 and beyond).
This preprint has been withdrawn.
Qian Shu et al.
Qian Shu et al.
Qian Shu et al.
Viewed (geographical distribution)
This study compares a few formulations of particle dry deposition algorithms in both a stand-alone process model and as implemented in CMAQ. Although the topic of particle dry deposition is an important one, this manuscript fails in many ways to significantly advance our understanding of this subject, and falls far short of the quality of science and its presentation that this reviewer has come to expect from the CMAQ group at EPA. In light of these failures, I recommend that this manuscript be rejected, in the hope that both the science and the presentation can be improved for future resubmission.
Overall, the writing and presentation in this paper is confusing and poorly done. Given the complexity of the subject, it is vital that the writing and English usage not make it even more difficult for the reader to understand the work performed and the choices that were made in the study. It would be too time consuming to provide a detailed list of all instances of awkward wording or phrasing, but a particular example is lines 199-206 on p. 6, which is difficult, if not impossible, to comprehend. And, in fact, this description is extremely germaine to the focus of the entire paper! Thus, my first suggestion is a complete overhaul of the manuscript and detailed editing to make it comprehensible.
From a science perspective, I question many of the details of the study and a lack of justification for some of the choices made. In particular:
- Why did the authors bother to include the Zhang et al. scheme in their analysis? They only compare it against the Pleim & Ran (PR) variations in the DepoBox process model and did not implement it into CMAQ. The only conclusion that can be drawn from this comparison, is that, yes, it produces very different deposition velocities than the PR scheme does. It seems clear that the real motivation for this work is to simply improve the PR scheme in CMAQ and not do any meaningful comparison between the Zhang and PR schemes, so why include it?
- In describing the current PR scheme in CMAQ (Section 2.1.2), they fail to explain many of the oddities of the scheme that have a large impact on the calculated deposition velocities. Some of these oddities include:
(i) the use of the 0.95 factor when calculating aerodynamic resistance (Eq. 16) - Why is this factor used?;
(ii) the use of the convective velocity scale, w*, is unique to the PR scheme - the value of this quantity has a huge impact on the calculated Vd, especially over a diurnal cycle, but the importance of this fact is not addressed (except for the assumed values presented in Figure S-2);
(iii) the value of 400 in the impaction efficiency formula is said to have been chosen "to better represent aerosol deposition to heavily vegetated regions" - Is this based on a previous (uncited) analysis or what?
(iv) the interception efficiency is ignored in the PR algorithm because "it is difficult to specify realistic estimates of these parameters over the area of typical grid cells" - Ignoring a term because it is difficult to estimate is a very weak justification;
(v) the use of Eq. 21 as the original impaction efficiency is said to have been "in order to facilitate incorporation of the integrated Stokes number" - It's not clear exactly what this means; was it done to make the modal integration easier? Or something else? As presented in the supplement, other choices for the impaction efficiency formula could have been made to make the integration simpler, but they all are somewhat arbitrary;
- In Section 2.1.3 Proposed Schemes, there are several issues with either insufficient (or unintelligible) discussion or unjustified model modifications:
(i) As already mentioned, lines 199-206 on p. 6 are nearly incomprehensible, but are key to understanding the results the paper presents. I have only a vague understanding of what was done with the impaction efficiency in the "OFF" scheme and why it was modified;
(ii) For the VGLAI scheme, yet another form of the impaction efficiency is given (Eq. 22), this one with a value of 1 instead of 400 in the denominator. No justification for this change is given;
(iii) A leaf area index factor is introduced into the formulation of the boundary layer resistance (Eq. 24) for VGLAI, but no justification for the form of this factor is given or discussed. Where did this come from and what is its intended effect?
- In line 223 on p. 7, it is laughable to state that "a _comprehensive_ evaluation of particle dry deposition schemes discussed in Section 2.1" has been conducted. Three observational studies over only three land use types is not "comprehensive" in my way of thinking.
- The comparisons in Figs. 1-3 of the DepoBox process model with observations do not inspire much confidence in the performance of _any_ of the schemes and make comparisons of the CMAQ implementations with observations in Figure 9 rather questionable. Comparison of results from the PR-based schemes in CMAQ are potentially interesting, but without a more detailed (indeed, comprehensive!) evaluation against observations, it is hardly more than a numerical exercise to see what all of the arbitrary changes made to the algorithms result in changes to Vd. The authors' selection of the VGLAI scheme as the "most applicable for predicting particle dry deposition over grass and coniferous forests", while potentially true, is not justified by the results presented in this manuscript.