the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 May 2026
BoxMART v1.1 – Box Model for Atmospheric Research and Teaching version 1.1
Abstract. A new open-source chemistry and aerosol box model BoxMART (Box Model for Atmospheric Research and Teaching) has been developed, aiming for easy and flexible configuration for various simulation cases, as well as suitable to teach and learn complex chemistry and aerosol processes. The model is written in Fortran and stored as a gitlab repository. The chemistry module is handled by KPP (Kinetic PreProcessor) which converts the chemistry scheme in an adapted format to Fortran code. The aerosol module currently includes coagulation, condensational and dissolutional growth, evaporation and aerosol thermodynamic processes. The MOSAIC (Model for Simulating Aerosol Interactions and Chemistry) model has been implemented to calculate the non-equilibrium dynamic gas-particle partitioning of non-volatile and semivolatile species, including H2SO4, HNO3, HCl, NH3, and organic compounds to multiple particle size bins. The current ion system includes H+, NH4+, Na+, SO42–, NO3–, Cl– and H2O, where pH value in each size bin is also calculated. The particle size distribution is discretized with a fully stationary method. In order to provide good examples for evaluating BoxMART, as well as future model development, a number of benchmark simulation cases from previous studies ranging from simple to complicated cases, have been revised in details and simulated to verify different processes.
- Preprint
(3566 KB) - Metadata XML
-
Supplement
(104 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-2381', Anonymous Referee #1, 15 Jul 2026
-
RC2: 'Comment on egusphere-2026-2381', Anonymous Referee #2, 02 Sep 2026
This manuscript presents an open-source atmospheric chemistry and aerosol box model called BoxMART, which couples gas-phase chemistry (processed through KPP into Fortran code) with the aerosol module MOSAIC to achieve integrated atmospheric chemistry and aerosol simulations. The primary intended applications of this model are atmospheric chemistry research and teaching. Overall, while substantial effort has been invested in the engineering implementation of the model code and the compilation of benchmark cases, the manuscript still exhibits notable deficiencies in scientific novelty, model architecture clarity, and teaching applicability. In its current form, I recommend a major revision.
(1) The title is overly broad and fails to convey the core functional features that distinguish BoxMART from other box models. Furthermore, after describing the model's capabilities, the abstract does not clearly articulate the original contributions of this work. The current manuscript gives the overall impression of being an integration and repackaging of existing code rather than an independently innovative model development effort; the authors need to explicitly define BoxMART's original contributions.
(2) The authors mention that BoxMART is developed based on SOSAA-Box, but they neither provide any basic introduction to SOSAA-Box nor explain the key differences between BoxMART and SOSAA-Box. It is recommended that a paragraph be added to the introduction to clarify the inheritance and distinctions. Additionally, although the abstract claims that the model can handle gas-particle partitioning for multiple species including organic compounds, the introduction focuses primarily on inorganic thermodynamic modules such as E-AIM and ISORROPIA, while providing almost no review of recently developed organic aerosol modules. The authors should supplement the introduction with a description and comparison of organic models such as SOAP, CSVA, and UNIPAR, to clarify BoxMART's position within the existing model landscape and its limitations in organic aerosol treatment.
(3) The abstract states that MOSAIC is used to calculate non-equilibrium dynamic gas-particle partitioning, while the methods section indicates that condensational/dissolutional growth follows Jacobson's APC/APD schemes. This creates a serious inconsistency. According to the original design, the ASTEM submodule within MOSAIC is specifically responsible for the kinetic solution of gas-particle mass transfer. If BoxMART employs APC/APD instead of ASTEM, then what specific function does MOSAIC serve in BoxMART? Is it only the thermodynamic module being called to calculate activity coefficients, water content, and pH, while the kinetic mass transfer is entirely handled by APC/APD? Moreover, the authors do not clarify whether the three modules (gas-phase chemistry, gas-particle mass transfer, and thermodynamic equilibrium) are coupled online or offline, nor do they specify the calling sequence within the main time step, the type of ODE solver, time step settings, or computational efficiency. The authors should add a section or expand the flowchart to detail the module calling sequence, data exchange format, and mass conservation checks.
(4) The authors use 14 cases to verify model performance, and it is recommended that the purpose of each case be clearly indicated in the corresponding headings. In Case J1997_APC_TEST in Section 3.1, the species C1 and CORE are introduced without specifying their exact chemical nature; although this is an idealized test, a clear explanation would still facilitate reader understanding.
(5) On page 3, the authors mention that a unit conversion error in the Kelvin equation in the original MOSAIC code has been corrected, which represents an important technical contribution of this work. However, the description is too brief. It is recommended that the mathematical expressions before and after the correction be compared in the main text or appendix, and that a quantitative figure be added to illustrate the impact of this bug.
(6) In Section 3.2, the APD test does not provide a numerical comparison table with the results from the original Jacobson reference, unlike the APC test; this should be supplemented.
(7) The results of Case Z2008_CASE14_NEW show that under high RH (85%) conditions, particles smaller than 25 nm remain solid, which contradicts conventional understanding and reveals a missing physical mechanism in the current model regarding phase transitions at the nanoscale. The authors should thoroughly discuss the origin of this limitation, clarify its impact on the reliability of Aitken-mode simulations, and outline future improvement directions in the conclusions.
(8) Although the introduction qualitatively compares models such as ARCA, PyCHAM, MAFOR, and ADCHAM, no quantitative comparison is presented in the main text. It is recommended that the authors conduct a quantitative comparison with at least one or two existing open-source box models under identical benchmark cases, in terms of both numerical results and computational efficiency, to demonstrate BoxMART's relative advantages.
(9) The authors repeatedly emphasize that BoxMART is suitable for teaching in the title, abstract, and introduction, but the manuscript provides no concrete materials or design features to support this claim. It is unclear whether self-explanatory input files or features specifically designed to lower the learning curve are provided. If the current version does not yet include such teaching support, the authors should clearly outline the future development plan in the conclusions and appropriately moderate the relevant claims in the abstract and introduction to avoid creating unrealistic expectations for readers.
Citation: https://doi.org/10.5194/egusphere-2026-2381-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 260 | 132 | 47 | 439 | 49 | 43 | 45 |
- HTML: 260
- PDF: 132
- XML: 47
- Total: 439
- Supplement: 49
- BibTeX: 43
- EndNote: 45
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript presents BoxMART v1.1, an open-source box model combining gas-phase chemistry handled through KPP with aerosol dynamical and thermodynamic processes. The aerosol module includes coagulation, condensation, dissolution, evaporation, and the MOSAIC framework for size-resolved dynamic gas–particle partitioning and inorganic aerosol thermodynamics. The model is evaluated using a substantial collection of benchmark cases derived from earlier studies by Jacobson and co-workers and by Zaveri et al.
The development of a documented and openly available implementation of these processes may be useful to the atmospheric modelling community. In particular, the effort to reconstruct benchmark cases that were incompletely documented in the original publications, to provide corresponding input files, and to identify or correct errors in earlier descriptions or implementations could constitute a valuable community resource. The correction of the Kelvin-effect calculation inherited from the MOSAIC implementation may also represent an important technical contribution.
However, in its present form, I do not think the manuscript yet establishes sufficiently clearly the novelty and specific role of BoxMART relative to existing aerosol box models. Most of the physical and numerical methods implemented in BoxMART are taken from previously published algorithms, and several existing models already provide closely related capabilities. A particularly important omission is SSH-aerosol, which is directly relevant to the scope of this manuscript and also includes sectional aerosol dynamics, coagulation, dynamic condensation and evaporation, thermodynamic partitioning, the Kelvin effect, gas-phase chemistry, and numerous benchmark cases.
In addition, some of the benchmark results still show significant numerical diffusion or differences from the published reference results, and the manuscript itself states that further verification is required for several cases. The distinction between verification against analytical solutions, reproduction of previously published model results, and evaluation of reconstructed cases therefore needs to be made more carefully.
I consider the model and benchmark effort potentially publishable, but substantial revisions are required to clarify the contribution of BoxMART, position it objectively within the existing modelling landscape, quantify its numerical performance, and support the claims made regarding research and teaching applications.
Major comments
The manuscript repeatedly presents BoxMART as a new chemistry and aerosol box model, but the specific novelty of the work remains difficult to identify. The aerosol algorithms are largely based on Jacobson et al. for coagulation and condensation/dissolution, while the aerosol thermodynamic and dynamic partitioning calculations are based on MOSAIC. The gas-phase chemistry is handled using KPP, and the overall model is developed from SOSAA-Box.
This does not mean that the work lacks value. Integrating these components into an accessible standalone framework may be useful. However, the manuscript should explicitly distinguish between: new physical or numerical developments; a new software architecture or interface between existing components; corrections made to existing code; reconstruction and standardisation of historical benchmark cases; features introduced specifically for teaching or model development.
At present, these different types of contributions are combined under the general statement that a “new model” has been developed. I recommend adding a clear paragraph at the end of the Introduction stating precisely what is new in BoxMART v1.1 and what is inherited from existing models.
2. The comparison with existing aerosol box models is incomplete
The Introduction discusses ARCA, PyCHAM, MAFOR, and ADCHAM, but omits other directly relevant models. In particular, SSH-aerosol should be cited and discussed:
Sartelet et al. (2020), “SSH-Aerosol v1.1: A Modular Box Model to Simulate the Evolution of Primary and Secondary Aerosols”, Atmosphere, 11, 525.
Sartelet et al. (2026), “Advanced modeling of gas chemistry and aerosol dynamics with SSH-aerosol v2.0”, Geoscientific Model Development, 19, 389-421.
SSH-aerosol is a modular sectional box model that includes gas-phase chemistry, coagulation, dynamic condensation and evaporation, inorganic and organic thermodynamics, the Kelvin effect, nucleation, and size-bin redistribution. It also provides a substantial suite of test cases and may be operated as a standalone model or coupled to three-dimensional models.
Importantly, the urban coagulation case based on Zhang et al. (1999), reproduced here as J1994_CASE5, is also already used as a benchmark in SSH-aerosol. The use of a common benchmark is not problematic (in fact, common benchmarks are highly valuable) but previous implementations should be acknowledged, and the results could usefully be compared across models.
I suggest adding a comparison table including at least BoxMART, SSH-aerosol, MAFOR, ADCHAM, ARCA, and PyCHAM. Relevant comparison criteria could include: treatment of gas-phase chemistry;
sectional, modal, or particle-resolved representation; stationary versus moving size sections; coagulation; dynamic versus equilibrium gas–particle partitioning; inorganic and organic thermodynamics; particle pH; nucleation; mixing-state representation; particle-phase chemistry;
wall losses; coupling to three-dimensional models; availability of benchmark cases; graphical or educational tools; code language, installation procedure, and licence.
3. “Validation” and “verification” should be distinguished
The manuscript refers to the benchmark cases as validating the aerosol dynamic processes. Most of the cases instead constitute numerical verification tests, because the results are compared with analytical solutions or previously published model calculations rather than independent experimental observations.
I recommend consistently distinguishing among:
4. Several benchmark cases remain unresolved
The manuscript acknowledges that BoxMART results differ from those of Jacobson and Turco (1995) for several F5 and F6 sub-cases and concludes that further verification is still needed. This is an important limitation because these cases are presented as part of the verification of the condensation and coagulation implementation.
The authors should investigate these discrepancies further before making a general statement that the corresponding model processes have been verified. At a minimum, they should identify whether the differences arise from incomplete documentation in the original study, different coagulation kernels, different size-bin treatments, time-step choices, or implementation errors; provide complete numerical configurations for both the reference and BoxMART calculations; clearly label unresolved cases as such; avoid including them among successfully passed benchmarks unless quantitative justification is provided.
5. Numerical diffusion associated with the stationary-bin method requires a more systematic analysis
The results show that the fully stationary sectional method may introduce substantial numerical diffusion, particularly when particle growth dominates. In the complex MOSAIC cases, this diffusion appears to merge initially distinct modes and may even allow chemical interactions between particle populations that would otherwise remain separated.
This is not only a graphical issue: numerical diffusion can affect particle composition, gas–particle partitioning, pH, and the formation of salts. It therefore needs to be quantified systematically.
I recommend adding sensitivity tests addressing: the number of size bins; the volume ratio between adjacent bins; the aerosol-dynamics time step; the redistribution frequency; the treatment of the lower and upper size boundaries.
For selected cases, results could also be compared with a moving-section calculation or another remapping algorithm. The aim does not need to be to eliminate numerical diffusion entirely, but to establish the resolution required for acceptable accuracy and the computational cost associated with it.
6. The correction to the Kelvin-effect calculation should be documented more rigorously
The manuscript states that a unit-conversion error was identified in the Kelvin equation in the original MOSAIC code and corrected following communication with one of the MOSAIC developers. This may be one of the most important technical contributions of the study, but the description is currently too brief.
7. The claimed teaching function is not yet demonstrated
The stated purpose of BoxMART includes both atmospheric research and teaching. However, the manuscript primarily describes a Fortran model and a series of benchmark simulations. It does not yet demonstrate what makes the model particularly suitable for teaching compared with other open-source box models.
To support this claim, the authors should describe the pedagogical material included with the model.
8. The demonstrated model scope is narrower than suggested in the Abstract
The Abstract refers to a chemistry and aerosol model, including the dynamic partitioning of inorganic and organic compounds. However, the benchmark section focuses mainly on aerosol coagulation, condensation/dissolution, and inorganic thermodynamics. The manuscript provides little demonstration of the KPP chemistry interface or of the treatment of organic aerosol compounds.
The claims in the Abstract, Introduction, and Conclusions should be restricted to the capabilities actually demonstrated in the manuscript.