the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Jun 2023
Impacts of dynamic dust sources coupled with WRF-Chem 3.9.1 on the dust simulation over East Asia
Abstract. Dust emission refers to the spatial displacement process of soil particles with the influence of wind. The quantitative and accurate description of dust emission is the basis of dust simulation in the modeling. The previous studies always employed static land cover in the numerical models, ignoring dynamic variations in the surface bareness and leading to large uncertainties in the dust simulation. We build six sets of dynamic dust sources functions, which shows a pronounced monthly and annual variability with the influence of seasonal change. Compared that in July, the dynamic dust source in March shows an expanding pattern to the edge of the deserts. Moreover, the dust source function in the Taklimakan Desert and Gobi Desert decrease at an annual rate of 2.42 × 10-4 and 3.06 × 10-4. The Weather Research and Forecasting model coupled to Chemistry (WRF-Chem) coupled with dynamic dust sources can effectively reproduce the spatiotemporal distribution of aerosol within satellite and ground-based observations. Our results show that the surface bareness and topographic characteristics jointly control the spatial distribution and value of dynamic dust sources. Further, the dynamic change of dust source further affects the dust emission and dust cycle. This study highlights the importance of surface bareness and the topographic characteristics on the dynamic dust source, and effectively improves dust cycle simulation over East Asia.
- Preprint
(3588 KB) - Metadata XML
-
Supplement
(28 KB) - BibTeX
- EndNote
Status: closed
-
CC1: 'Comment on gmd-2023-81', Sandra LeGrand, 18 Jul 2023
Thank you for sharing your research.
1) P2 L61: “The accuracy of dust emission simulation mainly depends on the spatial distribution of dust sources” – do you have a reference for this statement? I’d suggest removing this comment otherwise. Dust emission simulation errors can emanate from several sources (e.g., erroneous wind or environmental forcing condition patterns, erroneous soil/terrain/vegetation characterization, poorly or overly tuned empirical constants, etc.).
2) Sect. 2.3: Which WRF-Chem dust emission module is this section describing? Based on the description provided, it appears the authors used the dust_opt = 13 WRF-Chem namelist option, which is an adapted version of the GOCART dust emission scheme coupled with the MADE/SORGAM and MOSAIC aerosol scheme. If so, please reference the model description and publication by Zhao et al. (2010). Usually, when WRF-Chem users reference the GOCART Ginoux et al. (2001) scheme (e.g., Table 1), they’re referring to the dust_opt = 1 setting. If the authors added new dust emission equations to the WRF-Chem code (i.e., if the authors did not use dust_opt = 1, 2, 3, 4, or 13), please state that in the manuscript.
Please note that if the authors did use dust_opt = 13, the dust emission code for this study deviates from the original formulation described in the paper by Ginoux et al. (2001). These modifications are described in detail in Sect. 3.1.2 of LeGrand et al. (2019). While the LeGrand et al. (2019) overview was written specifically for dust_opt = 1, most of these same changes also apply to the dust_opt = 13 setting (see function sorgam_source_du in module_aerosols_sorgam.F in the chem directory of the WRF-Chem source code). The most notable change is the switch from a 10m wind speed-based threshold (ut) to one derived in terms of wind friction speed (u*t). This modification leads to spurious dust emissions under very low wind speeds since u10m >> u*t. The dust_opt = 13 setting from WRF-Chem v3.9.1 also includes a degree of saturation value (θs) threshold (gwet in the code) that restricts dust emissions anywhere θsexceeds 0.2 (i.e., very dry conditions). This aspect of the dust emission module may be important to the authors' conclusions given the emphasis on green vegetation with NDVI.
Effectively, the dust emission flux (G) in dust_opt = 13 more closely resembles the following:
G = CSsp (u10m)3 if θs < 0.2; G = 0, otherwise.
3) This study used an older version of WRF-Chem. There was an important bug fix added to the WRF-Chem dust gravitational settling scheme in version 4.1 (see Ukhov et al. 2021 and https://github.com/wrf-model/WRF/commit/2ffdebf4ac311a5b1ef8cd0c639e0d857b550fdb). The error causes dust to fall out of the atmosphere too quickly. While redoing the experiment with a newer model version may not be necessary, how might this error affect the interpretation of the results?
References:
LeGrand, S. L., Polashenski, C., Letcher, T. W., Creighton, G. A., Peckham, S. E., and Cetola, J. D.: The AFWA dust emission scheme for the GOCART aerosol model in WRF-Chem v3.8.1, Geosci. Model Dev., 12, 131–166, https://doi.org/10.5194/gmd-12-131-2019, 2019.Ukhov, A., Ahmadov, R., Grell, G., and Stenchikov, G.: Improving dust simulations in WRF-Chem v4.1.3 coupled with the GOCART aerosol module, Geosci. Model Dev., 14, 473–493, https://doi.org/10.5194/gmd-14-473-2021, 2021.Zhao, C., Liu, X., Leung, L. R., Johnson, B., McFarlane, S. A., Gustafson Jr., W. I., Fast, J. D., and Easter, R.: The spatial distribution of mineral dust and its shortwave radiative forcing over North Africa: modeling sensitivities to dust emissions and aerosol size treatments, Atmos. Chem. Phys., 10, 8821–8838, https://doi.org/10.5194/acp-10-8821-2010, 2010.Citation: https://doi.org/10.5194/gmd-2023-81-CC1 -
RC1: 'Comment on gmd-2023-81', Anonymous Referee #1, 04 Aug 2023
Chen et al. present a vegetation-dependent – and hence dynamic – dust source function to use with the GOCART dust emission scheme in WRF-Chem. Introducing these dynamic dust sources, the authors aim to address a supposed long-standing neglect of variations in surface bareness in dust modeling (Abstract, lines 16-17). While the subject of the manuscript is highly relevant and there are still important unknowns concerning the representation of dust sources and dust emission in models, I unfortunately do not see much novelty in the presented manuscript for the reasons detailed in the following.
The claim that dust models usually neglect variability in surface bareness is not correct. Dust models have been considering variations in surface bareness, particularly due to vegetation coverage, for a long time, e.g. Tegen et al. (2002), Zender et al. (2003), …, Klose et al. (2021), Leung et al. (2023). However, the influence of dynamic vegetation is not necessarily implemented in a preferential dust source function as done in the present paper, but is used separately either as a multiplicative factor or as a correction function to the threshold friction velocity or friction velocity (drag partitioning). If vegetation is treated that way, a static dust source function is indeed sufficient for those schemes that use it. There are also schemes that do not use preferential source functions, but aim to explicitly represent the land-surface properties and their impacts on dust emissions.
Specific for use with the GOCART dust emission formulation by Ginoux et al. (2001) – which originally indeed does not consider dynamic vegetation – the new dynamic source functions may still be very useful. However, the dynamic source functions presented here are in fact not new, but have already been presented by Kim et al. (2013, 2017).
Applying the dynamic source functions, the authors then present sensitivity experiments investigating the impact of those functions on the dust cycle in East Asia. Unfortunately, the discussion of results of this part remains very descriptive and does not go into enough detail to provide new insights.
From a more technical perspective, very little information is provided about the simulation setup and it is not clear to me how the simulated dust deposition fluxes can be about two orders of magnitude larger than the dust emission fluxes. Those should typically be on the same order of magnitude.
I hope that the authors will keep their motivation to advance dust modeling in the future.
Citation: https://doi.org/10.5194/gmd-2023-81-RC1 -
RC2: 'Comment on gmd-2023-81', Anonymous Referee #2, 16 Sep 2023
Review Comments for the manuscript “Impacts of dynamic dust sources coupled with WRF-Chem 3.9.1 on the dust simulation over East Asia” by Chen et al.
The authors attempt to improve the dust emissions and transport capability of WRF-Chem for East Asia, by changing the characterization of dust sources. This is accomplished by using an NDVI dataset to estimate surface bareness, basically the way aridity is represented in the dust emission scheme.
I have many comments and concerns with the manuscript. My first concern is that the “dynamic dust source” that the authors refer to, is not substantiated in the manuscript. The dynamic nature of a model input can be temporal or spatial, or hopefully both. If the main advantage is the monthly variation of bareness from NDVI (lines 101-102), since the WRF-Chem simulations are essentially for one month only (March 2020), it is impossible to assess how this addition improved dust prediction and also can be named “dynamic dust source”.
Second, dust modeling requires a delicate description of how dust particles move horizontally (saltation) and vertically (sandblasting, entrainment, disintegration) and stay in the air, their origin (soil texture, and particle size classification), the dust particle size distribution during atmospheric transport, and, of course, atmospheric conditions. Most atmospheric models (global or regional) that simulate the dust cycle, use some characterization of the aridity of the area that changes temporally and spatially. I don’t see how this work can be considered model development, which is the core mission of GMD. The manuscript is mostly representing sensitivity simulations with WRF-Chem, by changing one input parameter that affects dust. If I have misunderstood the authors work, I argue that they should be more explicit on the contribution of their work towards model development.
Third, the manuscript lacks details on the dust emission scheme, specifically how the source function S is calculated, how are sp and ut estimated. My guess is that ut is the threshold friction velocity which is parameterized somehow. All these components must be clearly described in the text, to allow the reader to understand how the authors’ addition influences the dust emission scheme.
The evaluation of the dust simulation also lacks robustness. Even though AOD is a very important component, the evaluation must also include dust concentrations or emissions, at least some PM10 measurements that are more readily available. The calculation of AOD depends on how dust is emitted, but there are other aerosol optical characteristic components that dilute a direct evaluation of the dust emission scheme. The same stands for dust deposition.
Limited area models like WRF-Chem face other constraints, such as the lateral and initial boundary conditions that influence dust production and transport processes. How did those constraints influence the WRF-Chem simulations?
In line 150, the authors mention that “GOCART also has been widely welcomed by various numerical models and show excellent performance on dust emission over East Asia (Chen et al., 2014, 2017).” If the performance is excellent, what is the point of this work?
Finally, the manuscript needs thorough English grammar editing. There are many instances in the text that the tense is wrong, there is a use of plural instead of singular nouns (e.g. line 128 “WRF-Chem Models”) and other grammatical errors.
I believe the manuscript needs extensive revisions to reach the standards of GMD and be considered for publication. I urge the authors to follow the comments and suggestions and improve the quality of the paper.
Citation: https://doi.org/10.5194/gmd-2023-81-RC2 -
AC1: 'Comment on gmd-2023-81', Siyu Chen, 13 Nov 2023
The comment was uploaded in the form of a supplement: https://gmd.copernicus.org/preprints/gmd-2023-81/gmd-2023-81-AC1-supplement.pdf
Status: closed
-
CC1: 'Comment on gmd-2023-81', Sandra LeGrand, 18 Jul 2023
Thank you for sharing your research.
1) P2 L61: “The accuracy of dust emission simulation mainly depends on the spatial distribution of dust sources” – do you have a reference for this statement? I’d suggest removing this comment otherwise. Dust emission simulation errors can emanate from several sources (e.g., erroneous wind or environmental forcing condition patterns, erroneous soil/terrain/vegetation characterization, poorly or overly tuned empirical constants, etc.).
2) Sect. 2.3: Which WRF-Chem dust emission module is this section describing? Based on the description provided, it appears the authors used the dust_opt = 13 WRF-Chem namelist option, which is an adapted version of the GOCART dust emission scheme coupled with the MADE/SORGAM and MOSAIC aerosol scheme. If so, please reference the model description and publication by Zhao et al. (2010). Usually, when WRF-Chem users reference the GOCART Ginoux et al. (2001) scheme (e.g., Table 1), they’re referring to the dust_opt = 1 setting. If the authors added new dust emission equations to the WRF-Chem code (i.e., if the authors did not use dust_opt = 1, 2, 3, 4, or 13), please state that in the manuscript.
Please note that if the authors did use dust_opt = 13, the dust emission code for this study deviates from the original formulation described in the paper by Ginoux et al. (2001). These modifications are described in detail in Sect. 3.1.2 of LeGrand et al. (2019). While the LeGrand et al. (2019) overview was written specifically for dust_opt = 1, most of these same changes also apply to the dust_opt = 13 setting (see function sorgam_source_du in module_aerosols_sorgam.F in the chem directory of the WRF-Chem source code). The most notable change is the switch from a 10m wind speed-based threshold (ut) to one derived in terms of wind friction speed (u*t). This modification leads to spurious dust emissions under very low wind speeds since u10m >> u*t. The dust_opt = 13 setting from WRF-Chem v3.9.1 also includes a degree of saturation value (θs) threshold (gwet in the code) that restricts dust emissions anywhere θsexceeds 0.2 (i.e., very dry conditions). This aspect of the dust emission module may be important to the authors' conclusions given the emphasis on green vegetation with NDVI.
Effectively, the dust emission flux (G) in dust_opt = 13 more closely resembles the following:
G = CSsp (u10m)3 if θs < 0.2; G = 0, otherwise.
3) This study used an older version of WRF-Chem. There was an important bug fix added to the WRF-Chem dust gravitational settling scheme in version 4.1 (see Ukhov et al. 2021 and https://github.com/wrf-model/WRF/commit/2ffdebf4ac311a5b1ef8cd0c639e0d857b550fdb). The error causes dust to fall out of the atmosphere too quickly. While redoing the experiment with a newer model version may not be necessary, how might this error affect the interpretation of the results?
References:
LeGrand, S. L., Polashenski, C., Letcher, T. W., Creighton, G. A., Peckham, S. E., and Cetola, J. D.: The AFWA dust emission scheme for the GOCART aerosol model in WRF-Chem v3.8.1, Geosci. Model Dev., 12, 131–166, https://doi.org/10.5194/gmd-12-131-2019, 2019.Ukhov, A., Ahmadov, R., Grell, G., and Stenchikov, G.: Improving dust simulations in WRF-Chem v4.1.3 coupled with the GOCART aerosol module, Geosci. Model Dev., 14, 473–493, https://doi.org/10.5194/gmd-14-473-2021, 2021.Zhao, C., Liu, X., Leung, L. R., Johnson, B., McFarlane, S. A., Gustafson Jr., W. I., Fast, J. D., and Easter, R.: The spatial distribution of mineral dust and its shortwave radiative forcing over North Africa: modeling sensitivities to dust emissions and aerosol size treatments, Atmos. Chem. Phys., 10, 8821–8838, https://doi.org/10.5194/acp-10-8821-2010, 2010.Citation: https://doi.org/10.5194/gmd-2023-81-CC1 -
RC1: 'Comment on gmd-2023-81', Anonymous Referee #1, 04 Aug 2023
Chen et al. present a vegetation-dependent – and hence dynamic – dust source function to use with the GOCART dust emission scheme in WRF-Chem. Introducing these dynamic dust sources, the authors aim to address a supposed long-standing neglect of variations in surface bareness in dust modeling (Abstract, lines 16-17). While the subject of the manuscript is highly relevant and there are still important unknowns concerning the representation of dust sources and dust emission in models, I unfortunately do not see much novelty in the presented manuscript for the reasons detailed in the following.
The claim that dust models usually neglect variability in surface bareness is not correct. Dust models have been considering variations in surface bareness, particularly due to vegetation coverage, for a long time, e.g. Tegen et al. (2002), Zender et al. (2003), …, Klose et al. (2021), Leung et al. (2023). However, the influence of dynamic vegetation is not necessarily implemented in a preferential dust source function as done in the present paper, but is used separately either as a multiplicative factor or as a correction function to the threshold friction velocity or friction velocity (drag partitioning). If vegetation is treated that way, a static dust source function is indeed sufficient for those schemes that use it. There are also schemes that do not use preferential source functions, but aim to explicitly represent the land-surface properties and their impacts on dust emissions.
Specific for use with the GOCART dust emission formulation by Ginoux et al. (2001) – which originally indeed does not consider dynamic vegetation – the new dynamic source functions may still be very useful. However, the dynamic source functions presented here are in fact not new, but have already been presented by Kim et al. (2013, 2017).
Applying the dynamic source functions, the authors then present sensitivity experiments investigating the impact of those functions on the dust cycle in East Asia. Unfortunately, the discussion of results of this part remains very descriptive and does not go into enough detail to provide new insights.
From a more technical perspective, very little information is provided about the simulation setup and it is not clear to me how the simulated dust deposition fluxes can be about two orders of magnitude larger than the dust emission fluxes. Those should typically be on the same order of magnitude.
I hope that the authors will keep their motivation to advance dust modeling in the future.
Citation: https://doi.org/10.5194/gmd-2023-81-RC1 -
RC2: 'Comment on gmd-2023-81', Anonymous Referee #2, 16 Sep 2023
Review Comments for the manuscript “Impacts of dynamic dust sources coupled with WRF-Chem 3.9.1 on the dust simulation over East Asia” by Chen et al.
The authors attempt to improve the dust emissions and transport capability of WRF-Chem for East Asia, by changing the characterization of dust sources. This is accomplished by using an NDVI dataset to estimate surface bareness, basically the way aridity is represented in the dust emission scheme.
I have many comments and concerns with the manuscript. My first concern is that the “dynamic dust source” that the authors refer to, is not substantiated in the manuscript. The dynamic nature of a model input can be temporal or spatial, or hopefully both. If the main advantage is the monthly variation of bareness from NDVI (lines 101-102), since the WRF-Chem simulations are essentially for one month only (March 2020), it is impossible to assess how this addition improved dust prediction and also can be named “dynamic dust source”.
Second, dust modeling requires a delicate description of how dust particles move horizontally (saltation) and vertically (sandblasting, entrainment, disintegration) and stay in the air, their origin (soil texture, and particle size classification), the dust particle size distribution during atmospheric transport, and, of course, atmospheric conditions. Most atmospheric models (global or regional) that simulate the dust cycle, use some characterization of the aridity of the area that changes temporally and spatially. I don’t see how this work can be considered model development, which is the core mission of GMD. The manuscript is mostly representing sensitivity simulations with WRF-Chem, by changing one input parameter that affects dust. If I have misunderstood the authors work, I argue that they should be more explicit on the contribution of their work towards model development.
Third, the manuscript lacks details on the dust emission scheme, specifically how the source function S is calculated, how are sp and ut estimated. My guess is that ut is the threshold friction velocity which is parameterized somehow. All these components must be clearly described in the text, to allow the reader to understand how the authors’ addition influences the dust emission scheme.
The evaluation of the dust simulation also lacks robustness. Even though AOD is a very important component, the evaluation must also include dust concentrations or emissions, at least some PM10 measurements that are more readily available. The calculation of AOD depends on how dust is emitted, but there are other aerosol optical characteristic components that dilute a direct evaluation of the dust emission scheme. The same stands for dust deposition.
Limited area models like WRF-Chem face other constraints, such as the lateral and initial boundary conditions that influence dust production and transport processes. How did those constraints influence the WRF-Chem simulations?
In line 150, the authors mention that “GOCART also has been widely welcomed by various numerical models and show excellent performance on dust emission over East Asia (Chen et al., 2014, 2017).” If the performance is excellent, what is the point of this work?
Finally, the manuscript needs thorough English grammar editing. There are many instances in the text that the tense is wrong, there is a use of plural instead of singular nouns (e.g. line 128 “WRF-Chem Models”) and other grammatical errors.
I believe the manuscript needs extensive revisions to reach the standards of GMD and be considered for publication. I urge the authors to follow the comments and suggestions and improve the quality of the paper.
Citation: https://doi.org/10.5194/gmd-2023-81-RC2 -
AC1: 'Comment on gmd-2023-81', Siyu Chen, 13 Nov 2023
The comment was uploaded in the form of a supplement: https://gmd.copernicus.org/preprints/gmd-2023-81/gmd-2023-81-AC1-supplement.pdf
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 447 | 154 | 28 | 629 | 29 | 16 | 20 |
- HTML: 447
- PDF: 154
- XML: 28
- Total: 629
- Supplement: 29
- BibTeX: 16
- EndNote: 20
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
