Performance and evaluation of remote sensing satellites for monitoring dust weather in East Asia

General Comments

The study presents an interesting approach by integrating various satellite-based aerosol products to assess their utility in near real-time dust monitoring. Comparing products that observe the same region and time frame offers valuable insight into their respective strengths. However, there are several concerns regarding the alignment between the stated research objective and the methodology.

The authors claim to evaluate the performance of remote sensing satellites; however, the methodology does not fully support a comprehensive performance evaluation. While parameters such as AOD, AAI, and dust indices are indeed correlated with PM10 concentrations, such correlations can be highly variable depending on meteorological conditions, aerosol microphysical properties, and vertical distribution, which are not fully considered in the current analysis. The study applies different dust detection criteria to each satellite product and to the ground-based PM10 observations, which introduces inconsistencies in the evaluation process. As the detection thresholds are not standardized, the comparison cannot be considered an “apple-to-apple” evaluation. The performance metrics may significantly change depending on the threshold settings for each product. Furthermore, the study is limited to a short time frame during which specific dust events occurred and is confined to regions with existing ground-based monitoring station. As such, the findings lack generalizability and may not reflect the performance of these satellite products under broader spatial or temporal conditions. The robustness of the evaluation would benefit from a more comprehensive and standardized analysis across a wider range of events and locations.

Through the analysis presented in this study, I believe that the applicability of the findings could be enhanced by proposing a comprehensive approach that utilizes various aerosol products observed from different platforms (GEO/LEO) to monitoring dust over East Asia. In this regard, I recommend authors incorporating and analyzing additional data such as SSA and Aerosol Layer Height provided by current satellites like TROPOMI, GEMS, and EPIC.

Specific Comments

Abstract: Full name of acronym (FY4A, MODIS, and POCD)

Lines 27-29: “When dust weather occurs, … (Mahowald, 2011)”: More references should be added here.

Lines 45-47: “After the 1970s, with the rapid development of various earth observation…": Instead of listing references, I suggest briefly introducing key previous studies closely related to this research and linking them individually to the references. This approach should be applied to many other sections of the manuscript as well.

Lines 53-77: This paragraph introduces the various sensors (distinguishing between GEO/LEO), algorithms, and the variables provided by each algorithm. I recommend to revise it to discuss their strengths and weaknesses and explain the rationale behind the selection of products in this study. For example, this study uses DT and DB products from three available MODIS aerosol products (DT, DB, and MAIAC). VIIRS continues MODIS observations, but the products are not used here. Although Korean Geostationary Satellites such as GEMS, GOCI, and AMI provide aerosol information, they are not used here. Additionally, references are needed for each aerosol product.

Lines 86-87 “Its aerosol classification monitoring and vertical structure are currently the most comprehensive and accurate aerosol product” : On what basis is this defined?

Line 94 “The strongest dust weather”: On what basis is this defined? There are many instances of dust being transported across the Pacific Ocean.

Lines 101-104 “However, the accuracy, stability, and reliability of these satellite remote sensing retrieval products are not clear for dust weather monitoring” : That is not true. Here is a quick example: https://doi.org/10.1002/2015JD024103

Line 116 “Spring (March-May) is the season when dust weather occurs frequently in East Asia”: Need reference.

Lines 116-117 “the frequent activity of cold air in northern East Asia in spring provides a driving force for the formation of dust weather” : What does the ‘cold air activity’ refer to?

Figure 1: It is difficult to distinguish the dust scene from the background surface in this image. Why not provide a 'background RGB image' from a pristine day as a reference?

 2.2.1 FY-4A: Could you summarize the definitions of the various dust indices in a formula or table to allow for a clearer comparison? Which wavelength is used for deriving IDDI?

2.2.2 MODIS: Please revise Modis into MODIS in the section.  

Line 177: remove ‘the United State’

Line 181: remove ‘for free’

Line 183: MODIS provides multiple aerosol products (DT, DB, MAIAC…) and each algorithm provides Lv2 and Lv3 products.

Lines 187-188: “…while the DB algorithm was mainly designed to overcome the poor retrieve results of the DT algorithms in areas with high reflectance.”: The DT algorithm is designed based on its theoretical background to target dark soil and vegetation surface. The Deep Blue algorithm was developed to overcome uncertainties in bright surfaces by utilizing observations from the deep blue channel. As a result, both the DT and DB algorithms perform complementary roles in global aerosol observations.

Line 194: Collection 6.1 is the latest version, though it was not released recently. A new DT GEO-LEO combined products is available here :

ladsweb.modaps.eosdis.nasa.gov/missions-and-measurements/applications/geoleo/

Lines 207-209 “It can effectively observe trace gas components in the atmosphere around the world…”: TROPOMI/S5P instrument provides hyperspectral measurements in UV visibie, NIR and SWIR, which are also advantageous for retrieving aerosol absorptivity (like SSA) and aerosol layer height.

Line 221: aerosol optical thickness  aerosol optical depth

Line 249-251: The sentence here is unclear. What does “the value at 470 nm is higher than the value at 640 nm” means? “…using the spectral dependence of surface reflectance…”: Does it means the surface reflectance relationship suggested in Kaufman et al. (1997)?

2.3 Method: I would recommend summarizing the criteria for strategy for detecting dust pixels in a table or diagram.

Line 261 “…its size”: ‘magnitude’ might be better than ‘size’.

Line 281 “ground environmental monitoring stations”: Authors need to provide the characteristics of the PM10 observation (ex. Retrieval frequency, accuracy, sensor calibration…)

Line 307 “Figure 3”: Maybe Figure 2?

Figure 2: Please indicate the region mentioned in the text on the figure. This will make it easier to follow the discussion.

Figure 2 and Figure 3: I recommend aligning the projection areas to facilitate comparison between the dust index and PM₁₀ concentration. Why not consider combining Figure 2 and Figure 3?

Figure 4: It is recommended that each figure be accompanied by a title. It is questionable whether analyzing trends in this figure is appropriate, given that POCD and POFD are unlikely to vary continuously across space and time.

Lines 355-358 “In order to better prove the application … the atmospheric dust detection capabilities of the FY-4A DSD product.”: This appears to be an unnecessary statement. Removing it will improve conciseness and flow without losing important content.

Lines 370-371 “misjudgment and omission of atmospheric dust detection by satellite remote sensing are inevitable”: The current phrasing incorrectly generalizes limitations to all satellite remote sensing. This should be narrowed to specifically address the FY-4A dust products being discussed.

Line 409 “Generally speaking, when large-scale dust weather occurs, the main pollutant in the atmosphere is dust.” I guess it depends on season, time, and location.

Line 420 “Therefore, compared with the DB algorithm, the DT algorithm is not suitable for AOD retrieval in areas with high surface reflectance.”: The current statement presents an overly simplistic and potentially misleading view of the algorithms' capabilities. While the DB algorithm performs better in areas with bright surface, the DT algorithm provides accurate aerosol products over dark soil, vegetation and ocean surfaces. Each algorithm has its own strengths and limitations, with DT being more limited when performing retrievals over brighter surfaces.

Figure 6: Author need to clarify collocation criteria for the satellite aerosol products and ground-based PM10 observation.

Line 494 “Therefore, the AAI calculation method is not a classic aerosol retrieval method.”: What is the classic aerosol retrieval? The field of aerosol remote sensing employs numerous diverse approaches and algorithms for retrieving aerosol information, each with specific applications, advantages, and limitations.

Lines 519-520 “However, due to the limitation of the satellite observation range, Himawari-8 cannot effectively monitor dust activities in central and western Xinjiang, China.”: It would like to recommend revising it to “Due to the limited field of view from geostationary orbit, Himawari-8 has reduced observational capability in central and western Xinjiang, China.”

Lines 562-563 “However, the performance of the Himawari-8 AOD product was worse than that of several other products.”: It is hard to tell from the results shown here.

Reference: Suggest reviewing the manuscript to ensure that the references are appropriately used.

This study evaluated the performance of multiple satellite remote sensing products in monitoring a dust weather event in East Asia from March 13 to 20, 2021, using PM10 concentration data from 925 ground stations in China for validation. And it analyzed the spatial and temporal consistency of these products, and highlighted the strengths and limitations of each product, identifying issues such as non-standardized detection thresholds and the impact of meteorological and surface factors on dust detection accuracy. However, the manuscript still has the following issues that need to be addressed:

1. A single dust weather event is insufficient to characterize the accuracy and stability of different products in identifying dust storms. It is recommended that the authors expand the study’s temporal scope to include dust event cases across multiple seasons, analyzing the impact of seasonal surface changes, such as vegetation cover, soil moisture, and surface temperature, on the accuracy and stability of dust identification by different remote sensing products, (e.g., the high-dust storm period in spring season versus the low-dust storm period in other seasons), thereby evaluating their cross-seasonal stability and adaptability.
2. In the manuscript, there is an issue with the citation format, for example, on the line 86 of page 3, “which can distinguish dust from complex atmospheric environments. (Liu et al., 2008).”, and line 88 of page 3, aerosol-cloud interaction and climate effects, etc. (Gui et al., 2022; Jia et al., 2018; Wang et al., 2023).”, and etc.
3. It is recommended that the authors overlay these satellite remote sensing product data with PM10 concentration data from ground observation stations spatially (e.g., displaying satellite dust distribution and PM10 concentration points simultaneously on an image) to more intuitively demonstrate the consistency between product detection results and ground observations, thereby enhancing the readability of figures and the clarity of result interpretation, specifically between Fig. 2 and Fig. 3.
4. It mentioned that “when the DST judgment result was dust, the IDDI result may not be dust, which was more obvious over desert areas” on line 377-378, how to explain this situation, and the results in Fig. 2 on March 16, IDDI exhibited this difference compared to DST over the North China Plain. How can this phenomenon be explained?
5. Lines 421-423 state, "it was found that DB is significantly better than DTB in describing the details of dust weather." How can this statement be explained, and it is recommended to provide a detailed explanation.
6. How was the POCD for all products in Figure 12(a) calculated from March 13 to 20? Was it an average of daily results or another method? Referring to Figure 6, the average POCD for MODIS over the whole dust weather process cannot reach 91%.