the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 23 Oct 2023
First Atmospheric Aerosol Monitoring Results from Geostationary Environment Monitoring Spectrometer (GEMS) over Asia
Abstract. Aerosol optical properties have been provided from the Geostationary Environment Monitoring Spectrometer (GEMS). It is the world’s first geostationary earth orbit (GEO) satellite instrument designed for atmospheric environmental monitoring. This study describes improvements to the GEMS aerosol retrieval algorithm (AERAOD). These include spectral binning, surface reflectance estimation, cloud masking, and post-processing. Furthermore, the study presents validation results. These enhancements are aimed at providing more accurate and reliable aerosol monitoring results for Asia. The adoption of spectral binning in the lookup table (LUT) approach reduces random errors and enhances the stability of the satellite measurements. In addition, we introduce a new high-resolution database for surface reflectance estimation based on the minimum reflectance method adapted to the GEMS pixel resolution. Monthly background aerosol optical depth (BAOD) values are used to consistently estimate the hourly GEMS surface reflectance. Advanced cloud-removal techniques are implemented to significantly improve the effectiveness of cloud detection and enhance the quality of aerosol retrieval. An innovative post-processing correction method based on machine learning is introduced to address artificial diurnal biases in aerosol optical depth (AOD) observations. This study investigates specific aerosol events. It highlights capability of GEMS to monitor and provide insights into hourly aerosol optical properties during various atmospheric events. The performance of the GEMS AERAOD products is validated against the Aerosol Robotic Network (AERONET) and Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) data for the period from November 2021 to October 2022. The GEMS AOD demonstrates a strong correlation with the AERONET AOD (R = 0.792). However, it exhibits bias patterns including underestimation of high AOD values and overestimation in low AOD conditions. Different aerosol types (highly absorbing fine, dust, and non-absorbing) exhibit distinct validation results. The GEMS single scattering albedo (SSA) retrievals agree well with the AERONET data within reasonable error ranges, with variations observed among the aerosol types. For GEMS AOD exceeding 0.4 (1.0), 42.76 % (56.61 %) and 67.25 % (85.70 %) of GEMS SSA data points fall within the ±0.03 and ±0.05 error bounds, respectively. Model-enforced post-processing correction improved the GEMS AOD and SSA performances, thereby reducing the diurnal variation in biases. The validation of the GEMS aerosol layer height (ALH) retrievals against the CALIOP data demonstrates a good agreement, with a mean bias of -0.225 km, and 55.29 % (71.70 %) of data within ±1 km (1.5 km).
- Preprint
(3699 KB) - Metadata XML
-
Supplement
(1109 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on amt-2023-221', Anonymous Referee #1, 17 Nov 2023
Using observations from GEMS, the authors presented methods for retrieving aerosol optical properties, including AOD, SSA, ALH, UVAI and VisAI. Retrieved AOD, SSA and ALH data were evaluated against AERONET and CALIOP data. The concepts included in the study, including aerosol retrievals from UV and VIS observations, as well as using a machine learning method for noisy data removal, are not new. Still, this paper has some merits by applying the above mentioned methods to GEMS data. Still, there are major issues in this study that need to be addressed.
- The post-processing step involves refining/correcting retrieved AOD and SSA values using AERONET data (1-30 days data before a given date) and through a machine leaning method (RF). This creates a potential issue, as non-trivial autocorrelations may exist in AERONET AOD and SSA data for a given AERONET station. Thus, by use the same AERONET site for training and testing, the results of the study may be biased toward AERONET sites. It is unknown the performance of the retrieved data over regions without AERONET data. I would suggest the authors pick some AERONET sites as the testing sites, and AERONET data from these test sites shall not be used for training purposes for the machine learning method.
- Also, cloud contamination exists prior to the post-processing step (Figure 7) yet is suppressed by the post processing step (Figure 8). Any reason for that? Could this be potentially causing an issue? Those cloud contaminated pixels should be excluded in the study.
- Version 3, level 1.5 AERONET data were used in this study. I would recommend the authors use version 3, level 2 AERONET data as it is quality assured. There is a reason why the AERONET team spent efforts creating level 2 data from level 1.5 data. The additional data included in the level 1.5 AERONET data may likely be problematic retrievals.
- Aerosol physical and optical properties are needed for the retrieval process but yet are not discussed in the study. This is needed as the low bias in retrieved AODs (Figure 7) and high bias in SSA (Figure 9; Figure 5) may be introduced by inaccurate aerosol properties used in the retrieval process.
- The paper needs to be carefully proof-read. There are quite a few places that I tried to guess what the authors were trying to say.
- It would be interesting to show seasonal and regional variations of AOD, SSA and ALH as a function of local time (diurnal patterns). It shall not be a difficult task as the authors have the data handy for the task.
Other issues.
Page 1, lines 30 and 34. “GEMS AOD”. Wavelength for the mentioned AOD?
Page 2, line 46, “While significant diurnal variations in AOPs have been observed”. Provide references.
Page 3, line 82,”Considering the solar zenith angle”. What do the authors mean? Considering the sun position changes??
Page 3, lines 96-97, I am not sure what the authors try to express. Please try to rewrite.
Page 3, line 115, “In this paper, we report the first aerosol monitoring results”?? Aerosol retrievals?? Please rewrite.
Page 4, lines 125-126 make no sense to me. Please try to rewrite.
Page 4, equation 1. Where is SZA? Downward solar radiation shall be a function of SZA.
Page 5, line 166, “The preliminary GEMS AERAOD”.. Shall be “An early version of GEMS AERAOD”??
Page 5, line 190. Define “the Levenberg-Marquardt equation”. Or provide a reference.
Page 6, line 208, “the calculations were performed using the Mie theory” Be specific. I assume the authors computed optical properties using the Mie code and applied the computed optical properties in RTM calculations. What about needed parameters for Mie and RTM simulations? But please be precise with your discussions.
Page 6, line 211, provide references for the GEMS spectral response function. Also, GEMS has a spectral resolution of 0.6 nm. Why do the authors resample the spectral data (from RTM) to a spectral resolution of 0.2 nm?
Page 6, line 226, “minimum reflectance method” Need a reference here.
Page 7, line 242, this equation doesn’t make sense. Please check.
Page 7, Section 2.1.4, may need plots to demonstrate cloud detection steps.
Page 10, line 379. Define Q value. Be specific about GCOS requirement.
Citation: https://doi.org/10.5194/amt-2023-221-RC1 -
AC1: 'Reply on RC1', Jhoon Kim, 26 Mar 2024
The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2023-221/amt-2023-221-AC1-supplement.pdf
-
RC2: 'Comment on amt-2023-221', Anonymous Referee #2, 10 Jan 2024
Reviewer’s comments on manuscript titled “First Atmospheric Aerosol Monitoring Results from Geostationary Environment Monitoring Spectrometer (GEMS) over Asia”
In this paper, authors presented the first results of a suite of operational aerosol products, including AOD, SSA and ALH, from Geostationary Environment Monitoring Spectrometer (GEMS), through evaluating their performance in monitoring air pollution events over Asia, and through validation against AERONET and CALIOP data. The objective of this paper is very clear, which intends to show the performance of operational GEMS aerosol product under different scenarios. The methodology in this paper is sound, solid and moderately innovative. The presentation of this paper is relatively clear, but could be largely improved. The results are significant enough to be published after some revisions. However, the manuscript needs to be reorganized to increase the clarity, some missing/misplaced figures needs to be fixes, also contains many editorial errors. It is suggested to be accepted but after some revisions.
Some comments and questions are given as follows.
Major comments:
- The core of this paper, in my opinion, is to demonstrate the performance of the GEMS operational aerosol retrieval algorithm, which went through some updates on the original version. The updates include new hourly surface reflectance database derived from GEMS observations by using the minimum reflectance in 30 days and the background AOD derived from AERONET measurements, new cloud screening method and postprocessing approach using machine learning technique to reduce not only biases but also time-dependency of the biases (As shown in Figure 1). Therefore, descriptions of those updates (bold box in Figure 1) and GEMS algorithm itself deserve to be in more details. The authors have described/touched all aspects of the updates, but is slightly lack of details. The questions should be answered in more details include how these updates are designed and applied, and most importantly what are the impacts on the retrievals. Some of the contents in section 5 can be moved here and show the impact/improvement after algorithm updates. For example, section 2.13, 2.14 and 3, all should add examples to show the improvement. And, It is suggested that section 4 and 5 will solely focus on the evaluations of results from the latest version of algorithm.
- Quality of the Figures needs to be improved, especially Figure 2, Figure 5 and 6. In addition, Figure S1, S2, S3 and S4 are described in text and the actually figures are missing.
- It is strongly suggested to go through editorial revision, including language, terms and sentence structures etc.
Details:
Details are referred to the attached file with reviewer’s comments
-
AC2: 'Reply on RC2', Jhoon Kim, 26 Mar 2024
The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2023-221/amt-2023-221-AC2-supplement.pdf
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 439 | 168 | 26 | 633 | 41 | 19 | 22 |
- HTML: 439
- PDF: 168
- XML: 26
- Total: 633
- Supplement: 41
- BibTeX: 19
- EndNote: 22
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1