the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Increasing Aerosol Optical Depth Spatial And Temporal Availability By Merging Datasets from Geostationary And Sun-Synchronous Satellites
Abstract. This comprehensive study analyzed aerosol observations from six Low Earth Orbit (LEO) and Geostationary Earth Orbit (GEO) sensors. LEO sensors like MODIS and VIIRS, providing 1–2 daily global measurements, were contrasted with GEO sensors (AHI, ABIs), offering high-frequency data (~10 minutes) over specific regions. The Dark Target aerosol retrieval algorithm was applied to six sensors (3 LEO and 3 GEO), and their Level 2 aerosol optical depth (AOD) data were grided and merged into a quarter-degree latitude-longitude grid with a 30-minute temporal resolution. Validation of AOD at 550 nm against AERONET measurements across global locations showcased the merged product's robustness, revealing a global mean bias of approximately ±0.05, and 65.5 % of retrievals fell within an expected uncertainty range with a correlation coefficient of 0.83, underlining the reliability of the dataset. The new grided level 3 dataset significantly improved daily global coverage to nearly 45 %, overcoming the limitations of individual sensors, which typically range from 12 % to 25 %. Furthermore, the study emphasized the unprecedented ability of the merged dataset to approximate the diurnal cycle of AERONET AODs, offering insights into unexpected diurnal signatures. The resulting dataset's high spatiotemporal resolution and improved global coverage, especially in regions covered by GEO sensors (Americas and Asia), make it a valuable tool for diverse applications. Tracking aerosol transport from phenomena like wildfires and dust storms gains precision, enabling enhanced air quality forecasting and hindcasting. Additionally, the study positions the merged dataset as a significant asset for evaluating and inter-comparing regional or global model simulations, previously unattainable in such a gridded format. The dataset and fusion framework layout in this study has the potential to include data from recently (future) launched other GEO (FCI, AMI) and LEO (PACE, VIIRS-JPSS) sensors.
- Preprint
(11770 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on amt-2023-259', Anonymous Referee #1, 29 Feb 2024
title: Review of "Increasing Aerosol Optical Depth Spatial And Temporal Availability By Merging Datasets from Geostationary And Sun-Synchronous Satellites"
authors: Gupta et al.
summary:
The authors apply the dark target aerosol optical depth algorithm to six satellite instruments (3 geo, 3 leo). They produce a quarter degree gridded product with statistics for each instrument and an ensemble average. They use the gridded product for intercomparison and validation against AERONET.response:
The manuscript is a significant contribution that fits well within the scope of AMT. I have only minor comments that focus on methodological clarity and minor editorial comments. The resultant data product will likely be extremely valuable to multiple air pollution disciplines.line-by-line:
- pg1, 22-23, I suggest moving the correlation before the percent within EE because it makes it could be read that the correlation is related to that subset.
- pg1, 44: grid[d]ed
- pg2, 16: my copy shows a strikeout that should be addressed.
- pg2, 16: SNPP and Aqua seem close in time, but Terra seems like a meaningfully different overpass time.
- pg2, 20-21: As written, this excludes the main reasons for missing pixels and then concludes nearly complete... The no clouds *and otherwise retrievable* seems weird.
- pg3, 30: (ATBD, 2023[)]
- pg4, 18: Can you be more specific about "after some time"? Are we talking about Phase F or something earlier?
- pg4, 25: Section 3 really only addresses LUT updates. Are algorithm adjustments always LUT updates? Or are there any more substantial updated?
- pg5, 14: It would be good for Table 2 or the text to explicitly mention overpass times.
- pg5, 34-35: Are any of the AERNET not collocated with leo orbits?
- pg6, 18: viewing "angle" will vary by product.
- pg6, 34: Are you saying finer pixel measurements at nadir are aggregated so that the pixel size range is smaller? Is that what the jumps are in Figure 2?
- pg6, 38: "box gridding" is not a term I am used to. Is this referring to binning pixels based on their centroids being within a quarter degree cell (nearest neighbor based on centroids)?
- pg6, 39: "spatial filling method" as described sounds like "averaging pixels whose footprint overlaps a grid cell".
- pg7, line 21: Visible discontinuity at the scale displayed seems like an unreasonable metric. We'd expect the discontinuity to be larger for a single scene when zoomed in.
- pg7, 37-38: This seems like a weird choice. I agree that it likely doesn't change the conclusions, but a 1 in 30 sample seems like an unnecessary simplification.
-pg8, 4: The g17 also looks at the arid west where aod comparisons have revealed higher uncertainty. I think it is important to note that it isn't just US vs Asia, but within countries as well.
Citation: https://doi.org/10.5194/amt-2023-259-RC1 -
RC2: 'Comment on amt-2023-259', Anonymous Referee #2, 18 Apr 2024
This manuscript presents advances in the monitorization of aerosol optical depth from satellite measurements by increasing the temporal resolutions at 0.25º grid. That permits studies of aerosol evolution during the day. To do so, the authors apply the Dark Target algorithm to six different satellites, three Low Earth Orbit (LEO) and three with Geostationary Earth Orbit (GEO). They also proposed a new product that merges data from the different sensors. Authors also present intercomparisons between the different sensors, validation versus AERONET and eventually the diurnal AOD cycle from satellite observations. I believe that all these issues are very interesting and will permit advances in aerosol studies. However, in the current state the manuscript requires revision before its publication in AMT.
To me the methodology for the creation of the new merged AOD product is not clear. The methodology proposed seems not to exploit the potential of combining all measurements in a single retrieval. Indeed it seems that it is just a statistical approach combining the same retrieval (dark target) on different sensors. I recommend the authors to further explain the new methodology. For example, box-diagrams for caveat 1 and caveat 2 in section 3.3 could help to further understanding
I believe that the current analyses of diurnal cycles are poor. Authors present the diurnal cycle for all possible AERONET stations and no remarkable diurnal cycle is observed. In previous AERONET studies (see Smirnov et al., 2002) diurnal cycle is site-dependent due to the different aerosol regime in each station. Also, in Smirnov et al., 2002 diurnal cycles are studied in percentage deviation versus daily means, that allow to identify diurnal patterns. Following the same approach could serve to see the potentiality of increasing the number of daily measurements in satellite observations. Studies cases would give clarity.
Apart of the two major concerns, I also have other minor concerns:
Introduction:
Page 1, Lines 25-27: This claim about diurnal cycles in AOD does not correspond with the analyses in the manuscript (see major comment).
Page 2, Lines 4-5: The statement that there is information on aerosol microphysical properties from Dark Target is not correct. To my knowledge Dark Target does not provide aerosol microphysics as an output. Indeed, the algorithm assumes predefined aerosol models and uses look-up tables.
Page 2, Lines 19-20: It is trivial that geostationary observations have more chance of getting measurements. It needs to be re-phrase
Page 2, Line 19: Again, the larger temporal resolution of geostationary observations is straightforward.
Page 2, Lines 26-27: It is straightforward the presence of aerosol diurnal variability from any ground observations. Why not remarking the limitations of LOE observations?
Page 2, Lines 33-34: I guess that authors want to mention that studies of AOD diurnal patterns have been made using AERONET observations.
Page 3, Lines 4-6: Why covering the globe with multiple geostationary sensors is not the same as viewing the entire globe with MODIS-like sensors? I do not get the point. Could this be related with the different spatial resolutions? Also, why having two identical sensors do not provide the same aerosol results unlike MODIS ?
Page 3, Line 13-14: If Dark Target is applied to different sensors I assume it needs different parameterizations.
Results and Discussions
Figure 1 seems to remark global coverage by combining the six satellite sensors. However, it does not remark the main goal of improving temporal coverage with high spatial resolution.
Figure 2: Why is this Figure relevant?
Figure 3: This Figure needs further discussion, particularly for AOD below 0.1.
Figure 5: Further discussion is need about the fact that larger deviations are observed for LOE sensors.
Figure 7: It is straightforward that the merged product has more measurements. Also, Europe/Africa region is less represented, but there are also geostationary observations for these regions.
Figure 10: Is there a better agreement with AERONET for AOD merged products? It is not clear. Indeed, the 1:1 results look quite similar.
Conclusions:
Page 11, Line 26: High quality aerosol observations? To my knowledge the only reliable product is AOD.
Page 11, Line 32: Does this study implement Dark Target? The algorithm and its application are well known in the scientific community.
Page 12, Line 14-15: There is no demonstration of the merged product for tracking aerosol transport / variability in a single place.
Page 12, Line 18-19: From the result presented in the manuscript I have not seen an unexpected strong diurnal signature from the global composite of AERONET AOD during northern Spring
Citation: https://doi.org/10.5194/amt-2023-259-RC2
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
286 | 89 | 13 | 388 | 11 | 15 |
- HTML: 286
- PDF: 89
- XML: 13
- Total: 388
- BibTeX: 11
- EndNote: 15
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1