the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 09 Apr 2024
Retrieval pseudo BRDF-adjusted surface reflectance at 440 nm from Geostationary Environmental Monitoring Spectrometer (GEMS)
Abstract. In remote sensing applications, enhancing the precision of level 2 (L2) algorithms relies heavily on the accurate estimation of the surface reflectance across the ultraviolet (UV) to visible (VIS) spectrum. However, the mutual dependence between the L2 algorithms and surface reflectance retrieval poses challenges, necessitating an alternative approach. To address this issue, many satellite algorithms generate Lambert Equivalent Reflectance (LER) products as a priori surface reflectance data; however, this often results in an underestimation of these data. This study introduces a novel approach to surface reflectance retrieval, termed background surface reflectance (BSR), which leverages a semi-empirical Bidirectional Reflectance Distribution Function (BRDF) model to simulate surface reflectance based on BRDF components. This study pioneered the application of the BRDF model to hyperspectral satellite data in the UV-VIS region, aiming to provide more realistic preliminary surface reflectance data. In this study, the Geostationary Environment Monitoring Spectrometer (GEMS) data was used, and a comparative analysis of the GEMS BSR and GEMS LER revealed an improvement in the relative Root Mean Squared Error (rRMSE) accuracy of 3 %. Additionally, a time-series analysis across diverse land types indicated a greater stability exhibited by the BSR than by the LER. For further validation, the BSR was compared with other LER databases using ground-truth data, yielding superior simulation performance. These findings present a promising avenue for enhancing the accuracy of surface reflectance retrieval from hyperspectral satellite data, thereby advancing the practical applications of remote sensing algorithms.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(6863 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(6863 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-601', Anonymous Referee #1, 24 Apr 2024
The paper is generally well-written and makes a valuable scientific contribution. However, I suggest including additional details and making minor revisions to improve clarity and aid readers' understanding.
After reading through the entire paper, I found a few areas that could benefit from further clarification. It would be helpful to expand on the description of the overall process outlined in lines 138-141 to make it easier for readers to follow. Initially, I assumed Figure 1 was a flowchart illustrating the algorithm, with the bottom row representing gap filling. This led me to believe that the method primarily uses BSR while incorporating LER and TOC for gap filling. It took me some time to realize that my interpretation of the flowchart was incorrect. It would be useful if the flowchart clearly indicated the structure of the algorithm, and the text explained that BSR is compared with LER and TOC to validate the BSR results. Even as I write this review, I'm not entirely sure if my understanding of the flowchart is accurate.
Additionally, if this method uses LER for gap filling, does it create any discontinuity between the pixels that use BRDF modeling and those where LER is applied? It would be beneficial for the authors to address this and discuss any potential issues with continuity if applicable.
Regarding the title of Section 4.4, GEMS LER is mentioned as one of the LER databases, but it's unclear whether it was used in Section 4.4. I tried to find a comparison with GEMS LER in this section but couldn't locate any reference to it. If this omission is accurate, it might be better to remove GEMS LER from the section title.
Lastly, I'm uncertain if it's appropriate to describe this method as a "novel concept" or "novel approach" in the abstract and conclusion. Although this study uses the new hyperspectral sensor GEMS, the methodology (BRDF modeling, LER) itself doesn't seem particularly innovative. Thus, I'm not sure if these terms accurately represent the uniqueness of the approach.
Below is a list of typos or text sections that may need to be revised. Thank you for considering these comments, and I look forward to seeing the updated version of your paper.
Line 140 : GEMS TOC -> GEMS Top of Canopy (TOC)
Line 173 : paragraph is duplicated in line 185
Line 212 : does ‘15 d’ means 15 days?
Line 239 : does ‘SFC’ meas surface reflectance?
Line 373 : I guess author was intending ‘GEMS TOC’ not ‘GEMS, TOC’
Citation: https://doi.org/10.5194/egusphere-2024-601-RC1 - AC1: 'Reply on RC1', Kyung-Soo Han, 07 Jun 2024
-
RC2: 'Comment on egusphere-2024-601', Meng Gao, 13 May 2024
In this work, a Bidirectional Reflectance Distribution Function (BRDF) model called background surface reflectance (BSR) is proposed. It based on a semi-empirical Roujean BRDF model, which represent total surface reflectance as a combination of isotropic, geometric, and volumetric scattering components. The model includes two physical kernels (f1 and f2) and three empirical coefficients. Geostationary Environment Monitoring Spectrometer (GEMS) data was used to derive the BSR model, with results compared with GEMS Lambertian Equivalent Reflectance (LER) with higher accuracy by 3% in terms of relative root mean square error, where the LER is derived from minimum reflectance method under the assumption of a Lambertian surface.
The authors also shows a better stability of BSR than LER in a time-series analysis on various land types. Similarly, the BSR was compared with other LER databases using ground-truth data with better performance of BSR than LER. This work is really interesting due the interconnection of surface and atmospheric retrievals. Any potential improvement of surface model will help better characterization of atmospheric system, such as aerosol. This work is comprehensive with careful comparisons with BRDF datasets from different sensors, ground validation points etc. However, I am not clear in the impacts of the BSR approach, as the percentage improvement seems small (3%), and also how the whole study fit into the UV-VIS range if only 440nm is focused here. Providing more details on its impacts to the atmospheric products over the UV-VIS range would be useful to understand the actual impacts. Please find my suggestive comments below:
- In the abstract, “This study pioneered the application of the BRDF model to hyperspectral satellite data in the UV-VIS region, aiming to provide more realistic preliminary surface reflectance data”. But in the title and the manuscript only 440nm (Page 3, Line 85) is discussed. Then how the authors can demonstrate the application to the boarder spectral range of UV-VIS?
- Page 3, Line 80 “This approach reflects both the high temporal resolution of GLER and the advantages of DLER’s own BRDF”. Can you explain more what advantage of DLER is preserved?
- Page 4, Line 113, “In this study, we utilized 550 nm AOD data to perform atmospheric correction.” Why choose 550nm to conduct atmospheric correction if 440nm is the focus of this study?
- Page 4, Line 115, why Pandora is better?
- Page 6, Line 119, how the aerosol information is determined for the atmospheric transmittance? Is it fixed?
- Page 6, Line 119, there are several transmittance variables used here, are they all defined in the same way or not? Namely whether they are for radiance or irradiance?
- Page 6, Line 119, How the spherical albedo is defined?
- Page 7, Table 1, continental aerosol, how it is defined and determined? Do you use an optimization procedure to retrieve the aerosol properties, such as AOD? Do you assume a surface model when determine the aerosol properties?
- Page 8, Line 195-202, the whole paragraph is duplicated from above!
- Page 8, Equation 5, is there spectral dependency of those parameters?
- Page 8, Line 207, how time frame matters here?
- Page 10, Line 231, what is the age variable? Where is it in the equation?
- Page 10, Line 246, “The GK-2A AMI albedo output establishes the standard for good quality BSR, requiring seven or more observations and a BRDF RMSE of 0.07 or lower” How the number of observations impact accuracy? Does this relate to the coverage of the angular range of the BRDF? Is the geometry limited for a geostationary satellite?
- Page 11, Figure 3 shows good agreement. How does the comparison relate to solar geometry? Do you have examples what the BRDF looks like?
- Page 13, Figure 4, what is the solar and view geometry here?
- Page 13, Line 296, how you separate the four land types? Could you show them in a map?
- Page 15, Line 312, how AOD is determined?
- Page 16, Line 398, how the surface reflectance is observed from ground site? What is the spatial coverage? Any particular viewing geometry the reflectance is defined?
- Page 17, Figure 7, it seems the surface reflectance variability is quite narrow comparing with the one derived from satellite. Is this because the ground site covers a small spatial range, while satellite measures a larger range?
- Page 20, line 396, “This study introduced the novel concept of BSR as an alternative output to resolve the output precedence dilemma between land surface reflectance and other L2 outputs applied to GEMS, a hyperspectral satellite observing in the UV-VIS range. “ Again, this study only discuss one wavelength, so the conclusion to the application to UV-VIS may be not complete.
- Page 21, line 404, “The simulation performance of the GEMS BSR was 3% more accurate than that of the GEMS LER data in terms of the rRMSE over the entire study period based on TOC”. Is 3% significant? How does this compare with the measurement uncertainty from GEMS?
Citation: https://doi.org/10.5194/egusphere-2024-601-RC2 - AC2: 'Reply on RC2', Kyung-Soo Han, 07 Jun 2024
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-601', Anonymous Referee #1, 24 Apr 2024
The paper is generally well-written and makes a valuable scientific contribution. However, I suggest including additional details and making minor revisions to improve clarity and aid readers' understanding.
After reading through the entire paper, I found a few areas that could benefit from further clarification. It would be helpful to expand on the description of the overall process outlined in lines 138-141 to make it easier for readers to follow. Initially, I assumed Figure 1 was a flowchart illustrating the algorithm, with the bottom row representing gap filling. This led me to believe that the method primarily uses BSR while incorporating LER and TOC for gap filling. It took me some time to realize that my interpretation of the flowchart was incorrect. It would be useful if the flowchart clearly indicated the structure of the algorithm, and the text explained that BSR is compared with LER and TOC to validate the BSR results. Even as I write this review, I'm not entirely sure if my understanding of the flowchart is accurate.
Additionally, if this method uses LER for gap filling, does it create any discontinuity between the pixels that use BRDF modeling and those where LER is applied? It would be beneficial for the authors to address this and discuss any potential issues with continuity if applicable.
Regarding the title of Section 4.4, GEMS LER is mentioned as one of the LER databases, but it's unclear whether it was used in Section 4.4. I tried to find a comparison with GEMS LER in this section but couldn't locate any reference to it. If this omission is accurate, it might be better to remove GEMS LER from the section title.
Lastly, I'm uncertain if it's appropriate to describe this method as a "novel concept" or "novel approach" in the abstract and conclusion. Although this study uses the new hyperspectral sensor GEMS, the methodology (BRDF modeling, LER) itself doesn't seem particularly innovative. Thus, I'm not sure if these terms accurately represent the uniqueness of the approach.
Below is a list of typos or text sections that may need to be revised. Thank you for considering these comments, and I look forward to seeing the updated version of your paper.
Line 140 : GEMS TOC -> GEMS Top of Canopy (TOC)
Line 173 : paragraph is duplicated in line 185
Line 212 : does ‘15 d’ means 15 days?
Line 239 : does ‘SFC’ meas surface reflectance?
Line 373 : I guess author was intending ‘GEMS TOC’ not ‘GEMS, TOC’
Citation: https://doi.org/10.5194/egusphere-2024-601-RC1 - AC1: 'Reply on RC1', Kyung-Soo Han, 07 Jun 2024
-
RC2: 'Comment on egusphere-2024-601', Meng Gao, 13 May 2024
In this work, a Bidirectional Reflectance Distribution Function (BRDF) model called background surface reflectance (BSR) is proposed. It based on a semi-empirical Roujean BRDF model, which represent total surface reflectance as a combination of isotropic, geometric, and volumetric scattering components. The model includes two physical kernels (f1 and f2) and three empirical coefficients. Geostationary Environment Monitoring Spectrometer (GEMS) data was used to derive the BSR model, with results compared with GEMS Lambertian Equivalent Reflectance (LER) with higher accuracy by 3% in terms of relative root mean square error, where the LER is derived from minimum reflectance method under the assumption of a Lambertian surface.
The authors also shows a better stability of BSR than LER in a time-series analysis on various land types. Similarly, the BSR was compared with other LER databases using ground-truth data with better performance of BSR than LER. This work is really interesting due the interconnection of surface and atmospheric retrievals. Any potential improvement of surface model will help better characterization of atmospheric system, such as aerosol. This work is comprehensive with careful comparisons with BRDF datasets from different sensors, ground validation points etc. However, I am not clear in the impacts of the BSR approach, as the percentage improvement seems small (3%), and also how the whole study fit into the UV-VIS range if only 440nm is focused here. Providing more details on its impacts to the atmospheric products over the UV-VIS range would be useful to understand the actual impacts. Please find my suggestive comments below:
- In the abstract, “This study pioneered the application of the BRDF model to hyperspectral satellite data in the UV-VIS region, aiming to provide more realistic preliminary surface reflectance data”. But in the title and the manuscript only 440nm (Page 3, Line 85) is discussed. Then how the authors can demonstrate the application to the boarder spectral range of UV-VIS?
- Page 3, Line 80 “This approach reflects both the high temporal resolution of GLER and the advantages of DLER’s own BRDF”. Can you explain more what advantage of DLER is preserved?
- Page 4, Line 113, “In this study, we utilized 550 nm AOD data to perform atmospheric correction.” Why choose 550nm to conduct atmospheric correction if 440nm is the focus of this study?
- Page 4, Line 115, why Pandora is better?
- Page 6, Line 119, how the aerosol information is determined for the atmospheric transmittance? Is it fixed?
- Page 6, Line 119, there are several transmittance variables used here, are they all defined in the same way or not? Namely whether they are for radiance or irradiance?
- Page 6, Line 119, How the spherical albedo is defined?
- Page 7, Table 1, continental aerosol, how it is defined and determined? Do you use an optimization procedure to retrieve the aerosol properties, such as AOD? Do you assume a surface model when determine the aerosol properties?
- Page 8, Line 195-202, the whole paragraph is duplicated from above!
- Page 8, Equation 5, is there spectral dependency of those parameters?
- Page 8, Line 207, how time frame matters here?
- Page 10, Line 231, what is the age variable? Where is it in the equation?
- Page 10, Line 246, “The GK-2A AMI albedo output establishes the standard for good quality BSR, requiring seven or more observations and a BRDF RMSE of 0.07 or lower” How the number of observations impact accuracy? Does this relate to the coverage of the angular range of the BRDF? Is the geometry limited for a geostationary satellite?
- Page 11, Figure 3 shows good agreement. How does the comparison relate to solar geometry? Do you have examples what the BRDF looks like?
- Page 13, Figure 4, what is the solar and view geometry here?
- Page 13, Line 296, how you separate the four land types? Could you show them in a map?
- Page 15, Line 312, how AOD is determined?
- Page 16, Line 398, how the surface reflectance is observed from ground site? What is the spatial coverage? Any particular viewing geometry the reflectance is defined?
- Page 17, Figure 7, it seems the surface reflectance variability is quite narrow comparing with the one derived from satellite. Is this because the ground site covers a small spatial range, while satellite measures a larger range?
- Page 20, line 396, “This study introduced the novel concept of BSR as an alternative output to resolve the output precedence dilemma between land surface reflectance and other L2 outputs applied to GEMS, a hyperspectral satellite observing in the UV-VIS range. “ Again, this study only discuss one wavelength, so the conclusion to the application to UV-VIS may be not complete.
- Page 21, line 404, “The simulation performance of the GEMS BSR was 3% more accurate than that of the GEMS LER data in terms of the rRMSE over the entire study period based on TOC”. Is 3% significant? How does this compare with the measurement uncertainty from GEMS?
Citation: https://doi.org/10.5194/egusphere-2024-601-RC2 - AC2: 'Reply on RC2', Kyung-Soo Han, 07 Jun 2024
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 350 | 94 | 32 | 476 | 22 | 26 |
- HTML: 350
- PDF: 94
- XML: 32
- Total: 476
- BibTeX: 22
- EndNote: 26
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Suyoung Sim
Sungwon Choi
Daeseong Jung
Jongho Woo
Nayeon Kim
Sungwoo Park
Honghee Kim
Ukkyo Jeong
Hyunkee Hong
Kyung-Soo Han
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(6863 KB) - Metadata XML