the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Mar 2023
OLCI A/B tandem phase: Evaluation of FLEX like radiances and estimation of systematic differences between OLCI-A and OLCI-FLEX
Abstract. During the tandem phase of Sentinel-3A and -3B in summer 2018 the Ocean and Land Color Imager (OLCI) mounted on Sentinel-3B satellite was reprogrammed to mimick ESA’s 8th Earth explorer the Fluorescence explorer (FLEX). OLCI in FLEX configuration (OLCI-FLEX) had 45 spectral bands between 500 nm and 792 nm. The new data set with high resolution measurements (band width: 1.7–3.7 nm) serves as preparation of the FLEX mission. Co-registered measurements of both instruments will be used to describe the atmosphere and the surface. For such combined products, it is essential that both instruments are radiometrically consistent. We developed a transfer function to compare radiance measurements from different optical sensors and to monitor their consistency.
In the presented study, the transfer function shifts information gained from high-resolution "FLEX-mode" settings to information convolved with spectral response of the normal (lower) spectral resolution of the OLCI sensor. The resulting reconstructed low resolution radiance is representative for the high resolution data and it can be compared with the measured low resolution radiance. This difference is used to quantify systematic differences between the instruments. Applying the transfer function, we could show that OLCI-A is about 2 % brighter than OLCI-FLEX for most bands. At the longer wavelengths OLCI-A is about 5 % darker. Sensitivity studies showed that the parameters affecting the quality of the comparison of OLCI-A and OLCI-FLEX with the transfer function are mainly the surface reflectance and secondarily the aerosol composition. However, the aerosol composition can be simplified as long it is treated consistently in all steps in transfer function.
Generally, the transfer function enables direct comparison of instruments with different spectral responses even with different observation geometries or different levels of observation. The method is sensitive to measurement biases and errors resulting from the processing. One application could be the quality control of the FLEX mission.
-
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
(1204 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1204 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-306', Anonymous Referee #1, 07 Apr 2023
This paper presents a method to compare the OLCI-A and OLCI-FLEX radiometry from the tandem phase acquisitions of Sentinel-3A and Sentinel-3B, paving the way for comparing OLCI and FLEX measurements when FLEX is launched in tandem formation with Sentinel-3. The method is generic and can be extended to other missions, it is well described and the results are well presented.
Compared to the OLCI-A/OLCI-B tandem phase analysis in Lamquin et al., the focus is here made on the spectral range of interest for FLEX between 500 and 800 nm as well as on the vegetated targets of specific interest for FLEX. The commonalities and differences between the results from the two studys are well discussed and the arguments are clear, especially when focusing on the accuracy of spectral characterization and potential failure of the processing of OLCI-FLEX in the NIR. It could be emphasized a better continuity at camera interfaces although a direct comparison should provide a clearer picture.
The method presented in this paper provides an improvement for the precision and uncertainty assessment of the radiometric comparisons when considering vegetated targets. There could be more emphasis on this point since, on a practical aspects, the method is more complex than the one from Lamquin et al.
I recommend this paper to be published after these minor points are considered, as well the points listed below.
Thank you.
line 3 « with high spectral resolution », add « spectral »
line 4 « spatially co-registered measurements…» add « spatially »
line 5 « to describe the atmosphere and the surface» is a bit vague, if you can find something a bit more explicit it’s better
line 10 : « The resulting reconstructed low resolution radiance is representative for the high resolution data and it can be compared with the measured low resolution radiance », you could detail it into « The resulting reconstructed low resolution radiance is representative for the high resolution data (OLCI-B measurements in FLEX mode) and it can be compared with the measured low resolution radiance (OLCI-A measurements) ».
line 12 : « for most bands of the OLCI-FLEX spectral domain »
line 12 : « at the longer wavelengths (> 700nm) », or change 700 as desired but provide a boundary
line 18 : « One application could be the quality control of the FLEX mission. » à « One application could be the quality control of the FLEX mission, presently it is also useful for the quality control of the OLCI-FLEX data ». If I understand correctly.
line 23 : « Currently, two twin Sentinel-3… », or add information above that there are multiple instances of S3 satellites which are similar by design but not strictly similar after factory (for instance in spectral characterization)
line 28 : the correct reference for the comparisons of radiances while in tandem is the « part 1 » paper, not the « part 3 » :
Lamquin, N.; Clerc, S.; Bourg, L.; Donlon, C. OLCI A/B Tandem Phase Analysis, Part 1: Level 1 Homogenisation and Harmonisation. Remote Sens. 2020, 12, 1804. https://doi.org/10.3390/rs12111804
line 37 : « For a meaningful usage of the OLCI-B data in FLEX configuration », precise OLCI-B
line 42 : « It is applied for vegetated cloud free land pixels, as the main objective of FLEX mission is to retrieve fluorescence emitted by plants » : this is a very important methodological point because Lamquin et al. (part 1) did comparisons on different targets and concluded that clouds and deserts are the « easiest » targets for the exercise (the least contaminated by dependency with transmission and other spectral effects). Hence I would add in the following :
line 43 : « Lamquin et al. (2020) showed a systematic bias between OLCI-A and OLCI-B in the tandem constellation data » -> « Lamquin et al. (2020) showed a systematic bias between OLCI-A and OLCI-B in the tandem constellation data, with slight discrepancies depending on the nature of the targets which are likely due to residuals from the spectral alignment of the compared measurands ».
line 44 : « will be estimated by using our transfer function » à « will be estimated by using our transfer function on vegetated pixels» (If I understand correctly, otherwise remove my comment)
line 55 : I suggest rather « similar spatial resolution and observation geometry » instead of « the same spatial resolution and similar observation geometry, since later you speak of « similar spatial resolution » (l 60), also I think « similar » is more appropriate as you may be able to compare slightly different resolutions without having much side effects
line 57 : « to calculate »
line 59 : « instruments flying in tandem » is more general and preferable as tandem could be temporary only and not be a « tandem mission » as OLCI/FLEX
line 69 : « we developed a transfer function. » I would suggest to describe a bit more the purpose as I did not get it at first, following the graph it seems like you could say : « we developed a transfer function allowing to compare gas-corrected TOA radiances on the same spectral setting (resolution and wavelength) », something like that to your convenience…
line 75 : « The reconstructed OLCI-A spectrum will be referred to as OLCI-AR from now on », since the reconstructed spectrum is originally from OLCI-B measurements there can be a confusion with the fact that you align spectrally to OLCI-A as well (or do you align both OLCIs on the same spectral wavelength, what is actually done ?). Maybe calling it « OLCI-B2AR » as in B-to-A-reconstructed could help keeping track of the fact that the original radiometry (and its residual difference against OLCI-A) is from OLCI-B.
Or put some info at line 90-91 like: « The resulting OLCI-AR radiance is representative for the OLCI-FLEX measurement (hence originally from the OLCI-B calibrated radiance)»
And/or : « The difference between the reconstructed and measured OLCI-A radiance quantifies the bias between the two data sets (hence the bias between OLCI-A and OLCI-B radiance, which is why our results can be compared to Lamquin et al.)
After all, I did not get exactly what spectral setting the two datasets are aligned to, the OLCI-A one ? Or a central-wavelength smile-corrected one, identical for both ? Did you try a sensitivity analysis on this (especially for the O2 bands) ?
line 93 : « Besides the L1b radiance of OLCI-A and OLCI-FLEX, …», add the comma at the end
line 98 : « 520 detector rows are aligned along track » : for the spectral dimension I guess as OLCI is push-broom, could you precise ?
line 103 : « The central wavelength are taken from the temporal evolution model of the wavelength characterization » à The central wavelengths (s added)
Table 4 : step could be added on the table for visibility, eventually avoiding to describe it in the text
line 330 : « Only at longer wavelength 330 (780 nm) OLCI-FLEX is darker » à « Only at longer wavelength 330 (780 nm) OLCI-FLEX is brighter »
line 370 : « 370 camera 5 explains the difference of the median over the camera compared to the other cameras seen in Fig. 5. », you could add « , it is interesting to see that a better continuity between cameras 4 and 5 is observed compared to Lamquin et al. ». I’m wondering what could be the cause : updated spectral characterization ? The method itself ? Did you try comparing over cloudy pixels « raw » from your statistics (without using transfer method) ? It could help understanding why.
Note : you could relate this comment to the discussion part
line 371 : « Further camera effects can be observed for all cameras at 708.75 nm. This band is influenced by water vapour absorption » this is not observed in Lamquin et al. specifically in this channel. Could it be due to the spectral resampling of OLCI-FLEX into OLCI « nominal » configuration coupled to H2O absorption sensitivity ? Is it what you mean ?
line 468 : « affects » instead of « effects » ?
line 481 : « of more than », not « then »
line 515 : « as long as »
line 597-599 : « deserts » not « desserts » J
line 607 : « In this article, we showed systematic differences between OLCI-A and OLCI-FLEX during the tandem phase of Sentinel 3A and 3B » you could add « , consistent with known radiometric differences between OLCI-A and OLCI-B».
line 624 : « of the exciting satellites », Earth Observation science is indeed exciting but I guess you meant « existing » right ? J
Citation: https://doi.org/10.5194/egusphere-2023-306-RC1 - AC1: 'Reply on RC1', Lena Jänicke, 17 Apr 2023
-
RC2: 'Comment on egusphere-2023-306', Anonymous Referee #2, 14 Apr 2023
The reviewed manuscript is devoted to developing an approach for inter-comparison of the space-borne instruments with different spectral response functions. In particular, using measurements from OLCI A/B tandem phase, the transfer function was developed for mimicked FLEX high spectral and OLCI A low resolution co-registered measurements.
The manuscript provides the detailed description of the approach for the transfer function. Applying the transfer function on OLCI-A and OLCI-B/FLEX mimicked measurements, estimated systematic difference was found to be similar to the one observed previously for OLCI-A and OLCI-B with their original band settings.
The developed approach can be used for future constellation of FLEX, OLCI A and B instruments where radiometric consistency is crucial for advanced aerosol and surface characterization.
Specific comment.
It is stated that the developed approach for the transfer function “enables direct comparison of instruments with different spectral responses even with different observation geometries …”.
It was shown in the paper that the surface reflectance is the main source of uncertainties of the approach. It is also known that land surface can show very strong angular dependence of the reflectance. In this regards, the discussion about the dependence of the transfer function on surface reflectance angular anisotropy (BRDF effect) is necessary.I recommend the manuscript for publication after minor revision.
Citation: https://doi.org/10.5194/egusphere-2023-306-RC2 -
AC2: 'Reply on RC2', Lena Jänicke, 17 Apr 2023
Dear Reviewer,
thank you for your comment on our manuscript. You asked us to specifiy the need for a angular dependent description of the surface in case of comparing two measurments with different viewing geometries. We agree, that in this case the surface reflectance should be described using the BRDF. We mentioned this in the conclusion of the manuscript (line 620 in original/ line 638 in updated manuscirpt).
Best regards,
Lena Jänicke
Citation: https://doi.org/10.5194/egusphere-2023-306-AC2
-
AC2: 'Reply on RC2', Lena Jänicke, 17 Apr 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-306', Anonymous Referee #1, 07 Apr 2023
This paper presents a method to compare the OLCI-A and OLCI-FLEX radiometry from the tandem phase acquisitions of Sentinel-3A and Sentinel-3B, paving the way for comparing OLCI and FLEX measurements when FLEX is launched in tandem formation with Sentinel-3. The method is generic and can be extended to other missions, it is well described and the results are well presented.
Compared to the OLCI-A/OLCI-B tandem phase analysis in Lamquin et al., the focus is here made on the spectral range of interest for FLEX between 500 and 800 nm as well as on the vegetated targets of specific interest for FLEX. The commonalities and differences between the results from the two studys are well discussed and the arguments are clear, especially when focusing on the accuracy of spectral characterization and potential failure of the processing of OLCI-FLEX in the NIR. It could be emphasized a better continuity at camera interfaces although a direct comparison should provide a clearer picture.
The method presented in this paper provides an improvement for the precision and uncertainty assessment of the radiometric comparisons when considering vegetated targets. There could be more emphasis on this point since, on a practical aspects, the method is more complex than the one from Lamquin et al.
I recommend this paper to be published after these minor points are considered, as well the points listed below.
Thank you.
line 3 « with high spectral resolution », add « spectral »
line 4 « spatially co-registered measurements…» add « spatially »
line 5 « to describe the atmosphere and the surface» is a bit vague, if you can find something a bit more explicit it’s better
line 10 : « The resulting reconstructed low resolution radiance is representative for the high resolution data and it can be compared with the measured low resolution radiance », you could detail it into « The resulting reconstructed low resolution radiance is representative for the high resolution data (OLCI-B measurements in FLEX mode) and it can be compared with the measured low resolution radiance (OLCI-A measurements) ».
line 12 : « for most bands of the OLCI-FLEX spectral domain »
line 12 : « at the longer wavelengths (> 700nm) », or change 700 as desired but provide a boundary
line 18 : « One application could be the quality control of the FLEX mission. » à « One application could be the quality control of the FLEX mission, presently it is also useful for the quality control of the OLCI-FLEX data ». If I understand correctly.
line 23 : « Currently, two twin Sentinel-3… », or add information above that there are multiple instances of S3 satellites which are similar by design but not strictly similar after factory (for instance in spectral characterization)
line 28 : the correct reference for the comparisons of radiances while in tandem is the « part 1 » paper, not the « part 3 » :
Lamquin, N.; Clerc, S.; Bourg, L.; Donlon, C. OLCI A/B Tandem Phase Analysis, Part 1: Level 1 Homogenisation and Harmonisation. Remote Sens. 2020, 12, 1804. https://doi.org/10.3390/rs12111804
line 37 : « For a meaningful usage of the OLCI-B data in FLEX configuration », precise OLCI-B
line 42 : « It is applied for vegetated cloud free land pixels, as the main objective of FLEX mission is to retrieve fluorescence emitted by plants » : this is a very important methodological point because Lamquin et al. (part 1) did comparisons on different targets and concluded that clouds and deserts are the « easiest » targets for the exercise (the least contaminated by dependency with transmission and other spectral effects). Hence I would add in the following :
line 43 : « Lamquin et al. (2020) showed a systematic bias between OLCI-A and OLCI-B in the tandem constellation data » -> « Lamquin et al. (2020) showed a systematic bias between OLCI-A and OLCI-B in the tandem constellation data, with slight discrepancies depending on the nature of the targets which are likely due to residuals from the spectral alignment of the compared measurands ».
line 44 : « will be estimated by using our transfer function » à « will be estimated by using our transfer function on vegetated pixels» (If I understand correctly, otherwise remove my comment)
line 55 : I suggest rather « similar spatial resolution and observation geometry » instead of « the same spatial resolution and similar observation geometry, since later you speak of « similar spatial resolution » (l 60), also I think « similar » is more appropriate as you may be able to compare slightly different resolutions without having much side effects
line 57 : « to calculate »
line 59 : « instruments flying in tandem » is more general and preferable as tandem could be temporary only and not be a « tandem mission » as OLCI/FLEX
line 69 : « we developed a transfer function. » I would suggest to describe a bit more the purpose as I did not get it at first, following the graph it seems like you could say : « we developed a transfer function allowing to compare gas-corrected TOA radiances on the same spectral setting (resolution and wavelength) », something like that to your convenience…
line 75 : « The reconstructed OLCI-A spectrum will be referred to as OLCI-AR from now on », since the reconstructed spectrum is originally from OLCI-B measurements there can be a confusion with the fact that you align spectrally to OLCI-A as well (or do you align both OLCIs on the same spectral wavelength, what is actually done ?). Maybe calling it « OLCI-B2AR » as in B-to-A-reconstructed could help keeping track of the fact that the original radiometry (and its residual difference against OLCI-A) is from OLCI-B.
Or put some info at line 90-91 like: « The resulting OLCI-AR radiance is representative for the OLCI-FLEX measurement (hence originally from the OLCI-B calibrated radiance)»
And/or : « The difference between the reconstructed and measured OLCI-A radiance quantifies the bias between the two data sets (hence the bias between OLCI-A and OLCI-B radiance, which is why our results can be compared to Lamquin et al.)
After all, I did not get exactly what spectral setting the two datasets are aligned to, the OLCI-A one ? Or a central-wavelength smile-corrected one, identical for both ? Did you try a sensitivity analysis on this (especially for the O2 bands) ?
line 93 : « Besides the L1b radiance of OLCI-A and OLCI-FLEX, …», add the comma at the end
line 98 : « 520 detector rows are aligned along track » : for the spectral dimension I guess as OLCI is push-broom, could you precise ?
line 103 : « The central wavelength are taken from the temporal evolution model of the wavelength characterization » à The central wavelengths (s added)
Table 4 : step could be added on the table for visibility, eventually avoiding to describe it in the text
line 330 : « Only at longer wavelength 330 (780 nm) OLCI-FLEX is darker » à « Only at longer wavelength 330 (780 nm) OLCI-FLEX is brighter »
line 370 : « 370 camera 5 explains the difference of the median over the camera compared to the other cameras seen in Fig. 5. », you could add « , it is interesting to see that a better continuity between cameras 4 and 5 is observed compared to Lamquin et al. ». I’m wondering what could be the cause : updated spectral characterization ? The method itself ? Did you try comparing over cloudy pixels « raw » from your statistics (without using transfer method) ? It could help understanding why.
Note : you could relate this comment to the discussion part
line 371 : « Further camera effects can be observed for all cameras at 708.75 nm. This band is influenced by water vapour absorption » this is not observed in Lamquin et al. specifically in this channel. Could it be due to the spectral resampling of OLCI-FLEX into OLCI « nominal » configuration coupled to H2O absorption sensitivity ? Is it what you mean ?
line 468 : « affects » instead of « effects » ?
line 481 : « of more than », not « then »
line 515 : « as long as »
line 597-599 : « deserts » not « desserts » J
line 607 : « In this article, we showed systematic differences between OLCI-A and OLCI-FLEX during the tandem phase of Sentinel 3A and 3B » you could add « , consistent with known radiometric differences between OLCI-A and OLCI-B».
line 624 : « of the exciting satellites », Earth Observation science is indeed exciting but I guess you meant « existing » right ? J
Citation: https://doi.org/10.5194/egusphere-2023-306-RC1 - AC1: 'Reply on RC1', Lena Jänicke, 17 Apr 2023
-
RC2: 'Comment on egusphere-2023-306', Anonymous Referee #2, 14 Apr 2023
The reviewed manuscript is devoted to developing an approach for inter-comparison of the space-borne instruments with different spectral response functions. In particular, using measurements from OLCI A/B tandem phase, the transfer function was developed for mimicked FLEX high spectral and OLCI A low resolution co-registered measurements.
The manuscript provides the detailed description of the approach for the transfer function. Applying the transfer function on OLCI-A and OLCI-B/FLEX mimicked measurements, estimated systematic difference was found to be similar to the one observed previously for OLCI-A and OLCI-B with their original band settings.
The developed approach can be used for future constellation of FLEX, OLCI A and B instruments where radiometric consistency is crucial for advanced aerosol and surface characterization.
Specific comment.
It is stated that the developed approach for the transfer function “enables direct comparison of instruments with different spectral responses even with different observation geometries …”.
It was shown in the paper that the surface reflectance is the main source of uncertainties of the approach. It is also known that land surface can show very strong angular dependence of the reflectance. In this regards, the discussion about the dependence of the transfer function on surface reflectance angular anisotropy (BRDF effect) is necessary.I recommend the manuscript for publication after minor revision.
Citation: https://doi.org/10.5194/egusphere-2023-306-RC2 -
AC2: 'Reply on RC2', Lena Jänicke, 17 Apr 2023
Dear Reviewer,
thank you for your comment on our manuscript. You asked us to specifiy the need for a angular dependent description of the surface in case of comparing two measurments with different viewing geometries. We agree, that in this case the surface reflectance should be described using the BRDF. We mentioned this in the conclusion of the manuscript (line 620 in original/ line 638 in updated manuscirpt).
Best regards,
Lena Jänicke
Citation: https://doi.org/10.5194/egusphere-2023-306-AC2
-
AC2: 'Reply on RC2', Lena Jänicke, 17 Apr 2023
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 165 | 44 | 13 | 222 | 4 | 3 |
- HTML: 165
- PDF: 44
- XML: 13
- Total: 222
- BibTeX: 4
- EndNote: 3
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Lena Katharina Jänicke
Rene Preusker
Marco Celesti
Marin Tudoroiu
Jürgen Fischer
Dirk Schüttemeyer
Matthias Drusch
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1204 KB) - Metadata XML