Impact of the revisit frequency on cloud climatology for CALIPSO, EarthCARE, Aeolus, and ICESat-2 satellite lidar missions
- Space Research Centre, Polish Academy of Sciences, 00-716 Warsaw, Poland
- Space Research Centre, Polish Academy of Sciences, 00-716 Warsaw, Poland
Abstract. Space profiling lidars offer a unique insight into cloud properties in Earth’s atmosphere, and are considered the most reliable source of total (column-integrated) cloud amount (CA), and true (geometrical) cloud top height (CTH). However, lidar-based cloud climatologies suffer from infrequent sampling: every n-days, and only along the ground track. This study therefore evaluated four lidar missions, namely CALIPSO (revisit every n = 16 days), EarthCARE (n = 25), Aeolus (n = 7), and ICESat-2 (n = 91), to test the hypothesis that each mission provides accurate data on CA and CTH. CA/CTH values for a hypothetical daily revisit mission were used as reference (data simulated with Meteosat 15-minute cloud observations, assumed to be a proxy for ground truth). Our results demonstrated that this hypothesis is invalid, unless individual lidar transects are averaged over an area 10×10° in longitude and latitude (or larger). If this is not the case, the required accuracy of 1 % (for CA) or 150 m (for CTH) cannot be met, either for a single-year annual or monthly mean, or for a >10-year climatology. A CALIPSO-focused test demonstrated that the annual mean CA estimate is very sensitive to infrequent sampling, and that this factor alone can result in 14 % or 7 % average uncertainty with 1° or 2.5° resolution data, respectively. Consequently, applications that use gridded lidar data should consider calculating confidence intervals, or a similar measure of uncertainty. Our results suggest that CALIPSO, and its follow-on mission EarthCARE, are very likely to produce consistent cloud records despite the difference in sampling frequency.
Andrzej Z. Kotarba
Status: closed
-
RC1: 'Comment on amt-2022-80', David Winker, 26 Apr 2022
This paper examines whether the impacts of sparse sampling from a nadir-viewing satellite lidar varies with the revisit time of the satellite orbit. Orbits of several different existing satellite lidars are chosen as examples. Parameters of interest are cloud amount, cloud top height, and cloud optical depth. SEVIRI cloud retrievals are taken to be truth. Lidar sampling errors are then simulated by sampling SEVIRI retrievals along the ground tracks of the various lidars. Lidar sampling error is measured by the difference between statistics based on SEVIRI retrievals sampled along the orbit track of each of the modeled lidars and SEVIRI retrieval statistics sampled by a hypothetical lidar with one day revisit time.
The paper is well organized and clearly written, for the most part. I have one major concern and a few specific comments.
My major concern is that Section 3.3 and Section 4 (Table 4) seem to come to opposite conclusions. Section 3.3 shows that shifting the initial day of the CALIPSO 16-day orbit cycle (essentially, shifting the orbit tracks observed on a given day) can be a major source of uncertainty. On the other hand, Table 4 shows that at the annual scale, with 10x10 grid cells, accuracy requirements can be met for most locations. Are all the results in Table 3 for 1x1 degree grid cells? Figure 5 shows that sampling uncertainties decrease when size of the grid cells increases but the uncertainties seem to be larger than what is indicated by the results in Table 4. But the metrics shown in the two sections are different and difficult to compare. Are results in the two sections consistent or do results in Table 4 ignore uncertainties due to initial day of the cycle? Please explain.
Minor comments:
I did not find the latitudinal extent of SEVIRI CLAAS dataset in the text. Figure 2 seems to show the CLAAS data extends from about 70S to 70N. This is important to mention in the text, to make clear that lidar sampling of the high Arctic is not being evaluated in this study.
Line 133 states that Aeolus is in an equatorial (0-degree inclination) orbit. This is not correct. Aeolus is in a 97-degree inclination orbit.
In Section 3.3 it is not clear what grid cell size is used in generating the statistics which are reported. Other than Figure 5, do all statistics refer to 1 degree grid boxes? What grid cell size is shown in Figure 5 c, d, g, and h?
Line 432: “spatial resolution above 10 degrees” is ambiguous. Does this mean “spatial resolution better than 10 degrees” ?
Line 438. Please explain why confidence intervals are preferred over means and medians in this circumstance. Also, provide a reference on how to compute confidence intervals.
- AC1: 'Reply on RC1', Andrzej Kotarba, 20 Jun 2022
-
RC2: 'Comment on amt-2022-80', J. Yorks, 08 Jun 2022
This paper investigates the satellite sampling of cloud amount and cloud top height from several sun-synchronous satellites hosting lidar instruments compared to SEVERI. The paper is well written, clear, and provides results that are important for designing future space-based lidar mission architectures. It deserves to be published after a few minor revisions that I believe will strengthen the paper.
My 2 major comments are:
- Diurnal variability: It is not clear to me what role diurnal variability plays in this sampling study. Based on #3 on line 156, I believe only Meteosat data from the time of the satellite overpass was used in the "truth" dataset. Is that true? If so, your monthly and annual averages are biased to the times of the overpasses and thus diurnal variability will not be accounted for. That is fine for assessing the sampling at the equatorial crossing times of each satellite, but that needs to be explicitly stated in the paper. If the SEVIRI “truth” dataset DOES include clouds from all times of day, then the geographical distribution of absolute differences in CA/ CTH estimations (Figure 2) will be influenced by diurnal variability of CA or CTH. This pattern (1-day climatology was most accurately reproduced by n-day data at high latitudes but with lower accuracy at lower latitudes) would also be more consistent with where the largest diurnal variability was reported by Noel et al. (2018).
- Big picture impacts: The author does a great job discussing what these results mean for future lidar missions in the Conclusion. But after reading the paper, I found myself asking - what does this means for current data users? For example, if I want to use the data from these missions to compute global, annual cloud radiative effects based on CA and CTH, I can do that confidently. However, if I want to compute radiative effects at seasonal/monthly and regional/finer spatial scales, the CA and CTH from these lidar datasets may be biased based on Table 4. That is an important point for data users and will make this paper worth citing for future authors. I suggest adding a few sentences on this topic to the Discussion or Conclusion.
The 4 minor comments to be addressed are:
- Table 1: Do all these satellites have a 98-degree inclination angle? If not, I suggest adding an inclination angle column to this table since it impacts the repeat times.
- Line 66: I suggest citing Yorks et al 2016 for CATS as it is more of an overview paper. The citation is: Yorks, J. E., M. J. McGill, S.P. Palm, D. L. Hlavka, P.A. Selmer, E. Nowottnick, M. A. Vaughan, S. Rodier, and W. D. Hart (2016), An Overview of the CATS Level 1 Data Products and Processing Algorithms, Geophys. Res. Let., 43, doi:10.1002/2016GL068006.
- Line 110-111: Are there any papers that reports the accuracy of CA and CTH from the SEVIRI products? If so, I suggest adding a sentence to report those accuracies and cite those papers. I know that doesn’t impact the results of the study, but I found myself wondering what the accuracies are as I read the paper.
- Line 445-446: This is a highly relevant point for future architecture designs. Did you consider looking at the ISS to provide a reference point for a lower inclination angle? I know it would be more work, but I think it would really strengthen the paper to add the ISS to this analysis. At the very least, it would be beneficial to add a sentence or two about the ISS revisit time and where it may fall compared to the satellites you studied.
- AC2: 'Reply on RC2', Andrzej Kotarba, 20 Jun 2022
Status: closed
-
RC1: 'Comment on amt-2022-80', David Winker, 26 Apr 2022
This paper examines whether the impacts of sparse sampling from a nadir-viewing satellite lidar varies with the revisit time of the satellite orbit. Orbits of several different existing satellite lidars are chosen as examples. Parameters of interest are cloud amount, cloud top height, and cloud optical depth. SEVIRI cloud retrievals are taken to be truth. Lidar sampling errors are then simulated by sampling SEVIRI retrievals along the ground tracks of the various lidars. Lidar sampling error is measured by the difference between statistics based on SEVIRI retrievals sampled along the orbit track of each of the modeled lidars and SEVIRI retrieval statistics sampled by a hypothetical lidar with one day revisit time.
The paper is well organized and clearly written, for the most part. I have one major concern and a few specific comments.
My major concern is that Section 3.3 and Section 4 (Table 4) seem to come to opposite conclusions. Section 3.3 shows that shifting the initial day of the CALIPSO 16-day orbit cycle (essentially, shifting the orbit tracks observed on a given day) can be a major source of uncertainty. On the other hand, Table 4 shows that at the annual scale, with 10x10 grid cells, accuracy requirements can be met for most locations. Are all the results in Table 3 for 1x1 degree grid cells? Figure 5 shows that sampling uncertainties decrease when size of the grid cells increases but the uncertainties seem to be larger than what is indicated by the results in Table 4. But the metrics shown in the two sections are different and difficult to compare. Are results in the two sections consistent or do results in Table 4 ignore uncertainties due to initial day of the cycle? Please explain.
Minor comments:
I did not find the latitudinal extent of SEVIRI CLAAS dataset in the text. Figure 2 seems to show the CLAAS data extends from about 70S to 70N. This is important to mention in the text, to make clear that lidar sampling of the high Arctic is not being evaluated in this study.
Line 133 states that Aeolus is in an equatorial (0-degree inclination) orbit. This is not correct. Aeolus is in a 97-degree inclination orbit.
In Section 3.3 it is not clear what grid cell size is used in generating the statistics which are reported. Other than Figure 5, do all statistics refer to 1 degree grid boxes? What grid cell size is shown in Figure 5 c, d, g, and h?
Line 432: “spatial resolution above 10 degrees” is ambiguous. Does this mean “spatial resolution better than 10 degrees” ?
Line 438. Please explain why confidence intervals are preferred over means and medians in this circumstance. Also, provide a reference on how to compute confidence intervals.
- AC1: 'Reply on RC1', Andrzej Kotarba, 20 Jun 2022
-
RC2: 'Comment on amt-2022-80', J. Yorks, 08 Jun 2022
This paper investigates the satellite sampling of cloud amount and cloud top height from several sun-synchronous satellites hosting lidar instruments compared to SEVERI. The paper is well written, clear, and provides results that are important for designing future space-based lidar mission architectures. It deserves to be published after a few minor revisions that I believe will strengthen the paper.
My 2 major comments are:
- Diurnal variability: It is not clear to me what role diurnal variability plays in this sampling study. Based on #3 on line 156, I believe only Meteosat data from the time of the satellite overpass was used in the "truth" dataset. Is that true? If so, your monthly and annual averages are biased to the times of the overpasses and thus diurnal variability will not be accounted for. That is fine for assessing the sampling at the equatorial crossing times of each satellite, but that needs to be explicitly stated in the paper. If the SEVIRI “truth” dataset DOES include clouds from all times of day, then the geographical distribution of absolute differences in CA/ CTH estimations (Figure 2) will be influenced by diurnal variability of CA or CTH. This pattern (1-day climatology was most accurately reproduced by n-day data at high latitudes but with lower accuracy at lower latitudes) would also be more consistent with where the largest diurnal variability was reported by Noel et al. (2018).
- Big picture impacts: The author does a great job discussing what these results mean for future lidar missions in the Conclusion. But after reading the paper, I found myself asking - what does this means for current data users? For example, if I want to use the data from these missions to compute global, annual cloud radiative effects based on CA and CTH, I can do that confidently. However, if I want to compute radiative effects at seasonal/monthly and regional/finer spatial scales, the CA and CTH from these lidar datasets may be biased based on Table 4. That is an important point for data users and will make this paper worth citing for future authors. I suggest adding a few sentences on this topic to the Discussion or Conclusion.
The 4 minor comments to be addressed are:
- Table 1: Do all these satellites have a 98-degree inclination angle? If not, I suggest adding an inclination angle column to this table since it impacts the repeat times.
- Line 66: I suggest citing Yorks et al 2016 for CATS as it is more of an overview paper. The citation is: Yorks, J. E., M. J. McGill, S.P. Palm, D. L. Hlavka, P.A. Selmer, E. Nowottnick, M. A. Vaughan, S. Rodier, and W. D. Hart (2016), An Overview of the CATS Level 1 Data Products and Processing Algorithms, Geophys. Res. Let., 43, doi:10.1002/2016GL068006.
- Line 110-111: Are there any papers that reports the accuracy of CA and CTH from the SEVIRI products? If so, I suggest adding a sentence to report those accuracies and cite those papers. I know that doesn’t impact the results of the study, but I found myself wondering what the accuracies are as I read the paper.
- Line 445-446: This is a highly relevant point for future architecture designs. Did you consider looking at the ISS to provide a reference point for a lower inclination angle? I know it would be more work, but I think it would really strengthen the paper to add the ISS to this analysis. At the very least, it would be beneficial to add a sentence or two about the ISS revisit time and where it may fall compared to the satellites you studied.
- AC2: 'Reply on RC2', Andrzej Kotarba, 20 Jun 2022
Andrzej Z. Kotarba
Andrzej Z. Kotarba
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
268 | 59 | 15 | 342 | 5 | 7 |
- HTML: 268
- PDF: 59
- XML: 15
- Total: 342
- BibTeX: 5
- EndNote: 7
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1