Technique for comparison of backscatter coefficients derived from in-situ cloud probe measurements with concurrent airborne Lidar
- Atmospheric Sciences, University of North Dakota, Grand Forks, 58202, United States of America
- Atmospheric Sciences, University of North Dakota, Grand Forks, 58202, United States of America
Abstract. Jet engine power loss due to ice particle accumulation is a recognized aviation hazard occurring in cloud conditions difficult to forecast or visually recognize. High-altitude cirrus clouds can have ice particle concentrations high enough to be dangerous; therefore, pilots must be informed when aircraft enter such environments. One approach to determining ice particle concentration is an onboard Lidar system. Concurrent Lidar measurements are compared to backscatter coefficients derived from particle size distributions obtained from wing-mounted, in-situ probes during four case studies consisting of sixty-second flight segments at different temperatures; +7 °C and +4 °C for water droplet analysis, -33 °C and -46 °C for ice particle analysis. Backscatter coefficients derived from external cloud probes (ECP) are correlated (0.91) with measurements by an airborne Lidar system known as the Optical Ice Detector (OID). Differences between OID and ECP backscatter coefficients range from less than 1 to over 3 standard deviations uncertainties. The backscatter coefficients are primarily in agreement for liquid clouds and disagreement for the -33 °C and -46 °C cases, with ECP derived backscatter coefficients lower than the OID for three out of the four cases. Measurements over four research flights show that measured total water content is correlated (0.74) with the OID backscatter coefficient, which indicates that the OID is a useful instrument for determining ice particle concentrations over a broad range of environments, including at ice water contents as low as 0.02 g m-3. Additionally, concurrent measurements from cloud imaging probes and an airborne Lidar test system provides improved knowledge of cloud conditions and can help in understanding cloud processes.
- Preprint
(3182 KB) -
Supplement
(1320 KB) - BibTeX
- EndNote
Shawn Wendell Wagner and David James Delene
Status: final response (author comments only)
-
RC1: 'Comment on amt-2022-87', Anonymous Referee #1, 14 Jun 2022
The manuscript reports on a comparison between cloud Particle Size Distribution (PSD) and optical measurements of the backscatter coefficient. The former is provided by cloud probes (ECP), the latter from a short-range cloud Lidar (OID), both present onboard. Backscatter coefficients computed with Mie theory are compared to lidar-derived ones on four case studies, where 60 s flight segments, with each segment selected to represent a different meteorological condition (i.e. two liquid and two ice clouds), have been used.
Of the four case studies, one (on a warm cloud) show a satisfactory comparison, while the remaining three are less satisfactory, albeit the two backscattering determinations remain within the Three Sigma Rule for all of the time when warm cloud are considered, and for part of the time when cold clouds are considered.
The disagreement is more marked when changes in mean particle dimension and concentration (and maybe shape? The authors look at that as well) settles in. Surprisingly, in three out of the four case studies (those where the author excludes saturation problems in the OID data), the computed ECP backscatter result lower than the measured one. In fact, scattering from nonspherical particles can significantly differ from that of volume or surface equivalent spheres, and this is particularly true in backscattering, where the lack of positive interference effects of the waves propagating inside the particle generally leads to a depression of the backscattering for aspherical particles. Hence, the Mie theory should provide an upper limit to the backscattering from nonspherical particles. So, it is surprising how the Mie computation resulted systematically lower than the measured values. An inaccurate inversion of the OID data may play a role in it, albeit minor. In particular, the extinction used. Chen et al. (Appl. Opt, 2002) reported a variability of Lidar Ratio in cirrus clouds from 20 to 40 sr (at 532 nm). Given the extreme values of the backscattering coefficients for ice clouds, around 0.1 km-1 sr-1, this would deliver an attenuation changing, in the 10 meters path traversed by the OID signal, from 95 to 80%. This may be quoted, along with a more thoughtful description of the OID data inversion procedure, but does not explain the magnitude and sign of the mismatch. The authors honestly acknowledge an unaccounted source of systematic error, and investigate the effects of different methods to define an equivalent spherical particle (surface equivalence in the manuscript, fast circle in the supplementary material), possible biases in concentration, possible presence of more than one phase in clouds. That would dramatically change the computation of the backscattering for the cold cloud cases.
This analysis thus proposes possible causes for the mismatch but does not reach definitive conclusions. Furthermore, there is no discussion on the limits of applicability of the theory of Mie to aspheric particulate matter. I believe this can usefully be added.
Overall the work is interesting, it provides regressions between backscattering (calculated and measured) and TWC in the clouds, and deserves to be published. However, the authors may consider expanding it according to the directions I have highlighted here, and more specifically, as reported below.
(132) “The backscatter coefficient is calculated…”. This β has contribution both from molecules and particles. Given the very high particle β measured, the molecular contribution could be neglected but have to be mentioned.
(144) “…the primary error source is likely the inversion of the range-resolved Lidar signal to estimate extinction.” This is probably true and cast its shadow on the following. Suppose a Lidar Ratio (LR) of some tens sr, given the highest β values reported in the study, the extinction coefficient ε=LR*β (by the way, why use σ instead of ε which is more common in the literature?) would be larger than 10 km-1 and the attenuation from even a distance as short as 10 meters could be significant, and could explain some of the mismatch between computed and measured β, reported afterward. The authors should dwell more on how do they invert their lidar signal, what are the hypothesis done on the LR they use, what is their - at least qualitative - impact on the uncertainties. As instance, are they using the same LR for liquid and ice clouds? Unfortunately, the quoted reference Lolli et al (2013) is of no help since it deals with the determination of colour ratio of rain droplets, explicitly neglecting extinction effects.
(259) “Backscatter efficiency values are calculated using MiePlot for diameters distributed log-normally between 1 μm and 30 mm.” Not clear what “distributed lognormally” means here. Do you mean that the calculated efficiencies were calculated for radii equally spaced on a logarithmic scale from 1 μm to 30 mm?
(260) “Backscatter efficiencies are averaged for all particle diameters within each channel” Where they arithmetically averaged? Was an attempt made to choose the mean value of the radius in the bin so that it was perhaps more representative? For example, by weighing the average of the radii with an estimate of the concentration of the particles at those radii, which can be derived for example from the estimated slope of the PSD in that bin (the arithmetic average assumes that the distribution of particles in the bin is unform). Could this make things better?
(270) Figure 3 is quite interesting as it shows an increase of two orders of magnitude of the backscattering efficiencies for large particles, despite a relatively small change of the refractive index, from ice to water values. This was quite unexpected for me. I have taken the liberty of checking this result with one of the avatars of the BHMIE program which is at the core of the MiePlot package used in this work, and reproducing the same result. Still puzzled, I contacted Philip Laven (the author of the MiePlot package) who confirmed, with independent computation based on the Debye series approach, the correctness of the results of the paper. He explained that the 10th order rainbow is responsible for the increase in backscattering at 905 nm when the real part of the refractive index n = 1.3263 (the value chosen for water in the paper). The authors could underline the peculiarity of the factor 100 difference backscattered intensity at 905 nm between ice and water. In a sense, it is quite unfortunate that the choice of the 905 nm wavelength lead to such dramatic change in the backscattering from ice and liquid water, thus making the assumptions on the particulate phase very critical and impacting for the result. The reviewer thanks Philip Laven for the enlightening mail exchanges.
(390) Figure 10 lower panel is not sufficiently addressed in the text. There it appear two regimes in the TWC-OID backscattering regression. The authors should dwell more on that, and perhaps define two different regression lines.-
AC1: 'Reply on RC1', Shawn Wagner, 20 Jun 2022
The authors would like to express their appreciation for the level of detail of the review and effort of the referee. The comments are thoughtful and provide excellent insight into areas of the manuscript which can be improved. Each of the initial points made will be incorporated into a text revision: the reference to Chen et al. 2002 and the importance of attenuation over the 10 m OID signal path with more information on the OID inversion procedure, and more discussion on how Mie fundamentally applies to spherical particles rather than aspherical particles.
Replies to the line specific comments are given below in italic after repeating the comment:
(132) “The backscatter coefficient is calculated...”. This β has contribution both from molecules and particles. Given the very high particle β measured, the molecular contribution could be neglected but have to be mentioned.
(132) It is true that the molecular backscatter contribution to β is negligible in comparison to the particles. From Anderson et al., 2015: “The OID is not sensitive enough to measure molecular scattering that is commonly used to calibrate cloud lidar.” Hence, it seems reasonable to add a statement near Eq. 1 similar to “While the backscatter coefficient includes scattering from both molecules and cloud particles, the OID is not sensitive enough to measure molecular only scattering (Ray and Anderson, 2015); therefore, it is assumed that all backscatter is from cloud particles.”
(144) “...the primary error source is likely the inversion of the range-resolved Lidar signal to estimate extinction.” This is probably true and cast its shadow on the following. Suppose a Lidar Ratio (LR) of some tens sr, given the highest β values reported in the study, the extinction coefficient ε=LR*β (by the way, why use σ instead of ε which is more common in the literature?) would be larger than 10 km-1 and the attenuation from even a distance as short as 10 meters could be significant, and could explain some of the mismatch between computed and measured β, reported afterward. The authors should dwell more on how do they invert their lidar signal, what are the hypothesis done on the LR they use, what is their - at least qualitative - impact on the uncertainties. As instance, are they using the same LR for liquid and ice clouds? Unfortunately, the quoted reference Lolli et al (2013) is of no help since it deals with the determination of colour ratio of rain droplets, explicitly neglecting extinction effects.
(144) Attenuation does affect backscatter; however, since the OID uses a pulsed laser, close returns can be compared to returns from further away to assess if attenuation is a significant issue. This type of analysis indicates that attenuation affects the +7 case but not the other cases analyzed. We are working on how to address this point in more detail, which we willll incorporate into the revised manuscript.
(259) “Backscatter efficiency values are calculated using MiePlot for diameters distributed log-normally between 1 μm and 30 mm.” Not clear what “distributed lognormally” means here. Do you mean that the calculated efficiencies were calculated for radii equally spaced on a logarithmic scale from 1 μm to 30 mm?
(259) In this case, “distributed lognormally” is intended to mean that with progressively larger diameter particles, the intervals between diameters used in the calculation increases. Intervals between the smallest diameters used in the calculation (starting with 1 µm) begin at 0.0001 µm and increase to intervals of 3 µm at the largest diameter (30 mm). While this was described within the caption of Fig. 3, we will make it clearer within the text at line 259 by adding more details in the text.
(260) “Backscatter efficiencies are averaged for all particle diameters within each channel” Where they arithmetically averaged? Was an attempt made to choose the mean value of the radius in the bin so that it was perhaps more representative? For example, by weighing the average of the radii with an estimate of the concentration of the particles at those radii, which can be derived for example from the estimated slope of the PSD in that bin (the arithmetic average assumes that the distribution of particles in the bin is unform). Could this make things better?
(260) The backscatter efficiencies are arithmetically averaged. As evident from Figure 3, the averaged backscatter coefficient efficiency changes very little from one bin channel to the next. Weighting the averaged efficiency by how the particles are distributed within the channel would move the efficiency slightly (we would estimate 10 %) to smaller sizes in the case of ice, where the efficiency mostly decreases with increasing size and hence may increase the overall backscatter. The maximum percentage difference between scattering efficiency changes from one channel to the next is 17%; hence, 10% of this would be 1.7%, which is small compared to the observed difference between ECP and OID backscattering.
(270) Figure 3 is quite interesting as it shows an increase of two orders of magnitude of the backscattering efficiencies for large particles, despite a relatively small change of the refractive index, from ice to water values. This was quite unexpected for me. I have taken the liberty of checking this result with one of the avatars of the BHMIE program which is at the core of the MiePlot package used in this work, and reproducing the same result. Still puzzled, I contacted Philip Laven (the author of the MiePlot package) who conformed, with independent computation based on the Debye series approach, the correctness of the results of the paper. He explained that the 10th order rainbow is responsible for the increase in backscattering at 905 nm when the real part of the refractive index n = 1.3263 (the value chosen for water in the paper). The authors could underline the peculiarity of the factor 100 difference backscattered intensity at 905 nm between ice and water. In a sense, it is quite unfortunate that the choice of the 905 nm wavelength lead to such dramatic change in the backscattering from ice and liquid water, thus making the assumptions on the particulate phase very critical and impacting for the result. The reviewer thanks Philip Laven for the enlightening mail exchanges.
(270) The authors appreciate the extensive work which went into the verification of the results in Figure 3. The differences described were of some interest, and it is appreciated that there is an explanation. It is agreed that this is something which could be further acknowledged and explained within the Methodology section rather than waiting to be simply acknowledged within the Discussion section. This explanation will be added in paper revision.
(390) Figure 10 lower panel is not sufficiently addressed in the text. There it appear two regimes in the TWC-OID backscattering regression. The authors should dwell more on that, and perhaps define two different regression lines.
(390) The authors agree that the Fig. 10 could be addressed further, particularly the lower panel. There does seem to be two difference regimes in the OID backscatter coefficient regression: one for the cold cases (primarily ice), and one for the warm cases (primarily water). This would seem to be an important distinction as this is clearly not accounted for in the ECP backscatter coefficient shown in the top panel. We will add additional regression lines for water and ice to Fig. 10, as well as discussion within the text.
-
AC1: 'Reply on RC1', Shawn Wagner, 20 Jun 2022
-
RC2: 'Comment on amt-2022-87', Anonymous Referee #2, 20 Jun 2022
This study compares the cloud particle optical and microphysical properties by cloud probe (ECP) and a concurrent airborne Lidar (OID) under different temperature regimes. The manuscript also illustrates some inconsistencies between the two measurement methods and then gives several potential reasons for the differences. This is a very useful measurement in understanding the gap between model computation and measurements for cloud particle optical and microphysical properties. Several comments are listed below.
1. In line 244, what is “ni”? I think you refer to eta_i in equation (4). A similar issue appears in equation (5).
2. For the upper right plot (B) in Figure 7, I think the median value is more reasonable than the mean value for the particle diameter measured by the ECP. Also, what is the definition of backscatter per second in Figure 5, the mean or median value?
3. The Figure 8 caption said the least square fit is the black line, but it is not black in Figure 8.
4. Figure 10 needs some help. Two different color dots are heavily overlapped. Would it be possible to change to partly transparent to better distinguish cold and warm particles, or reduce the marker size? Furthermore, the figure with the same range for the x-axis and y-axis may be better to compare.
5. The study points out that the biased low calculated backscattering from ECP. The backscattering is calculated by the measured effective diameter in this study. I am not very clear how the effective diameter is determined in the measurement. You also mentioned a “fast-circle diameter method”. However, I did not find the related description in the supplement materials either. I think it is better to describe the measure and convert process more.
-
AC2: 'Reply on RC2', Shawn Wagner, 28 Jun 2022
The authors would like to thank the referee for taking the time to review the manuscript and make helpful suggestions toward improvements. Replies to the line specific comments are given below in italic after repeating the comment:
1. In line 244, what is “ni”? I think you refer to eta_i in equation (4). A similar issue appears in equation (5).
1. This is a formatting error where “ni” should be “ηi”. This will be updated within the text at lines 244 and 279.
2. For the upper right plot (B) in Figure 7, I think the median value is more reasonable than the mean value for the particle diameter measured by the ECP. Also, what is the definition of backscatter per second in Figure 5, the mean or median value?
2. While the median will be different than the mean of the spectrum and could be viewed as a more reasonable way to represent the spectrum, the merged spectrum data currently do not contain the spectrum diameter median; hence, additional software development would be necessary. The point of Figure 7b is to show relative difference in spectrum diameters between the four cases and to show when/if size changes occur during the time period. For this purpose, the mean works as well as the median. Additionally, Figure 7c shows the particles spectrums so the reader can interpret how the mean and medium are different. Hence, we respectfully feel that it is not important for the reader’s understanding to present the spectrum median instead of the spectrum mean in Figure 7b.
The backscatter per second for the OID is the result of 20 kHz measurements aggregated to 5 Hz raw data. The mean of the 5 Hz raw data is then taken to match the 1 Hz ECP data, noted in lines 137 – 138.
3. The Figure 8 caption said the least square fit is the black line, but it is not black in Figure 8.
3. This is a formatting error where the “black” line should be labeled as “teal” (a change that was missed after the color scheme was updated to be more viewer friendly). This will be updated within the text at line 394.
4. Figure 10 needs some help. Two different color dots are heavily overlapped. Would it be possible to change to partly transparent to better distinguish cold and warm particles, or reduce the marker size? Furthermore, the figure with the same range for the x-axis and y-axis may be better to compare.
4. To increase the readability of Fig. 10, the size of the scatter points has been reduced, and a level of transparency has been added to the markers on both the top and bottom of the figure. The x-axes of the top and bottom plots have been made to match. The y-axes have been kept separate to maintain the one-to-one ratio of the top plot, as well as prevent a large amount of empty space for the bottom plot. These changes can be seen in an attached figure. For the revised manuscript as fit to the cold data will also be added.
5. The study points out that the biased low calculated backscattering from ECP. The backscattering is calculated by the measured effective diameter in this study. I am not very clear how the effective diameter is determined in the measurement. You also mentioned a “fast-circle diameter method”. However, I did not find the related description in the supplement materials either. I think it is better to describe the measure and convert process more.
5. The diameter shown in the main text is found using the area-equivalent processing method, in which the total circular area of the pixels contained within an imaged particle are used to determine the associated diameter. The fast-circle processing method used in the supplemental material calculates the imaged particle diameter by encompassing the imaged particle entirely within a circle. The diameter of the resulting circle is assumed to be the diameter of the particle. Currently there is no generally agreed upon method for calculating the effective diameter, so we are presenting the two most accepted methods. Details regarding the methods for determining particle diameters will be added at lines 205 to 210 within the manuscript.
-
AC2: 'Reply on RC2', Shawn Wagner, 28 Jun 2022
- RC3: 'Comment on amt-2022-87', Anonymous Referee #3, 27 Jun 2022
Shawn Wendell Wagner and David James Delene
Data sets
Analysis of Concurrent Backscatter Coefficients from In-situ Cloud Probes and Airborne Lidar Shawn Wagner; David Delene https://doi.org/10.31356/data015
Shawn Wendell Wagner and David James Delene
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
349 | 75 | 13 | 437 | 23 | 2 | 3 |
- HTML: 349
- PDF: 75
- XML: 13
- Total: 437
- Supplement: 23
- BibTeX: 2
- EndNote: 3
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1