Review of “Investigating the role of typhoon-induced gravity waves and stratospheric hydration in the formation of tropopause cirrus clouds observed during the 2017 Asian monsoon” by Pandit et al.

This study analyzes backscatter sonde and optical particle counter measurements from balloons over Hyderabad, India to characterize properties of subvisible cirrus layer observed near the tropopause.  They then investigate the formation mechanism of these cirrus cloud layers near the tropopause using a combination of back trajectories, satellite observations, and ERA5 reanalysis data.  They conclude that overshooting convection from a typhoon created a hydration patch which was transported towards Hyderabad and ascended (via gravity waves) to form the cirrus layer.  While the analyses of balloon data to characterize the cirrus cloud properties were interesting, the discussion on the formation mechanism of those clouds, particularly its ties to gravity waves, needs further development and clarification.  Detailed comments are provided below.

Main Comments:

Although the combined use of various observations is generally a good idea, the frequent comparisons between measurements from different data sources at different locations and times (e.g., Table 2, Fig. 7, 10, 13) were difficult to interpret.  This is especially problematic for this study since the phenomena of interest (i.e., overshooting convection, gravity waves) are small scale features that require coincident measurements.  I suggest refining and limiting the selection of observations to only those that meet rigorous time and space matching criteria.  I would also present the comparisons in a manner that is easily interpretable by the reader.

I also found the discussion of the trajectory results to be somewhat incoherent.  If the main purpose of using the backward trajectories is to show the influence from deep convection on cirrus clouds measured by the balloon, the most important quantity to show is the temperature and relative humidity evolution from the convection encounter to the measurement location.  Trajectory results should be more concisely presented instead of across multiple figures.  I would similarly reorganize other parts of the paper and make the paper more concise overall.

Detailed Comments:

• P4: The temperature impact by tropical storms in Biondi et al. studies is fairly localized, and as such, I would not describe it as “a change in large-scale temperature fields in the TTL”
• P11: Have the authors looked in the effects of trajectory uncertainties on the results? I’m assuming these are kinematic trajectories, whose dispersive nature may be concerning especially when calculating forward trajectories from UTLS levels.
• P12: Is the use of ERA5 cloud cover fraction necessary? Cloud fields in reanalyses are model products (Wright et al., 2020) and vary depending on prescribed boundary conditions, physical parameterizations (e.g., convection), data assimilation approach, and assimilated data.  It is likely unsuited as “cloud observations”.  Rather, plotting the 100 hPa temperature (or cold point temperature) field may be more reliable.
• P15: What do you mean by running HYSPLIT trajectories from different altitudes of the CL5? Are the trajectories run at multiple altitudes within the ~2 km thick cloud layer estimated from the balloon measurements?  And these trajectories intersect the CATS orbit several hours later at the same altitude as the cirrus detected from CATS?  Can you be more specific?
• Fig. 5a: Can you explain what is meant by “day fraction (UTC)”?  This is confusing to me since the dots are supposedly showing backward trajectories from 23 Aug 2017 at 20 UTC.  Time since balloon measurement may be easier to interpret.
• Fig. 5: What is the temperature history of these trajectories?  Are the parcels cooling as they reach the balloon measurement point when cirrus clouds were formed?  Figs. 11 and 12 are discussed much later, but this is a key parameter to look at as is RH with respect to ice.
• Fig. 5: How are the parcels that descend backward in time between 115-125 degE in panel (a) represented in panel (b)? In the plan view, it looks like they are wrapped in the typhoon at 125-130 degE, which does not seem to be consistent with panel (b)?
• P21: How exactly is the convective anvil top altitude determined from the brightness temperatures to quantify the convective influence? The text mentions that this is discussed in Section 2.4.1, but I don’t see any description of this methodology.
• P21: Pixel resolution certainly impacts temperatures within a convective cloud, but that fact alone is insufficient to claim that “the highest overshooting tops likely exceeded 18 km height”. More concrete evidence of this is needed to make such a claim.
• Table 2: While I understand the attempt to describe the temperature evolution of the parcel sampled, the disparate data sources (radiosonde, GNSS-RO), locations and times of these measurements make it difficult to say anything about the actual temperature evolution/effect, especially since we expect temperature perturbations on small spatial and temporal scales that greatly affect the cirrus cloud formation. It would make more sense to look at the temperatures along the backward trajectories to describe the evolution (like Fig. 11) and see if those agree with temperature observations at matching times and locations.
• Fig. S3: (a) It would be good to include latitude and longitudes. (b) How realistic is the ERA5 water vapor (compared to observations like MLS)? Are you claiming that the wet anomalies are from the typhoon injected water and ice?    (c) is x-axis the latitude?
• Fig. S4: Two temperature profiles are compared, but they are again from two different sources at two different locations at two different times.  It is overreaching to claim from this figure alone that “tropopause temperature and tropopause altitude might be substantially reduced by the overshooting convection”.  Or is this meant to approximate the cold-point tropopause altitude?
• Fig. 6: Do the back trajectory segments shown here exactly match the trajectories in Fig. 5b? It would be better if you could somehow combine these plots into one since they are approximately showing the same information.
• Fig. 7: Where are the MLS observations (track) on the map?  It would be helpful to show those to interpret panel (b).  Due to the deep averaging kernel of MLS, it is not clear that the 6.5 ppmv water at 82.5 hPa actually represents enhanced water above the CPT as you suggest in p23.  They could be contributed from enhanced moisture near the CPT.
• Fig. 9: Is (a) identical to Fig. S4?  If so, this needs to be noted (and S4 could be removed).  If not, please explain.  How do the MLS values compare with the mean ERA5 water vapor shown in (b)?  Rather than color coding the dots by the date, consider coloring the dots with the WV mixing ratios?
• Fig. 10: It would better to plot panels (b) to (e) in chronological order to see the decreasing magnitude (or select one representative location and combine the four profiles in one plot).
• P28: Have the authors checked to confirm that supersaturation (and how high of a supersaturation) is achieved along the trajectories to allow for ice nucleation?
• Fig. 12: What resolution analyses are used for these trajectories? Is the resolution high enough to capture gravity waves? I see that the use of MERRA-2 reanalysis data is mentioned later on p31, but this should be discussed earlier along with Fig. 12.  Also, how exactly are the vertical wind speeds derived along the trajectories in (b)?  I presume the vertical winds are used to compute the kinematic trajectories.
• Fig. 13: The purpose of this figure is unclear. The text seems to suggest that the purpose is to show the cold anomalies using high resolution data (and their impact cirrus), but it is difficult to interpret where and when these measurements are taken in relationship to the balloon measurements and trajectories.  The time and lat/lon are noted in the figure caption, that without a map or time series to place them in larger context, it is very difficult to interpret this figure.  It is even unclear whether the measurements plotted in one panel are coincident in time and location (they don’t appear to be).
• Fig. 14: Temperatures were anomalously cold at ~17.9 km near the CPT on 23 Aug, but the tropopause altitudes presumably vary across all the profiles used to calculate the mean. Therefore, panel (a) alone is not sufficient to claim that the CPT was unusually cold on that day.  It might make more sense to calculate the anomalies in tropopause relative coordinates.  I also do not follow the argument that the variations above 15 km are likely due to waves (“wave-like pattern”).  Cooling is likely associated with cirrus formation, but the suggested role of waves from Fig. 14 vertical profiles seems circumstantial at best.  The authors mention a “Hodograph analysis of radiosonde data” to show evidence of gravity waves and derive their characteristics, but no description of the analysis is provided.

References:

Wright, J. S., Sun, X., Konopka, P., Krüger, K., Legras, B., Molod, A. M., Tegtmeier, S., Zhang, G. J., and Zhao, X.: Differences in tropical high clouds among reanalyses: origins and radiative impacts, Atmos. Chem. Phys., 20, 8989–9030, https://doi.org/10.5194/acp-20-8989-2020, 2020

See attached supplement