In this study, Gloeckner et al. present the BEACH dropsonde dataset from the ORCESTRA field campaign, based on HALO dropsonde deployments over the Atlantic ITCZ during the PERCUSION and MAESTRO operations. The manuscript outlines the processing from raw AVAPS output to quality controlled and gridded profile products, and it also includes mesoscale diagnostics from circular flight patterns, including divergence and vertical velocity. Overall, the dataset should be a useful community resource for studying ITCZ structure and variability using a large and coordinated set of dropsonde releases. The product design supports a wide range of applications, from standard thermodynamic and wind profile analysis to mesoscale diagnostics that can be linked with other campaign measurements and used for model evaluation. Making both the dataset and the processing code publicly available improves transparency and reproducibility.

Comments

Line 32: The citations in this sentence feel mismatched. The text states that Yanai (1961) applied the methods and produced some of the first estimates of W, but the supporting citations given are Reed and Recker (1971) and Yanai et al. (1973). Please clarify which study provides the first estimates and align the citations accordingly.

Line 77-78: The notation “ca 40 min" is unclear. Please clarify what “ca” means and what the 40 minutes refers to. Also explain how 40 min maps to ≈140 km diameter (state the assumed speed or conversion).

Line 175-180: The QC step that removes measurements where gpsalt exceeds the aircraft altitude is described, but does not specify what aircraft altitude reference is being used or how it is made consistent with the dropsonde gpsalt.

Line 15, 78, 83, 270: The repeated use of phrasing such as Windmiller and authors and “Bony and authors” is nonstandard and reads informal. 

Line 182-183: In Section 3.5, the manuscript explains that JOANNE’s profile fullness (sat-test) was reformulated as a profile sparsity metric based on missing values relative to a hypothetical perfect sonde. However, there is no clear connection of this reformulation to the selected threshold used in Section 3.2.1 (passing if >20% of theoretically available data are present). As written, it is difficult to interpret what level of completeness this corresponds to.

The manuscript “BEACH:  Barbados and Eastern Atlantic High-altitude dropsonde datasets” by Gloeckner et al. is a good description of the dropwindsonde data available from that particular field experiment.  I do have some concerns about the clarity of the description, which I mention below:

First, the output from Aspen, both netCDF and FRD, contain vertical velocity measurements.  I think this is important since the study seeks to calculate large-scale vertical velocity.  The dropwindsonde measures vertical velocity at the exact location of the instrument, whereas the study seeks to calculate averages within the aircraft circles, so the measurements are different.  Still, they may provide some additional information for the study.  Vertical velocity is mentioned starting on line 162, but is not mentioned after that.

In addition, the vertical velocity can be used to assess the validity of other observations from the dropwindsondes.  Some instruments are fast falls, or partial fast falls, and some of the observations may be questionable due to the parachute not properly being deployed.  It is not mentioned in the text whether this is used, though Aspen flags fast falls.  However, not all fast falls are found by Aspen, so research-quality datasets require some manual checks.  Line 143 suggests that a launch detect means that the parachute opened properly, but it only means that the parachute opened.

Line 150 starts a description of Aspen.  It isn’t clear whether the default settings of the Aspen configuration were used or not.  The default settings were designed for operational quality control so that data can be assimilated into models and used for situational awareness.  Different settings can be used to create research-quality datasets.  It isn’t mentioned whether any changes were made.

Line 151 suggests that Aspen removes post-splash data.  This is not the case; Aspen flags cases in which post-splash data may exist and asks the user to manually set the end of drop.  Further, Aspen frequently misses cases that do have post-splash data.  A good check is to look at the vertical velocity near the surface.  If the vertical velocity in the processed data is large and positive near the surface, the product likely contains post-splash data.

Lines 152-165:  The characteristics of the sondes can be changed in the configuration, so that the correct vertical velocity can be recovered.

Line 175:  The altitude can be removed via the configuration settings.

Line 179:  Sondes frequently move upward, but this probably doesn’t happen outside of deep convection.  https://doi.org/10.1175/MWR-D-18-0041.1

Lines 224-224:  Is there a time difference between two good points larger than which interpolation is not done?

Line 405:  It may be useful to cite https://doi.org/10.1175/2010JCLI3496.1

The main reference for the GPS dropwindsonde used here is https://doi.org/10.1175/1520-0477(1999)080<0407:TNGD>2.0.CO;2.  Characteristics of the NRD41 sonde specifically are presented in https://doi.org/10.1175/BAMS-D-22-0119.1.