Synopsis: The manuscript describes a novel upper air balloon observing system, consisting of a double balloon setup that allows for measurements during a long drift phase in the stratosphere and during the descent  phase. The start of the drift phase and of the descent phase can be triggered remotely and steering the height is possible as well. The manuscript also describes the potential for using the system for targeting observations and how assimilation of the data improves analysis and forecasts. 

Overall comment: It is certainly desirable to make better use of weather balloons than is currently the case with many conventional radiosondes where only data collected during the ascent phase are used. The longer residence of the balloons in the lower stratosphere may be useful for observing certain features there, e.g. gravity waves, with more detail. The authors claim a cost advantage of the new system launched at relatively few stations compared to maintaining or even enhancing the relatively dense Chinese radiosonde observation network for targeting if severe weather is approaching. While it is encouraging that the data of the new observing system have already been assimilated by weather forecast models in China, the impact on forecasts has been relatively weak but seemingly consistent. The papers referenced in this context (e.g. Wang et al. 2023) are in Chinese and thus impossible for me to follow. I did not try to use automatic translation for this. The results are based on relatively short validation periods (30 days) or on a case study.  Overall I do not consider the results presented as rigorous proof that the additional measurement data from the drifting balloons improve the quality of analyses and forecasts. While the figures generally support the statements in the text their technical quality is partly poor and should be improved. 

This leads to the following assessment 

Scientific significance: fair

Scientific quality: fair

Presentation quality: fair

As such I require the following major revisions:

1) Some figures are practically unreadable, please redraw or omit:

Fig. 12 is unreadable

Fig. 11a): Please use thicker lines or make figure sharper. Red color is used for both showing the simulated vertical speed and the circle for highlighting the descent phase. This is confusing. 

Fig. 9: What do the dots exactly mean? For me it would be logical if red is the start of the ascent phase, green is the start of the drift phase and blue is the start of the descent phase. The end of the black lines would then be the location where the payload reaches the surface. However this is not consistent with how the colors are labelled. 

Fig. 8: The T-logp diagrams are almost unreadable and they also have different scales. For a publication in a serious journal these must be redrawn. 

Fig. 7: Is it possible to zoom in? Most of the information East of Wuhan appears unimportant. The red writing in the chart is unreadable. Should it be "Wuhan"?

Fig. 6: The 3D-plot is not helpful. Is it possible to draw the same trajectory informationinto panel b), using a multi-colored polygon with the color scheme indicating the height of the balloon?

Fig. 4: Same suggestion as for Fig. 6, draw the info of panel b) into panel a). While looking fancy, the extreme exaggeration of the vertical coordinate compared to the horizontal ones is somewhat misleading. 

2) The added value of the RDSS compared to a sounding system that has just an ascent and a descent phase is unclear and appears limited. ECMWF reported about successfully assimilating descent data https://www.ecmwf.int/sites/default/files/elibrary/102021/20225-newsletter-no-169-autumn-2021_1.pdf,

similar to what is reported in this manuscript. There is very little information about the quality of data collected during the drift phase, except for the caveat that radiation errors may be larger than during ascent and descent due to lacking motion of the balloon relative to the atmosphere. However I wonder if the information during the drift may be useful for analyzing gravity wave activity, for example. This aspect is getting increasing attention. 

The authors appear to see the drift phase more as an opportunity to steer the balloon, less as a measurement period. 

However the chance to steer the balloon depends a lot on favorable wind conditions. So for targeting one needs to launch the balloon at the right position so that it can descend later into a weather system of interest. Wouldn't that be possible also with conventional radiosondes?

The question is really if the money invested into the drifting capability would not be better invested into e.g. balloons that can reach as high up as possible (e.g. https://www.researchgate.net/publication/384155227_Seasonal_and_geographic_viability_of_high_altitude_balloon_navigation)

or into better humidity sensors that can measure reliably under cold low-pressure conditions. 

3) There is little information about availability of these data to the public. They are potentially valuable for weather centers around the globe or for atmospheric climate reanalysis. It would be important for the reader to know whether these data can be accessed and used under a general public license.

Apart from that I have a few minor comments:

Table 3: Please explain what "Airspeed" means. Is it the data transmission rate in bits per second?

l295ff: It is mentioned that the effect of radiation in the drift phase is greater than during ascent/descent. Can you quantify that, ideally with a plot or table showing the increased measurement uncertainty during the drift phase. It is unclear to the reader how the CFD correction model works and how it performs. 

4.2: The assessment is rather crude, particularly for humidity. Having just one value for the whole troposphere is insufficient by today's standard. Can you give a profile of the humidity measurement uncertainty?

l368f: .. track the occurence of the eintire convective system and the changes .. in real time... is a bit strong wording, given the fine grained structure visible in the RADAR picture which is by no means resolvable with a few radiosondes, even with drifting capability. 

l425: results from one month are not conclusive. There should be at least three months (e.g. used as minimum by ECMWF when upgrading their system), ideally from different seasons, to cover any dependency on the annual cycle. 

l534: The reduction of the forecast trajectory error in one typhoon event must be considered anectotal. While encouraging it is certainly not science-grade evidence of an improvement. 

l550: It is not obvious from the material presented that the RDSS really reduces costs compared to conventional radiosondes. This needs more detailed explanation. 

l567ff: This statement on data availability is unacceptable by today's open science standards. 

The Round-trip Drifting Sounding System (RDSS) proposed in this study demonstrates significant innovation and holds promising application potential. It is recommended for publication after minor revision. The RDSS’s key advantage lies in completing a continuous atmospheric observation cycle of "ascent (1 hour) - drifting (4 hours) - descent (1 hour)" with a single balloon launch. Notably, the drifting phase provides continuous observation of the lower stratosphere, while the descent phase provides atmospheric vertical profiles over remote areas far from the launch site. This effectively addresses the gap in traditional sounding observations at 06 and 18 UTC. Furthermore, by utilizing long-distance drifting, RDSS expands the capability for vertical atmospheric sounding in remote regions through the descent observations.

Field experiments conducted in the Yangtze River Basin successfully tackled several key technical challenges. Compared to China's previous-generation L-band sounding system, RDSS exhibits reduced observation errors, particularly in tropospheric wind fields and humidity. Additionally, RDSS possesses a degree of mobility, suggesting potential for targeted observation campaigns. The study also presents preliminary results from RDSS data assimilation and forecast impact experiments, indicating a significant improvement in the skill of 12-24 hour accumulated precipitation forecasts initialized at 06UTC and 18UTC.

Specific Recommendations for Revision:

1. Table 1: The horizontal lines in the three-line table vary in thickness. It is recommended to standardize the line weight.
2. Table 4: Words are hyphenated across lines. It is recommended to adjust the table format to prevent word breaks.
3. Figure 6(a): The 3D plot lacks clarity. It is recommended to replace it with a 2D plot, using different colors to represent trajectory altitudes.
4. Figure 7: This figure solely displays radar reflectivity for a specific weather event and has weak relevance to the main focus of the paper. It is recommended for deletion.
5. Figures 8(a)-(c): The three T-logP diagrams are unclear. Furthermore, they have differing vertical axis ranges and inconsistent sizes, which is unsuitable for formal journal publication. It is strongly recommended to redraw these figures uniformly.
6. Figure 9:
   • (a): This panel merely illustrates the experimental domain and provides limited information. It is recommended for deletion; the location can be described textually in the main body.
   • (b): The symbols are confusing and undefined. It is recommended to recreate this figure with clear labeling of all symbols in the legend. The caption should provide a detailed explanation of the figure's content.
7. Figure 12: The image is unclear and primarily demonstrates the trajectory visualization software's output. It has minimal connection to the paper's core innovations and technical content. It is recommended for deletion.
8. Data Assimilation Experiments: The assimilation application experiments using RDSS data in this paper are relatively limited. It is recommended that subsequent research focuses on: (a) Further refining assimilation techniques for RDSS second-level data (e.g., data thinning methods); (b) Conducting more extensive numerical forecast assimilation impact experiments utilizing additional observational data; (c) Conducting a thorough evaluation of the forecast skill improvement offered by RDSS compared to China's conventional L-band sounding system.
9. "Zhuang Z R, Wang R C (2019), Wang J C, et al." - Literature duplication.
10. The two attached papers also studied and confirmed the impact and value of the RDSS data on numerical forecasting. It is recommended to include them.

ZHANG Xin, WANG Qiuping, MA Xulin, et al. 2025. The Influence of New Round-Trip Drifting Sounding Observation on the Quality of Numerical Prediction in the Middle and Lower Reaches of the Yangtze River [J]. Chinese Journal of Atmospheric Sciences, 49(1): 245−256. doi:10.3878/j.issn.1006-9895.2304.22224

Zhang, X., Sun, L., Ma, X., Guo, H., Gong, Z., Yan, X. Can the Assimilation of the Ascending and Descending Sections’ Data from Round-Trip Drifting Soundings Improve the Forecasting of Rainstorms in Eastern China? Atmosphere 2023, 14, 1127. https://doi.org/10.3390/ atmos14071127

Review of Development and application of the Round-trip Drifting Sounding System (RDSS) 

by Cao et al 

General

I have just completed a rereview of "The Beidou Navigation Radiosonde Observation Experiment and Data Evaluation" 

by Lebao Yao et al (with Xiaozhong Cao as a co-author) submitted to Atmospheric Research. 

When the request for review from Copernicus arrived I clicked accept thinking that this was the other paper.  

The two papers do not seem to have been coordinated as I would expect: in terms of content - there is some overlap, 

the name of the sounding system differs and there is no reference to Yao et al in this manuscript. 

I find the 'Round-trip' in the title a misleading name. In English a round-trip finishes at the point of origin - 

not the case with the RDSS. Yao et al use the term Beidou Navigation Radiosonde. In places this manuscript refers 

to the 'ADD' (Ascent-Drift-Descent) system instead of RDSS. ADD seems a more accurate name to me. 

More explanation of the overall design would be useful. What is the purpose of the 'drift' stage? 

What determines how long the drift stage is? How much does it add to the batteries required? 

Will the drift and descent stages be reported on the GTS? If so what BUFR format will be used for the drift stage? 

At recent WMO meetings (eg TECO 2022) there has been discussion of reducing the environmental impact of 

radiosondes and other instruments. The reduced weight of the new radiosonde is a step in that direction, 

but has there been any effort to reduce plastic use or the toxic elements in circuit boards? 

Are there plans to refurbish and reuse some of the radiosonde instruments? 

There should also mention of the WMO Global Basic Observing Network (GBON, https://community.wmo.int/en/activity-areas/wigos/gbon ) as CMA is planning to use these drift radiosondes operationally 

GBON technical regulations specify that there should be soundings to 30 hPa and a subset to 10 hPa. 

To what extent would this be met in the tests so far?  

One could read this manuscript and think that NWP still revolves almost entirely around radiosonde profiles - not true. 

There is one mention of aircraft reports and almost nothing about satellite observations. 

In practice satellite observations are now hugely important and radiosondes are less so although still very useful 

(partly for validation/verification). Below are three references (of many) to support this view. 

Bauer, P., Thorpe, A. & Brunet, G. The quiet revolution of numerical weather prediction. Nature 525, 47-55 (2015). https://doi.org/10.1038/nature14956

Bormann N, Lawrence H, Farnan J (2019) Global observing system experiments in the ECMWF

assimilation system. ECMWF Technical Memorandum 839, 24 pp. https://doi.org/10.21957/sr184iyz

https://community.wmo.int/en/meetings/8th-wmo-impact-workshop-home (2024, final report and presentations)

Detailed comments

17 "Meteorological sounding primarily refers to the balloon-borne radiosonde" 

The satellite community would disagree. Rewrite. 

21 "zero-pressure dual-mode meteorological balloon" what does zero-pressure mean?

18 "uppe-rair" > "upper-air"

46-47 "Since the 19th century, the balloon-borne radiosonde has served as the primary tool for direct measurements of

47 upper-air meteorological elements below 30 km and is extensively utilized globally (Gallice et al., 2011)." 

The use of radio with balloon soundings only started in the 1930s or so (20th century), with early operational systems 

in the 1940s. A better reference would be Pettifer (2009). 

Pettifer R (2009) From observations to forecasts, part 2. The development of in situ upper air

measurements. Weather 64:302-308. https://doi.org/10.1002/wea.484

48 "For over a century" - "For about eighty years"

49 "one measurement" - "one profile"

54 "as Russia's decrease from twice-daily" - "as Russia's temporary decrease from twice-daily"

See https://www.ecmwf.int/sites/default/files/elibrary/2016/18147-global-radiosonde-network-under-pressure.pdf 

61-63 "The United States and other countries have successfully assimilated airborne sonde data into their model 

systems, resulting in a 20%-40% reduction in errors in hurricane trajectory predictions (Stephen et al., 2013; 

Wang et al., 2015)." 

"20%-40% reduction in errors" seems too high to me for modern NWP systems with good assimilation of satellite data, 

Majumdar (2016) already referenced is more balanced. 

68 "The upper-air balloon system, known as ValBal, developed by the Stanford Space Program in the United States, 

has accomplished multi-day flights at altitudes ranging from 10 km to 25 km (Sushko et al., 2011)." 

I hadn't heard of ValBal, I suspect that it was a short-lived research system. 

On the contrary the WindBorne system (Johnson et al, 2024) is active and currently providing data on the GTS 

Johnson, A., Wang, X., Hutchinson, T., & Creus-Costa, J. (2024). Impact of WindBorne observation assimilation on prediction of a TPV merger case from THINICE. Journal of Geophysical Research: Atmospheres, 129, e2024JD041395. https://doi.org/10.1029/2024JD041395 

81 "descent data" see also Ingleby et al (2022) 

Ingleby, B., Motl, M., Marlton, G., Edwards, D., Sommer, M., von Rohden, C., Vömel, H., and Jauhiainen, H.: On the quality of RS41 radiosonde descent data, Atmos. Meas. Tech., 15, 165-183, https://doi.org/10.5194/amt-15-165-2022, 2022. 

95 'more than a century' - 'about 80 years' (prior to the 1940s there were tiny numbers of research flights that 

mostly relied on someone finding and returning the instrument package and  recorded data) 

106 "ascends at approximately 400 meters per minute" convert to metres per second

186 "The ascent phase of meteorological sounding generally lasts less than one hour." 

It is usually 1.5 to two hours for most radiosondes. Or is this a statement about RDSS? 

193 "hydrogen loss"  It should have been stated earlier that hydrogen is the lifting gas. 

232 "The positioning accuracy of SoC is 0.8 meters horizontally and 1.3 meters vertically" 

1.3 meters vertically sounds better than I would expect - please provide some evidence for this. 

Is the accuracy worse near the surface (due to reflection or ducting of the signals)? 

242-243 "The GTH3 participated in WMO UAII2022(Upper-Air Instrument Intercomparison Campaign organized by the 

World Meteorological Organization (WMO) and co-organized by the Deutscher Wetterdienst (DWD) in 2022)" 

Their table lists 

 Tianjin Huayuntianyi Special Meteorological Sounding Tech. Co., Ltd.  HT-GTS(U)2-1 (GTH3)

I think it is worth giving the manufacturer name here and probably the alternate radiosonde name (HT-GTS(U)2-1). 

Is the current "GTH3" the same as that tested in UAII2022 or a development of it? 

Table 4 caption "The planetary boundary layer (PBL) ranges from 2 to 7 kilometers" - should say "Free Troposphere" 

other parts of the caption also need changing. 

317-318 "Comparative tests led to the selection of conical parachutes" 

Please provide more details of the parachute, such a a diagram, dimensions and any vents. 

319 "As a result, RDSS achieves a stable descent speed of approximately 6 m/s ± 1 m/s" 

Is this true even in the stratosphere? Please show some example profiles and/or mean and standard deviation profiles. 

Ingleby et al (2022) report much faster descent speeds in the stratosphere 

(even with a parachute) and a lot of variation from one flight to another.

"4.1 Field experiment ..." 

What is the distribution (above ground level) of the lowest descent point received? 

This depends on the orography and how close to a receiving station it lands. 

359 "the fifth generation of ECMWF (ERA5) was used to evaluate the QC data of RDSS (Minola et al. 2020; Roja et al. 2011)," 

"the fifth generation of ECMWF reanalysis (ERA5, Hersbach et al, 2020) was used to evaluate the quality of RDSS data," 

I'm not sure how relevant Minola et al is. Roja et al predates ERA5 and doesn't appear in the references. 

Hersbach H et al (2020) The ERA5 global reanalysis. Q J R Meteorol Soc 146:1999-2049. https://

doi.org/10.1002/qj.3803

Table 5. 'EAR5' should be 'ERA5'; 'troposphere top' should presumably be 'tropopause', units of RH are "%RH". 

'{SD of] radiosonde data and ERA5' should be 'radiosonde data minus ERA5'

IMPORTANT: Please provide mean differences for temperature. 

What quality control (including comparison with ERA5) was used to exclude outliers? 

What 'saturation vapour pressure' equation is used for RH? As discussed with Lebao Yao, I think there are 

differences in the definition of RH between the radiosonde and ERA5. It is cleaner to start from ERA5 

specific humidity and temperature and then make sure you are comparing like with like. 

Note. The 'drift' temperature differences seem too large to be usable (despite the processing described above). 

It would be useful to know the mean wind speeds for the different phases. 

Figure 7. Wuhan should be better marked (there is something written in red that I can't read). 

Perhaps a coloured X or other symbol, described in the caption would be better. 

Figure 9b. A lot of green 'descent' dots are close to the 'ascent' point. 

I wondered if green is actually the start of the drift phase and blue the start of descent? 

Some people are red-green colour blind, changing one of the colours would help them. 

(Unlike reviewer 2 I find the orientation given by Fig 9a useful.) 

411-412 "With the four-dimensional ensemble forecast error is introduced into the CMA global data

assimilation system, and the H-4DEnVar assimilation scheme is developed." 

The English here is strange/unclear. What is the H in H-4DEnVar? 

Figure 10 caption. Stage that positive values indicate benefit when descent data is used. 

Precipitation thresholds in mm? 

449-452 "Targeted observations ... " This is overstated. They have "sometimes been a frontier field" 

The impact of targeted observations is almost always less than one or two years work on improving the baseline 

NWP system (Lorenc pers comm) - and improving the baseline tends to reduce the benefit of targeted observations. 

"Majumdar & Sharanya, 2016" should be "Majumdar, 2016" 

Two quotes from his 'Conclusions and Discussion': 

"The fact that many results are inconclusive is not surprising, for several reasons. ..." 

"In summary, the initial vision that the Global Observing System would ultimately be supplemented by networks on an adaptive basis, or even optimized into a fully adaptive network, has not been realized." 

500 '5.3.3 Targeted observations experiment of Typhoon' 

I was surprised by this section because surely the biggest impact on TC forecasting comes from improving the 

analysis and forecast one, two or three days before landfall. 

In the online discussion a response was "The RDSS can cover the South China Sea area" where would they be launched from? 

For another view see 

Magnusson, L., Majumdar, S.J., Dahoui, M.L., Bormann, N., Bonavita, M., Browne, P.A., et al. (2025) The role of observations in ECMWF tropical cyclone initialisation and forecasting. Quarterly Journal of the Royal Meteorological Society, 151(768), e4924. Available from: https://doi.org/10.1002/qj.4924

A few quotes form this:

"The strongest impact came from withholding of microwave radiance observations ..." 

"Scatterometer data improved the forecasts ..."

"While the aircraft-borne data (dropsonde, flight-level data) were mostly beneficial, further investigations on how to best use this data in the TC core are needed."

552 'century-old' - '20th century'

556 'RDSS is essentially a mature system.' I would say it is close to maturity (but not there yet). 

No data have been provided on the GTS yet as far as I know. 

Some of my questions in the general comments need answering/clarification. 

References

Where the papers are in Chinese please indicate this. 

Many readers will not be able to read such papers, I suggest removing some of the less relevant ones. 

632 "Majumdar, S. J.: A review of targeted measurements" this is the whole of the tile, delete 

": targeted measurements to improve numerical forecasts of high-impact weather events over the past two decades, 

particularly during the THORPEX era (2005-14), are evaluated"

660 "Tan et al ...  doi:10.3969/j.issn.1004-4965.2006.01.003." I couldn't find this despite the doi.