General Comments

This manuscript presents a comprehensive analysis of cloud vertical structure (CVS) across China using a national network of 80 MMCRs for the year 2024. The authors develop an improved CVS retrieval algorithm that incorporates height-dependent radar sensitivity and variance-based clutter suppression. They systematically characterize the spatial, seasonal, and diurnal variations of CVS metrics (cloud base height, cloud top height, cloud thickness, and cloud layer number) and quantify the regulatory effects of thermodynamic (lower-tropospheric stability, specific humidity at 850 hPa), dynamical (wind shear at 700 hPa), and land-surface factors. The findings provide valuable observational benchmarks for evaluating and improving cloud parameterization schemes in weather and climate models. Overall, the manuscript is well written and the concepts are clearly presented. However, several major and minor issues require further improvement. Therefore, I recommend that the authors revise their manuscript to address the following comments.

Major Comments

1. The discussion of cloud thickness (CT) appears in Sections 3.5.3 (single-factor correlations) and 3.6.3 (bivariate analysis). While both add value, some redundancy exists. Consider streamlining the single-factor section to focus on broad patterns and reserving detailed interaction effects for the bivariate section.
2. Supplementary figures (S2, S3) are mentioned but not described in the main text. Please add a brief summary of their key patterns (e.g., seasonal CTH and CT distributions) to make the main text more self-contained.
3. ERA5 represents large-scale atmospheric conditions at 0.25° resolution (~31 km), whereas MMCR observations are point-based. The mismatch in spatial scales may introduce uncertainties, particularly in regions with complex terrain or strong local gradients. Please add a brief discussion of this limitation.

Minor Comments

1. Lines 117-120: Introduction. The statement “no previous study has ever provided a national-scale, high-frequency characterization of CVS across China using a coordinated ground-based radar network” may be overly absolute. Please revise it.
2. Line 348, Section 3.2 paragraph 2: "favouring higher cloud bases." → "favoring" (consistent with American English spelling used throughout the manuscript).
3. Reference list: Slingo, J. M. (2007) → The year in the text (Slingo, 1987) does not match the reference year; please correct the reference or the in-text citation.
4. The term “lifting condensation level (LCL)” is used multiple times in the text. After its first definition, please use only “LCL” thereafter to avoid repetition.
5. Figure 3. Please unify the longitude range in Figure 3 with that in the other figures..
6. Figure 15. The MMCR has been defined in Figure 1 already. Therefore, you don not need to define it here.
7. In Section 3.7, “afternoon single-layer clouds” are defined as those occurring during peak daytime heating, but the exact time window is not specified. Please define “afternoon” (e.g., 1400–1800 LT) explicitly to ensure reproducibility.
8. Line 651-668: There is a numbering error: “3.7.1 All-cloud conditions” is followed by “3.7.1 Afternoon single-layer clouds.” The second subsection should be numbered “3.7.2.” Please correct.

This paper uses one year of observations from a newly established national Ka-band cloud radar network to characterize cloud vertical structure across China, including cloud occurrence, layer number, CBH, CTH, and CT, and further relates these properties to humidity, lower-tropospheric stability, wind shear, and land cover. The dataset is clearly valuable, especially because the 1-min temporal resolution and 30-m vertical resolution allow the authors to examine seasonal and diurnal variations that are difficult to resolve from satellites or radiosondes. The paper’s main contribution is therefore not only the national-scale description of cloud vertical structure, but also the attempt to connect these observed structures with thermodynamic, dynamical, and land-surface controls. Overall, the study has strong observational potential and could provide useful benchmarks for model evaluation, but several issues need to be addressed before the physical conclusions can be considered fully convincing.

Specific comments:

1. One limitation is the short time period. All major statistics are based on 2024, but the manuscript often presents the results as a national-scale benchmark. Cloud vertical structure over China can be strongly affected by monsoon variability, large-scale circulation anomalies, regional drought or wet conditions, and unusual synoptic patterns in a given year. The authors may discuss how representative 2024 is, for example by comparing cloud fraction, precipitation, humidity, stability, or circulation conditions with multi-year satellite or reanalysis records. 

2. The retrieval algorithm may need more validation. The height-dependent sensitivity threshold and clutter-removal method are reasonable, but the final cloud occurrence, layer number, CBH, CTH, and CT depend on several empirical thresholds, including the variance criteria, the −20 dBZ filter, and the thin-layer merging/removal rules. These choices may affect thin clouds, weak high clouds, multilayer clouds, and precipitation-contaminated cases. Is it possible to include systematic validation against independent observations, such as ceilometer, radiosonde-derived cloud layers, lidar, or active satellite products?

3. The spatial representativeness and scale matching should be discussed more carefully. Although the network is impressive, it is still a set of point measurements rather than an area-covering observing system. In addition, the MMCR observations are matched with ERA5 fields at 0.25° resolution, which may not represent local cloud environments well, especially over complex terrain or regions with strong surface heterogeneity. The authors may consider the sensitivity of key results to station distribution, sparse regions, complex terrain, and the point-to-grid matching method. 

4. The physical attribution is still mainly based on correlation. The reported relationships between q850, LTS, WS700, and cloud properties are plausible, but the current analysis does not fully separate the effects of season, region, cloud type, and large-scale weather regime. For example, the positive relationship between q850 and CBH may partly reflect the co-occurrence of high humidity, deeper boundary layers, and convective regimes in summer, rather than a direct moisture effect on cloud base. 

5. CT should be treated carefully, since it is defined as the distance between the highest cloud top and lowest cloud base and may include clear gaps in multilayer cases. Clarifying this definition would make the interpretation of CT correlations more reliable.

6. The use of fixed pressure-level environmental variables needs caution, especially over the Plateau area. The manuscript uses q850 and an LTS definition based on 700 hPa and 1000 hPa, but these pressure levels may be below ground or not physically representative at high-elevation sites. 

The manuscript led by Hui Xu and Jianping Guo presents a national-scale characterization of cloud vertical structure across China using a ground-based millimeter cloud radar network. The topic is important, and the dataset has potential value for understanding regional cloud climatology. The authors state that, to their knowledge, this is the first national-scale, high-frequency characterization of cloud vertical structure across China using a coordinated ground-based radar network, and that the study systematically quantifies thermodynamic, dynamical, and land-surface controls.

I have several major concerns and questions that need to be addressed before publication. In particular, the manuscript would benefit from a more explicit discussion of how these radar-derived statistics could be used (e.g., the connection between the observational findings and their implications for model evaluation or improvement, especially from only one year of data). I also think the authors need to be cautious about the interpretation of single-layer clouds and land-surface coupling.

Major: 

1.  I recommend adding a more explicit discussion of how these radar-derived statistics could be used. Please clarify how the findings in this study can help improve weather and climate predictions. At present, the manuscript describes the observational characteristics well, but the pathway from these findings to improved model prediction is not sufficiently clear.
2. The use of only one year of data needs more justification. The results appear to be based primarily on one year of observations, 2024. The authors should clearly state this limitation and explain how representative 2024 is relative to longer-term climatology. One year of data can provide valuable high-frequency observational statistics, but it may not be sufficient to draw robust climatological conclusions or to claim direct implications for model improvement. The authors should consider discussing whether 2024 was meteorologically typical or anomalous, and whether interannual variability could affect the reported cloud vertical structure statistics.
3. Interpretation of single-layer clouds and land-surface coupling. Please clarify whether the single-layer cloud category includes high-level clouds. If so, the discussion in Section 3.7 may need to be reconsidered. High-level single-layer clouds are not necessarily strongly coupled to land-surface processes, whereas low-level clouds are more directly influenced by boundary-layer thermodynamics and land-atmosphere interactions. One suggestion could be: separating single-layer clouds by cloud-base or cloud-top height before discussing land-surface controls. Otherwise, the relationship between single-layer cloud occurrence and land-surface properties may be difficult to interpret physically.
4. Please explain the scientific motivation for separating two-layer clouds from multi-layer clouds. 

Specific comments:

1.Line 166: Please clarify what is meant by “valid observation days”. Does this refer to days with continuous/full-day radar observations, or days that meet a minimum data availability threshold? If a threshold was used, please specify it. 

2.Please clarify whether radar wind profiler data were used during precipitation periods. If so, how were precipitation-contaminated periods handled? The reliability of RWP retrievals can vary substantially under clear-sky, cloudy, and precipitation conditions. Please discuss how the authors assessed the trustworthiness of RWP observations under these different conditions, and whether any filtering or quality flags were applied before using the data in the analysis.

3.Line 187: The manuscript should provide more details on the quality control procedures applied to the observations. For example, what types of artifacts were removed? Were thresholds applied to signal-to-noise ratio, reflectivity, velocity, or other radar moments? Were insects, precipitation, ground clutter, or non-meteorological echoes filtered? Were quality flags provided by the instrument data stream used? Because cloud vertical structure statistics can be sensitive to radar sensitivity and QC choices, a more detailed description is needed to make the results reproducible and interpretable.

4.ERA5 data have an hourly temporal resolution. Did the authors average or aggregate the MMCR observations to hourly resolution before matching them with ERA5? Please describe the temporal matching procedure clearly. 

5.Line 223: The authors compare the detection limit of the China Ka-band cloud radars with ARM Ka-band radar observations. However, radar sensitivity depends strongly on hardware settings, calibration, operating mode, antenna configuration, pulse length, integration time, and processing algorithms. Please clarify how the authors confirmed that the detection limits are comparable between the ARM Ka-band radar and the China radar network. Were instrument specifications, calibration documents, or sensitivity estimates used? If not, the authors should be more cautious in making this comparison.

6.Line 295:I do not fully agree with the interpretation that the lower radar-derived cloud fraction relative to the MODIS long-term mean necessarily reflects a declining trend in total cloud fraction over East Asia. This difference could also be related to fundamental differences between observing systems..

7.Line 318: The authors correctly state that satellite products represent areal cloud fraction over large grid cells, whereas MMCR measurements capture local, vertically resolved cloud occurrence. This is an important point. However, this also means that the manuscript should be very cautious when comparing satellite cloud fraction with MMCR-derived temporal cloud fraction. These two quantities are not directly equivalent. I suggest revising the relevant discussions.

8.Line 326:The discussion in this section relies mostly on citing previous studies. I suggest adding more analysis. 

9.The authors state that cloud-base height is important for regional weather forecasting and climate change assessment, but this point needs clearer explanation. Please describe more specifically why CBH matters. 

10.Line 390: Please check the citation of Shupe et al. 2011. I guess that Shupe et al. focused primarily on Arctic (or high-latitude) cloud observations. If so, the citation may not be directly applicable to mid-latitude or China-region cloud regimes.