Please see the attached file.

General Comments

The manuscript entitled “Extended TCKF1D-Var framework for Mie–Raman Lidar Water Vapor Profiling in the Nocturnal Boundary Layer: Insights into Pre-precipitation Moisture Evolution” presents an extension of the previously developed TCKF1D-Var framework by incorporating Raman-channel observations from a ground-based Mie–Raman lidar network. The study develops a physically constrained lidar observation operator, integrates dynamically estimated covariance matrices within a Kalman filtering framework, and evaluates the retrieval performance using radiosonde observations prior to nocturnal heavy precipitation events across China.

Overall, the manuscript addresses a relevant and timely topic within the scope of AMT. The combination of Raman lidar observations with a variational retrieval framework for boundary-layer moisture analysis is potentially valuable for future nowcasting and data assimilation applications. The manuscript is generally organized, and the multi-case statistical evaluation represents a clear strength of the study. Compared with many retrieval-oriented studies, the present manuscript makes a commendable effort to connect the retrieved moisture structures with pre-precipitation boundary-layer evolution. In addition, the authors provide a relatively balanced discussion of current limitations and uncertainties, which improves the overall credibility of the work.

While the manuscript is already in reasonably good shape, several methodological and interpretative aspects could still benefit from additional clarification. In particular, some discussions related to the covariance estimation strategy, the innovation-based diagnostic metric, and the interpretation of the retrieval increments relative to ERA5 could be refined to improve clarity and avoid possible over-interpretation. I therefore recommend publication after major revision, although most of the required revisions appear achievable through clarification, expanded discussion, and moderate textual revision rather than substantial methodological redevelopment.

Major comments:

Additional clarification of the dynamically estimated covariance matrices should be demonstrated.

The dynamically estimated observation and process covariance matrices represent an important component of the proposed framework. The current manuscript already provides a clearer explanation than many similar retrieval studies, and the overall methodology appears reasonable. Nevertheless, readers may still benefit from a slightly more detailed discussion regarding whether the estimated covariance structures remain relatively stable under rapidly evolving pre-convective conditions.

In particular, since the manuscript retains off-diagonal covariance terms to represent vertically correlated uncertainties, it may be useful to provide one or two representative examples of the covariance matrices. This would help readers better understand the physical and numerical behavior of the retrieval system without substantially increasing the complexity of the manuscript.

The interpretation of the innovation-based diagnostic metric should be clarified further.

The innovation-to-forward-model-error ratio introduced in the manuscript provides a useful diagnostic perspective for interpreting the sensitivity experiments associated with different temporal averaging windows. I also appreciate that the authors clearly state that this quantity is intended primarily as an interpretative indicator rather than a rigorous information-theoretic metric.

However, some additional clarification regarding the intended scope and limitations of this quantity would still be helpful. For example, readers may wonder whether the metric is primarily framework-dependent, how directly it relates to retrieval information content, and whether it should be interpreted qualitatively rather than quantitatively. I do not believe a formal information-content analysis is necessary for the current study. Nevertheless, maintaining consistently cautious wording throughout the manuscript would help avoid possible ambiguity regarding the interpretation of this diagnostic quantity.

The interpretation of the pre-precipitation moisture structures need clarification.

One of the strengths of the manuscript is the effort to relate the retrieval increments to physically meaningful pre-precipitation moisture evolution. The presented results appear broadly consistent with enhanced lower-tropospheric moistening prior to heavy precipitation onset.

At the same time, it may be helpful for the manuscript to acknowledge somewhat more explicitly that part of the retrieved moistening signal could also reflect correction of ERA5 prior biases under convective boundary-layer conditions. Since the current validation is based primarily on radiosonde comparisons, a brief clarification regarding the distinction between retrieval enhancement relative to the prior and independent confirmation of atmospheric moisture evolution would help readers interpret the results more carefully by moderating several statements related to physical attribution to ensure that the conclusions remain closely aligned with the evidence currently presented.

Physical interpretation of the precipitation-category-dependent structures need to be strengthened.

The manuscript presents interesting differences in the MB and RMSE structures among the three precipitation-intensity categories, particularly for the ≥30 mm cases. The revised discussion is improved, especially regarding possible observational limitations under strong precipitation conditions.

However, the interpretation still remains descriptive. I encourage the authors to expand the discussion further, particularly regarding the possible role of enhanced turbulence and vertical mixing, cloud contamination and attenuation effects on Raman lidar observations, and how rapidly evolving moisture gradients may influence retrieval uncertainty. A more physically grounded interpretation would increase the broader meteorological relevance of the study.

Minor comments

Figures 7–9 are informative. However, precipitation onset timing could still be indicated more explicitly within the figures themselves, rather than only in the text description.

Please clarify more explicitly whether the reported 30 m vertical resolution refers to retrieval grid spacing, native lidar resolution, or effective retrieval resolution after filtering and inversion.

The discussion regarding computational efficiency and operational applicability is useful, but a brief estimate of computational cost per retrieval cycle would further improve the practical relevance of the manuscript.

Some grammatical and stylistic issues remain throughout the text. Although none are severe, an additional language editing pass would improve readability. For example, “All authors delcare no competing interest” should be “All authors declare no competing interests”, and “within boundary layer preceding” should include “the”.

The conclusions are generally balanced. However, several statements concerning future NWP assimilation applications may still benefit from slightly more cautious wording, given that no direct assimilation experiments are presented in the current study.