the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 09 Jul 2026
Adaptive Observation Weighting in TCKF1D-Var for Ground-Based Multi-Sensor Thermodynamic Retrievals Prior to Nocturnal Heavy Precipitation over China
Abstract. Ground-based microwave radiometers (GMWRs) and Mie–Raman lidars (MRLs) provide valuable thermodynamic observations for atmospheric profiling, but conventional variational retrieval frameworks typically rely on static observation weighting assumptions that may not adequately represent varying observation quality under precipitation conditions. To address this limitation, an adaptive observation weighting framework based on the Thermodynamic-Constrained Kalman Filter 1D-Var framework (TCKF1D-Var) is developed and evaluated using 107 nocturnal heavy-precipitation cases. The proposed method dynamically estimates observational contributions during the retrieval process and is applied to GMWR, MRL, and GMWR–MRL synergistic retrievals. Retrieval performance is assessed against radiosonde observations and compared with that of a conventional static-weighting TCKF1D-Var framework. Results show that the adaptive weighting approach consistently improves retrieval accuracy, with the largest benefits found for water vapor mass mixing ratio profiles. For both GMWR and MRL retrievals, reductions in mean bias and root-mean-square error are obtained relative to the static-weighting framework. The synergistic retrieval further improves moisture-profile retrievals and generally achieves the best overall performance among all experiments. Diagnostic analyses reveal that the adaptive framework dynamically adjusts the utilization of observational information according to sensor characteristics and atmospheric conditions, while redistributing observational influence between GMWR and MRL measurements during synergistic retrievals. These results demonstrate that adaptive observation weighting provides an effective strategy for improving thermodynamic profile retrievals under heavy-precipitation pre-onset conditions.
- Preprint
(5224 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-3341', Anonymous Referee #1, 09 Jul 2026
-
AC2: 'Reply on RC1', Qi Zhang, 08 Sep 2026
Dear Reviewer，
Thank you very much for your positive evaluation of our manuscript and for your recognition of the work presented in this study. We sincerely appreciate your time and consideration.
Following the suggestions provided by the other reviewer, we have carefully revised the manuscript and made corresponding improvements to further clarify the methodology, strengthen the discussion, and improve the overall presentation of the study.
We are grateful for your support and encouragement. Should you have any further comments or suggestions regarding the revised manuscript, we would sincerely appreciate your guidance and constructive feedback.
Thank you again for your time and consideration.
Best regards,
Qi Zhang
On behalf of all co-authorsCitation: https://doi.org/10.5194/egusphere-2026-3341-AC2
-
AC2: 'Reply on RC1', Qi Zhang, 08 Sep 2026
-
RC2: 'Comment on egusphere-2026-3341', Anonymous Referee #2, 05 Sep 2026
This manuscript presents an adaptive observation weighting framework based on the TCKF1D-Var method for synergistic retrieval of atmospheric thermodynamic profiles from ground-based microwave radiometer and Mie–Raman lidar observations before nocturnal heavy precipitation events. The authors introduce dynamically optimized observation weights into the variational retrieval framework and demonstrate the performance using 107 nocturnal heavy precipitation cases over China. The study further investigates the contribution of individual GMWR channels and the vertical distribution of MRL observation weights.
The manuscript addresses an important scientific problem in atmospheric remote sensing and data fusion. Compared with traditional retrieval methods that assume fixed observation uncertainties, the proposed adaptive weighting approach provides a potentially useful strategy to dynamically adjust the relative contributions of multi-source observations under varying atmospheric conditions. The diagnostic analysis of observation weights also provides additional insights into the information content of different observations.
Overall, the manuscript is well organized, the methodology is clearly described, and the results demonstrate the potential advantages of the proposed framework. However, several aspects require further clarification before publication. In particular, the interpretation of diagnosed weights, and the limitations of the current validation strategy should be strengthened. Some conclusions regarding the physical meaning of adaptive weights should also be moderated because the current experiments mainly demonstrate statistical improvements rather than directly quantify observational information content. I therefore recommend minor revision.
Major Comments:
- The theoretical interpretation of adaptive observation weights requires further clarification.The central innovation of this study is the introduction of adaptive observation weights into the TCKF1D-Var cost function. However, the current manuscript does not sufficiently explain the theoretical meaning of these optimized weights.According to Eqs. (3)–(5), the observation weights are introduced as adjustable parameters during the minimization process. However, it remains unclear whether these weights represent the relative information contributions of different observations or the mathematical scaling factors introduced to optimize retrieval performance. The manuscript frequently interprets large weights as indicating higher observation information content. However, the relationship between optimized weights and formal information content is not rigorously demonstrated. For example, the manuscript states that certain GMWR channels with large weights provide dominant observational constraints and that higher-altitude MRL observations receive larger confidence. However, because the weights are determined jointly by observation residuals, background uncertainties, and the retrieval framework itself, these interpretations may not necessarily indicate the intrinsic information content of the observations. The authors should clarify the physical interpretation of the adaptive weights, whether the diagnosed weights are comparable among different instruments and observation types, and what assumptions are required when interpreting weights as information contributions.
- The validation strategy should better separate retrieval improvement from possible effects of the background field.The manuscript evaluates retrieval performance against radiosonde observations and compares adaptive weighting with static weighting. This approach is reasonable. However, because ERA5 profiles are used as prior information, the retrieved profiles may partly inherit characteristics from the background field. The manuscript should provide more discussion regarding whether adaptive weighting mainly adjusts observation influence or compensates for deficiencies in ERA5, and whether the improvements are consistent under different background uncertainties.
- The physical interpretation of precipitation-dependent performance requires further discussion.The manuscript shows that the advantage of adaptive weighting becomes weaker under precipitation exceeding 30 mm/hr, especially for GMWR-only retrievals. This is an interesting result; however, the explanation that increased observational uncertainty reduces the effectiveness of adaptive weighting remains speculative. The authors should clarify whether precipitation contamination effects were explicitly considered in the observation processing, and whether the signal attenuation and cloud effects influence the results.
Minor Comments:
- Clarify the novelty compared with previous TCKF1D-Var studies.The manuscript extends previous studies by Zhang et al. (2026a,b). However, the distinction between the previous TCKF1D-Var framework and the current adaptive weighting version should be highlighted more clearly in the Introduction. A concise comparison table showing the differences between previous method and current improvement would help readers understand the methodological contribution.
- Terminology of “dynamic weight” and “adaptive weight”.The manuscript uses both “adaptive observation weighting” and “dynamic weight” to describe the proposed method. Please consider using one consistent terminology throughout the manuscript. “Adaptive observation weighting” appears more appropriate because the method adjusts weights during optimization rather than necessarily evolving with time.
- Definition of observation weights should be added explicitly.The manuscript should provide a clear mathematical definition of the weight coefficient. For example, Is a weight of 1 equivalent to the original observation contribution? Does a weight smaller than 1 indicate reduced confidence? Can weights larger than 1 occur? These details are important for reproducibility.
- The manuscript uses MRL observations under heavy precipitation conditions. Since Raman lidar performance can be strongly affected by clouds and precipitation attenuation, the manuscript should provide more details about the cloud screening, the missing data treatment, and the signal quality criteria.
- Clarify the retrieval vertical resolution.The manuscript discusses different vertical resolutions of GMWR and MRL observations, but the retrieval framework appears to use a common vertical grid. Please clarify the vertical resolution of the final retrieved profiles and how vertical resolution mismatch is handled.
- Some statements in the conclusion imply that adaptive weights reveal physical observation information content. Since this interpretation is not directly validated through formal information-content analysis, the wording “information content ” should be moderated in Lines 74, 81, 287, 309, 422, 645, 679.
- Line 19, “dynamically estimates observational contributions” can be replaced with “dynamically estimates the relative contribution of individual observations” for better understanding.
- Lines 22–23, “with the largest benefits found for water vapor mass mixing ratio profiles” should be “with the most significant improvements found for water vapor mass mixing ratio profiles”.
- Line 52, “a MRL” should be “an MRL”.
- Line 114, “The MRLs emits” should be “The MRLs emit”.
- Line 197, “sample amount” should be “sample size”.
- Lines 203–204, “leads to a improved retrieval performance” should be “leads to an improved retrieval performance”
- Lines 206 and 207, “framewrok” should be “framework”.
- Lines 226–227, “prior thermodynamic profiles generated from ERA5 reanalysis” should be “prior thermodynamic profiles generated from the ERA5 reanalysis”.
- Table 1, “Case Amount” should be “Number of Cases”.
- The title of Figure 11b should be “Hourly Precipitation ∈[10, 20) mm”, please make corrections.
- Lines 462–463, “The fact that the adaptive observation weighting framework consistently achieves” can be changed to “The consistent lower retrieval errors achieved by the adaptive framework suggest that” for better understanding.
Citation: https://doi.org/10.5194/egusphere-2026-3341-RC2 -
AC1: 'Reply on RC2', Qi Zhang, 08 Sep 2026
Dear Reviewer,
Thank you very much for your careful and constructive review of our manuscript. We sincerely appreciate the time and effort you have devoted to evaluating our work and for providing such detailed and insightful comments.
We have carefully considered all of your comments and suggestions and have addressed each of them point by point in the revised manuscript and the accompanying response letter. We have also made corresponding revisions throughout the manuscript to clarify the methodology, strengthen the discussion, and better explain the limitations and scope of the proposed approach.
Please find the detailed point-by-point responses in the attached response letter. We hope that our revisions and responses adequately address your concerns and suggestions.
Thank you again for your valuable time and constructive comments, which have helped us substantially improve the manuscript.
Best regards,
Qi Zhang
On behalf of all co-authors
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 213 | 56 | 28 | 297 | 24 | 24 |
- HTML: 213
- PDF: 56
- XML: 28
- Total: 297
- BibTeX: 24
- EndNote: 24
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The authors have answered all my questions properly.