Direct Assimilation of Radar Reflectivity Data Using Ensemble Kalman Filter Based on a Two-Moment Microphysics Scheme for the Analysis and Forecast of Typhoon Lekima (2019)
Round 1
Reviewer 1 Report
Review of “Direct Assimilation of Radar Reflectivity Data using Ensemble 2 Kalman Filter based on a Two-Moment Microphysics Scheme 3 for the analysis and forecast of Typhoon Lekima (2019)” for Remote Sensing
Recommendation: Accept with minor revisions
This paper addresses the innovative use of a new data assimilation scheme in order to use the radar reflectivities and velocities from a coastal radar for a landfalling typhoon. I have only minor comments for the authors to address, as it appears that the work advances the science and is well-formulated. One limitation is that this is just for one case, though the authors acknowledge this in the summary. My comments are coming from the perspective of a tropical cyclone forecaster, not a data assimilation expert. It is hoped that a reviewer with expertise in data assimilation was also obtained.
All of the model simulations suffer in not getting the initial intensity correct. Please discuss possible reasons why including errors in the best track database.
Given that coastal radars are only available right before landfall, their incorporation has limited ability in assisting typhoon forecasts. How can this work be extended to be more useful with potential new observing systems in the future?
Line 55: “over” – should be “other”?
Line 163: “none” – should be “non”?
Line 206: Please provide a reference for the CMA best track database.
Figure 4a and 4b: Please use a different color line for the coastline. Also neither show a double wind max structure that would be suggested from Figure 3's concentric eyewall structure.
Line 247: Should be “1200 UTC 9 August to 0000 UTC 10 August"
Line 248: "significant" would imply statistical significance. Surely not the case with the sample of 1 forecast.
Figure 5: What are the landfall times for each case?
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This is a well written manuscript addressing the issue of assimilation of radar reflectivity data, using the EnSRF data assimilation method to initialize 3 km WRF model and to investigate the impact of assimilated data on the short term forecast of typhoon Lekima (2019). The results indicate a good and expected performance of the data assimilation, resulting in best typhoon track and intensity forecasts when the data assimilation system including all the model state variables is used (Z-DA). I suggest accepting the manuscript for publication, subject to the following minor comments.
Minor comments:
(1) Page 4 lines 170-174: What was the reason for using random perturbations to define initial WRF ensembles? Have you tried using GFS ensembles instead?
(2) Page 4, line 175: Please explain how the lateral boundary conditions are defined? Are they the same for all ensembles?
(3) Page 6, Figure 4 a-b: The experiment Z-DA (Figure 4b) seems to have small scale noise in the pressure field. Please provide a possible explanation for that.
(4) Page 9, Figure 8: Please include experiment CNTL into the figure, as a reference for the expected minimum noise level.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Review of “Direct Assimilation of Radar Reflectivity Data using Ensemble Kalman Filter based on a Two-Moment Microphysics Scheme for the analysis and forecast of Typhoon Lekima (2010)”, by Luo, J., Li, H., Xue, M., and Zhu, Y.
This paper aims to evaluate the impact of directly assimilating radar reflectivity data using ensemble Kalman filter (EnKF) based on a DM MP scheme in WRF model data assimilation (DA) framework for the landfalling typhoon Lekima. They compared the Z-DA results which assimilate radar reflectivity data with the baseline control experiment. The intensity and track forecast errors results were discussed, as well as the RMS and PS tendency.
One major question for this paper is that the methodology of development of radar reflectivity assimilation in GSI-based EnKF is not clear. The authors need to investigate the impacts of the different hydrometeor state variables for direct assimilation of reflectivity. The authors also want to show the equation how the equivalent reflectivity dBZ is obtained. It is recommended to add a methodology section.
Second, the results from this paper are based on one single case and lacks scientifically in depth discussions. The track and intensity forecast comparison between the CNTL and Z-DA is not statically significant due to the very limited sample size. The RMSI of simulated reflectivity in the experiments doesn’t show any significant improvement with adding Z-DA either.
Third, there is issue with the quality control procedure used in this study. What are the thresholds being used in the experiment? Observations within certain thresholds likely is not able to distinguish precipitation from ground clutter, birds, etc. The authors need to include the horizontal and vertical localization scales that were applied to EnKF as well. What method is used to account for the deficiency of the spread of the first guess ensemble in the EnKF? I also did not see any gross error check being used in the QC. In many other publications, observations within 5 and 10 dBZ were not assimilated, as the reflectivity within these thresholds likely is not able to distinguish precipitation from other sources. The authors need to carefully address all these potential QC issues which will have a large impact on the assimilation results.
Fourth, the reflectivity analyses and forecast discussion is not completed. It is very hard to draw any conclusion from Fig. 3, the reflectivity observation at 15-min time interval during the forecast period from both CNTL and Z-DA is needed for visualization and comparison. The authors need go through the results in great detail, and add some in depth discussions.
Overall, the level of scientific contribution of this work has not met the Remote Sensing journal. The authors are encouraged to resubmit the manuscript after the aforementioned issues are addressed.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
Review of revised manuscript titled “Direct Assimilation of Radar Reflectivity Data using Ensemble Kalman
Filter based on a Two-Moment Microphysics Scheme for the analysis and forecast of
Typhoon Lekima (2010)”, by Luo, J., Li, H., Xue, M., and Zhu, Y.
I am pleased to see that my previous comments have been addressed and the level of scientific contribution of the modified manuscript has now met the Remote Sensing journal with minor changes needed.
Page number and comments:
P3, 2.4: It is not clear to the reader if any conventional observations were assimilated. The authors may want to add one or two sentences to address this.
P15: Please rearrange the references in alphabetical order of the authors’ last names.
