All comments are found in the attached supplement.

Recommendation: Major revisions

--- Overview

This study develops a new radar observation operator within a JEDI ensemble-variational data assimilation framework. The new operator is consistent with the forecast microphysics scheme and demonstrates notable advantages in a high-impact squall line event.

The authors provide an excellent evaluation of the new radar operator, including detailed comparisons with the still-operational GSI system, as well as operational forecasts from HRRR and the experimental RRFS. I appreciated the discussion of how radar increments project onto different hydrometeor species, along with the combination of subjective and objective verification statistics. The results from the new radar operator in JEDI are promising and represent an important step toward broader adoption of JEDI workflows.

I only have two major comments, which are summarized below, and followed by minor suggestions.

--- Major comments

1. The introduction text up to line 75 should be considerably shortened, as it does not directly relate to the research topic of this paper. Instead, the authors should focus the narrative on the subsequent text which pertains to the type of radar DA issues addressed in this manuscript.

2. The results in Fig. 4 are promising. However, unlike radial velocity, the reflectivity-based innovation statistics in Fig. 4 do not show signs of convergence toward stable values, indicating that six cycles may not be sufficient. In the DA literature, fair comparisons traditionally discard these spin-up periods during which DA diagnostics exhibit mean trends. Since the paper focuses on a single case, I believe extending the experiment to additional cycles is worth pursuing and would provide a more convincing demonstration of the benefits of the new radar operator developments in JEDI.

--- Minor comments

Please check whether acronyms for ensemble-variational methods are used consistently throughout the paper. For example, lines 50–51 state that Wang and Wang (2017) developed an En3DVar system; the correct acronym should be 3DEnVar.

Line 114: This refers to the radar reflectivity factor, not simply reflectivity.

Fig. 1 and surrounding discussion: The revised text should clarify how the WRF reflectivity values are calculated.

Section 3.2.2: It would be helpful to write out the full DA equations to make this paper self-contained.

Line 235: The NODA experiment does not assimilate radar data. What about other conventional observations? Please clarify in the revised manuscript.

Lines 256–258: Can you briefly comment on what causes the positive biases?

When referring to states in spatial maps, it would be helpful to label them directly on the maps. It would also be helpful to include title labels for each plot; for example, experiment names in Fig. 12.

Fig. 9: In addition to underestimating surface temperatures, it should be noted that the DA experiments produce a stronger outflow. This behavior is consistent with cold pool dynamics (stronger cold pools correspond to stronger winds).

Line 463: “The NODA experiment … with weak, spatially inconsistent probabilities” — The NODA probabilities appear to be much higher than those of any other experiment; please reword this statement.