Review of ‘A bias correction scheme for FY-3E/HIRAS-II observation data assimilation’ by Chen and Guan.

Chen and Guan address biases between FY-3E/HIRAS-II thermal infrared radiances and model-simulated radiances. They propose two correction steps: (1) a scan-angle bias correction, which effectively removes a systematic viewing-angle dependent discrepancy, and (2) an air-mass bias correction using four predictors correlated with observation-minus-background (O–B) differences. While both corrections reduce bias, the underlying causes and physical interpretations remain insufficiently explained.

Major issues

1. Scan-angle bias origin unclear.
   
   The paper shows a clear scan-angle bias in HIRAS-II data but does not clarify whether this originates from the satellite sensor, the radiative transfer model (e.g., RRTOV), or atmospheric mismatches between observations and simulations. Given that TOA radiances naturally depend on viewing angle (due to path length, surface emission angle-dependence, and atmospheric heterogeneity), the authors should systematically analyze and discuss these potential causes.
2. Air-mass bias concept not well defined.
   
   The introduction should better distinguish scan-angle bias (instrumental/geometric) from air-mass bias (systematic model error in representing specific air mass types). The use of predictors to correct O–B needs a clearer explanation: what physical aspects are being corrected—surface temperature, tropospheric composition, or something else?
3. Inconsistent description of data assimilation setup.
   
   The abstract mentions NCEP-GSI, but the experiments use WRFDA v4.4. This inconsistency must be resolved. Also, the timeline of the assimilation period is contradictory (17–31 August vs. 1–31 August 2023). This needs clarification to maintain credibility.
4. Lack of physical explanation for air-mass correction.
   
   The correlation between the chosen predictors and O–B is shown, but there is no justification for why these predictors lead to better TOA radiance simulation. The study would benefit from showing how the air-mass correction improves key state variables (e.g., surface T, pressure, or humidity) and thereby reduces bias.

Minor corrections:

Title: FY-3E/HIRAS-II (no space after /)

L26: “satellite radiations” --> satellite radiance.

L34-35: “The satellite launched …time window”. This sentence is unclear. FY-3E has an early morning overpass, so how would this fill a time gap within a 6-hour assimilation window? IASI also has a morning overpass.

L46: “destroying the global NWP system” … this is oddly phrased. The system will surely not be destroyed, but the model state may be negatively influenced when biases persist in satellite observations. This should be rephrased.

L108: Period missing after Saunders-reference.

L135: “exceeds”--> exceeding

L148: rephrase to ‘For each channel, the global …position is calculated as:’

L300: “The scatters” --> The scatter

L380: the colour bar in Figure 7 is unfortunate with green hue indicating no bias. Suggest to use a cold-white-warm colour scheme, where white corresponds to zero bias.