This manuscript integrates an ensemble Kalman filter-based non-linear data assimilation method with an atmospheric chemistry transport model (CTM). The primary advancement is the coupling of the hybrid Kalman-Nonlinear Ensemble Transform Filter (KNETF) with an adaptive forgetting factor to the CTM model. The method was tested using a real-world dataset, with experiments varying ensemble sizes and evaluated against multiple metrics. The presentation quality is good, though some minor issues need to be addressed:

Line 15: Replace "difficulty" with "challenge".

Line 147: The term "level-2" is not adequately introduced. Consider moving the reference from Lines 152-153 to the beginning of the paragraph for clarity.

Line 215: Properly cite online resources instead of directly inserting hyperlinks.

Figure 3: The coloring of Domain 3 is difficult to distinguish.

Line 395: Remove the word "deeply".

Figure 4: It is unclear from the manuscript whether the experiments were run multiple times, particularly for plot d). The stochastic nature of this method may introduce variation in running time. Indicate whether the presented values are the mean of multiple runs or include the mean and uncertainty band for multiple runs.

By addressing these points, the manuscript will be clearer and more robust.

This paper builds on earlier developments in the NAQPMS-PDAF model. My main concern is that the writing reads very awkward in many places, please make sure the use of words is bringing what you wish to convey. Several sentences were conflicting or confusing because of ambiguous or inappropriate choice of words.

L13, 32 the first lines of the abstract and introduction are too similar.

L14: please note the difference between accuracy and precision… accuracy (accurately) fits the context more than precisely

L15: change difficulty to challenge, or a difficulty to challenging.

L33: delete diversely, it’s unclear if the influence is diverse or the subjects are diverse

L37: “insufficient in interpreting PM2.5 chemical components” This is untrue or at least in literal text, this is claiming that measurements are not enough to interpret PM2.5 chemical components. However, observations have shown different aerosol composition (see AMS data). Please consider revising this claim. It doesn’t seem to be what you want to say.

L43: What kind of “biases relative to real situation”? How?

L91: Delete Besides or use another word.

Section 2.1 please expand the description for people unfamiliar with the model. This is a GMD paper after all.

L116: “PDAF has offline and online modes.”

L117: “, which is easy to write code” does not fit here.

L119: “instead of twice independently” What does this mean? what needs twice independently?

Figure 1: please write the description in the captions and discuss it in the text. 

L153-171: consider using a flow chart to illustrate the steps.

L172: Configurations

L217-218 The target PM2.5 chemical components are NH4 + , SO4 2- , NO3 - 217 , OC, and EC, and the perturbed emission species correspondingly 218 include SO2, NOx, VOCs, NH3, CO, PM10, PM2.5, EC, and OC,” You have a target of 5, and corresponding to 9 species… please be specific.

L343: what do you mean by superiority and how is it measured and considered?

L352: situation ->scenario or test

L358: “The last test was like the first but with a different situation” This is too colloquial 

Figs 6-10: Figures are unreadable. Too small. Please re-plot.