The revisions significantly improved Zhao et al manuscript. The choice of mixing Results and Discussion improves the readability. The drawback is that some of the statements are not supported by observations. This was already the case in the previous version of the manuscript. I encourage the authors to read through their manuscript and remove or rephrase some of their claims that are not supported by their observations. Specific examples are listed in the comments below. For example I would remove the section/paragraph about tidal mixing and future state that are not linked to the current results.
L168: Do you have supporting observations for this assumption?
L199-200: Mixing Results and Discussion doesn't mean you can extrapolate your results without supporting observations. The wording is important. In this sentence a better wording could be: "The freshwater bulge also featured a relatively weak bloom, with Chl a concentrations of ~10 μg L-1 and DO of ~250 μmol kg-1 (equivalent to a DO saturation level of ~ 125 %). Similar blooms were previously associated with high nutrient concentrations and a long water residence time in the plume (Lu and Gan 2015)."
L201-202: That does not seem like a relevant explanation for the lower water temperature. 1) there is no indication that the lower surface temperature is the direct result of the air temperature, 2) air temperature is lower only at the end of the leg for a short period of time and 2) air temperature could be the same in the other areas.
"during data collection", "at the time of sampling" would be better than "visits".
L204: Bottom shelf waters are found in z>10-20m below plume waters. They are not "intruded". The plume is attached to the bottom in the inshore area (0<z<20) according to Figure 4.
L208-209: please rephrase "a region... measurements"
L212-213: how did you calculate these areas?
L215-216: please remove this statement. You cannot draw long term conclusions by comparing 2014 and 2018. Also, the sampling area was different.
L224-225: That is not supported by your observations from Fig S1. Air temperature at the Hong Kong observatory was >28 at this time.
L229-230: Same comment here, be careful with the wording. The bloom is within the plume and therefore more likely associated with the riverine nutrient input.
L239-241: "Although... pycnocline": This is not yet supported, please remove.
L249-250: provide supporting observations or remove this statement
L265: What is the depth of this subsurface (mid-) layer shown in Fig. S2?
L265-266: "likely... winds". again, this is not shown
L272: "expanded... isobath": where is this shown?
L274-277: Again, this is not supported by your observation. It can be discussed (as in a regular discussion) but you cannot make this type of conclusion based on your observations.
L282: This section title is a bit redundant with the previous subsection titles. In part 3 you look at the temporal evolution of hypoxia (as stated in the section/subsection headers) and in part 4 you look into the mechanisms that lead to the temporal evolution observed in part 3. This should be reflected in the section headers.
L292: The time series is only 26h long so you shouldn't draw too many conclusions. The end of the time series indicate that the location of the plume varies rapidly in space/time
L292-298: all hypoxic centers occur in cold/salty waters during leg 1.
The terminology "intrusion" is misleading as it tend to indicate that offshore bottom waters are advected onshore, whereas it seems that it is the bottom-attached plume that prevents offshore bottom waters to reach the z<20m area.
L317-340: You could shorten this paragraph to avoid redundancy with section 3.
L337: "with a limited spatial extent beneath the surface plume": What do you mean? vertical extent?
L345-360: You speculate here, this paragraph should be reduced to one short sentence where you mention that tide could influence your estimates of hypoxia extent.
L362-370: this paragraph is not necessary, you should focus on your results. In general, it is better to start a section by stating your results and then develop.
L371: "On the condition of a precedent restoration of density stratification": can you rephrase for clarity?
L385-394: The main center of hypoxia off Modaomen occurs where there is no O2 mixing (Figure 8d), but you seem to suggest that this area is influenced by the upwelled waters SW of Hong Kong. The two centers of hypoxia (off Modaomen and off Hong Kong) seem to have different dynamics.
L405: what do you mean here?
L416-418: please rephrase, this was not observed in your study.
L421: "only the western hypoxic centre was located"
This is linked to the comment above (L385-394): it suggests different mechanisms for hypoxia formation between the two hypoxia centers?
L439-441: Is there a reason? it would be interesting to mention it
L457: See previous comments about intrusions
L444-462: This paragraph is not very clear and I am not sure that it adds anything to your story.
L472-473/480-481: The hypoxia formation time scale (6-12days) is shorter than the estimate of residence time. Also 15 days in Li et al (2020) is a depth-averaged estimate so residence time in bottom waters could be significantly longer.
L475: "shoreward-intruded": same comment as before, "shelf bottom waters" is more appropriate
L498-501: rephrase for clarity.
what does that mean: "had the potential to overwhelmingly destroy the stability of the water column"? How is this defined?
L508-509: This sentence is odd.
L511: suggestion: "Although strong winds constrain the development of hypoxia, tropical cyclones potentially..."
L515-516: The reference is in chinese and therefore not accessible to most reader, please remove. You can provide a time series of rainfall in the supporting material if necessary or add it to Figure 1. However, what you mention here is river discharge and it would be more appropriate to show a time series of freshwater discharge in the PRE. This would fit well in Figure 1.
L518: suggested change: "... PRE did not increase significantly". See comment above about the time series of river discharge.
L521-522: remove "as shown in Fig. 3 that strong blooms occurred in the surface plume along the coast with much higher Chl a concentrations during Leg 2 than that during Leg 1."
L526: Figure 4 show that O2 is decreasing in bottom waters but hypoxia did not re-establish yet.
L542: the decreasing frequency is in the tropics but not necessarily further north, e.g. a recent study with similar conclusions: https://advances.sciencemag.org/content/6/51/eabd5109
L551: For clarity you could say that the development of hypoxia depend on a trade-off between between storm intensity and frequency.
L557-568: this is more a summary than conclusions.
L564: suggested modification: "This hypoxia timescale is shorter than water residence time but comparable to the disturbance of hypoxia..."
L566-568: please remove
L572: http://www.geodata.cn: the website doesn't seem to be available on english. Will the data be available to the non-chinese speaking readers?
Figure 9c: why is there negative time intervals?
L191/217/244: Suggestion: add the leg # for each phase in the header, i.e. "(Leg 1)" for section 3.1
L204 and all occurrences elsewhere: bottom seem more appropriate than benthic
L209: suggestion: "The general spatial pattern of hypoxic centers..."
L228: suggestion: Large phytoplankton blooms were identified in the surface plume between...
L235: remove "however"
L550: remove "enduring"
Figure 7: it should be: "Profiles of temperature (°C) (blue dashed lines)"
Figure S2: DO: the color bar is not complete
L311: (e.g., stations A8 and A11, Fig. 7a)