This manuscript presents a comprehensive investigation of the interannual variability of summer shelf circulation in the Northern South China Sea (NSCS) from 2000 to 2022 based on ROMS modeling. The study makes significant contributions to understanding the differential impacts of ENSO and Pearl River Estuary (PRE) freshwater runoff on NSCS circulation dynamics. While the paper is generally well-structured and scientifically sound, several aspects require clarification and improvement before publication.

Comments:

1. Line 57-66, I recommend incorporating this study's findings and conclusions to more explicitly identify the shortcomings of existing research in quantifying the impacts of ENSO and river runoff on cross-shelf transport in the northern South China Sea.
2. Section 2, the authors have presented comparison results between observations and model simulations for temperature, salinity, and sea level. However, they did not mention the more crucial current observation data. How does the model perform in simulating shelf circulation? Of course, I understand that current observation data is relatively scarce, and model simulations of circulation may have greater errors. It would be preferable to provide some validation of circulation observations here.
3. Section 3.1, the MVEOF analysis serves as a critically important tool in this study. However, for readers unfamiliar with this methodology, the current paper provides insufficient explanation. I had to search online to understand the fundamental principles of this method before being able to properly follow the article's discussion. The introduction of MVEOF should be moved to Section 2 (Methods), and requires more detailed explanation.
4. Line 147, there is an extra closing parenthesis ")" in this line.
5. All variables in the equations should be clearly defined in the text. For example, the H and tao in Equ. 2.
6. As a researcher specializing in shelf material transport, I find the subject of this paper particularly compelling. The study provides in-depth dynamic analysis of ENSO and runoff impacts on cross-isobath transport, but lacks quantitative information that would be most useful for practical applications. For instance: （1）Missing baseline metrics: What is the approximate summer cross-shelf transport velocity in the northern South China Sea? This could be calculated from Figure 4b. Without concrete values, the results are difficult for other researchers to directly utilize. ​(2) Quantification of forcing contributions: Can the relative contributions of ENSO versus river discharge to cross-isobath transport be quantified?Including these quantitative results in the abstract or conclusions would significantly enhance the paper's citation potential and appeal to a broader scientific audience.

Review of "Interannual variability of summertime cross-isobath exchanges in the northern South China Sea: ENSO and riverine influences" by Song et al.

This manuscript studies the interannual variability of shelf circulation in the Northern South China Sea (NSCS). Low-frequency variability in the ENSO and the Pearl River Estuary (PRE) runoff are shown to have important effects in the spatial patterns of cross-isobath transport. Distinct regimes exist inshore and offshore of the 100 m isobath, respectively associated with bottom friction+nonlinear effects and stratification-dominated effects. The regions east and west of the PRE's outflow are also starkly different, being respectively dominated by variability in the PRE plume's volume and in the Kuroshio's intrusions.

The text, figures and tables could use minor improvements but read generally well, and the reasoning is easy to follow. The major issues I see with the manuscript are in terms of a couple of subjective choices, namely the regression analysis methodology and the criterion for identifying large-outflow years. I request that the authors address the following points:

Major points

M1 (lines 218-229): I am not sure I follow the need for the regression step in the two-stage regression approach. My understanding is that this analysis is a conditional average of the anomaly fields for each variable at the times when each variable's MVPC1/PC1 was in a positive phase, is that correct? I do not follow where the linear slopes calculated from the least-squares analysis are actually used. The time series of the MVPC1/PC1 should contain the relevant temporal variability of the leading EOF mode. Please clarify this paragraph.

To justify the need for a more elaborate method, the authors also need to compare it to the simplest one. How do all results in the paper compare to doing the same analyses just by conditionally-averaging the anomaly fields over years of positive MVPC1/PC1 phase?

M2 (line 197-199): Is there an objective criterion for choosing large-runoff years, like an outflow volume threshold? I think it is important to have one, and it should be described here.

Because the choice of the threshold is also arbitrary (e.g., it could be the years where the outflow was greater than the 75th or 90th percentile), a second step is to study the sensitivity of the results to this choice as well.

M3: I think a key result worth emphasizing is the identification of the different dynamical regimes in terms of their response to different ENSO/PRE plume drivers (inshore PGF_{y*}^b-dominated/offshore JEBAR-dominated and west Kuroshio intrusion-dominated/east PRE plume-dominated). I think adding a schematic/cartoon-type figure illustrating these would be a good way to summarize the results in a mechanistic way and make them more visible to readers studying other regions influenced by Western Boundary Currents, wind-driven upwelling, and large river outflows.

Minor points

m1 (lines 39-41): Topographic effects should be more important in locations with more curved isobaths such as in the NSCS' widened shelf area, as previous work in the NSCS shows. So I don't follow why nearly shore-parallel isobaths should result in enhanced cross-isobath flow.

m2 (lines 82 and 119): Comparing the model resolution to the local first deformation radius derived from the model stratification is important here, especially inshore of the 100 m isobath (where the nonlinear terms are shown to be more important in the depth-averaged vorticity balance).

m3 (line 87): A couple of example references using the Mellor-Yamada scheme could be added here, in addition to the original paper describing the scheme.

m4 (Fig 1): It would be helpful to add the 100 m and 200 m isobaths to this figure for reference. In Figs. 4, 6, and A1, it would also help to have them labelled on the figure itself.

m5 (Fig 1's caption): Are the geostrophic currents shown as pink arrows the surface geostrophic velocity derived from the model sea surface slopes? This could use clarification. Differentiating between "geostrophic" and "shelf" currents is also confusing because there are geostrophic currents both on the shelf and offshore.

m6 (line 247): How are the degrees of freedom estimated (e.g., from integral timescales derived from the time series of the velocity components at each grid point)? It would be good to describe it here.

m7 (Fig. 5) The discussion relies on different stratification regimes, which appear to be both salinity- and temperature-driven. It would therefore help to overlay isopycnals of the conditionally-averaged density fields on each panel.

m8 (line 290-291): It is more objective to include some metric of the smallness of the GMF term, for example, what is its size relative to the next-largest term in the balance? This ratio will also vary spatially, so I suggest the authors include a figure with the GMF term's spatial structure and its relative size in the Appendix.

m9 (Fig. 8 caption): Are these each of the JEBAR terms' contributions to the total PGF_{y*}^b/f anomalies in cm/s (like Fig. 6)? Please add the units to the caption like in Fig. 6, or to the colorbar labels.

Typos/minor edits

Line 23: Intrusion -> intrusions

Line 91: Large amount of freshwater influx -> a large freshwater influx

Line 119: Smaller scaled -> smaller-scale

Line 119: Could be further detailed -> are not fully resolved

Line 196: Streamflow -> runoff/outflow

Line 214: Missing space before "Interannual"

Line 294: Within the -> inshore of

Line 377: In the -> in