General comments

This is a numerical model study to identify the contribution of three sources of nitrogen, which is from the land, from the open sea, and from the seafloor, to DIN in a semi-enclosed coastal sea (Seto Inland Sea: SIS). Although this information is important for the prevention of eutrophication in the sea, the calculation method has some problems as shown below.

Specific comments

1. Why do you include seafloor sources as a nitrogen source in addition to open sea and terrestrial sources?

Unlike the case of phosphorus, nitrogen leached from the seafloor is the result of mineralization of “new” sediments, so nitrogen originating from the seafloor may be included in the open sea nitrogen and land nitrogen.

1. Why is dissolved organic nitrogen not included in the calculation?

Dissolved organic nitrogen, which accounts for about 90% of the total nitrogen in SIS, is not included in the calculation. The Ministry of the Environment's total load reduction for SIS is also based on total nitrogen in its calculations.

1. Are the DIN boundary conditions used in the numerical model reasonable?

The boundary condition is that the open sea origin DIN is zero at the seafloor and landward.

If there are no biochemical processes in SIS, no DIN supply from land or seafloor, only physical diffusion, then the DIN concentration in SIS is equal to the open boundary DIN (DIN from the open sea) and SIS is filled with DIN from the open sea. In other words, DIN = 0 does not occur on the seafloor surface or on the landward shore.

Technical correction

1. It should be noted that the land load in this report is an underestimate.

The terrestrial nitrogen load for SIS is published every five years by the Ministry of the Environment  of Japan (MEJ). It is necessary to state the values of the terrestrial load by MEJ and the terrestrial load in this report.

In SIS, which experienced eutrophication in the 1970s, the majority of domestic and industrial wastewater is treated at treatment facilities on the waterfront and discharged directly into the sea in recent years. Therefore, there is a large difference between the DIN flow via rivers and the total nitrogen flow actually entering the sea (especially in the eastern Seto Inland Sea).

In SIS, river discharge is significantly lower in winter, resulting in large seasonal variations in DIN flow from rivers, whereas there is little seasonal variation in DIN flow from domestic and industrial sources. This affects the seasonal variation of DIN concentration in the SIS.

1. Section 3.2.

 It is important to indicate the time required for the numerical model to become stationary; the DIN flow path during the set-up period is not the flow path when the model becomes stationary.

1. Actual measurements of the amount of nitrogen and phosphorus entering SIS from the open sea were made by several organizations in the 1980s to 2000s, and it has been shown that the amount of nitrogen entering from the open sea is equivalent to the amount of land-based load during the summer months. It is desirable to cite these papers.

1. Line 114

 It should be noted that the seasonal variation of the nitrogen load from rivers is due to the seasonal variation of the river flow. Unlike Europe, the SIS receives a little precipitation in winter.

General comments:

Information about the hydrodynamics of the SIS is missing, specifically the main currents should be described and compared between your model and e.g. literature references, since it is essential that the advective transport is realistically captured, which is not obvious from validating the DIN and DIP concentrations alone. Also comparing salinity to observations might help to check whether the mixing ratio between riverine and oceanic water masses is realistically captured in the model.

You state that your nitrogen loads are smaller than previously reported values, as you neglect industrial and land-based sources as well as particulate forms of riverine nitrogen. It seems you also ignore atmospheric deposition? Since the main result of the paper, which is the oceanic fraction of the nutrients in the ecosystem, will be strongly dependent on the terrestrial loads you put in, please give a quantitative estimate on how large this uncertainty/error in the loads is.

Sediment DIN flux: Your sediment model is very simplistic and maybe a bit too simplistic for your application. You assume constant DIN fluxes from the sediments in a study where you state that your goal is to understand the temporal dynamics of eutrophication. You ignore a positive feedback loop in which enhanced nutrient loads lead to more settling PON, to higher reactive TN concentrations in the surface sediment and subsequently to higher DIN release from the sediments. Please at least discuss the potential implications of this strong simplification in your discussion section. This is especially critical since sediment-water DIN fluxes are not easily observable. They tend to show substantial small-scale variation depending on e.g. the presence of bioturbating or bioirrigating macrofauna. Please give more information on what the uncertainty of the benthic flux estimates is.

You consider sedimentary DIN as a "source". Actually, sediments are not a source for nutrients, but just a temporary storage or a permanent sink. The nutrients stored in the sediment are originally mostly from riverine or oceanic origin. Even if this may be somehow clear for most readers, I think it is still worth mentioning.

Another point maybe worth discussing is that the "oceanic" DIN can be of riverine origin, just added to the Japanese coastal waters from rivers outside the SIS. Or is it the "open" Pacific Ocean signal that is really controlling the conditions at the borders of your model domain? 

Please give some references why it is reasonable to exclude dinitrogen fixation as a relevant N source in the SIS and neglect it in the model. (in other coastal seas it is a majour source)

Section 4.2 is lacking information on how the figures presented in the article relate to previous estimates of the nitrogen budget of the SIS.

Section 4.3 occurs very unexpectedly. If nutrient load reduction experiments are performed, this should be mentioned in the methods section and the results section and not appear for the first time in the discussion section. Anyway, the model with its assumed constant sedimentary N fluxes seems not appropriate for nutrient load scenarios, since here the sediment feedback is essential. Your model implicitly assumes that as soon as some riverine N reaches the sediment in particulate form, its influence is gone. In reality, specifically in shallow near-coastal sediments, fresh organic matter that reaches the sediment can me remineralized quickly and (in case that this does not happen due to denitrification) become available for primary production again. So maybe leave just leave out this section (it adds a side-story to the main story line of the article) or move it to the online supplement?

Minor comments:

Line 30: "regulated" -> "influenced"? (climate change has no "regulating" effect)

Line 33:  "presenting a different seasonal variation" -> "so their import has a seasonality that is different"

Line 56: Abbreviation "COD" is not defined

Line 58: "concern about oligotrophication was raised for it" is unclear, please rephrase

Line 59: meaning of "As the first step" is unclear. Are you doing a multi-step approach, or do you indicate that you are the first who try to understand these changes?

Line 91: "from a daily dataset" is too unspecific, please give a few more details

Line 93: Please specify where your hydrodynamic boundary conditions come from. 

Line 112: "The spatial variation" -> "Spatial variation"

Line 133: Wang 2002 actually only cites the method from Ariathurai and Krone (1976), please give the original reference.

Line 172-177: Please state more clearly which fluxes you define at the boundaries. You state you define "zero concentration" but that is puzzling. At the land-sea and sediment-water boundaries you should have identical fluxes as for DIN for one of the tagged state variables and zero flux for the others. For the open boundary condition, this should be the same during times of inflow, but during times of outflow (in the upwind scheme) the DIN_??? should be exported according to the ratio DIN_???/DIN. Please clarify.

Line 186: Why do you use observations in 50 m depth as "bottom value" for areas deeper than 50 m? Please clarify.

Section 3.1: While Fig. 2 and Fig. 3 are good for showing how well the model captures the spatial signal, it is really hard to see by eye whether it also resolves the seasonal patterns. I suggest adding 

a few climatologies from the model compared to observations, for a few stations representative for different subareas of the model domain. This is probably sufficient in the supplement.

Line 234-238: "have already occupied most areas of the SIS": it would be better to calculate the ratio (DIN_ocean+DIN_river+DIN_sediment)/DIN. If that is close to one everywhere in the model domain, you can estimate that your spin up period for the tagging is completed.

Line 373: "whose ratio is 1.4:1": The ratio between what? Subsequently more occurences.

Line 454: "the management can also be applied to the sediments": It is very unclear how you would "manage" sedimentary nutrient release, you cannot easily modify it. If this is a serious option please give more details, e.g. will you add substances to capture some of the escaping nutrients?