General comments

The article shows the exhaustive work of data validation and quality check performed before the analysis and climatological fields computation. Explanations are easy to follow and the entire work could be replicated by other people in other marine areas. The analysis of results brings out the main differences and agreements between this new computed climatologies and the existing ones (WOA2018, medBFM) with carefully explained details level by level and basin by basin.

The lack of data in certain areas is a well-known problem in the Mediterranean and researchers have to face it when they try to find complete descriptions or intermediate and deep sea layers behaviors. The problem is greater when they want to obtain information about nutrients at intermediate or deep water measurements. In this sense, these climatologies could help in future works to partially palliate this lack of data by offering reference values and error fields. Future updates an re-evaluation of these climatologies will be welcomed by the scientific community.

 Specific comments

Regarding the GEBCO-30sec  (aprox 920mx920m grid) dataset used by the authors, I would put forward that the current EMODnet Digital Terrain Model (DTM), released in Sept2018 has a grid resolution of 1/16 minute x 1/16 minute (=115m x 115 m) and new DTMs are planned to be published in short. As GEBCO, it is freely available to any interested people. Note also that from the start of the EMODnet Bathymetry projects, the followed approach is based in the pre-gridded DTMs. They have been locally computed by the data providers according to the chosen grid and origin. This facilitates the precision of the final computed DTM. This bathymetry, perhaps, could improve the presented climatologies at some levels and can be is a line to explore in future works.

In relation with the Deep Water formation in the Gulf of Lion, the importance of Tester's team and work (line53) must be recognised, and the great impulse to the knowledge of this phenomenon achieved with the new instrumental advances, but I would like to draw the attention of the authors to the extensive bibliography published along the time, to which they could also refer. I mention some articles as an example below, but they are not the only ones.

• MEDOC Group. Observation of formation of Deep Water in the Mediterranean Sea. Nature, 227 (1970), pp. 1037-1040
• Canals, P. Puig, X.D. de Madron, S. Heussner, A. Palanques, J. Fabres. Flushing submarine canyons Nature, 444 (2006), pp. 354-357, 10.1038/nature05271
• Durrieu de Madron, L. Houpert, P. Puig, A. Sanchez-Vidal, P. Testor, A. Bosse, C. Estournel, S. Somot, F. Bourrin, M.N. Bouin, M. Beauverger, L. Beguery, A. Calafat, M. Canals, C. Cassou, L. Coppola, D. Dausse, F. D'Ortenzio, J. Font, S. Heussner, S. Kunesch, D. Lefevre, H. Le Goff, J. Martín, L. Mortier, A. Palanques, P. Raimbault. Interaction of dense shelf water cascading and open-sea convection in the northwestern Mediterranean during winter 2012: Shelf Cascading and open-sea convection. Geophys. Res. Lett., 40 (2013), pp. 1379-1385, 10.1002/grl.50331

In the QC of the analysis fields (line 339) the authors refer that the residual values are NaN. This could be a computational output, but it could be explained in another way. It will be preferable if they explain that these computational points are not considered by different reasons (out of domain, or whatever). In my humble understanding,  NaN is merely the way in which the computational tool used (julia) encodes the criteria adopted by the authors. It will be very welcomed if the authors consider to reformulate the paragraph.

Technical corrections

line 226. Please add the link / reference to Diva User Guide

line 697.  Please type Western Mediterranean Transition (WMT)

line 779. Please, add the corresponding DOI to reference

This is a well-written, high-quality manuscript providing a significant long-term climatological dataset for the biogeochemical parameters, their spatio-temporal change and the processes involved in the Western Mediterranean. The data product is based on in-situ observations from various cruises over the period 1981 – 2017.

1. The abstract is informative, although I would like to see 1-2 statements on data quality, on the comparison with previous datasets and the overall value of this data product.
2. In Introduction, lines 31-32, I believe upwelling is not relevant in the context of this sentence.
3. In Introduction, the main physicochemical processes and circulation affecting the distribution of nutrients in the WMED is shown. However, the role of rivers supplying nutrients in the EMED and WMED is not discussed. Please add a paragraph discussing the main riverine fluxes of nutrients, mostly from the Nile, Rhone, Po and Ebro, and their trends in the various basins and sub-basins.
4. Table 1 illustrates the existing two nutrient climatologies for the WMED. The Table is informative and well-structured; it is unclear however, if the present work contains also the data of the past climatologies. Please explain. Also, the WOA18 contains only bottle data, while you imply to also use data from argo-floats. It is unclear what are the percentages between bottle data and data from Argos. Finally, please provide the units used to report the nutrient data per climatology.
5. One deficiency of the database built is the gap in observations between 1997 and 2003. Although mentioned and presented graphically in Figure 2a, it is not very clear how authors deal with this gap in their analysis.
6. Another deficiency is the bias towards summer period. Again, it is not very clear how mean-annual fields are produced from these biased data. The statement “Adjustments were applied to measurements when bias was detected” should be further elaborated.
7. Authors do not report the analytical methods followed for nutrient analysis, per database used. Perhaps a Table including the analytical techniques and instrumentation used will be useful.
8. For the data quality check, perhaps a flow chart will guide the readers on the step-by-step procedures followed. In WOA18, other checks are also followed, like the Range and Gradient Check and the Representativeness of the data check. It is unclear if QA/QC followed here is the same as the compared climatologies.
9. In Eq (1) the term (μi) in the observational constraint term is not explained. What is its range and how it is evaluated. Similarly, although known, it is better to describe the φ(xi, yi)-term.
10. Authors state that they applied the fourth-dimensional DIVAnd method. What is DIVA's response when there are gaps in time as in this dataset. Perhaps analysis should be divided in two periods, prior and after the period lagging data, and apply DIVA in each separate dataset.
11. Also, please explain if DIVA was used to extrapolate data in areas with data gaps and towards the coast?
12. Line 261. For which parameter do these values refer? for which depth?
13. Lines 277 – 279. Probably here you refer to the horizontal Lc variability. How do you comment on the lower values of Lc for silicates compared to nitrate and phosphate?
14. Lines 280 – 288. Text is mixing the diagrams and Lc-parameter from horizontal to vertical. Better write one paragraph describing the horizontal and another for the vertical Lc.
15. Line 330. “A score is assigned to each observation”. Please elaborate on the score assigned per observation. What is the score range and the increments used on the scaled error.
16. Figure 6. Please explain the dashed blue line.
17. Lines 389-391. You should also refer to the rivers supplying nutrients to each area. The e-HYPE database from SMHI could be helpful to assess the riverine fluxes of nutrients.
18. Since nitrate is the dominant N-species in the WMED, authors could produce the stoichiometric N:P:Si ratios and discuss their mean and standard deviation values per sub-area and for the whole WMED.
19. Line 506. Please consider that medBFM assimilates satellite and argo data and includes terrestrial inputs of N and P from 39 rivers.
20. Lines 516 – 522. If I understand well, you re-gridded the BGC-WMED from 0.25 deg to 1 deg, to compare with WOA18 and the medBFM fro, 0.063 deg to 0.25 deg to compare with NGC-WMED. Please explain better this process.
21. Lines 542-543. Could the largest difference seen in the Alboran Sea be attributed to the occurrence of more frequent upwelling events during the WOA18 period?
22. Line 761-762. This is a valid point you are making on the decline of river discharge. What abou the nutrient fluxes over this period?