Interactive comment on “ Copernicus ( CMEMS ) operational model intercomparison in the western Mediterranean Sea : Insights from an eddy tracker ”

We thank the reviewer for the their helpful input. We also wish to apologise to the reviewer regarding the incomplete state of our initial submission. This we discovered was because we submitted the wrong pdf, and neither we nor OSD picked up on that mistake. The fault is entirely ours. We appreciate that this made reviewing the submission rather challenging for both the reviewers. We mark our replies below with ———— Anonymous Referee #2 Received and published: 3 April 2019 General Comments: In


Interactive comment
Printer-friendly version Discussion paper this manuscript, the authors applied an eddy tracker and an eddy compositing technique, to three forecasts models: (i) the Mediterranean Forecasting System (MFS), (ii) the Global Mercator model (GLO ), and (iii) the Iberia-Biscay-Ireland system (IBI ), provided by the Copernicus Marine Service (CMEMS) website.They used the py-eddy tracker, an eddy detection and tracking method similar to the procedure proposed by Chelton et al. (2011).This method is widely used by the scientific community.The eddy compositing analysis is applied to analyse the 3-D structures of three gyres of the Alboran Sea (WAG, EAG, CRT), an interesting area of the Western Mediterranean, characterized by strong density gradients and high mesoscale activity.The authors compared the models results to evaluate their performance and conclude the paper with some technical suggestions in order to improve these CMEMS products.This paper is well structured and the methodological approach is correct for the purpose of the journal.Nevertheless a more detailed description of the methods and additional results are needed to support the interpretation and conclusion.Therefore, at this stage, a major revision is required before acceptance.Please see below for more details: Major remarks: 1) I found the Section 2.4 (Eddy compositing) hard to follow.The methodology used for the 3-D eddy composites and for the computation of the anomalies of T and S is not adequately explained.The authors refer to the article of Mason et al., 2017 (see pg. 8 line 15) but it is applied in a different context and in a different area.----A full description is now included, which includes reference to figure 2.
2) The eddy properties, detected by py-eddy tracker applied to the three models and to the ALT data, are not adequately compared (section 3.1).In Section 3.1.1,the results on the numbers of eddies detected and tracked in relation to their lifetimes are not quantified.In Section 3.1.2,the authors have identified similarities "between the patterns" of eddy radius, amplitude and intensity (pg. 10, l.16, pg.11, l.1 and l.11).These may be due to the strong signal of the Atlantic Water flowing eastward, which generates the large and energetic Anticyclonic Eddies.This flow causes differences in terms of eddy properties between the northern and the southern part of the basin.

Interactive comment
Printer-friendly version Discussion paper Pessini et.al, 2018 identified these N-S differences as function of the area of formation and lifetime.Escudier et al., 2016 compared the eddies distribution per radius and lifetime, of altimetry and model data.I suggest the authors to follow the last approach since, in my opinion, more detailed comparison between the models and ALT data are needed.----We have added several lines to the Introduction about the reasons for the N-S differences in the WMED.
3) The authors did not describe how they calculated the mean and median coordinates/radii of the eddies, listed in Table 2.There are various methods to calculate them for example: (i) selecting all the eddies detected by the eddy tracker independently for lifetime, (ii) discriminating for lifetime (i.e.avoiding the shortest) and (iii) selecting only the eddies identified simultaneously by the three models in the same sub-area.The 3-D eddy composite analysis should depend on these variables (pg.3, l. 10 to l. 22) and therefore should be method-dependent.Some results (pg. 18, l. 11 to l.12) and conclusions (pg.24, l. 13 to l. 21) are based on vertical structure of T'.The authors ac-tually cannot demonstrate "how closely these eddy composite results may correspond to reality" (pg.26, l.10-11).For these reasons, a check to verify if the 3-D subregion eddy composite results are method-dependent should be done.Alternatively, I suggest the third method because is the most appropriate to compare the model results.----These statistics were calculated using the second method suggested by the reviewer, where the minimum eddy lifetime (5 days) is set as an input parameter to the eddy tracker.The third method is not practical because IBI does not have data assimilation so there is no meaningful correspondence at all with altimetric eddies.Lastly, we are working on providing a reference for the composite T and S profiles using Argo floats.4) In Table 2, please provide the mean and median coordinates and the properties of the eddy for the altimetry data (section 3.2).This will allow a comparison among the models and the altimetry data.----The altimetry information has been added to the table.

Interactive comment
Printer-friendly version Discussion paper 5) Table 3 is not acknowledge through the text.The variables Rmin, Rmax, Rmean, Rmedian, Rmad are not defined.Some values are missing and other seem to have no sense.Specifically, in "Product: MFS" at line "CRT" are listed the values: 111, 222,333, 444, 555 which seem to be out of scale.In Section 2 (Data and methods) the effective and the speed-based (inner) radii are indicated respectively with the symbols Le and L while in Table 3 are labelled as R. ----Table 3 has been removed as it was redundant.6) In Figure 10, the seasons are not mentioned.In winter and autumn, the deepening in the mixed layer depth (MLD) in the center of the anticyclonic eddy increase from GLO to IBI (fig.10).This seems coherent with the zonal and meridional relative vorticity of figures 7 and 8, having more intense anomaly for higher resolution models.I assumed that these differences in the MLD inside the anticyclonic eddy are due to rotational speeds, better simulated in these higher resolution models.Anyway the authors assert that "the incoming Atlantic Jet in IBI is suspected to be too strong such that these ζ values may be an overestimate (pg.16, l. 2-3)".The last sentence should be motivated and this part must be clarified also because these considerations could be used to evaluate the performance of the models (pg.24, l. 22) and to provide a more consistent conclusion.----The text in the MLD section (3.2.6) has been modified.Labels for the seasons in figure 10 have been added.Concerning the statements about the Atlantic Jet in IBI, we added a reference to the CMEMS MedSub project report which contains more information about this anomaly.

Interactive comment
Printer-friendly version Discussion paper IBI_REANALYSIS. . . .In the last case, please check the sentence "(based on a 3DVAR scheme)" in Section 2.1.1.For MED-SEA_REANALYSIS_..., it should be substitute with "(based on a OceanVAR scheme)".----No, we did not use the suggested *REANALYSIS* products.The products we used were the ones that were available at the start of the MedSub project in March 2016.The GLOBAL* _001_024 data are still available for the period 2006-2016 by using the motu_client software, rather than using the subsetting option on the CMEMS portal.----We changed the text to "based on an OceanVAR scheme" as suggested.