Interactive comment on “ DUACS DT-2018 : 25 years of reprocessed sea level altimeter products

The manuscript presents the overall enhancement of gridded and along-track altimetry products following the DT2018 reprocessing, in a way that is similar to the DT2014 reassessment published earlier. Methods and Processing for quality assessment are therefore established, and skill assessment has not been developped further, but this is acceptable to me. I believe it is a necessary step to publish such reassesment peridodically, and to synthetize skill metrics for the state-of-the-art altimetry products as proposed. I therefore support the publication of this manuscript, suggesting some modifications below. Title is appropiate.

Authors: Pujol et al.,2018 shows the new MSS15 is more extended at high latitude than the old one. (see also figure 1 below). This allows us to compute the OI with much more precision and stability in this region. The figure 4 of Pujol et al.,2018 shows the difference of the variance of SLA along HY2A tracks. These differences are major at high latitude.
Figure 2 (below) shows the difference of SLA variance with DT2018 and DT2014 gridded products from the same point of view as figure 1. The difference in spatial coverage of the two MSS explains the difference in quality of the SLA grid products in this area.  4) p8 line 11-12: "However, in the equatorial band (±20•N), the EKE in the DT2018 is less important (-17%). This is linked with the evolution of the noise measurement considered in the mapping process for all satellites." I'm getting really sick of this vague uninformative style: 'linked' and 'evolution'.
Authors: The sentence has been changed and details added. 5) p8 line 19-29: Discussion of table 3. This is an important part of this study, but lots of information is missing. Table 3 has just 2 values for each of 4 regions. Why trim it down to such a bare minimum of information? E.g. For the reference area |trackmap|ˆ2=1.4cmˆ2. This is for a 'low variability' region. But how low? Easy to answer: list the |track|ˆ2 and |map|ˆ2 values as well.
Authors: The low variability region has been introduced in Pujol et al.,2016. The authors found interesting to reuse it to have a reference area where observations errors are small. The SLA variability in this region is between 0 and 7 cm². This precision has been added to the Table 3. A figure (figure 5 in the new manuscript version) has been added in the manuscript to show the RMS difference (in % of RMS) between two-sat gridded products and along-track product for DT2018 and DT2014 versions.
We also added a discussion about improvements in the intertropical zone.
6) p8 line 19-29: Discussion of table 3. --also: this is just for the 2-sat product. What about the multisat product? I hear the answer already: "Because none of the data are withheld". My response: this does not stop you listing the map minus track stats, which are then measures of the closeness of fit (as distinct from map error). To estimate map error, pick a time with many good satellites and rerun the OI, withholding one (e.g. C2) for use as the error measurer.
Authors: This issue has been discussed p8 between line 23 to 25. The error described here must be considered as the upper limit. We choose not to describe in the manuscript a configuration with more than two satellites. However, the authors also studied the period 2016-2017, and the conclusions are similar with C2 as an independent along-track mission. (using Jason-2 and AltiKa for the mapping process).
The L4 all-sat validation is complemented by in situ drifter's comparison. 7) p9 line 1-2 "Positions and velocities of drifters are interpolated using a 3-day lowpass filter in order to remove high-frequency motions." I have 3 grumbles: i) don't use the passive voice ('are interpolated') -it leaves it to the reader to guess who did the interpolating -we assume it was you but we can't be sure. ii) this is a very brief 'Methods' section squeezed into the Results section iii) why remove 'high frequency motions?' A 3-day filter also removes a lot of low-frequency Eulerian velocity (a drifter can easily go 1/4 of the way around a well-resolved eddy in 3 days). So, instead of filtering then differencing, it is better to do differencing then filtering.
Authors: The authors added a relevant reference which explain the interest and the method used for 3-day lowpass filtering: Use of Altimeter and Wind Data to Detect the Anomalous Loss of SVP-Type Drifter's Drogue M.-H. Rio. 2012. The main objective of the filtering process is to discard the tide and the inertia in drifters' data.
We know that: -we don't filter enough between 10S and 10N to get rid of all the inertia -we filter a little too much at high latitudes, knowing that we don't want to go below 24 days for the tide. The 3-day period is a compromise between these two. The methodology still needs to be improved. 8) Fig. 6: It seems to me that 2 panels are missing: the ones showing the DT2018-DT2014 difference.
Authors: The authors have added the missing plot and related comments. 9) p9 line 4-5 "the comparison reveals that DT2018 altimetry products underestimate absolute geostrophic current." This statement is not supported by Fig. 6, Table 4, or by the mention that someone (we don't know who, because passive verb was used) has done a Taylor diagram (but kept the results to themselves -all we know is that the results are 'strong'). As in comment 5 above, list the variance of the drifter and altimetric velocities in order to prove that the altimetry underestimates the drifter velocities. 30-45N. Let's see some example time-series of errors for each product individually, not to mention the two signals being differenced (altim and TG) individually as well.
Authors: We know from Saraceno et al, 2018 (Estimates of sea surface height and near-surface alongshore coastal currents from combinations of altimeters and tide gauges) that coastal processes are more difficult to resolve with altimeter data, because of two types of problems. First, and most importantly, intrinsic difficulties affect the corrections applied to the altimeter data near the coast (e.g., the wet tropospheric component, high-frequency oceanographic signals, tidal corrections, etc.). Thus, data are usually flagged as unreliable within some distance of the coast. Second, the interpolation of along-track data collected by just one or two satellites provides only marginal resolution of mesoscale and smaller-scale structure in ocean circulation [Le Traon and Dibarboure, 2002;Leeuwenburgh and Stammer, 2002;Chelton and Schlax, 2003], which is dominant in the coastal region.
We did compare some time series for tide gauges on the Portuguese coast. It is difficult to draw conclusions about a particular time period over which comparisons are degraded. We were unable to correlate these degradations with periods when there are fewer data (fewer satellites in the constellation, or anomaly on a satellite).
We know that the new tide correction is particularly important in coastal areas, but again we have not been able to explain these degradations with this correction.
We are not in a position to explain the degradation observed in these well-located areas of the globe (West Coast of the USA, Portuguese coast, etc.).

3.4-onwards
Sorry, but I am not prepared to read any further. I think this paper has too many faults to be published in close to its present form.
This paper presents findings from assessment of the quality of the DT-2018 products versus DT-2014. I find that the most convincing improvement is near coast and in the Med Sea and the Black Sea. The interpretation of the open ocean performance is not compelling. The following are some specific comments: P.1 Introduction-I'd suggest adding some text on the history of altimetry missions over the past 25 years.
Authors: Done p.2 last line-Is the data from Hayaing-2 A incorporated in DT2018?
Authors: As shown in Figure1, Hayaing-2 A data are incorporated in DT2018. The particularity of the reprocessing is to integrate additional HY2A data that were not taken into account in the DT2014 production: data from March 2016 to February 2017. This paragraph has been rewritten to be more explicit.
p.3 first line-What about the data distribution by NASA?
Authors: L2P data are only distributed by CNES and EUMETSAT. The data distribution by agencies NASA, NSOAS, ISRO, ESA, EUMETSAT, CNES… are taken into account in L2 products. DUACS processing only uses L2P data. The sentence has been reformulated in the manuscript.
Line 6-is the altimetry community represented by the OSTST? If so, please mention it.
Authors: Done p.5 first line-cite Table 2 when the mean period (MP) is first introduced.
Authors: These "upstream measurements" correspond to the L2P products that have been presented previously. The sentence has been rewritten.
Line 17-give a reference for the MSS.
Line 24 -What is "Theoretical Track"? Authors: The authors added a reference (Dibarboure et al., 2011) which provide appropriate details: "Altimetry satellites generally use repetitive orbits: after 10-35 days, the sensor flies over the same locations, hence the notion of cycles (time needed to revisit the same location) and the ability to colocate data. However, the satellite ground track cannot be perfectly controlled and is kept only in a band about 1 km wide. It is thus necessary to use an arbitrary and mission-consistent position for the co-location process. SSH measurements are then projected onto these co-location points." p.6 line 11-give reference for the MSS Authors: Done. p. 7 line 16-delete "at" after "be" Authors: Done.
line 29-30 -Is "additional variance for high variability regions in DT2018" an improvement? if so, why?
Authors: At this stage, this diagnostic is only used to characterize the impact of the new mapping process and new altimeter corrections. It is not presented as an improvement (It might as well also correspond to noisier DT2018 products). The only conclusion is that there is more variability in DT2018 products. It is only in a second step, by comparing with independent dataset and in-situ measurements, that we show that this gain of variability corresponds to an improvement.
p.8 line4-why is the difference of variance important? What does it mean?
Authors: The authors have reformulated this sentence.
Line 9-How is the EKE at the equator computed while geostrophy breaks down there?
Authors: The geostrophic current products disseminated to users are computed using a nine-point stencil width methodology (Arbic et al., 2012) for latitudes outside the ±5°N band. In the equatorial belt, the Lagerloef methodology (Lagerloef et al,1999) introducing a β plane approximation is used. The EKE is computed from this geostrophic estimation. This methodology did not changed since DT2014 version.
As at the equator the geostrophy breaks down, the ±5°N band is usually masked at the equator. Figure 5 has been corrected.
Line 11-What does it mean by "less important"?
Authors: The authors have reformulated this sentence.
Line 16-Given the issue of geostrophy near the equator, how would one interpret the equatorial EKE reduction as improvement?
Authors: The equatorial EKE reduction is a direct consequence of the increase of the noise measurements considered in the OI process: Observation errors have been increased in the equatorial belt, so the SLA signal is smoother and less energy is observed in this region. It has been noted that in DT2014 products, there was too much noise at the equator.
In the ±5°N band, near the equator, the EKE has been masked.
p.9 line 4-Is the fact that DT2018 products underestimate absolute geostrophic current an improvement? If not, what is the interpretation?
Authors: It is presented as a fact, not an improvement. Main reasons are that absolute geostrophic current from altimeter are smoother (fewer small scales) than with drifters, there is probably still some ageostrophic signal left in drifters' data. Authors: The authors have corrected this mistake.
line 13 -What does it mean by "improvement is clearly visible in the intra-tropical band" while the regions are blocked in Fig 6? Authors: The sentence has been reformulated to take into account that the ±5°N band is masked.
p.10 line7-Please quantify the global reduction of the variance.
Authors: Global reduction of the variance is around 0.6%. it has been added in the document. Authors: The authors have reformulated the sentence. The first estimate using along-track measurements of the reference mission only  is not display here.
p.11 lines 13-15-I think the information of Table 5 is sufficient and Fig 9 can be deleted. It does not convey much additional information.
Authors: The authors have replaced the figure with the difference of the root mean square of the SLA minus independent Tope/Poseidon along-track SLA, using successively DT2018 and DT2014 gridded product. The authors thought that the spatial information conveyed by this comparison would be more relevant. We have added a description of this new figure in the body of the manuscript.
Line 26-Please quantify the overall improvement shown in Fig 10. Authors: Overall reduction of the variance for Mediterranean product is around 0.4%.
Interactive comment on "DUACS DT-2018: 25 years of reprocessed sea level altimeter products" by Guillaume Taburet et al.

Anonymous Referee #3
Received and published: 26 March 2019 Authors: We warmly acknowledge Rev.#3 for his review. All comments and remarks have been considered. In the next paragraphs we present the reviewer's comments followed by our point-bypoint reply.
General Comment : **************** The manuscript presents the overall enhancement of gridded and along-track altimetry products following the DT2018 reprocessing, in a way that is similar to the DT2014 reassessment published earlier. Methods and Processing for quality assessment are therefore established, and skill assessment has not been developped further, but this is acceptable to me. I believe it is a necessary step to publish such reassesment peridodically, and to synthetize skill metrics for the state-of-the-art altimetry products as proposed. I therefore support the publication of this manuscript, suggesting some modifications below. Title is appropiate. * As a suggestion : I believe the whole manuscript could be summarized on a single figure, in the form of a target or taylor diagram showing skill metrics for the different products (along-track, gridded SLA, gesotrophic currents) and scales (regional, global coastal, global offshore, climatic, etc ..) showing DT2014 postions and DT2018 positions. This is a mere suggestion, but I think it would provide a very efficient overview of the DT2018 update. Unless there are good justifications why this can not be done (at least for part of the datasets presented), I think it would be relevant for the manuscript to consider issuing this figure. Specific Comments (I start with question mark "?" to denote a suggestion) Authors: The authors do agree that this suggestion is a good idea. We have tried to compute such figure reusing existing results, and particularly Table 3 to 5. However, the result does not appear to us to be sufficiently successful to be published. It would deserve much more substantive work. The authors keep the idea and will try to implement it in future quality document associated with the DT2018 products and for future reprocessing. *************** * Abstract: P1L19 : I understand the reason for providing quantitative metrics in the abstract, but the term "errors" is too vague in the present abstract. Please precise.
Authors: The authors specified that these values have been computed using independent and in-situ measurements. In particular, the difference in variance of difference between altimetry and independent dataset allows to characterize this error. Authors: Done P4L18 : It would ease the read to define "geoditic" and "drifting" mission, and help nonspecialized readers to grasp the challenges of altimetry processing.
Authors: The authors replaced the terms "geodetic" and "drifting" by "non-repetitive mission". Improvements summary DT2018 vs DT2014 for global products (in red), regional Mediteranean Sea products (in blue) and Black Sea products (in green) -variance reduction in percent -P4L23 : please define more clearly the "percentage of data recovery" Authors: The authors have reformulated this sentence which was very confusing. There was no data in DT2014 products and now validated measurements are available. P6L9 : Does "selection" applies on 1) altimeter data for along-track data product generation or 2) along-track product for gridded products generation ?
Authors: it is for gridded product generation. The explanation has been clarified. is a period that has already been studied in Pujol et al, 2016, so we thought it would be interesting to continue over this "reference" period. We also did the study on another more recent year (2017) and the conclusions are similar.
P8L23: The author avoided the nomenclature "two-sat"/"all-sat" up to this point. Can it be also avoided here ? (I think it is the only place where it is used).
Authors: The Taylor skill score (Taylor, 2001)  Where R 0 is the maximum correlation attainable (hereafter R 0 = 1), R is the correlation coefficient between the model and the observations, σ mod and σ obs are respectively the model and the observations standard deviations.
So it is more correlation and standard deviation than variance and rms.
Authors: Done P9L26 "in the" repeated Authors: Done P10L4 : Why "maximum" correlation ? Does that refer to a selection amongst the neighboring pixels ?
Authors: The processing is detailed in Valladeau et al.,2012. The method is based on a criterion of maximal correlation between tide gauge time series and altimeter gridded products, where the most consistent state of the ocean between both data time series is considered within 300km around tide gauge. The main advantage of this method is to reduce the effect of oceanic variability and the error on the MSS with respect to the same altimeter point.
Authors: It has been added.
P12L14 "large" -> "largeR" Authors: Done p12l22 "lager" -> "larger" Authors: Done P13L8 "for" -> "from" **************** * Are appropriated and all useful in general. * Small to very small coordinates, axes and colorbar title. Please ensure readability. If considered essential, should the figgure be reprocessed with larger bins ? It does not provides many information as for now, except : "more data in the 20km coastal band", "lot of noise in the center" and " a strange, uncommented blue track in the center of East Med". Unless justified otherwise, i suggest to remove this figure.
Authors: The authors have decided to remove this figure.   Authors: This caption refers to an old version of the figure. It has been corrected. * References : ************ * There are many references to work 'in prep.', including on to "In prep. to be submitted to OD in 2016" (Lyard et al.) . Please check with editorial office on the policy as regards reference to unpublished works.
Authors: The authors have contacted the editor.
Here is the answer: In general, please note that "submitted to", "in preparation", "in review", … can be left as is. During typesetting of your manuscript our Typesetters will check all references related to Copernicus Publications for an update. If an update is available our Typesetters will insert it and inform you accordingly.
* The reference style is not homogeneous, with years being given some times at the end, some times after the authors. Please homogenize. Several changes werehave been implemented in the DT2018 processing in order to improve the product quality of the products. New altimetery standards and geophysical corrections werehas have been used, refined data selection was refined and has been implemented and Optimal Interpolation (OI) parameters werehave been reviewed for global and regional map generation.
Through this paper describes the, an extensive assessment of DT2018 has been carried outreprocessing. The error budget 20 associated withto the DT2018 products at global and regional scales washas been refined defined and the improvements on the previous version were compared with the previous version quantified (DT2014; Pujol et al., 2016). The DT2018 mesoscale errors at mesoscaleswere estimated using independent and in-situ measurements. They and have beenare reduced by nearly 3 to 4 % for global and regional products compared to the DT2014. This reduction is even greatermuch more important in coastal areas (reduction is up to 10%) where it is directly linked to the altimeter geophysical corrections 25 appliedused toin the DT2018 processing. The cConclusions are very similar concerning geostrophic currents, where error wasis globally reduced by around 5% and as much asup to 10% in coastal areas.

Introduction
Since 1992, high precision sea level measurements have been provided by satellite altimetry. They have largely contributed to better understand both the ocean circulation and the response of the Earth's system to climate change.has been able to 30 2 provide high precision for mesoscale and large scalelarge-scale monitoring. It has become a key indicator for climate change studies (ref CCI) and a variable of interest for scientist for data assimilation. Following Topex-Poseidon in 1992, the constellation has grown from one to six satellites flying simultaneously (see Figure 1Figure 1figure 1Fig.ure 1). The combination of these altimetersmissions permits to resolve the ocean circulation both on a mesoscale and global scale and on different time scales (annual and inter-annual signals and decadal trends)main space and timescales of the ocean circulation 5 in particular the mesoscale ocean circulation. This has been made possible thanks to the DUACS altimeter multi-mission processing system, initially developed in 1997.In this sense and in order , in this sense and to merge homogenous and intercalibrated altimetery missions, the multi-mission processing system for altimetry data known as the DUACS system haswas developed emerged in 1997.
The multi-mission processing system for altimetry of altimeter data so called known as DUACS (Data Unification and 10 Altimeter Combination System) exists has existed sincewas developed in 1997. Ever sinceSince then, it has been producinged altimetry products for the scientific community in either Near Real Time (NRT), with a delay ranging fromof a few hours to one day, orand Delayed Time (DT), with a delay of a few months, altimetric products for the scientific community. The processing unit has been redesigned and regularly upgraded as the knowledge of altimetry processing has been refined (Le Traon et al., 1998;Ducet et al., 2000;Dibarboure et al., 2011;Pujol et al., 2016). Every few years, a 15 complete reprocessing is performed through DUACSll that includesDUACS data are reprocessed including all altimetry missions and that uses, taking into account the latest up-to-date improvements and recommendations from the international altimetry community. a full reprocessing is performed by DUACS including all missions and taking into account recent improvements and recommendations from the altimetry community. This paper presents the latest reprocessing of DUACS DT reprocessing datareanalysis (written hereafter DT2018) and 20 focuses on improvements that have been conducted implemented since the last preceding version DT2014 (Pujol et al., 2016). Previously reprocessed productsFormer reprocessing (including DT2014) werehave been distributed bythrough Aviso from 2003 to 2017. Since May 2015, the European Copernicus Program (http://www.copernicus.eu/) has taken responsibility for allthe whole the processing, along with the operational production and distribution of along-track (level 3) and gridded (level 4) altimetryer sea level products. have been taken over by the European Copernicus Program 25 (http://www.copernicus.eu/). The L3 products for Sentinel-3's altimetry mission altimeter mission L3 products are processed at CLS on behalf of EUMETSAT, funded by the European Union.
The timeseries of the daily DT2018 products time series starts from January 1 st , 1993 and the temporal extensions of the sea level record areis regularly updated with a delay ofnearly nearly six 6-months delay with present day. Multi-mission products are based on all the altimetry satellites representing a total of 76 mission-years and 20 missions as shown in Figure  30 1 Figure 1. The DT2018 reprocessing is characterized by important major changes in terms of altimeter standards and data processing compared to the DT2014 version. These resultschanges, are highlighted in section 2, and have a significant impact on the quality of the sea level products quality. Two different types of gridded altimetery sea level products are available in the DT2018 version. The first one is dedicated to the retrievingal of mesoscale signals in the context of ocean 3 modeling and analysis of the ocean circulation on a global or regional scale. This requires the most accurate sea level estimation at each time step with the best spatial sampling of the ocean by using all mission available. This type ofSuch dataset is produced and distributed within by the Copernicus Marine Service (CMEMS). The second is dedicated to the monitoring of the long-term evolution of the sea level, for use in both climate applications and the analysis of oOcean/cClimate indicators (such as the evolution of the global and regional Mean Sea Level (MSL) evolution). This 5 requires a homogeneous and stable sea level record and a steady number of two altimeters is used. This second type ofSuch dataset is produced and distributed within by the Copernicus Climate Change Service (C3S). More details on the differences between the products distributed by these twoboth Copernicus Services can be found in section 2.4.
The paper is organized as follows: section 2 considers the DUACS processing , from the level 2 altimeter standards to the inter-mission calibration (level 3) and the mapping procedure (level 4).from the altimeter standards to L3 and L4 products is 10 considered in section 2. Sections 3 and 4 focus respectively on the quality of the global and regional products at different spatial (coastal, mesoscales) and time scales (climate scales) scales. Finally, section 5 discusses the key results and future prospectsperspectives are covered in section 5.

Altimetery standards 25
DUACS system takes Level 2P (L2P) altimetery products as its input data. These data are disseminated by CNES, CLS and EUMETSAT. L2P products are poweredsupplied by L2 products that are distributed by different agencies: NASA, NSOAS, ISRO, ESA, CNES, EUMETSAT. They include the geophysical altimetry standard, that is algorithms and parameters used to retrieve the sea level anomalies from the altimeter measurementsstandards that allow the calculation of sea level anomalies , (i.e.(i.e. instrumental, geophysical and, environmental corrections together with, Mean Sea Surface ( -MSS)), as 30 well as and a validity flag that is used to remove spurious measurements ..

4
Indeed, the altimeteer measurement is affected by various disturbances (atmospheric, instrumental...) that must be estimated to correct it. Specific corrections are also applied to remove high frequency signal that cannot be taken into account in the DUACS processing (Escudier et al., 2017). The Dynamic Atmospheric Correction (DAC) and ocean tide correction are the two main examples. The DUACS DT2018 global reprocessing was an opportunity to take into account new recommendations and new corrections from the altimetry community (Ocean Surface Topography Science Team, OSTST). 5 The altimetery standards have beenwere carefully selected in order to be as consistent and homogeneous as possible between the different various missions, whatever their purpose use (in particular the retrieval of mesoscale signals or climate applications). This selection washas been made possible between 2014 and 2017 in the framework of the phase II of the ESA's Sea Level Climate Change Initiative (SL_cci) project. between 2014-2017. Within thesePart of the project activities included selecting a restricted number of a tight altimetery standards selection has been carried out (Quartly et al., 2017;10 Legeais et al., 2018a). Table 1Table 1 presents the altimetery standards that have been used in the DT2018 and the changes that occurred compared with the previous version (written in bold format). Major changes from the previous version (DT2014) include the implementation of the new GDR-E orbit standard. The Oorbit standards from Jason-1, Jason-2, Cryosat-2, AltiKa, Jason-3 and Sentinel-23A altimeter missions were upgraded from a GDRPrecise Orbit Estimation (POE)OE--D to a new POEGDR-E (Precise Orbit Estimation -D or E standard). The nNew GDRPOE-E standards are 15 reaching of a very goodhigh quality (Ollivier et al., 2015;AVISO, 2017b),. In this version, the main improvementdevelopments concernswe can note among others, the following improvement: the evolutions of gravity field model that has a positive impact on regional MSL error and greatly reducethe important reduction of geographicallycorrelated errors that enable to improve the L2 products.

Developments in Evolution of the DUACS processing
The DUACS processing includes involves an initiala first preprocessing step during which data from the various altimeters are acquired and homogenized.to acquire and homogenize the data from the different altimeter. Then Next, along-track 30 products (L3) and multi-missions gridded products (L4) can be estimated. Finally, the derived products are computed and disseminated to the users. This section is not intended does not aim to detaildescribe the entire data processing system in

Acquisition and preprocessing
The DUACS Pprocessing sequence in DUACS can be divided into several multiple steps: acquisition, homogenization, input data quality control, multi-mission cross calibration, along-track SLA generation, multi-mission mapping and final quality 5 control.
The acquisition and homogenization processesstage consists in retrieving altimeteerr and ancillary data and applying to them those data with the most recent corrections, models and references recommended by expert (as described in section 2.1 and 2.2). This up-to-date selection is available in Table 1Table 1.
The Input Data Quality Control is a process linked related with to the calibration/validation activities carried out for CNES, 10 ESA and EUMETSAT. It is composed of several editing processes designed to detect and fix spurious measurements and to ensure thea long-term stability of L2P products. The up-to-date editing process is described in annual Cal/Val reports for each mission (AVISO, 2017c). Since 2014, and learning from experts' experience, great efforts have been performed made to refine this global global process and notably to adapt tailor some parts to specific regions such as: high-latitude and coastal areas. At high latitudes the idea is to filter an altimeter parameter which has a straight specific signature foron ice, 15 compared to the ocean, and then to flag associated data as ice. But such a filtering solution is affects all datag,lobal andwith the risk that potentially disturbed compromised data outside of icecy areas can be inaccuratelybadly flagged as ice. The proposed updated evolution development consists in using a mask where so that the chosen filtering solution always provides relevant results (Ollivier et al., 2014). The mask is based on the Ssea ice concentration product offrom the EUMETSAT Ocean and Sea Ice Satellite Application Facility (OSI SAF, www.osi-saf.org) and gives us a maximum 20 estimation of ice extent.
In coastal areas, along-track SLA measurements for non-repetitive missions were rejected for L2P DT2014 products, mainly due to a the reduced lower quality ofof the mean sea surface (MSS)MSS afolong closer than 20 km to the coast, all alongtrack SLA measurementr areas less closer than 20 km fromto the coast for geodetic and drifting missions were rejected for non-repetitive missions drastically rejected in the L2P DT2014 products (Pujol et al., 2016). In DT2018 benefits from a 25 solution for , with improved quality MSS solutionquality (Pujol et al., 2018), so efforts were done made to keep retain as muchmany as possible valid measurements as possible close to near the coast. The data selection strategy is based on a median filter applied in a 30km wide strip off the coastline band from the coast (Ollivier et al., 2014). Number of valid data usable in DUACS system is now increased in a substantial proportion, especially for geodetic measurements. As a result, substantially more valid data can be used in DUACS, especially for geodetic measurements. Figure 2 presents an example of 30 the gain of measurements for the Cryosat-2 geodetic mission in DT2018 over the Mediterranean Sea. 100% of the measurements of geodetic missions are recovered in the 20km band near the coast (all rejected in DT2014 version).
6 Finally, Tthe cross-calibration step makes ensures that all data from all satellites provide consistent and accurate information (Pujol et al., 2016). Even if L2 data have been homogenized, they are not always coherent because of various geographically correlated errors ranging from instrument, processing or orbit standards. The first step ensures mMean sea level continuity between altimeter missions is ensured by reducing global and regional biases for each transition ofbetween reference missions (TP-J1, J1-J2 and J2-J3). Then, and iIn order to minimize geographically -correlated errors, two algorithms using 5 empirical process methods are then usedapplied, namely: the Orbit Error Reduction (OER) and the Long Wavelength Error Reduction (LWER)). The OER is based on a global crossover minimization performed on mono and multi-missions crossovers (Le Traon and Ogor, 1998). The LWER is based on an optimal interpolation process and aim to remove local bias between neighboring for each satellite (Le Traon et al. 1998 andDucet et al., 2000).

Along-tTrack product generation 10
The along-track generation for repetitive altimeter missions is based on the use of a mean profile (MP) ( Table 2, Dibarboure et al., 2011 andPujol et al., 2016;and Dibarboure et al., in prepreview). These MPs are necessary in order to co-locate the sea surface heights of the repetitive tracks and to retrieve a precise mean reference in order to for the computeation of sea level anomalies. The methodology used to computefor the DT2018 MP computation wasis the same as infor DT2014. The .
Ddifferences come arise from the upstream measurements, with as new altimetery standards were used in DT2018 15 (described in section 2.2), along with new data selection (see section 2.3.1) and reviewed temporal periods for the different altimeters considered. Table 2Table 2 introduces presents the altimeter missions and time periods used to compute the four different MPs that are available along the following tracks: TopexPoseidon/Jason1/OSTM-Jason2/Jason3, TopexPoseidon Interleaved Phase/Jason1 Interleaved/Jason2 Interleaved, ERS-1/ERS-2/Envisat/Saral-AltiKa and Geosat Follow On tracks.
UnlikeCompared toFollowing the previous MPs version of the MP, additional measurements collected by OSTM/Jason-2 20 and SARAL/AltiKa between 2012 and 2015 were usedexploited for DT2018. They concern OSTM/Jason-2 and SARAL/AltiKa. Since March 2015, however, AltiKa has been considered as a drifting non-repetitive mission for Delayed-Time products. As a resultTherefore, no we do not take into account any measurements after that date were taken into accoundt when to computinge the ERS-1/ERS-2/EN/AL MP. beyond that date. To limit the error of ionospheric correction error inover the ERS-1/ERS-2/EN/AL mMean pProfilethis MP, no ERS-2 data collected from between January 2000 andto 25 October 2002 have not beenwere used to compute the MP because. Indeed, during this period, the ionospheric activity was much more intense during this period than between 1995 toand 2000.
New DT2018 MPs wereare defined as close to the coast as possible as illustrated in Figure 3  It should be noted that for the Sentinel-3A, it was impossible to estimate mission the estimation of a precise MP was not possible for this reprocessing, due to the short time period (i.e. a few months) available to compute it. Consequently, data from the Sentinel-3A mission wereare only interpolated ionto theoretical positions Theorical Track (Dibarboure et al., 2011), 5 then and the gridded MSS (Pujol et al, 2018) iswas removed. Since the reprocessingthen, an MP has been evaluated calculated (Dibarboure et al., in preprev.;and Pujol et al., 2018b) and the Sentinel-3A dataset will has been reprocessed in a future CMEMS version in 2019.
For non-repetitive missions (ERS-1 during its geodetic phase, Cryosat-2, Hayaing-2A, both Jason-1 and Jason-2 in their geodetic phase, Jason-2 geodetic phase,and SARALaral-AltiKa in its geodetic phase), no MP can be estimated. The SLA is 10 thenin this case derived along the real real altimeter tracks using the gridded MSS (Pujol et al., 2016;and Dibarboure et al., in preprev).
The finalLast step of the along-track processing consists ofin noise reduction using by loaw-pass Lanczos filtering, and subsampling. This process remains unchanged compared tofrom the DT2014 version (Pujol et al., 2016).
DT2018 Rreprocessing was also the opportunity to propose new products. New along-track products were tailored for 15 assimilation purposes andto provide users with the specific geophysical corrections, used to compute the sea level anomaly in the DUACS processing: DAC, ocean tide and LWER. As explained in section 2.2, these geophysical effects are taken into account in DUACS because their temporal variability is too high to be resolved by altimeter measurements and to be mapped using the OI method. 20

Gridded product generation: multi-mission mapping
The multi-mission mapping proceduress in DUACS is based on an optimal interpolation (OI) technique derived from LeTraon et al., 1998;Ducet et al., 2000 andLeTraon et al., 2003. This method aims is designedat to generate producing regularly gridded products offor Sea Level Anomalies by combining measurements from different altimeters. The main objective in the DT2018 reprocessing framework was to improve gridded altimetry products improvements were focused 25 mainly iin the tropics, in coastal areas and at mesoscale. To do so, The last reprocessing DT2014, have shown great improvement on the SLA signal reconstruction mainly offshore (Pujol et al., 2016). The reprocessing DT2018 focused on what had been less emphasized on the previous reprocessing: coastal scale and mesoscale, to do so e. Specific parameters in the DT2018 OI processing parameters werehave been optimized adjustedto this effect. 30 The sea level variability of the signal's spatial and temporal variability scales of the signal havewere been updatedmore accurately defined based, on the 25 years of available observations available. PA particular attention washas been paid put 8 toon coastal areas, where spurious peaks of high variability were able to behave been reduced. An optimized selection of along-track data was incorporated into OI processing by changing T the size of the suboptimal interpolation window, decreasing it by one third decreased by 33% in regions of high variability region and in the equatorial bandbelt. . OI Oobservation errors were increased in the equatorial bandbelt, as the impact of filtering and subsampling had been 5 previouslyhad beenwere underestimated in this area whichere they generates noise at small scales oin gridded products.
An optimized selection of the data has been implemented in DT2018 products. The impact is visible at different scales (mesoscale, coastal and climate scale) and over global and regional products. The observations errors have been refined.
Errors induced generated when using the gridded MSS werehave been updated with the use of the new MSS versionupdated replaced with the new one for missions that do not use a precise MP (Pujol et al., 2018a). In addition, the a-priori 10 knowledge of the signal variance has been updated based on the 25 years of available observations. Correlation scales were onlyremain unchanged for the global and Black Sea products, compared with the ones used in DT2014. They have been reviewed only for the regional Mediterranean products. While set to a constant values (100 km and 10 days) in the DT2014 version, a specific effort has been made to compute precise covariance and propagation models 15 werehave been computed tofor thise DT2018 regional mapping. Spatial scales now range from 75 km to 200 km whileand temporal scales remain atare set to 10 days. These changes have actively contributed to the improvingimprovement the retrieval of the mesoscale signals' retrieval in the Mediterranean regional products (see section 4).
For the Black Sea processing, OI parameters are now similar to the global ones parameters used for the global ocean processing, except for the correlation scales which are still set to 100km and 10days. 20

Different products for different applications
Two different types of altimeter sea level gridded altimetry products are available in the DT2018 version. The first ontypee, produced and distributed within within the Copernicus Marine Service (CMEMS), is dedicated to the mesoscale observation.
The other onetype, produced and distributed withinwithin the Copernicus Climate Change Service (C3S), is rather dedicated to the monitoring of the long-term evolution of the sea level for use in climate applications and forthe analyzingsis of 25 oOcean/cClimate indicators (such as the global and regional MSL evolution). Two types of altimeter processing configurations are exploited to build these two products.Different processing parameters are used to generate leads to these two products. The first difference of configuration is related to the number of altimeters used in the satellite constellation.
The mMesoscale observation requires the most accurate sea level estimation at each time step, along with the best spatial sampling of the ocean. All available altimeters are thus included in the CMEMS products, and the sampling can vary with 30 time depending on the constellation status. At the oppositeOn the other handIn contrast, the temporal stability of the surface sampling is more important when monitoringrather required for the long-term sea level evolution observation. A steady number (two) of altimeters (two) are thus used in the C3S products. This corresponds to the minimum number of satellites required to for the retrieveal of mesoscale signals in delayed time conditions (Pascual et al., 2006;and Dibarboure et al., 2011). Within the production process, the long-term stability and large-scale changes are built established onupon the records basis of records from the reference missions (TOPEX-Poseidon, Jason-1, Jason-2 and Jason-3) used in both CMEMS and C3S products. The Any additional missions (e.g. up to 5as many as five additional missions in 2017) are then homogenized with respect to the reference missions and contribute help to improve the sampling of mesoscale process 5 samplinges, provideing the high-latitude coverage and increaseing the product accuracy. However, the total number of satellites hasstrongly greatly variesd during over the altimetry era and some biases may appear develop whenwith the introduction of a new satellite flying on a drifting orbit is introduced., Each addition, which may affect the stability of the global and regional MSL from by several millimeters (data not shown here). AlthoughEven if the spatial sampling is reduced with when there are fewer satellites, the risk of introducing such anomalyies is thus also reduced in the C3S 10 products, resulting in and the stability is improved stability. In the CMEMS products, the stability is ensured by the calibration with the reference missions and the mesoscale errors are reduced due to the improved ocean surface sampling thanks to made possible by using the use of all the satellites available in the constellation.
As a second difference of configuration, the reference used to compute the Sea Level Anomalies for C3S products was is a Mean Sea Surface (MSS) for all missions in the C3S products whereas for CMEMS products, a mean profileMP of sea 15 surface heights is was used along the theoretical track of the satellites following with a repetitive orbit (see section 2.3.2) in the CMEMS products. The use of MP increases the local accuracy of the sea level estimation (Pujol et al., 2018a and Dibarboure et al., in prep) but fFor the C3S productions, a non-repetitive mission (Cryosat-2) has beenwas used for a short period of time. Considering the regional mean sea level temporal evolutionUnfortunately, the combined use of MSS and MPMPmean profile for successive missions in the merged product give rise to regionalcan be at the origin of c centimetric 20 bias when these products arefor regional products (data not shown here). SoConsequently, the systematic use of the MSS for all missions has been privileged in the C3S products to ensure contributes helpsto ensuringe the MSL stability in the C3S products; and and the use of MPmean profiles for repetitive missions has been selected in the CMEMS products to increase their accuracy of the CMEMS products is increased with by usingthe use of the mean profiles for repetitive missions.
The Ddifferences between CMEMS and C3S product quality are discussed aton a climateic scale in section 3.4. 25

DT2018 Global products quality
This sectione following chapter focuses on the quality of gridded (L4) products. We analyzedS sea surface heights and

Mesoscale signals in Aalong-Ttrack and gridded products
Optimizing theThe mapping process optimization (section 2.3.3) and incorporating the new altimetery corrections (section 2.2) haved a direct impact on the observation of ocean sea level and surface circulation dynamicsphysical content observed in the gridded products. To characterize this impact, the difference between DT2014 and DT2018 temporal variability is presented shown in Figure 4Figure 3. An Aadditional variance of between 2% and 5% is observed for high variability 5 regions in DT2018 products and is linked to the new OI parametrization. This represents between 2 to 5% of DT2018 the variance. This increase is mainly due to having changed the OIthe new variaibility of spatial and temporal scales of the signal used in the mapping process and+ decreased of the size of suboptimal interpolation window size. The OI selection window ,is more focused on closed observations (both spatial and temporal) -> more var of the signal. OIn coastal areas, an important substantial reduction in SLAof the variance of the SLA is observed; this is duebeing related to both the  2018a)) have underlined emphasized that the new gridded MSS shows lessa reduced degradation of SLA degradation near the coast. These improved standards contribute to a valuableimportant local reduction ofin the SLA error variance (up to 50% alongshore). At high latitudes, the difference of variance is important significant (±100cm² to ±200cm²), and) and is 15 linked due to the new MSS correction. Indeed, Pujol et al., ((2018a)) have shown that the CNES_CLS 2015 MSS improves both coverage in the Aarctic and resolution of the shortest wavelengths at high latitudes.
Compared to the DT2014, the new version revealshas more intense geostrophic currents in western boundary currents (geostrophic part). This has a direct impact on the Eddy Kinetic Energy (EKE) derived from these products. Figure 5 Figure   4 presents the spatial difference inof the mean EKE over global ocean between DT2018 and DT2014 products, along with 20 products and also their temporal evolution. As observed before infor the differences of SLA variance, we clearly see a higher energy is evident in high variability areas. This represents corresponds to a 2% increase in EKE in DT2018. However, in the equatorial band belt (±20°N), the EKE in the DT2018 is lowerless important (-17%). This is a direct consequence of the noise measurement that is taken into consideration in the mapping process for all satellites: observation errors increasedprescribed during OI in the tropical belt have been increased, so the SLA signal is smoother and less energy is 25 observed in this region. This is linked with the evolution of the noise measurement considered in the mapping process for all satellites. The consistency between altimeter geostrophic current and independent measurements is significantly improved in this area as discussed in section 3.2. IOn coastal areas, the DT2018 version version presents less fewer spurious peaks of high EKE (Figure 4Figure 4 b). As already stated, this is linked related to with the improved altimetricy correction and lower the variance SLA reductionvariance. Considering the mean EKE time series, a global reduction of 26 cm² (17%) is observed 30 for dataset the DT2018 dataset. ItThis is directly linked due to the lowerwith the equatorial tropical EKE reduction.

11
products is loweress peaky than in DT2014. This illustrates that EKE variations are less important, there isare fewer isolated anomalies (and these are mostly coastal) in the new DT2018 products.
The gridded SLA accuracy of the gridded SLA wasis estimated by comparisonng SLA with independent along-track measurements. Maps produced by merging only two altimeters (C3S products) arewere compared with SLAs measured along -track from the tracks of another mission that was kept independent ofromf the mapping process (see Pujol et al., 2016 5 for full methodology). Topex-PoseidonTP interleaved iswas compared with gridded products that mergeds Jason-1 and ENnvisat over the year 2003-2004. It is therefore then important to notemust be pointed out that these results are much more representative of "two-sat-merged" gridded products combining two altimetry missions. PThe "all-sat-merged" products combining all available missions can usually benefit from an improved sampling when three to six altimeters are used. Thus, the errors described here should thus be considered as the upper limit. Table 3Table 3 summarizes the results of the 10 comparisons over different areas. Figure 5 shows the percentage of the difference in variance between gridded products and TP independent along-track measurements for DT2018 and DT2014 products. The gridded product error for mesoscale wavelengths ranges between 1.4 cm² (for a low variability area) and 37.7 cm² (for a high variability region). The improvements in of DT2018 compared with DT2014 isaffect all areas global.: Offshore, the improvement is quite fairly low (around 3%) and is associated with the enhanced version of the OI mapping parameter of the OI. In coastal areas, the 15 improvements are more significant (around 10%) and linked relatedcaused by to with the use of the new Tide tidal correction (FES2014) and, to a lesser extent, with to the MSS and MPs. In the tropical belt, improvements are also significant (around 9%) and related to the observation errors that were increased in this area for the OI processing.

Geostrophic current quality
DT2018 aAbsolute geostrophic currents for DT2018 were has been assessed using drifter data for the time period 1993-20 2017 time period. The AOML (Atlantic Oceanographic & Meteorological Laboratory) database has beenwas used for the comparison (Lumpkin et al. 2013). These in-situ data are were corrected fromor Ekman drift (Rio et al., 2011) andbut also from wind if a drifter's' drogue hads been lost (Rio et al., 2012) so as to be comparedable with the altimetry absolute geostrophic currents. Drifters The pPositions and velocities of drifters arewere interpolated using a 3-day low-pass filter in order to remove high-frequency motions (Rio et al., 2011). The aAbsolute geostrophic currents derived from altimetry 25 products are were then interpolated onto drifter positions for comparison.
The distribution of the current's intensity shows an overall underestimation of currentmagnitude in altimetry products compared to drifter observations (data not shown).
As the previous version (Pujol et al., 2016), the comparison reveals that DT2018 altimetry products underestimate absolute geostrophic current. Figure 6Figure 6 shows the RMS difference between the DT2018 geostrophic current and that of 30 drifters. The mMean RMS is nearly 10 cm/s and the main errors are located nearshore and in high variability region with peaks higher than 20 cm/s. Taylor skill scores (Taylor, 2001) werehave been computed for the zonal and meridional components of the current in DT2018. This assessment lookenablestook into consideration both the signal's variance The Table 4Table 4 summarizes the mean rms RMS of the differences between geostrophic current maps altimeter maps and drifter measurements over different areas for the versions DT2018 and DT2014 version. DT2018 products are more consistent with drifter measurements than the DT2014 version products. The improvement is clearly visible in the intra-5 tropical bandbelt. The vVariance of the differences with drifters is reduced around by 7% 20 to 40% in this area. Additional noise-like signals, presentviously introduced in the DT2014 version ahad reduced nd leading to a degradation of the consistency with drifter measurement (Pujol et al., 20146). This degradation was is now corrected forin the by DT2018 version. This is directly linked to the change inof the mapping parameters used for this updated version (see section 2.3.3). A Ssignificant improvement canis also be observed in coastal areas, where with a reduction of the variance of the differences 10 with drifter measurements is reduced byreaching nearly 15% (Table 4). Elsewhere, this reduction in thee variance of difference reduction ranges between from 4 and to 7%.

Coastal areas
As described in sections 2.3.1 and 2.3.2 the new DUACS DT2018 processing has a keyan important impact on coastal areas.
The clearest impact is the major gain of points from every non-repetitive missions and missions not having a MP. We gain 15 all points no further than 20 km of the coast for these six missions over 16 years in total. There is also an improvement for repetitive missions since in average we gain points nearshore (Figure 3). and Ooverall, all missions have more measurements available in DT2018 compared to the previousDT2014 version.
Specific efforts were done in the DT2018 processing to improve the products quality near the coast. Choice of up-to-date standards, specifically ocean-tide and MSS (see section 2.2), clearly contribute to the quality of the altimeter measurement 20 near the coast. Additionally, refined data selection (see section 2.3.1) significantly increase the data availability in the in the band 20km close to the coast. Finally, review of the mapping parameters (section 2.3.3) also contribute to the improved quality of the gridded products in the coastal area.
Previous comparisons between gridded maps and independent measurement underlined the positive impact of the DT2018 processing in the coastal area. Compared with results obtained with DT2014 version, we observe with DT2018 a reduction 25 of the variance of the differences between gridded SLA products and independent along-track measurements by nearly 10% (Table 3, Section 3.1), and a reduction of the RMS of the differences between altimeter geostrophic current and drifter measurement by nearly 15% (Table 4, section 3.2).
The assessment of the gridded products in coastal areas includedwas completed with a comparison with tide-gauges (TG) measurements. We have used mean monthly mean TG measurements from the PSMSL network (Permanent Service for 30 Mean Sea Level, PSMSL, 2016) from 1993 to 2017. We considered used only long-term monitoring stations with a lifetime of moregreater than two2 years. Sea surface hHeight measured by TG iswas compared with gridded SLA by considering the maximum correlation with the nearest neighboring pixel (Valladeau et al., 2012;and AVISO, 2017a). In Figure 7, Figure 7 Mis en forme : Police :Non Italique Mis en forme : Police :Non Italique 13 the variance of the difference between DT2018 altimetricy products and TG measurements is compared with that obtained from the differences using DT2014 altimetricy products. The results show a global reduction inof the variance (0.6%) when DT2018 data are used. There is a clear improvement along the Indian coast, Oceania and northern Europe. A lLocal degradation can be observed along the coast of Spain and along the United States' Western coast of United States. These degradations, which that are not observed in other diagnoses such aslike independent along track measurements still need to 5 be further investigatedare not yet understood yet..

Climate scales
The global mean sea level (GMSL) is a key indicator of climate change since it reflects both the amount of heat added in the ocean and the land ice melt coming mainly from Antarctic and Greenland ice sheets and glaciers. Three different altimeter products can be used to compute three GMSL estimates: and can be computed from tthe time series of the box-averaged 10 along-track measurements of the the reference missions only , . The global MSL can also be derived from the DUACS L4 merged gridded sea level products from CMEMS and C3S distributed by both marine and climate Copernicus services (e. g. Figure 8Figure 8, left). Considering For the same products versions and computation periods of computation, these three GMSL estimates (box-averaged mono-mission and two gridded products) of the global MSL are considered to be equivalent since almost the same altimetery standards are used to compute the sea level anomalies and for 15 all products, the long-term stability for all products is ensured by using the same reference missions. The remaining observed global GMSL differences observed (~0.17mm/year) are not significant given the uncertainty considered on different scales (uncertainty in the GMSL trend is approximately of 0.5 4 mm/yr.ear at the 90% confidence level given by Legeais et al., 2018bAblain et al. 2019. Note that as aforementioned (section 2.4), differences can be found between the two different Copernicus gridded products (CMEMS/C3S) when computing regionally-averaged MSLsthe situation is not the same on a 20 regional scale where differences can be found according depending onto the product used (CMEMS/C3S) for the MSL computation.
When computing area-averaged MSL time series, users are advised that the DUACS products are not corrected for the effect of the Glacial Isostatic Adjustment (GIA) due to the post glacial rebound. A and a GIA model should be used to estimate the associated sea level trends. 25 In addition, between 1993 and 1998, the globalG MSL ishas been known to have been be affected by an instrumental drift in the TOPEX-A measurement,s which has been as quantified by several studies (Watson et al., 2015Watson et al., 2015Beckley et al., 2017Beckley et al., 2017. The altimeter sea level altimetry community agrees that it is necessary to correct the TOPEX-A record for the instrumental drift to improve the accuracy and reduce the uncertainty inof the total sea level record. However, there is not yet consensus so far on the best approach to estimate the drift correction at 30 global and regional scales. The DUACS altimeter sea level altimetry products are not corrected for the TOPEX-A drift, waiting pendingfor the on-going TOPEX reprocessing by CNES and NASA/JPL but the users can apply their own 14 correction. Adjusting for this TOPEX-A anomaly create a GMSL acceleration of 0.10mmyr -2 for the 1993-2017 time span that does not otherwise appear otherwise (WCRP 2018).  Legeais et al., (2018b). The map of the differences offor the local MSL trend derived from the latest and previous products versions (Figure 8Figure 8, right) displays a pattern predominantly associated with the differencet of orbit standards used in the two b productoth versions of the products (GDR-E versus GDR-D, see Table 1Table 1). Such a result is confirmed by the comparisonng of the altimetery products with the independent measurements of dynamic height s measurements derived from in-situ Argo profiles (Valladeau et al., 2012;Legeais et al., 2016). 10

SLA field quality
As previously discussed for the Gglobal ocean products, the quality of the regional gridded SLA products is estimated bythrough comparison with independent altimeter along-track and tide gauge measurements.
The Figure 9Figure 9 shows the spatial distribution of the RMS of the differences between regional DT2018 SLA gridded 15 products and independent along-track measurements (Topex/Poseidon iInterleaved along-track measurements over the period [2003][2004] period). The Mmain statistics onfor these comparisons, as well as a comparison with the previous DT2014 version, are also given in Table 5Table 5. In contrast withContrary to the processing applied for Gglobal products assessment, the evaluation of regional products cannot include the mesoscale signal analysis: the short length of the main part of the tracks segments available over these the regional Sseas does not allow us to accurately filtering of the signal in 20 order to focus specifically on mesoscale signals. The results obtained show that fFor the DT2018 Mediterranean product, the main errors are located oin coastal areas and in the Adriatic and Aegean Seas, with RMS values ranging from 6 to 9 cm. The Black Sea products present also show higher errors oin coastal areas (results not shown here). The mean rms Variance of the differences between gridded products and along-track measurements is reaches nearly 17 cm² and (23 cm²) over the Mediterranean Sea. and the ( Black) Sea. This value is higher than the mean error observed over low variability areas in the 25 Gglobal ocean (Table 3Table 3), mainly due to the different wavelengths addressed in these comparisons. Compared to the previous regional version DT2014 version, the error is reduced by 4.2% (3.5%) for the Mediterranean Sea and 3.5% for the (Black) Sea. It is important to note that these results are representative of the quality of the gridded products quality when only two altimeters are available. These products can be considered to beas degraded products for mesoscale mapping since they use minimal altimeter sampling. 30 Compared to the previous version, cConsistency with monthly TGTide Gauges measurements (Figure 10Figure 10) is improved locally with in the regional DT2018 Mediterranean gridded product from the Balearic to Ligurian Seas as well as

Geostrophic current quality in the Mediterranean Sea
DT2018 regional absolute geostrophic current in the Mediterranean basin has beenwas assessed using drifter data for the period  period. The data were collected from dDrifters released in the Mediterranean Sea as part ofin the frame of AlborEx (Pascual et al., 2017) Table 6Table 6 summarizes the main statistical results for the whole basin. The DT2018 regional product presents a correlation coefficient with drifter data 4% larger greater than that obtained when using the DT2014 regional product. 15 Moreover, the errors in the former later version are slightly reduce lowerby 1%, whilst its improvement in the explained variance reaches is as high as 14%.
We repeated tThe analysis was then repeated but for the different dynamical sub-regions of the basin (see Figure 11Figure 11.a) reported by Manca et al. (2004). This differentiation is based on the typical permanent features in the upper 200 m of the water column. Overall, comparisons between geostrophic velocities derived from the DT2018 regional product and 20 absolute surface velocities retrieved by the drifters (Figure 11Figure 11. be) present reveal a correlation coefficient larger greater than 0.40 in most of the boxes. Correlations larger greater than 0.50 are mainly located in the southernmost part of the basin where a stronger mesoscale activity occurs; namely the Alboran Sea (DS1), the Algerian Basin (DS3 and DS4), the Sardinian Channel (DI1), the Sicily Strait of Sicily (DI3), the Ionian Sea (boxes DJ7, DJ8 and DJ5), and the Cretan passage (DH3). The overall RMS difference between both datasets ranges between 8 and-11 cm/s, although whilst it reaches 20 25 cm/s in DS1 due this area'sto the strong dynamics of this area. Slightly larger errors are obtained when comparing the DT2014 product with drifter observations (figure not shown here). Furthermore, drifter data collected in boxes DS1, DS3, and DS4 present have the largest variability due to the aforementioned mesoscale activity. This fact is also reflected in the two altimetry products, which present havethere the largest variance values in the Mediterranean basin.
Overall, the correlation coefficient between the DT2018 regional product and in-situ drifter data is improvesd by between  10% with respect to that obtained when using the DT2014 product (Figure 11Figure  with respect to DT2014 lies ion the variance explained (Figure 11Figure 11.h), which presents values nearly 20% (10%) larger higher in the former later product in some places areas of the western part of the(eastern) basin and nearly 10% higher in the eastern part. This is due to a better captureing of the mesoscale activity. This improvement is not observed in the 5 northernmost part of the basin, where lessa lower mesoscale activity occurs.

Discussions and Conclusions
More than 25 years of Level-3 and Level-4 altimetery products have beenwere reprocessed and delivered as theversion DT2018 version. This reprocessing takes into account the most up-to-date altimetery corrections and also includes changes in the parameters involved in the mapping processing parameters. These changes impact the SLA signals at multiple 10 temporal and spatial scales.
A notable important change concerns the gridded altimeter sea level altimetry products that are available in the version DT2018 version. They are produced and distributed through two different Copernicus Services that correspond to different applications. Through CMEMS distributes, maps that includede all the available altimeter missions available are distributed.
These mapsy provide give the most accurate sea level estimation with the best spatial and temporal sampling of the ocean at 15 all times. Through C3S, maps that include only two satellites are used to compute the most homogeneous and stable sea level record though over time and space. Sea lLevel C3S products are dedicated to the monitoring of the long-term sea level evolution of the sea level for climate applications and the analyzingsis of Oocean/cClimate indicators (such as the global and regional MSL evolution).
Other changes werehave been implemented in the DT2018 processing;: the altimetery standards and geophysical corrections 20 were brought are up-to-date with expert recommendations, and mapping parameters, have been refined including spatial and temporal correlation scale and measurements errors were refined. We also focused on the improvingement of coastal editing to gain obtain many relevant sea level data, mainly forom drifting altimeters. Additional sea level data have beenwere incorporated into used compared to the DT20148, especially in particular Sentinel-3A measurements that takenhave been used over a 6-month extensionded period. 25 Having discussedDissussingDiscussing these important key changes, we have then focused on the describingption their of the impact on gridded sea level products. The SLA variability has been increased in energetic areas (from 5 to 10%) and reduced decreased locally along the coasts (up to 50%). A 10% EKE decrease in the equatorial beltand has is also been observed and linked related to the refined reduced measurements errors prescribed for OI in thise area.
To realize achieve independent comparisons, we have used unrelated in-situ measurements. G geostrophic currents have 30 beenwere examined d and are still underestimated compared to the in-situ observations. Nevertheless, c Compared to the version DT2014 version, offshore improvements (+4-5%) particularly in the tropics (+5-10%) and coastal improvements 17 (+10%) have been shown demonstrated using independent drifters' data. An Iindependent along-track sea level comparison and Ttide Ggauges comparisons have strengthened these conclusions.
Regional products are also improved enhanced with DT2018, taking advantage of the altimeter new standards and processing. The SLA gridded product errors in the regional products are have decreaseding by from 3% to 4% when estimated using independent along-track measurements. 5 The lLimitations exposed by Pujol et al. (2016) are still valid and the errors observed in the retrievingal of mesoscale features also highlight the L4 product's spatial resolution capability. To estimate the spatial resolution of the gridded products, anan evaluation washas been carried out done based on a spectral coherence approach. A full description of this approach can be found in Ballarotta et al., (in prep.2019).
Many products applications are derived from these global and regional gridded products and are strongly greatly benefit 10 from affected byaffected by the the productsir quality: the Lagrangian products (FSLE d'Ovidio et al. 2015), or and eddy tracking application (Delepoulle et al., 2018) are a prominent examples.
Medium t-term developments concern new Level-3 products that will be dedicated to data assimilation and the CMEMS Monitoring Forecasting Centre. These new products will be new in Delayed-Time mode. The Mean Dynamic Topography will also be updated, and the Black Sea area will be integrated. Finally, a new regional European regional product will 15 substitute to the current Mediterranean and Black Sea products.
In the coming years, DUACS will face important major challenges with the arrival of new altimeter missions. SWOT, for example, will observe fine-scale dynamics, with swath SSH observations (Morrow et al., 2018), that will need to be integrated into the DUACS system. To do so, theThe next step, therefore, will consist in moving towards a higher resolution for along-track and gridded products. New mapping techniques should also be taken into consideration and are currently 20 being studied such as dynamical advection (Rogé al., 2017, Ubelmann et al., 2016.
The L3 products for Sentinel-3's altimetry mission are processed at CLS on behalf of EUMETSAT, funded by the European  Solid earth tide Elastic response to tidal potential [Cartwright and Tayler, 1971], [Cartwright and Edden, 1973] Mean Sea     [-22, -8°N] and corresponds to a very low-variability area (between 0 and 7 cm²) area in the South Atlantic subtropical gyre where the observed errors are small.l Table 4: Variance of the differences between gridded geostrophic current (L4) DT2018 products and independent drifter measurements (unit = cm2/s2). In parenthesis: variance reduction (in %) compared with the results obtained with the DT2014 products. Statistics are presented for latitude selection (5°N<|LAT| < 60°N).

Zonal Meridional
Reference area* 44.        successively DT2018 and DT2014 SLA gridded products. We used mean monthly TG measurements from the PSMSL networkMonthly Tide Gauges come from PSMSL network. Negative values represent reducedmean that the SLA differences between DT2018 altimetry gridded SLA altimetry and TGs are reduced when considering DT2018 products. The statistic is expressed as a percentage of RMS of TG measurements. : Difference of the RMS of the difference between gridded regional Mediterranean Sea (left frame) and regional Black Sea (right 10 frame) SLA products and independent Topex/Poseidon interleaved along-track SLA measurements, using successively DT2018 and DT2014 version. Negative values represent reduced differences between DT2018 altimetry products and independent along-track measurements. The statistic is expressed as a percentage of RMS of the independent along-track product.RMS of the difference between regional Mediterranean Sea (left frame) and regional Black Sea (right frame) gridded DUACS DT-2018 sea level anomaly and independent TP along-track measurements over the period [2003][2004]  : Difference of the variance between regional Mediterranean gridded products (upper frame) and regional Black Sea products (lower frame) SLA products and TG, using successively DT2018 and DT2014 gridded products. We used mean monthly TG measurements from the PSMSL network. Negative values represent reduced differences between DT2018 altimetry gridded SLA and TG. The statistic is expressed as a percentage of RMS of TG measurements. The statistic is expressed as a percentage of RMS of the independent along-track product.Difference of the variance of the altimeter SLA minus TG SLA differences, using