Saily WR

MedFS

CMCC's Mediterranean Forecasting System, from Copernicus Marine: hourly current, tide computed in the model, a grid of about 4 km, over 10 days.

Now in Saily WR

Model Type Status Run Frequency Resolution Horizon
ItalyMedFSCurrentUp to date15/11 00z1×/day0.0417°8 d 23 h

Spatial coverage

From a pilot project to EAS10

The Mediterranean Forecasting System began as a European pilot project, from 1998 to 2001, whose aim was to show that ten-day forecasts of the whole basin could be made in near real time. It set up lines of ships dropping temperature probes every 12 nautical miles, a moored observing station, the use of satellite altimetry as it arrived, and a basin model at 1/8°. From January 2000 it produced an analysis and a 10-day forecast once a week. Its quality document still dates the system's analyses and forecasts from 2000.

A second project, MFSTEP, ran from March 2003 to June 2006. Its model had a grid of 1/16°, about 6.5 km, and 72 levels, in a large Atlantic box closed at its edges. It went into operational mode during a demonstration from September 2004 to March 2005 and, as Pinardi and Coppini wrote in 2010, had not stopped since, with a ten-day forecast every day. In 2009 Oddo and colleagues reported a twin experiment over 2004–2007: the same Atlantic box, opened and nested in a global ocean model, improved the salinity of the Atlantic water entering the Mediterranean and the seasonal rise and fall of its sea level.

Then came the waves, first from the WAM model in 2010, which gave the ocean its drag coefficient, then a two-way coupling with WaveWatch-III in 2013, and in 2015 the service under Copernicus Marine. The step to the present grid came in 2017: 1/24° in the horizontal, 141 levels against 72, 39 rivers. The Dardanelles became an open boundary in 2019, and the atmosphere moved to ECMWF's 1/10° fields in December 2020. The tide came in May 2021, with version EAS6, recorded as a major change of the modelling system. Coppini and colleagues, assessing the system up to 2020, still put part of its salinity error down to the lack of tides at Gibraltar.

Since 2022 a new version has come each November. EAS7, in 2022, improved the tide and changed the friction in the straits of Gibraltar and Messina. EAS8, in 2023, moved to NEMO v4.2 and a newer WaveWatch-III, joined by the OASIS coupler, and began assimilating in situ data in the Atlantic box. EAS9, in 2024, took daily river flows from EFAS, gliders and finer altimetry. EAS10, on 25 November 2025, brought a new forecast chain, a new assimilation scheme and a new boundary at the Dardanelles. The Mediterranean centre of Copernicus Marine is led by CMCC, with HCMR, OGS and CINECA as consortium members.

The science: a small tide, narrow straits, an Atlantic box

MedFS is a regional configuration of NEMO v4.2, on a grid of 1/24° and 141 levels, its time step 180 seconds. The surface moves freely and non-linearly, and the levels stretch with it (z-star coordinates); the depth-averaged flow is computed apart, on shorter steps, so that fast waves at the surface are resolved. Vertical mixing depends on the Richardson number, friction on the sea floor follows a logarithmic layer. The sea floor starts from GEBCO's 30-second grid, filtered and then changed by hand in critical places: near the Gibraltar and Messina straits, among the islands of the eastern Adriatic and at the edge of the Atlantic box.

The tide is computed in the model, not added to it. The tidal potential acts across the whole domain for the 8 main constituents of the Mediterranean, and at the Atlantic boundaries the tide enters as heights from FES2014 and currents from the TUGO model. SMOC works another way: it adds a tide from FES2014 to a circulation computed without one. The tide here is small, typically 0.2 to 0.3 m in the western basin, but it matters: in a twin experiment on an earlier version of this configuration, with and without tide, McDonagh and colleagues found the winter mixed layer deeper over most of the basin, notably where deep and intermediate water forms, and more deep water formed in the western Mediterranean.

The Atlantic box is there for the Mediterranean's sake: the quality document says the model extends into the Atlantic to better resolve the exchanges at the Strait of Gibraltar. It is the model's joining zone with the global system, nested in that system's daily fields at 1/12° and 50 levels. That same document gives its coverage as excluding the Bay of Biscay and the Black Sea, while the distributed grid has water in the south of the Bay of Biscay, up to 45.15°N. At the other end, the Sea of Marmara is not modelled: since EAS10, a high-resolution model of the Turkish straits gives temperature, salinity and sea level at the Dardanelles, and a daily climatology from a simpler Marmara model gives the velocities.

The ocean is coupled both ways with WaveWatch-III v6.07, on the same grid and the same time step, every hour. NEMO sends the difference between air and sea temperature and its surface current, which the waves refract on. The wave model sends back one field, the neutral drag coefficient, used to work out the wind's stress on the water. The surface current matches the model's first level, at 1.02 m: Saily WR found it within 0.0001 m/s of the top level of the 3D fields, over 24 hours in November 2026, at Gibraltar, at Messina and west of Gibraltar.

How a run is computed

The daily cycle starts at 08:00 UTC. It reruns the three days before with the observations, one assimilation every 24 hours, and the analysis it ends on starts a 240-hour forecast covering the current day and the next nine. Copernicus Marine's target for delivery is 16:00 UTC; on 6 November 2026 the hourly surface current was written at 13:05 UTC. A weekly cycle starts on Mondays at 12:00 UTC and is delivered on Tuesdays at 16:00 UTC: it goes back 15 days to take the best altimetry available, reworking the first 7 as a best analysis and the last 8 as an adjusted one, and it too ends in a 10-day forecast.

OceanVar2, CMCC's 3D-Var scheme, corrects the model with temperature and salinity profiles from Argo floats, gliders and ships' XBT probes, and with sea level anomaly along the tracks of the Altika, Cryosat, Jason, HY-2 and Sentinel-3 and 6 satellites. Its background errors vary each month at each grid point, drawn from a 30-year reanalysis. This version's new operator carries altimetry closer to the coast. In the Atlantic box only the in situ profiles are taken, no altimetry. Satellite sea surface temperature comes in another way: it corrects the heat fluxes. No measured current appears among the assimilated data: moorings and HF radars are kept apart, for validation.

Above the sea, ECMWF's fields at 1/10° drive the system through bulk formulae, every hour for the first 3 days of forecast, every 3 hours for the next 3, every 6 hours for the last 4 and for the analysis. Past the third day, then, an hourly current is not driven by an hourly wind; the wave model takes the same fields. River water comes from 39 rivers: daily flows from EFAS v5.0 for 38 of them, a monthly climatology for the Nile. On the Atlantic side the boundaries take the global system's daily fields, with the tide added.

The product holds hourly means, which the manual says are centred every half-hour, quarter-hourly snapshots at the surface up to 4 days ahead, daily means with the tide removed, and monthly means, with a rolling archive of two years. All fields are delivered at the centre of the grid cell after interpolation. The rectangle has 489,060 grid points: as Saily WR counted in November 2026, about 225,300 of them are water, and a third of those, 75,653, lie west of 6°W in the Atlantic.

At sea

MedFS covers the whole Mediterranean, from the Alboran Sea to the Levant, with the Adriatic and the Aegean, and in the Atlantic the coasts of Portugal and Galicia, the Cantabrian coast up to Arcachon, and Madeira. The Canaries and Brittany are outside it, and it stops at the Dardanelles. Its grid draws more of these waters than a global one: in a square of ±0.08°, as Saily WR measured in November 2026, MedFS has 14 water points at Gibraltar against SMOC's 4, 12 off Cape Corse against 2, none at the Dardanelles against 4. From 6 November 12:00 UTC to 8 November 12:00 UTC, the strongest current in such a square was 4.50 knots at Gibraltar against 3.07 for SMOC's total current, 1.00 off Cape Corse against 0.34, but 0.63 in the Sicily Channel against 1.01. These figures show a gap between two models, not which is right.

The producers' validation, over 2022, is firmest on the tide as a height of water. At 98 tide gauges the root mean square misfit is 1.03 cm for the main constituent, M2, 0.54 cm for S2, 0.41 cm for K1 and 0.29 cm for O1; at the 35 gauges used in earlier studies the mean distance for M2 is 1.21 cm, against 1.60 and 1.53 cm for two earlier models. Modelled M2 amplitudes run low where M2 is strongest (slope 0.86), and in the Gulf of Gabes they differ from the global TPXO9 tide by more than 4 cm where M2 exceeds 25 cm. These figures tell of the rise and fall of the sea, not of its stream.

The stream itself rests on far less. Near the surface, daily means differ by a root mean square of 0.13 m/s (0.25 knots) from 12 coastal moored buoys and 0.15 m/s (0.29 knots) from 4 HF radars, for speeds mostly under 0.3 m/s, with correlations of 0.46 and 0.27; the model falls short of the stronger observed speeds (slopes 0.69 and 0.48). The producers call the statistical robustness of this analysis poor. A daily mean smooths out the tide; on their maps no instrument stands at Gibraltar, Messina or Bonifacio, nor in the Atlantic box, which has no row in the accuracy tables either; and no skill is given by forecast range. At Gibraltar the net transport over 2019–2022 agrees with measurements from 2004 to 2009 (0.042 Sv against 0.038), but each way the model carries over a third more: 1.10 Sv against 0.81 eastward, 1.06 against 0.78 westward.

A grid of about 4 km cannot draw the narrowest straits. Messina narrows to about 3 km over a sill near 100 m deep; on a mesh going down to 50 m, Cucco and colleagues computed up to 2.4 m/s at spring tides there, and describe up to 2.5 m/s, about 4.9 knots, with eddies behind the capes. Imray gives up to 5 knots, with eddies and whirlpools. MedFS's strongest hourly mean within ±0.08° of Messina was 2.82 knots over the two days above, a window that says nothing either way about spring tides. In the producers' plots of tidal phase, the gauges of the Messina area are among the farthest from the line. In a strait or close to land, tidal atlases and pilot books remain the reference.

MedFS in Saily WR

Saily WR takes MedFS's hourly surface current from Copernicus Marine's public store, on its own grid, over 240 hours from 00:00 UTC on the day of each run. A server downloads each new run once it is published, and the Data page shows live which MedFS run the app is on, its horizon and the rectangle it covers. The store labels each hourly mean with the hour it begins, so each hour written is the half-sum of the two hourly means around it, and a peak is a little smoothed. Against the quarter-hourly snapshots, over 24 hours in November 2026, this cut the root mean square gap at Gibraltar from 0.22 to 0.09 m/s and at Messina from 0.10 to 0.02 m/s. As the Copernicus Marine licence requires, the data are credited wherever they are shown: "E.U. Copernicus Marine Service Information (CMEMS)".

There is nothing to choose: inside its waters a route sails in MedFS by itself, over SMOC. Past its 240 hours, or where none of the four MedFS points around the route has water, the route takes SMOC. A grid cell touching land keeps MedFS, drawn from its points with water alone: the route switches to SMOC, on a coarser grid, only in a harbour or a pass MedFS does not see. Copernicus' manual does not mention wave drift in this current, where SMOC adds it explicitly.

At the domain's edge, notably at 45.15°N in the Atlantic and at 17.29°W, a route passes from one current to the other with no blending, so the current it meets can jump there. Between 45.15°N and AMM15's 46°N it sails in SMOC. The route's summary names each current it crossed, in order, for instance "MedFS, then SMOC"; on a route sailed in several, the route log, on screen and printed, names the source of each leg, and the CSV has a "Current source" column. On the routing map, the Current layer shows MedFS inside its waters and SMOC around it.

If Copernicus stops publishing, the last run stays in use until its last step, even days old. On a long trip, whose time step lasts hours, a leg keeps one value of the current for its whole length although a tidal stream can turn within it, so the time gained in a strait can be overstated. In very light wind against the stream, a route can also stop short of its destination. And Saily WR has not compared MedFS's current with observations. A router exploits any more detailed field, right or wrong: nothing here says that a route sailed in MedFS is closer to the truth than one sailed in SMOC, and the Atlantic box is the model's joining zone, not an area where its current has been validated.

Sources

  1. The Mediterranean ocean forecasting system: first phase of implementation (1998–2001), Pinardi et al., Annales Geophysicae (2003)
  2. Operational oceanography in the Mediterranean Sea: the second stage of development, Pinardi and Coppini, Ocean Science (2010)
  3. A high-resolution free-surface model of the Mediterranean Sea, Tonani et al., Ocean Science (2008)
  4. A nested Atlantic-Mediterranean Sea general circulation model for operational forecasting, Oddo et al., Ocean Science (2009)
  5. A 1/24 degree resolution Mediterranean analysis and forecasting physical system for the Copernicus Marine Service: description and skill assessment, Clementi et al., EGU General Assembly abstract (2018)
  6. The Mediterranean Forecasting System – Part 1: Evolution and performance, Coppini et al., Ocean Science (2023)
  7. The characteristics of tides and their effects on the general circulation of the Mediterranean Sea, McDonagh et al., Ocean Science (2024)
  8. Hydrodynamic modelling of coastal seas: the role of tidal dynamics in the Messina Strait, Western Mediterranean Sea, Cucco et al., Natural Hazards and Earth System Sciences (2016)
  9. Product User Manual for MEDSEA_ANALYSISFORECAST_PHY_006_013, issue 2.5, Copernicus Marine Service (2025)
  10. Quality Information Document for MEDSEA_ANALYSISFORECAST_PHY_006_013, issue 2.6, Copernicus Marine Service (2025)
  11. Mediterranean Sea Physics Analysis and Forecast, Copernicus Marine Service (2026)
  12. Currents, 2D, hourly: the dataset's catalogue entry, Copernicus Marine Service (2026)
  13. Mediterranean – Monitoring Forecasting Centre (MED MFC), Copernicus Marine Service (2026)
  14. Product User Manual for GLOBAL_ANALYSISFORECAST_PHY_001_024, issue 2.4, Copernicus Marine Service (2026)
  15. Service commitments and licence, Copernicus Marine Service (2026)
  16. Mastering the Strait of Messina: Navigating the Mediterranean's most famous currents, Imray

Plan my route

All the data routes are computed on