Mediterranean Forecasting System MFS

CMEMS MED-MFC and the downscaling MONGOOS effort

Giovanni Coppini1, M. Drudi2, G. Korres3, C. Fratianni2, S. Salon4, G. Cossarini4, Emanuela Clementi2, A. Zacharioudaki3, A. Grandi2, D. Delrosso2, J. Pistoia2, P. Di Pietro2, C. Solidoro4, N. Pinardi2,3,5, R. Lecci1, P. Agostini1, S. Cretì1, G. Turrisi1, F. Palermo1, A. Konstantinidou3, A. Storto1, S. Simoncelli2, S. Masina1, S.A. Ciliberti1, M. Ravdas3, M. Mancini1, G. Aloisio1, S. Fiore1, G.Bolzon4, M. Buonocore1 and M. Ravdas3

And MONGOOS modelling working group: Georg Umgiesser, Pierre Garreau, Agustín S. Arcilla, Emanuela Clementi, Stefano Salon, Michalis Ravdas, Ivan Federico, George Zodiatis, Christian Ferrarin, Giorgia Verri, Gianpiero Cossarini, Marcos Garcia Sotillo, Andrea Cucco, Roberto Sorgente, Baptiste Mourre, Ivica Vilibić, Simone Sammartino, Giovanni Coppini, Enrique Alvarez Fanjul CMEMS - MFCs

The CMEMS is Arctic centrally Baltic coordinated by North West Mercator Ocean. It is Shelves build on a network of integrated sub- components that are either thematic (TACs) or Global geographically Ocean defined (MFCs), all contributing to the Black Sea monitoring and forecasting of the Mediterranean Essential Ocean Variables accessible Eastern North for the users by a Atlantic - IBI Web Portal: marine.copernicus.eu/ CMEMS Med-MFC organization & Service timing

The CMEMS Med-MFC is a consortium of 4 research institutes: CMCC (Leader of the consortium) INGV (Responsible for the Physical product) OGS (Responsible for the Biogeochemical product) HCMR (Responsible for the Wave product)

The CMEMS Med-MFC operational service started in May 2015 and is continuously improved following users requests and suggestions (user forum): a new upgraded release is delivered every year.

The V1 system in operation since the beginning of the service has just been replaced, the 13th April 2016, by an improved new version, namely V2.

May 2015 April 2016 April 2017 April 2018 Med-Currents products in CMEMS Catalogue

Hourly mean – Daily mean Analyses and Forecast (1/24° 141 Vertical levels) • 2D Sea Surface Height • 3D Salinity • 3D Potential Temperature • 3D Zonal/Meridional currents • 2D MLD • 2D Bottom Temperature • 2D Stokes Drift & Wavenumber Daily mean – Monthly mean MEDITERRANEAN SEA PHYSICS REANALYSIS (1987- 2014) PHY Reanalysis (1987-2015) (1/16° 72 Vertical levels) • 2D Sea Surface Height • 3D Salinity • 3D Potential Temperature • 3D Zonal/Meridional currents Monthly mean

MEDITERRANEAN SEA PHYSICS REANALYSIS (1955- 2014) PHY Reanalysis (1955-2015) (1/16° 72 Vertical levels) • 2D Sea Surface Height • 3D Salinity • 3D Potential Temperature • 3D Zonal/Meridional currents Med-Currents: analysis and forecast numerical model Atmo/Land forcings/OBC

ECMWF 1/8o atmospheric fields: - mean sea level pressure (MSLP) Temporal resolution: - cloud cover For forecasts: 3hr time resolution for the first 3 - 2m relative humidity days and 6 hours time resolution fields for the - 2m air temperature remaining 7 days - 10m zonal and meridional winds For analysis: 6 hours time resolution components - Precipitations

The land river runoff: vertical boundary condition for major rivers: green dots represent the 7 river sources included in both V2 and V3.2 systems, red dots represent the 32 additional rivers implemented in V3.2 (climatological monthly mean seasonal cycle values) The Dardanelles inflow is also parameterized also as a river

Boundary conditions in the Atlantic daily real time analyses and forecasts from Global Ocean Forecasting System (GLO-MFC)` @ 1/12° horizontal resolution, 50 vertical levels Impact of the improved river input

Surface Salinity measured at ODAS coastal mooring (green line) in the Ligurian Sea and compared to V3.2 (blue line) and V2 (red line) systems during year 2015

Time series of daily volume averaged salinity evaluated for systems V2 (red line), V3.2 (blue line) and compared to monthly climatological values from WOA2013 V2 datasets (black line). Med-Currents Analysis and Forecast system: In-situ observations used in the Data Assimilation scheme ATMOSPHERIC FORCINGS ECMWF 1/8o forecast/analysis fields OBSERVATIONS time resolution: 3/6 hrs: OGCM model Along track satellite SLA ➢mean sea level pressure (MSLP) NEMO v3.6 ➢cloud cover from CMEMS SL-TAC, for ➢2m relative humidity all available satellites ➢2m air temperature RESOLUTION: ➢10m wind velocity o ➢Precipitation 1/24 hor. res 141 vert. levels Vertical profiles of T and S LAND RIVER RUNOFF: 39 rivers from CMEMS InSitu TAC: sources aligned with Med-BIO Argo XBT Gliders

BOUNDARY CONDITIONS in the Atlantic: daily analyses/forecasts Data Assimilation: from Global Ocean MFC @ 1/12° hor. 3D variational scheme res., 50 vert. lev. implemented in the Med-Sea following Storto et al. (2015) Satellite SST (L4) from CMEMS OSI TAC Data assimilated: SLA satellite data from SL-TAC + PRODUCTS In-situ observations from InSitu-TAC 3D daily/hourly fields of: T, S, UV; INITIAL CONDITIONS from WOA V2 climatologies 2D daily/hourly fields of: SSH, MLD, Tb 7 Med-Currents Analysis and Forecast system: In-situ observations used in the Validation assessment

Temperature (degC) In-situ moorings used for CLASS2 independent validation: for T, S, SL, UV

Validation with Gliders in the Balearic Sea Med-Currents: analysis and forecast Product Quality Annual (2016) mean EAN (BIAS and RMSD) for V3.2 T, S, SST, SSH, U, V (indep. data system for temperature (left) and salinity (right) from moorings: calval.bo.ingv.it/)

T, S, SLA (misfits of quasi- indep. data) medforecast.bo.ing v.it/mfs-copernicus-evaluation/

Every Every Every Every year, Day Week 3 months New version Opening of Messina Strait

Messina Strait transport and currents. Left: Net transport during year 2016 [Sv]; Center: Surface currents in Med-Currents V2 system (the strait is closed); Right: Surface currents in Med-Currents V3.2 system (the strait is open) Med-Currents products in CMEMS Catalogue

Hourly mean – Daily mean Analyses and Forecast • 2D Sea Surface Height • 3D Salinity • 3D Potential Temperature • 3D Zonal/Meridional currents • 2D MLD • 2D Bottom Temperature • 2D Stokes Drift & Wavenumber Daily mean – Monthly mean MEDITERRANEAN SEA PHYSICS REANALYSIS (1987- 2014) PHY Reanalysis (1987-2015) • 2D Sea Surface Height • 3D Salinity • 3D Potential Temperature • 3D Zonal/Meridional currents

Monthly mean

MEDITERRANEAN SEA PHYSICS REANALYSIS (1955- 2014) PHY Reanalysis (1955-2015) • 2D Sea Surface Height • 3D Salinity • 3D Potential Temperature • 3D Zonal/Meridional currents Med-Currents: Reanalysis Product 006_009 (1955-2015)

Parameter BIAS RMSE SST 0.2±0.3 0.59±0.16

SST RMSE and BIAS computed from Med- MFC REA and satellite reference SST on a monthly basis

The largest RMSE are located along: Northern Adriatic coast Negative BIAS and large positive RMS values appear in the upwelling areas of the Mediterranean Sea, such as Eastern Adriatic Sea and Southern Sicily, and in the Northern Adriatic Sea and Aegean Sea Med-MFC Physical-Biogeochemical components

1/16o (~6km) hor. res 72 vert. levels

Med-Currents NEMO-WW3 2-way Circulation-Wave model Data Assimilation: OceanVar

Med-Biogeochemistry OGSTM-BFM Transport-biogeochem model Data Assimilation: 3DVAR Med-Biogeochemistry: Analysis and Forecast system: In-situ observations used in the validation of products BIOGEOCHEMICAL Observations: Physical forcing MODEL (resolution: 1/24°, Multisensor and ESA-CCI U, V, T, S, SSH from Chlorophyll Satellite from Med-MFC Physical 124z levels) CMEMS OC TAC component OGSTM - transport model Daily 2D fields at 4km res. vvl formulation (non-linear free surface) NEMO 3.6 daily 3D fields at 1/24° and 130z levels Assimilation Biogeochemical Flux Model – 3DVAR-BIO 51 variables; cycle ofBFM C, N, P, Si, O; carbonate variational scheme; weekly Land & Atm. system; Plankton Functional Types formulation assimilation cycle Forcings yearly and monthly climatological data PRODUCTS for rivers; seasonal estimates for atm. 8 variables: chlorophyll, nitrate, phosphate, primary production, Boundary phytopl. biomass, oxygen, pCO2, Conditions Seasonal pH profiles in the Atlantic buffer VALIDATION with climatology zone from MEDAR/MEDATLAS and (WOA2013) and in situ data CarbSys climatologies from NODC-OGS dataset, CARB-SYS-OGS dataset and Initial Conditions BGC-Argo floats MEDAR/MEDATLAS and 5- y hindcast spin-up + 17-y reanalysis Med-biogeochemistry Analysis and Forecast system: In-situ observations used in the Validation

NODC-OGS dataset for nitrate, phosphate, oxygen (~6000 data, 2000-2011)

CARB SYS OGS dataset for pH, DIC, alkalintiy (~4000 data, 2001-2013: )

Validation with in-situ data provides an estimate of the model accuracy in reproducing basin-wide gradients and vertical profiles, and a reference uncertainty number for each sub-

CMEMS Marine Week: 25- basin and layer.15 29/09/2017 V3 : to be released V4 : to be released April 2017 March 2018 Phase I Phase I ✓The Med-waves forecasting system has been implemented in a nested configuration : Atlantic V3 April 2017 V4 March 2018 V3.1: Mediterranean Sea wave V4.1: Mediterranean Sea wave (coarse grid model) -> Mediterranean Sea (fine grid model (WAM Cycle 4.5.4) 1/24 deg, model (WAM Cycle 4.5.4) 1/24 32 frequency and 24 directional bins deg, 32 frequency and 24 model). directional bins V3.2: Atlantic ocean wave model V4.2: Atlantic ocean wave model (WAM Cycle 4.5.4) 1/6 deg, 32 (WAM Cycle 4.5.4) 1/6 deg, 32 frequency and 24 directional bins, frequency and 24 directional nested with the Mediterranean bins)nested with the Cliquez et modifiez wave model V3.1 Mediterranean wave model V4.1 le titre

V3.3 ECMWF analyses aligned to all V4.3 ECMWF analyses aligned to the latest ECMWF model outputs all the latest ECMWF model outputs V3.9 ECMWF analysis fields (Feb 2006 -> Dec 2016)

V4.5 Data Assimilation of satellite observations using the standard DA Optimal Interpolation scheme embedded into WAM Cycle 4.5.4 Atlantic (Coarse) wave Model characteristics WAM Cycle 4.5.4 Model domain -75E - 10E , 10N - 70N V3.5: release of the 5 day forecast V4.6 release of analysis and the 10 Horizontal Resolution 1/6o x 1/6o (18.5km approximately) occurs 15:00-16:00 hours after the day forecast should occur within 9 nominal start of the forecast hours from the forecast nominal 32 (logarithmically spaced) 0.04177 - start Frequency Bins 0.8018 Hz V3.6: Cal/Val with satellite and In V4.7 Cal/Val with satellite and In Directional Bins 24 (equally spaced) Situ (fixed buoys)wave observations Situ (fixed buoys)wave Propagation Time-step 300 sec observations V3.7 Cal/Val of ECMWF atmospheric V4.8 Cal/Val of ECMWF Type of run DEEP forcing (10m winds) with in situ Med atmospheric forcing (10m winds) ECMWF operational analysis and 5-day stations with in situ Med stations Forcing (10m winds) forecast (1/8 x 1/8) ECMWF operational analysis & 5 day Sea Ice forecasts (1/8 x 1/8) - Daily Surface currents CMEM Global MFC (1/12 x 1/12) Validation of the wave system

QQ-Scatter plots of: (left) ECMWF forcing QQ-Scatter plots of Med-waves wind speed U10 versus satellite U10 (JASON- output SWH (Hs) versus wave 2); (right) Med-waves SWH (Hs) versus buoys' observations, for the full satellite SWH (JASON-2 and SARAL), for the Mediterranean Sea, for 1 year full Mediterranean Sea, for 1 year (2014) (2014) period: QQ-plot (black period. crosses), 45° reference line (dashed red line), least-squares best fit line (red line). Downscaling of CMEMS products

• TSCRM (Sorgente et al., 2016) and WMED (Olita et al., 2013) in the Western Mediterranean sub-basins (POM). • SAMOA project of Puertos del Estado for assessing parameters (waves, currents, sea level) in the Spanish harbors (ROMS). • SANIFS (Federico et al., 2017) in the coastal waters of Southern- eastern of Italy, with a focus on the Apulia region (SHYFEM). • CYCOFOS forecasting system in Cyprus (WRF, SKIRON) provides circulation and wave forecasts (Zodiatis et al., 2015; 2016). • SELIPS nested into ALERMO in front of the Israel coast (POM). • The Balearic Rissaga Forecasting System is a system that tries to predict resonant meteo-tsunami in the Ciutadella harbor. Coastal models circulation

• The SAMPA model is using Lagrangian techniques for describing the Atlantic water intrusion in the Mediterranean Sea and in the Algeciras Bay (Strait of Gibraltar) (Sotillo et al., 2016). • In the North-Western Mediterranean Sea, along the French coast, a two-way zoom model has been applied, resolving the regional circulation close to the coast with a resolution from kilometer until 100 meters (Garnier et al., 2014; Herbert et al., 2014). • SANIFS V2 in the coastal waters and horbours of Southern- eastern of Italy. IBI & SAMPA Validation in ALBO Region

IBI SAMPA Medium term future activities

• the change of the numerical ocean and biogeochemical models for re-analyses production to 1/24 degree resolution and 141 vertical levels; • the inclusion of tides in the Physics model; • the inclusion of an Estuarine Box Model at the river mouths; • the usage of new Sentinel 3 Copernicus satellite data in the assimilation chain as soon as they will become available from the CMEMS TAC; • the inclusion of a data assimilation system to the Med-Waves NRT and RAN production systems • the nesting of the Mediterranean wave model to the CMEMS Global wave forecasting system Medium term future activities

• the production of a new WAVE, BIO and PHYS reanalysis using ERA5 atmospheric reanalysis • the assimilation of SST in the physical model; • high-frequency, interannual river runoffs and the implementation of an extuary boxmodel to evaluate the river discharge and salinity • the Dardanelles Strait implemented as an open boundary condition (instead of a river-like parameterization) • the inclusion of BGC-Argo float data in the biogeochemical assimilation chain; • the daily update of the biogeochemical forecast products; Long term future activities

• Improve of DA in the physics with the inclusion of daily cycle of assimilation in order to ingest all available timely observations. • Assimilation of data from high-resolution altimetry missions like SWOT and HF radars; • Development of operational high-frequency 3D interfaces between coastal models & regional MFC will be designed • better representation of the coastal/shelf dynamics improving the mesoscale and submesoscale circulation and the vertical circulation for the physical system; • increase of the model resolution in order to reach a kilometric scale; • improvement of methodologies for ensemble ocean forecasting for physics and waves; • development of new biogeochemical products such as phytoplankton functional types • development of a wave forecasting system forced with high resolution winds; MONGOOS downscaling conclusions

• The activities of the MonGOOS community dealing with operational modelling and process based studies have been presented. • These activities are on the core of the MonGOOS group and the article show the extreme bandwidth of applications that are being developed and maintained. • In the future, it would be important to attract even more contributions from the Mediterranean modelling community in order to establish a firm working group that could then take the lead in model development and project participations.