CMCC Black Sea analysis and Forecasting System BSFS

SHORT DESCRIPTION

CMEMS BS-MFC logo

The Black Sea analysis and Forecasting System (BSFS, EAS8 version) is the operational system that provides regular and systematic information on the physical state of the Black Sea region. It is developed and maintained by CMCC, which is responsible for the Black Sea Physics Production Unit as part of the Black Sea Monitoring and Forecasting Centre (BLK-MFC) in the framework of the Copernicus Marine Service (CMEMS, https://marine.copernicus.eu/). The core model is based on NEMO v4.2 online coupled with OceanVar, a 3D variational scheme for assimilation of in-situ and satellite data. The BSFS catalogue offers near-real-time products – analysis and forecast – for 3D temperature, salinity, currents and 2D sea surface height, mixed layer depth and bottom temperature. From 2025 with the inclusion of the Azov Sea there are two additional variables: sea-ice concentration and sea-ice thickness.

More information on the Black Sea Physics analysis and forecast product is available here: https://resources.marine.copernicus.eu/product-detail/BLKSEA_ANALYSISFORECAST_PHY_007_001/INFORMATION

SYSTEM

The BSFS Hydrodynamical Core

The BSFS core model is based on NEMO ocean model, version 4.2 (Madec et al., 2023), online coupled with OceanVar2 (Oddo et al. 2026).

The model covers the whole basin including the Azov Sea and a portion of the Marmara Sea to provide an optimal interface with the Mediterranean Sea through the Bosphorus Strait. The primitive equations are discretized over a horizontal grid with 1/40º x 1/40º resolution and over 121 z* levels. The bathymetric source is provided by the GEBCO_2019 30-arc-second gridded bathymetric dataset (https://www.gebco.net) combined with a very high-resolution dataset for the Bosphorus Strait and the Marmara Sea (Gürses, 2016). An optimal barycentric interpolation method has been used to interpolate the high-resolution scattered dataset onto a regular spatial grid. The coastline has been revised using the NOAA shoreline dataset (https://www.ngs.noaa.gov/CUSP/) to account for the coastal peculiarities and structures in the Black Sea.

figure-desc
Figure 1: The model bathymetry until 2200 m depth with location of main regions such as the Marmara Sea, Bosphorus and Kerch Straits, Batumi, Sinop and Sevastopol regions. On the top and right, detail of the Azov Sea showing 20 m maximum depth. Major rivers are marked in dark blue and minor ones are in cyan.

The model is forced by momentum, water and heat fluxes interactively computed by bulk formulation as used in the MedFS system (Pettenuzzo et al., 2010; Clementi et al., 2021) and applied in BSFS system, using ECMWF IFS analysis and forecast atmospheric forcing (including precipitation) at the highest resolution at today available – 0.01° in horizontal and 1–3–6 hours frequency in time. Atmospheric fields used by BSFS are zonal and meridional components of 10 m wind (ms-1), total cloud cover (%), 2 m air temperature (K), 2 m dew point temperature (K) and mean sea level pressure (Pa) and precipitation (kg m-2s-1).

Land forcing is accounted for by a total of 90 rivers, with daily discharge sourced from the EFAS v5 model. The Danube River has deserved a more dedicated study in order to improve its representation: it uses a distributed freshwater source to properly represent the main branches – the Chilia, the Sulina and the St. George arms – accounting for river daily discharges as provided by the National Institute of Hydrology and Water Management (NIHWM, partner of the BLK-MFC consortium). For the major rivers, a monthly climatological salinity is imposed as provided by SeaDataNet for the Black Sea at the closest river mouth location to represent the riverine salt contribution to the ocean. The BSFS implements lateral open boundary conditions at the Marmara Sea box: 3D temperature, salinity, currents and 2D sea surface height are provided by the new Unstructured Turkish Straits System (U-TSS), a Shyfem-based model implemented for the Dardanelles-Marmara Sea-Bosporus (Ilicak et al. 2021) whose spatial domain is represented in Figure 2.

figure-desc
Figure 2: The U-TSS spatial grid (Ilicak et al. 2021)

The BSFS data assimilation system

BSFS data assimilation scheme is based on OceanVar2 (Oddo et al., 2026), a 3D variational scheme initially developed by Dobricic and Pinardi (2008) and further improved by Storto et al. (2011). The background covariance matrix is modelled using a set of empirical orthogonal functions (EOF) that provides a variable transformation to pre-condition the cost function minimization. The system uses a spatially varying set of 45 EOF to describe the covariance of sea surface height and temperature and salinity in the water column. The EOFs are derived from a 11-year integration of the hydrodynamical core without DA separately for each month to account for seasonal variability. Horizontal correlations are modelled through a third-order recursive filter (Farina et al., 2015), specified as a function of the distance from coast, ranging approximately from 9 to 27 km.

The in-situ observational errors vary spatially in depth and change monthly. They are iteratively estimated using the method of Desroziers et al., (2015). BSFS assimilates observations operationally provided by Copernicus Marine Service INS, SST and SLA TACs. The observations assimilated in the BSFS include:

  1. i) in-situ temperature and salinity profiles from NRT products;
  2. ii) along-track L3 sea level anomalies, from AltiKa, Cryosat-2, Jason-2/3, Sentinel-3A, Sentinel-3B, HY2B and Sentinel-6A satellites;
  3. iii) sea surface temperature L3 satellite observations.
The assimilation of SLA employs the steady-state solution of a barotropic model to balance the density increments in the water column and sea level increments over the domain (Dobricic and Pinardi, 2008). The DA system runs once daily, utilizing a 24-hour assimilation time window.

The BSFS system evaluation

The quality assessment of the system is monitored weekly by the calculation of the root mean square statistics of difference between observations and model background fields (so-called misfits): https://bsfs-evaluation.cmcc.it/

The BSFS performs data quality control and rejection during the assimilation phase through an online procedure. In-situ data – temperature and salinity profiles – and satellite data – sea level anomaly and sea surface temperature – are checked with respect to

  1. i) time and position quality check;
  2. ii) valid range and quality flags.

The Processing chain

The BSFS processing system consists of two different cycles (Figure 3). One cycle is daily, during which the system produces 3-day analysis, 1-day hindcast and 10-day forecast every day. Once a week, the BSFS performs a 14-days analysis, 1-day hindcast and 10-day forecast to incorporate a large number of in-situ and satellite observations into the data assimilation. The system produces 3D temperature, salinity, currents and 2D sea surface height, mixed layer depth and bottom temperature fields, as daily and hourly means. The available time series starts in Jan 2017 and is daily updated.

figure-desc
Figure 3: The BSFS processing system

The BSFS catalogue DOI is here available: https://doi.org/10.25423/cmcc/blksea_analysisforecast_phy_007_001_eas4

Essential references