A marine physical biogeochemical model simulation was performed with the model MOM-ERGOM for the years 2003 to 2014 covering the Baltic Sea. Previously, MOM-ERGOM had been initialized for several decades without tagging until 1999 and, then, from 2000 to 2002 with tagging (see below). The model output has been validated with measurement data of the "IOW Baltic Monitoring and long-term data program" (https://www.io-warnemuende.de/iowdb.html IOW: Leibniz Institute for Baltic Sea Research Warnemünde) and from the HELCOM database (http://ocean.ices.dk/helcom/Helcom.aspx HELCOM: Helsinki Commission). A publication is in preparation. The model simulation was forced by coastDat2 COSMO-CLM data (doi:10.1594/WDCC/coastDat-2_COSMO-CLM). Atmospheric nitrogen deposition data of 0.1° x 0.1° spatial resolution were taken from the 2018 reporting of the European Measurement and Evaluation Programme (EMEP) as presented in EMEP (2018, url: http://emep.int/publ/reports/2018/EMEP_Status_Report_1_2018.pdf) and available from the Norwegian Meteorological Institute (2018, url: http://thredds.met.no/thredds/catalog/data/EMEP/2018_Reporting/catalog.html).

                Nitrogen from atmospheric deposition of nitrogen from livestock/agricultural emissions (estimated, see documentation) and from all emission sectors has been tagged in the model simulation according to a method by Menésguen et al. (2006, doi:10.4319/lo.2006.51.1_part_2.0591). Therefore, all nitrogen-containing model variables exist three times in the output: once as regular variables and once per tagged nitrogen source (total atmospheric and agriculturally-related).

                The simulation was performed at the North-German Supercomputing Alliance (HLRN, project id: mvk00054, zulassung.hlrn.de/kurzbeschreibungen/mvk00054.pdf). The model output data were processed and evaluated on servers provided by the project 'PROSO - Prozesse von Spurenstoffen in der Ostsee' (FKZ 03F0779A).

                Technical details:
                model run on MPP1 Cluster of the HLRN-III Konrad Comlpex
                each simulation performed on 527 cores distributed on 22 nodes (24 cores per node; one core on one node not used) configuration of one node:
                Processor: 2x Intel Xeon E5-2695v2 CPUs (12 cores per CPU), R_peak: 230 GFlop/s
                RAM: 64 GiB DDR3-1866
                OS: SuSE Linux Enterprise Server (SLES) version 11
                Interconnect: Cray Aries