Gullmarfjord Biogeochemical Model



As part of its contribution to the EU's OAERRE project, SMHI has applied the PROBE & SCOBI models to the Gullmarfjord system. PROBE is a one-dimensional numerical model to describe the vertical mixing in lakes or ocean basins, and gives the temperature and salinity fields, after calculating the effects of atmospheric heating and cooling, fresh water input, sea levels and vertical mixing. This model is combined with the SCOBI (Swedish Coastal and Ocean BIogeo- chemical) model, which describes the nitrogen, phosphorus and oxygen dynamics. SCOBI calculates the nutrient fields based on inputs from the land (due to river run-off), the sea, and also from the atmosphere (based on deposition calculations from SMHI's MATCH model).

Nutrient consumption and recycling within the phyto- and zoo- plankton communities, and the production of detritus are also included. References describing the PROBE and SCOBI models can be found at the bottom of this page. The particular version of the PROBE/SCOBI model applied to the four basins of the Gullmarfjord system was developed by Jonny Svensson of Thalassos Computations.

The Gullmarfjord system was modelled as four connected basins. These basins were the coastal ocean, central Gullmar fjord, Färlevfjord and Saltkällefjord. A map of the actual fjord system, showing the main basins and rivers, is shown below:

The Gullmarfjord system The Gullmarfjord system

Input Data

For the OAERRE project, the model was run to simulate the period between 01 Jan 2000 and 01 May 2003. The dates were limited to the OAERRE project period, and also by the availability of input data. Input data were as follows:

  • Weather data:
  • Atmospheric forcing data were supplied from the SMHI database, for the point 58°30'N, 11°30'E. Data consisted of 3-hourly mean values of

    • air temperature (in °C)
    • wind velocity (eastwards at 10 m, in m/s)
    • wind velocity (northwards at 10 m, in m/s)
    • cloudiness, between 1 and 0
    • Humidity between 1 and 0
    These data were available up until the end of 2002. For the first five months of 2003, the January - April data from 2002 were used again. The 2003 data are only available after the close of 2003.

  • Atmospheric deposition
  • These data came in the form of monthly mean values of reduced and oxidised nitrogen (NHx and NOx) deposited over each basin. Values were calculated by the SMHI model MATCH, and the values calculated for 1997 were used for each year in the simulation. Other years will become available.

  • Sea level
  • The coastal basin of the model uses daily sea level values to control whether on not to push saline water into the central Gullmaren basin. Hourly sea level data were taken from the SMHI tide gauge at Kungsvik (59°00'N, 11°08'E). These values were filtered using a Doodson X0 filter, to give a good estimate of the daily mean sea level. These data were then de-meaned, as the model requires only the daily sea level fluctuations. These data were available for the whole of the modelling period. An improvement would be to use the data from the Smögen tide gauge, which is closer to Gullmaren, although it is expected that any differences would be small.

  • Coastal nutrients
  • These data were available from measurements made within the SMHI marine monitoring programme. Measurements came from the station P2 (57°52'N, 11°18'E) which lies close to the mouth of Brofjorden. Data from other stations further out into the Skagerrak had been tried, but they did not give such a good representation of the high nutrient values that are observed in the surface waters of the Baltic current. From the observed data, the following variables were available:

    • temperature (°C)
    • salinity (psu)
    • Nitrate (µmol/l)
    • Ammonium (µmol/l)
    • Total available nitrogen (set to 10 µmol/l)
    • Phosphate (µmol/l)
    • Total available phosphorus (set to 9 µmol/l)
    • Oxygen (ml/l)
    • Chlorophyll a (µg/l)
    Data were extrapolated from the measurement depths to the model grid, using a simple nearest-neighbour type scheme. Data were not interpolated in time. The model uses the current nutrient values as the offshore condition until the next set of data become available.

  • Freshwater inflow and nutrients from land
  • The model uses daily mean values of the inflow into each basin from the following six rivers:

    • Örekilsälven
    • Taskeån
    • Färlevälven
    • Ammerödsån
    • Skredsviksån
    • Vikenbäcken
    The fresh water, and nutrients are added to each basin in the uppermost (surface) layer. Data were supplied by SMHI and also by the Västra Götaland's Länstyrelsen (County Administration). The most recent data were unavailable during the modelling, so the data from 1998 - 2000 were re-used to cover the period 2001-2003. This meant that the nutrient inputs had a reasonable seasonal cycle.

Model output

The model produces two output files for each basin. The first is an unformatted binary file, which contains all the output for each of the main variables at each timestep. The other file is ASCII, and is a subset of the other file. The following table allows you to browse the graphs of the output data, and also to download the result files.


Basin 1: Coastal sea

Basin 2: Central Gullmaren

Basin 3: Färlevfjord

Basin 4: Saltkällefjord



Benthic Nitrogen
Total Nitrogen

Benthic Phosphorus
Total Phosphorus

All data!

Comparison with observations

In his report, Svensson (in prep) simulated the Gullmaren system for the period 1993 - 2000, and compared results with field observations from Tröskeln outside of the fjord, and from Björkholmen and Alsbäck within. The model mimicked the observed temperatures well - suggesting that heat exchange across the surface, and down through the watercolumn was modelled correctly. Salinity values in these tests over estimated surface salinity, which was attributed to having used data from measurement stations in the offshore Skagerrak to drive the coastal part of the model. Oxygen levels in the deepwater were also over estimated. Lack of oxygen consumption in deep water, or to frequent renewal of the deep water were suggested as possible reasons for this.

These model runs were compared with data from the monitoring stations at Släggö, at the mouth of the fjord, Alsbäck in Central Gullmaren, and at Björkholmen in the inner fjord. Surface temperature at Släggö and Björkholmen was well modelled, although the extremes at Släggö, both winter and summer, were underestimated by the model. At the inner fjord, surface salinities are slightly higher than observed, but the short term fluctuations do not agree - probably because of the river input used to drive the model. At Släggö, the range of surface salinities also agree between the model and observations. Surface oxygen concentrations were in perfect agreement.

Modelled and measured surface temperature and salinity

Modelled (lines) and measured (points) surface temperature (red) and salinity (blue), for Släggö (fjord mouth) and Björkholmen (inner fjord)

Despite the use of measured river inflow and nutrient supply from land, modelled surface nitrate levels differed strongly from observations. The modelled seasonal nitrate cycle was much stronger than observed. Comparison between observed and modelled ammonium values was better. Total nitrogen and phosphorus values where fairly constant throughout the model domain, as these were fixed at the coastal boundary.

Chlorophyll a values - a measure of phytoplankton productivity - show good agreement between the model and observations. At Släggö, the model tended to miss the size of the spring bloom, but represented the variablility well. Within Gullmaren, the stronger blooms were well described by the model. The model did tend to overpredict chlorophyll concentrations in winter.

In deep water (50 metres), the model reproduced temperatures at Släggö well. Salinity values were close to the observed, with the right level of variability, but the observed time series were not reproduced by the model. Oxygen concentration was well modelled, although the autumn decline in oxygen concentration is more rapid than that produced by the model. Within Central Gullmaren, at Alsbäck, winter oxygen levels were well modelled at 50 metres, but the rest of the year was overestimated by about 1 ml/l. At 100 metres, the model faired less well, producing similar oxygen values to those at 50 metres, while the observations showed a much stronger seasonal cycle. Autumn and winter values were overestimated by up to 5 ml/l.

Modelled and measured Chlorophyll a

Modelled (lines) and measured (points) surface oxygen (red) and chlorophyll a (green) concentrations, for Släggö (fjord mouth, top) and Björkholmen (inner fjord, lower)

The future

This model was reasonably successful at reproducing the observed conditions in Gullmaren, even when some of the input data were estimates of the actual forcing. To develop the model further, we would like to:

  • Run the model operationally, with real-time input data from hydrological, atmospheric and meteorological models.
  • Validate the model against similar high frequency data, such as that from the Ocean Origo Seatramp system that is operated in Gullmaren.
  • Use the model to assist in the improvement of sampling strategies for the improved monitoring of coastal systems
  • Develop the model to incorporate a more realistic representation of the physics and biogeochemistry


  • Marmefelt, E., Arheimer, B., Lagner, J., 1999. An integrated biogeochemical model for the Baltic, Hydrobiologia, vol. 393, 45 - 56
  • Marmefelt, E., Håkansson, B., Erichsen, A.C., Sehlstedt-Hansen, I., 2000. Development of an ecological model system for the Kattegat and the Southern Baltic , SMHI Report Oceanography, No. 29, 2000
  • Omstedt, A., 1990. Modelling the Baltic Sea as thirteen sub-basins with vertical resolution, Tellus, 42A, 286 - 301
  • Svensson, U., 1998. PROBE: PROgram for Boundary layers in the Environment: System description and manual, SMHI Report Oceanography, No 24, 1998
  • Svensson, U., 2003. Applying an ecological model system at the Gullmar, SMHI Report Oceanography, in prep
SMHI reports are available from our library. Click on this link to see how.

Back to top