Roskilde Fjord setup

Roskilde Fjord is located on Zealand, Denmark, and is connected to southern Kattegat through northern Isefjord via a narrow channel. The fjord has a surface area of 125 km² and a mean depth of 3 m.

The computational mesh consists of 3842 polygons. The model comprises 15 vertical layers and a total of 13814 computational cells. Horizontal resolution ranges from 115 m in the shallow and narrow sections to 275 m in the more open parts of the fjord, such as Roskilde Bredning.

At the open boundary towards Isefjord in the northwest, the model is forced with sea surface height, velocity, temperature, and salinity interpolated from Copernicus CMEMS model output. Boundary temperature and salinity were adjusted using observations from the Danish National Monitoring Programme station VSJ10003. Atmospheric forcing was obtained from the WRF meteorological model. Point-source and diffuse runoff from the Danish QNP model were aggregated into 13 freshwater inputs. The model was run for the period 2019–2022 using a hydrostatic solution with a time step of 1.5 minutes.

Publications:

• Vibe Schourup-Kristensen et al. Hydrodynamic Control of Cadmium Retention and Bioaccumulation in a Temperate Estuary. Marine Pollution Bulletin, 2026. https://doi.org/10.1016/j.marpolbul.2026.120170
• Gaia Grieco. Designing with Dispersal: Connectivity-Informed Site Selection for Bivalve Recruitment. Marine Environmental Research, 2026. https://doi.org/10.1016/j.marenvres.2026.108152

Agent based model

The animation illustrates water movement in Roskilde Fjord using a Lagrangian Agent-Based Model (ABM) coupled to the hydrodynamic model. In the ABM, virtual particles are released throughout the fjord and transported by the simulated currents and turbulence. Unlike traditional Eulerian approaches, which describe flow at fixed locations, the Lagrangian approach follows individual particles as they move through the water, providing an intuitive visualization of transport pathways and water exchange.

The traces indicate particle movement and speed over time. Areas with stronger currents appear as elongated streaks, highlighting the main transport corridors through the fjord. In contrast, regions with weaker circulation exhibit limited particle movement, indicating longer water retention times and reduced exchange with surrounding waters. The animation reveals how water masses are transported between different parts of Roskilde Fjord and illustrates the strong control exerted by the narrow channels connecting the broader basins.

Such simulations are used to study water exchange, residence time, dispersal of plankton and larvae, transport of nutrients and pollutants, and the spread of non-indigenous species. They provide valuable insight into the physical processes that shape the fjord ecosystem and support both scientific research and environmental management.

Understanding how water is transported and renewed is essential for assessing ecosystem dynamics, nutrient cycling, and the fate of pollutants in coastal waters. To quantify water exchange in Roskilde Fjord, two complementary tracer-based methods were applied:

Water Age
The water age tracer estimates the average time elapsed since a water parcel entered the fjord from either the open boundary or a freshwater source. Newly entering water is assigned an age of zero, while the age of water already present in the fjord increases with time.
Dilution Time
Dilution time describes how quickly water in a given location is replaced by water originating from outside the fjord. To estimate this quantity, a passive tracer was initially distributed uniformly throughout the fjord, while tracer concentrations at the open boundary and freshwater sources were set to zero. Dilution time was then defined as the time required for the tracer concentration to decrease below 1/e (approximately 37%) of its initial value.

The figure shows the four-year mean distribution of (a) water age, (b) dilution time, and (c) salinity throughout Roskilde Fjord. The lowest water age and shortest dilution times occur near the connection to Isefjord, where exchange with outer waters is strongest. Both quantities increase towards the inner parts of the fjord, reflecting progressively weaker water renewal and longer retention of water masses. The salinity distribution exhibits a similar pattern, with more saline water entering from the north and becoming gradually diluted by freshwater inputs further south.

Understanding the sources, transport, and persistence of pollutants in estuarine systems is essential for assessing ecological risk and supporting effective management. Whilst cadmium (Cd), a persistent trace-metal in coastal waters, presents a challenge due to its toxicity and high capacity for bioaccumulation in benthic organisms, its cycling is well-understood compared to other heavy metals. Cd observations was combined with numerical modeling to develop a first three-dimensional, system-scale model of dissolved Cd dispersal and bioaccumulation in blue mussels (Mytilus edulis) in Roskilde Fjord.

The coupled hydrodynamic–biological modeling framework resolves spatial and temporal variability in Cd concentrations and explicitly links water residence time to patterns of Cd retention and bioaccumulation in mussels. Model results show that the inner fjord, characterized by long residence times, exhibits enhanced Cd accumulation in biota, despite relatively low contemporary inputs, suggesting a key role of hydrodynamic retention in shaping exposure. Simulated mussel Cd concentrations reproduce observed spatial gradients and fall within the range of available monitoring data.

The figure below show spatial plots of mean annual a) surface Cd concentration (ng L−1) and b) number of days per year (2019–2021) with concentrations above the OSPAR background value of 12 ng L−1. The brown circles are the industry and WWTP Cd sources.

Overview of model concept. The yellow annotations refer to the hydrodynamic model, with an open boundary and atmospheric forcing. The red annotations refer to the Cd dispersal model, with passive transport in the water, sinking and resuspension, and external Cd load from land, the open boundary and the atmosphere, based on measurements. The green annotations refer to the DEB- and mussel bioaccumulation model. A) External Cd load from streams, industry and WWTPs. B) Atmospheric deposition of Cd. C) Advection to model domain across open boundary. D) Cd Measurements in streams and WWTP effluent. E) Physical atmospheric forcing: Wind, temperature, precipitation.