MERCW: Modelling of Ecological Risks Related to Sea-Dumped Chemical Weapons




After World War II, at least 50.000 tons of chemical weapons were dumped in the Baltic Sea. The toxic warfare, packed in crates or containers, was often dumped in relatively shallow waters and areas of active fishing - in close proximity to densely populated coastlines.

The MERCW project aims to assess the risks that these dumped chemical weapons pose for the Baltic Sea. The research and technology developments carried out in the project focus primarily on the dumpsite close to Bornholm, where over 35.000 tons of chemical weapons were dumped. The aim is to model the transport pathways and migration spreading of toxic agents in marine sediments and the marine environment. The final goal is to assess the ecological safety for the ecosystem and people of the coastal states near the dumpsite.

Project partners


externFinnish Institute of Marine Research (FIMR), Tapani Stipa


Other partners:
see externMERCW contact page

University of Bonn Tasks

The task of the University of Bonn Computer Graphics Group is to develop visualization and Virtual Reality technology that allows for visual analysis and interactive exploration of all data that is collected, modeled, and used in the project.

The use of visualization within the MERCW project has two main targets: First, it provides the project participants with a quick and intuitive overview of the data, allowing them to better analyse the available information, to plan efficiently the future field investigations, and finally to perform risk analysis based on the data. Second, it is an important instrument to present the results of the project such as potential risk scenarios, resulting from the analyses and modeling performed in the project, to the public.

The data considered for visualization includes:

  • bathymetric maps of the Baltic Sea,
  • 2D vector maps (for example of the dumpsite boundaries),
  • tracks of research cruises and survey lines,
  • positions of discovered shipwrecks,
  • positions and analysis results of samples taken during surveys,
  • seismic profiles,
  • vector fields describing the sea currents under certain weather conditions,
  • traces of particles through the water volume over certain time periods obtained by modeling and simulation,
  • 3D volume data describing predicted concentrations of chemical warfare agents obtained through analyses and numerical experiments,
  • 2D grid data describing effects on the marine food web.

In order to achieve high-quality and high-performance visualization for these different types of data in an integrated application, we apply and enhance several scientific visualization techniques, especially methods of terrain, volume and indirect flow visualization. Special emphasis is put on fast methods that exploit the high parallel computing power of the Graphics Processing Units (GPUs) contained in current PCs.

In addition to visualization, a further task of the University of Bonn is the management of all data that is collected, modeled, and used in the project. This includes implementation of data structures, realization of necessary file converters, set-up of a common project database and compilation of data obtained and owned by project partners.

The MERCW Visualization System

The visualization system that was developed within the scope of the MERCW project combines the capabilities and flexibility of a 3D geographic information system (GIS) with the latest advances in various fields of scientific visualization. The visualization integrates data acquired during the research cruises of the project as well as relevant archival data and thus enables visual spatial analysis of this data. The visualization system presents the data either on an interactive 2D map using a geographic projection of the user’s choice, or on top of a detailed 3D terrain model. Even stereoscopic 3D viewing is supported to enhance the perception and understanding of complex 3D phenomena (e.g. the movement of particles through the water volume based on sea currents at certain weather conditions). The interface of the application was designed in such a way that it is easily usable also for non-experts and features intuitive navigation in 2D and 3D that allows the viewer to look at the data in its geographic context at arbitrary scales.

Images (Screenshots)

Combined visualization of measured and simulated data from the project. On the left: a large-scale view on the Baltic Sea – useful to observe the potential spreading of toxic compounds (visualized by streamlines) within a certain simulated scenario. On the right: a close-up to the Bornholm dumpsite area – revealing the locations where data was acquired during the MERCW project (the points indicate positions where sampling was performed). More details about the measured data will be recognizable when zooming in further, whereas most kinds of simulated project data are on a much larger scale. This shows that an efficient multi-resolution approach is necessary for the combined visualization of different types of project data. (Note that these figures show exported maps/screenshots from the application's 2D mode with UTM map projection and overlaid Lat/Lon grid.)

Visualization of topographic and bathymetric data

Visualization of the Baltic Sea bathymetry using different shading techniques and overlaid by differently spaced depth contour lines: texturing and dynamic lighting (left), coloring by depth (center), shaded relief (right) with contour lines every 21 m, 25 m, 33 m, respectively.

Visualization of sample data and vector maps

Left: Visualizing survey network lines and sample locations simultaneously with seismic profiles and particles of a hydrodynamics simulation. Right: Visualizing only sample locations and boundaries of the Bornholm dumpsite (shown as points and lines at a fixed height above the seabed).

Visualization of seismic profiles

3D visualization of seismic profiles in combination with the digital terrain model (drawn semi-transparently). Here, a subset of the large-scale survey network is shown, which covers the whole Bornholm dumpsite area (data by courtesy of Tine Missiaen, RCMG Ghent).

Visualization of multi-dimensional scalar fields

Left: Visualization of a 3D scalar field describing predicted environmental concentrations for a certain CW agent release scenario with the spreading of toxic compounds illustrated by streamlines (data by courtesy of Victor Zhurbas, Shirshov Institute). Right: Visualization of a time-dependent scalar field describing predicted concentrations of CW agents in the food web in the case of seals at a certain simulation time step (data by courtesy of Susa Niiranen, FIMR).

Visualization of hydrodynamics

Visualization of streamlines and animated particles (represented by small arrows) showing real-time simulated flow based on a modelled current velocity field. Right: The simulation can be controlled at run-time by defining and positioning rectangular areas where new particles are emitted.

In the shown example, we perform a steady flow simulation based on a fixed time step of a rectilinear geodetic 4D grid of current velocity vectors that is publicly available from the ERGOM project of IOW Warnemünde. In general, such vector fields of current velocity data can be obtained from an ocean circulation model, such as the Princeton Ocean Model (POM) or the Modular Ocean Model (MOM).


Note: All videos have been captured using the built-in video export functionality of the application while performing real-time interaction (and partially while playing pack a camera path recorded previously within the application). GUI elements are hidden from video export (even if they overlap the render window); thus only the resulting rendering can be seen here.

2D mode - multiresolution example  (DivX6 video, 36 MB)

Seismic surveys have been performed in the project at multiple different scales: at first, a course, large-scale network covering the whole Bornholm dumpsite area was surveyed; then two networks of smaller extent but finer spacing were surveyed within the primary dumpsite area; finally, for two small (potentially interesting) areas within the extent of the previous surveys, additional networks were surveyed with even finer spacing.

In this video, we demonstrate the benefits of continuous multiresoltion for inspecting such data of different scales: At first, we consecutively zoom into the large-scale, medium-scale, and small-scale survey networks. Then, we start the playback of a precalculated particle simulation (provided by Victor Zhurbas, Shirshov Institute), where particles are emitted at the center of the primary dumpsite area. Starting from the dumpsite area, the particles finally spread over the whole Baltic Sea. Thus we have to zoom out by multiple orders of magnitudes to be able to observe the course of this simulation. (The playback speed of the particle animation is increased by the user at some points during the recording.)

The video was recorded in the 2D UTM map mode of the application. In the first half of the video, the rendering is additionally overlaid by a Lat/Lon grid and by depth contour lines (spacing 10 m).

3D visualization of seismic profiles  (DivX6 video, 13.8 MB)

This video shows the seismic profiles of every second / third survey line of the large-scale network visualized in 3D, demonstrating the benefits of providing real-time 3D visualization and interaction (in addition to map-based visualization and interaction). Since the profiles reveal structural information below the seabed, the seabed itself must be visualized transparently to be able to inspect the profiles. For this purpose, the user can adjust the amount of transparency of the terrain visualization freely, as shown in this video.

This video also demonstrates another important benefit of our multiresolution approach: being able to handle huge amounts of data at interactive performance. Since the application uses an efficient hierarchical datastructure for handling the seismic profile textures and the corresponding survey line geometry, the user can on the one hand zoom very closely to the seismic profiles, being able to inspect them at full resolution, while on the other hand, he can also zoom out to view all survey lines and profiles at once - without decrease of performance.

Particle simulation interaction example  (DivX6 video, 10.2 MB)

This video shows how the real-time particle simulation can be controlled at run-time by the user. After starting the simulation, the user first selects an emission area (visualized as yellow rectangle). Then he moves and scales this area while observing how this influences the simulation. Afterwards, the user adjusts several parameters of the particle simulation (via the corresponding GUI): he changes the shape of the particles from spheres to arrows (which reveal the local flow direction), then reduces the size of these arrows, and finally changes the coloring of the particles.


The MERCW Visualization System may be used freely for non-commercial purposes. Please write to to get a download link for the application and some sample data from the project (Windows installer).


D. Ebert and J. Krüger (Editors)
In proceedings of Eurographics 2009 - Areas Papers, pages 27-34, Eurographics Association, Apr. 2009
In proceedings of The Fourth International Symposium on 3D Data Processing, Visualization and Transmission (3DPVT'08), pages 261-268, June 2008
Y. Chen, I. Cluckie, and K. Takara (Editors)
Natalia Goncharova, Pavel Borodin, and Alexander Greß
In proceedings of 2nd International Conference of GIS/RS in Hydrology, Water Resources and Environment (ICGRHWE '07), Sept. 2007
Clausthal University of Technology, Technical Report number IfI-06-11, Sept. 2006
E. Gröller and L. Szirmay-Kalos (Editors)
In: Computer Graphics Forum (Sept. 2006), 25:3(497-506)
In proceedings of The 2nd US/EU-Baltic International Symposium, May 2006
In proceedings of The 20th IEEE International Parallel and Distributed Processing Symposium (IPDPS '06), pages 45, Apr. 2006