Local Oil Spill Models for South-East Baltic (LOSSEB Project)
Objectives The LOSSEB project main objective is to raise awareness within local/regional authorities as well as other organisations which take care of the coastal areas in Klaipeda and Pomeranian Regions about operational forecasting tools used by marine oil spill combating services already for many years. The oil spill forecasting model, primarily directed to marine oil spill combating services, in its maturity is offering information on possible coast pollution by oil. Actually coastal zones of both regions are potentially affected by oil production disasters (oil rigs in Polish and Russian waters, oil terminals in Lithuanian, Polish and Russian zones). Most of the coasts in the south-eastern part of the Baltic are scenic sandy beaches, being the major tourist attraction and basis for tourism industry. Sandy beaches are especially vulnerable to oil spill pollution and as such should be protected to highest possible level. Modelling tools offer ability to forecast possible location on the coast, where potential oil may land, thus giving information in advance for preparedness of necessary protective or combating measures. So far such tools are used mostly only by highly qualified personnel of SAR and oil spill combating services. One of the obstacles is English language of existing models – it is not yet suitable for crisis management people and local level – one part of the project will be to setup web-based models in local languages for both regions. Another problem is that resolution of underlying meteorological, hydrodynamic and wave model in the coastal region is still to low to resolve all important coastal phenomena – this problem will be solved by establishing one regional forecasting centre, which basing on the global Baltic models will produce more accurate and custom-tailored forecasts of these parameters for both countries. Preventive and combating measures against oil spills are hard to transfer into a direct impact on tourism, etc. It is clear what impact can have an oil spill reaching a beach, on the society and environment. However, side results of the project, as more reliable meteorological and hydrological forecast, can have direct impact on tourism, human activities in the sea, navigation safety, etc. Besides these activities we plan to organise validation experiments in Lithuanian and Polish waters of our Neighbourhood Programme area, which at the same time will be training for possible model users, thus creating new user community – being at the same time lifting them to the same level of knowledge through both regions. This will generate synergy between people responsible for crisis management and will bring us closer together – both from sector point of view (science – marine combating teams – land crisis management teams) as well as spatial point of view (Lithuania and Poland). The training will also show how to use tools to simulate oil spills for crisis management teams to get more realistic scenarios for normal organisational and operational trainings. PARTNERS
Lead Partner: Center of Marine Research, Klaipeda Taikos pr. 26, LT-91149, Klaipeda
Financial Partner Maritime Institute in Gdansk Dlugi Targ 41-42, PL 80-830 Gdansk
Contact Person Mr Juliusz Gajewski e-mail
Partner The Maritime Search and Rescue Service Hryniewickiego 10, PL 81-340 Gdynia
Contact Person Mr Marek Reszko e-mail
MAIN ACTIVITIES
Main objective of the project is to raise awareness on potential oil spill impact on the coast to scientific, marine oil spill combating teams and land crisis management groups and demonstrate possible benefits of using oil spill model forecasts in mitigation and prevention of oil spills. This main objective will be realised through the following activities:
establishment of one regional forecasting centre (delivering more accurate local forecasts of meteorological and hydrodynamic parameter based on global Baltic forecasts),
establishment of local oil spill models (being national clone versions of other widely use in the Baltic models – like Seatrack Web, CAROCS or others),
performing of validation experiments (where besides testing actual performance and accuracy of models the organisational skills of land and sea combating teams will be checked),
extensive training given to local/regional response teams as well as NGOs if required,
preparation of national manuals how to use models to minimise negative impact of oil spills on the coastal community.
INTERMEDIATE RESULTS OF LOSSEB PROJECT REALIZATION
1. Meteorological model COAMPS
Local versions of COAMPS are in operational run. The problem of unsatisfactory run time will be soon solved by purchase of a 32-core computer, especially for the forecasts purposes.
2. Wave model WAM
The model is run operationally with the resolution of 1 nautical mile in the South East Baltic.
3. Hydrodynamic model HIROMB
During the period from the last report, the codes of the HIROMB model have been received free of charge from Swedish Meteorological and Hydrological Institute. Local version of HIROMB, covering not only Lithuanian and Polish Exclusive Economic Zones, but also the areas of influence (in a case of an oil spill in these areas), is in the implementation phase.
4. Oil spill drift model and the Web system
The oil spill modelling system SeaTrack Web was planned to implement `using local models forecast data. It could not be done during the project realization due to the lack of the agreement with SMHI. This problem has been described in the final report. Negotiations with SMHI, due also to circumstances not depending from SMHI and MIG, took a very long time and have not been finished yet. The Licence Agreement, finally concluded between SMHI and MIG on 3rd September 2009, is limited to PADM (PArticle Dispersion Model, http://seatrack.smhi.se/seatrack/STW_manual_Technical_documentation.pdf). It is the planned basis for the oil spill model according to our License Agreement with the SMHI. The Web system, however, has to be developed by the MIG specialists or by an external contract. The negotiations of the contract for the purchase of the PADM model seem to be in a final phase.
All this means again delay of the Project completion, however, MIG is determined to achieve the all Project’s goals. It was stated, e.g., during the 12th HIROMB Scientific Workshop (25-27 August 2009), 18th HIROMB Steering Group Meeting (27 August 2009) in Gdansk and EuroGOOS 16th Annual General Meeting (6-8 October 2009) in Sopot.
4.1. Preliminary results of oil spill drift calculations
The preliminary calculations, showing possibility of oil spill modelling in the South East Baltic, have been done on the basis of the HIROMB forecast of the period 3 October 2009, 00 UTC – 10 October 2009, 15 UTC.
The drift and evolution of the oil spill is simulated in this preliminary study by a number of oil “particles”, moving together with the large-scale advection (represented by velocities, calculated by the HIROMB programme in the nodes of its grid), and with the random motion of small scale, simulating subscale motion as turbulence, waves, etc.
It should be noted that the algorithm of HIROMB uses so-called Arakawa-C grid (A. Arakawa and V. Lamb. Computational design of the basic dynamical processes of the UCLA general circulation model, Methods in Computational Physics, vol. 17, pp. 174-267. Acad. Press, 1977), where (for the sake of numerical stability) velocity components are staggered in different nodes – it is taken into account in the numerical programme for oil spill drift. Boundary line in this study is represented by the broken line parallel to parallels of latitude or meridians, according to the domain and grid resolution of the model. It would be changed into a real shoreline in the final version, to enable particles to move along the complicated shape of the coast, e.g., in straits and in the vicinity of islands.
Start time and position of an oil spill. As it is required, it is any time moment and any location in the all area of the Baltic Sea, including lagoons as Curonian Lagoon and Vistula Lagoon. For the particular study it has been limited to South East Baltic: from the west by meridian 17°.49’E, and from the north by parallel of 57°N. Time of the study was defined from 3rd October 2009, 00 UTC to 10th October 2009, 15 UTC.
Data needed for the drift simulation. From the MIG database of HIROMB forecast the data of u- and v- velocity components for the mentioned area and time with one-hour time step of the data archiving have been extracted. Other data as wind and waves are not included yet, they are partly reflected in the field of large-scale currents and random subscale motion, and will be included in the final version of the model.
The data for one time level for this area takes about 780 kB, so the volume of the all current data for more than a week is 144 MB. In the final version, there will be less time levels as the forecast advance will be limited to the real advance of HIROMB forecast (however, HIROMB is also calculated on the basis of medium-range wether forecast fro ECMWF, but it seems that for combating oil spill purposes, reliable forecast for 24-48 hours is satisfactory. Due to finer grid in the SE Baltic area, the total volume of data needed for 24-48 hours advance oil spill drift forecast in any location within the Baltic Sea would be even larger.
Simulation of the oil spill. The velocity of the spill (each particle independent) is in this study linearly interpolated form the neighbouring HIROMB nodes at two neighbouring time levels, plus the random component simulating subgrid motion. The time step of the calculation is 2 minutes to enable proper simulation of random motion of particles. For only the large-scale modelling, the time step could be even 1 hour, as the HIROMB data are archived once for each hour (24 time levels a day).
The first step is to give the geographic coordinates and time of the spill. In our particular case the location has been introduced in the programme (but any location of the area within the mentioned limits can be given): 55.323333°N, 20.571667°E, i.e., location of D6 oil rig. In the input data file (wspolrzpocz.txt) the time limits have been given: 3 October 2009 00 00 min. UTC for the beginning, and 5 October 07 00 min. UTC.
The next step is to find two time levels corresponding to the initial moment of calculations – it is 2009.10.03.00 and 2009.10.03.01 UTC. These files create the basis for time interpolation of the current velocities during the first hour of simulations.
Then the initial oil spill position is located within the data arrays – four neighbouring values of eastward – u-component (“u-rectangle”) of the current and four values of northward – v-component (“v-rectangle”) are found, and simple linear interpolation algorithm is then used. For such small grid cell areas, the approximation of flatness and rectangle can be used. The algorithm is principally the same for u- and v- component. For example, it is:
lu – the angular distance from the western boundary of the “u-rectangle” to the particle
fu – the angular distance from the southern boundary of the “u-rectangle” to the particle
dla – the angular resolution in parallel direction (here dla = 100 sec. angle)
dfi – the angular resolution in meridional direction (here dfi = 1 min. angle)
us(i,j) – i,j-element of the matrix of u-component at previous time level (previous hour)
un(i,j) – i,j-element of the matrix of u-component at new time level (next hour)
i, j – indices along the parallel and meridional axes respectively
l1 – time in minutes passed from the previous full hour (i.e., the previous array of HIROMB data) to the time of calculation of the next 2-minutes interval
The next step is to count the distance of the particle gone during the time interval of dt = 2 minutes. It is simply s = Upart * dt. Then new coordinates of the particle are calculated and the process is being repeated iteratively till the end of the simulation. Every hour the particles’ coordinates are stored in an output file. A snapshot of the programme in FORTRAN language is attached to the report.
Figure 1. The track of the virtual oil spill
The case study shown in the Figure 1 demonstrates the simulation made for a virtual oil spill from the D6 oil platform, at midnight 02/03 October, 2009. During first six hours, prevailing meteorological and hydrodynamic conditions caused slow drift in S-SE directions. After that, the meridional component of the mean flow changed its sign towards the north and, continuing its eastward direction, the virtual spill gradually accelerated its movement in north-east direction, and finally reached the coast of Curonian Spit at 13 UTC, on 4th October, 37 hours after the spill.
Appendix 1.A snapshot of the FORTRAN programme, the main algorithm:
