Scottish Marine and Freshwater Science Volume 5 Number 3: The Science of Deepwater Oil Spills - Results from the 2013 Marine Scotland Science Workshop

The workshop on “The Science Of Deepwater Oil Spills – Modelling” held in

Aberdeen in September 2013 included more than 50 experts drawn from academia,

government, consultants and industry to discuss aspects of the science needed to

respond in an effec

Session 3 - Oil Dispersion Modelling

State Of The Art


1. Setting the scene (Jon Rees, CEFAS).
2. OSIS (Lucy Short, Oil Spill Response Ltd.).
3. Capabilities of OSCAR, an introduction to 3D oil spill and contingency modelling (Petter Rønningen, Sintef).
4. Research into oil dispersion modelling (Charlie Main, NOC-S).

An overview was given detailing the need for oil spill trajectory modelling, both in response to an incident and for contingency planning. The challenges facing modellers were covered, especially in challenging areas such as the FSC. Overviews of some of the most commonly used oil spill trajectory modelling tools were given, with example screen shots and videos demonstrating how the software can be configured. Example outputs were shown including statistical and probabilistic outcomes. An example of a deep sea plume in the FSC was shown with oil released above 600 m depth going NE, and oil release below 600 m going SW. The fact that there are many oil dispersion modelling tools, all somewhat different, was highlighted, as was the need for validation against any cases of opportunity, such as surface drifters, drifting buoys, real oil spills, etc. The benefits of real time operational oceanography, such as the work carried out by various organisations represented in the North West European Shelf Operational Oceanographic System ( NOOS), and collaborative research and sharing of data, were highlighted.

Items identified in the discussions (beyond introductory talks) to characterise the state of the art are:

  • Missing Models: Oilmap (Applied Science Associates) was not covered/reviewed by the talks in sufficient detail in this session (it did have more coverage in theme IV, however). GNOME ( NOAA) was also not mentioned, despite being a common model used in the USA (and is part of the in-house oil spill forecasting capability of MSS).
  • Data Assimilation: There needs to be more focus on how data sources can be used in real-time with oil spill trajectory modelling tools in order to improve the accuracy of their predictions.
  • Model- GIS Integration: A subject area not covered was how GIS and other map based software/systems can be used in conjunction with oil dispersion modelling tools to disseminate information to users.
  • Ice Modelling: Ice modelling was not covered. It is important to know how ice moves and how it interacts with oil. However, this is not relevant for the area under consideration by this workshop. With the expansion of the oil industry into the Arctic, this subject will be of increasing importance to UK industry.
  • User Competence: The competency of model users was mentioned but more emphasis on this is sometimes needed. Who interprets and uses the models is important.
  • Parameter Sensitivity: The sensitivity of models (such as OSCAR) to different parameters was not covered in the review talks.
  • Appropriate Model Inputs: The current state of the art oil spill modelling presentations did not consider whether the current set of hydrodynamic model inputs are appropriate for FSC or not. This is a fundamental question in this region that any model user should ask themselves at the beginning.
  • Data Formats and Web-Based Access: Data formats were not covered. Common data formats would help academics and responders work together more easily, although some standard formats such as netDCF partially fulfil this requirement. Web based access to data would also be useful.
  • Peer-reviewed References: References for oil spill models and circulation models were not reviewed. These should still form part of the QA of any oil spill model.
  • A Catalogue of Strengths, Weaknesses and Limitations: Some strengths and weakness of the models were covered, but a comprehensive list of the strengths, weaknesses and limitations of each individual model would be very useful. This would help with making a decision on selecting the most suitable model for a particular event. So there is a need to be honest about the limitations of each oil spill model. For example, one potential limitation of OSCAR identified is that data is hard to input.
  • Communicating to Responders: How to communicate model output to responders was not covered here. This is an important topic, and needs more focus.
  • Surface Effects: Wind driven mixing and the 'windage' factor were not really covered by the presentations - the FSC is an extreme environment and these surface processes need to be validated.
  • Link to Oceanography: Whether the oil spill models can take on the important oceanographic processes identified in Theme I was not covered.
  • Remote Sensing: Application of satellite tracking (remote sensing) to help with short term prediction was not covered.
  • "Bespoke" Models: Preparatory in-house models and academic models were not presented. These are very valuable as commercial models can be limited in terms of data inputs.
  • Model Uncertainty: The presentations did not address model certainty/uncertainty. Bayesian techniques can be used to address this.

Knowledge Gaps

  • Models in the System: There needs to be some kind of structured and systematic approach to how models can be incorporated into the emergency response and licensing processes.
  • Communications: Better communication between the model providers and researchers would be very beneficial. A science-response community interface would also be useful in an emergency. Currently this does not exist, and response exercises are very limited and insular in the communities they involve. For example, are there web based dissemination interfaces, or could they be developed?
  • Access to Data: There is often a scarcity of data sources and access to data can be problematic. Can data be released from industry during a response, but also beforehand for constructing models before and incident occurs?
  • Wave Data: Wave data and parameterisations within models. This is important as swell affects the breakdown of oil droplets, for example. Not enough research has gone into waves and the effect of waves in this region.
  • Pelagic Impacts: What are the impacts of dispersed oil subsurface on the pelagic ecosystem?
  • Model Resolution: Current hydrodynamic models have too coarse spatial resolution - higher resolution is needed for some oil spill models.
  • Coastal Effects: How can freshwater runoff be incorporated into oil dispersion models, and is this important near coasts?
  • Oil Physical Properties: How do the physical and chemical properties of oil change under different environmental conditions, and how can this be incorporated into models? For example, oil may behave quite differently 'on the day' than in the laboratory. This is especially true for the FSC region where environmental conditions can be extreme.
  • Fate and Effects of Dispersants: Guidance on the (ultimate) fate and effect of dispersants are needed. These should be based on enhanced knowledge and research as current understanding is limited.
  • Scenario Planning: Scenario planning should be extended in the FSC region, including, for example, experimenting with different well depths/pressures etc., to find out what the existing tools can tell us about a possible event west of Shetland ?
  • Communication: Better guidance is still required in order to include a much wider involvement and input into the response to an incident. Currently communication lines are too restricted.
  • Datasets: Compilation of data is needed in order to facilitate rapid access in the times of need
  • Resource Sensitivity: More information on the sensitivity of resources is needed.
  • Model Validation: More oil spill model calibration/validation is needed. In order to achieve this more data is required, including current profiles etc., but also data from real events or tracer experiments. Comparisons with reality and real oil releases needs to be done and methods to do this need to be developed. Hind cast or re-analysis runs of real (past) events would be very useful.
  • Model Internals: Which models are acceptable and provide the best tool for the job? Often commercial models are treated as a black box. We need to know more about the processes each model tries to include, and how they are parameterised.
  • Industry Standard: A "model industry standard" is needed, which describes the minimum standard of a model for particular situation (3D (deep) for example). This could help identify which tools should be used by consultants?
  • Access to Models: How can industry get access to the most up to date models, especially from academia and government research? This includes both oil trajectory models, but also the hydrodynamic flow fields.
  • Knowledge Sharing: Sharing knowledge between consultancies would be very valuable. For example sharing past event runs would be useful to get more confidence in results.
  • Standard Model Inputs: Is there a need for a standard input to models to be defined by the regulator (e.g. for scenario testing and planning purposes)? Is there a requirement for a minimum of model output and interpretation? This is in order to aid model comparison and validation.

Future Directions

  • Ensemble Modelling: A possibly better way to employ models, either using a single model run multiple times, or many different models. Ensembles provide a method of quantifying potential variability and uncertainty.
  • Focus on West of Shetland: Current exploration areas west of Shetland are very complex. There need to be a focus on models of these areas. There is a need for more FSC research and dispersion model development for FSC. The complexity of this region is not currently captured by models.
  • Subsurface Dispersants: We need to know more about the benefits/drawbacks of subsurface dispersants, for example Rosebank oil properties can change over time, especially when subsurface. We need this in order to be able to include their effect in dispersion models.
  • Bespoke Models: The oil dispersion models are generic models at the moment. Do we need models specific to a particular location?
  • Basic Physics: We need to understand more about the basic physics of environmentally important areas.
  • A Range of Models: A wider range of models are needed to be made available - this might stimulate more model development. Currently the choice of models is too limited. Can the academic community be involved here?
  • Secondary Tools: More development of trajectory analysis tools and secondary/post processing tools is needed.
  • Cross-discipline Communication: Cross disciplinary discussion needs to be promoted, to help decisions during a response.
  • Model Comparison Exercise: A inter-model comparison exercise is needed in the FSC in order to assess the performance of different models and model combinations. Note that NOOS do inter-model comparisons - both drifts (i.e. trajectory modelling) and comparison of hydrodynamics.
  • Drifter Comparison Contest: Perhaps there should be a contest in order to find out which models can best reproduce real time drifter tracks (a prize could be offered!).


Back to top