Backtracking oil slicks by using an iterative algorithm with a forward trajectory model. Implementation and verification.

Our goal was to investigate the performance of the in-house mathematical drift model using the oil spills according to the Marine Search and Rescue Service (MRCC) Riga data for the Eastern part of Baltic Sea 

There have been 15 cases in the Latvian territorial waters in 2016 when satellite imagery has identified potential marine pollution on the sea surface. Two additional reports of potential marine pollution have been received from ships. Satellite imagery from CleanSeaNet has identified 16 possible cases in the Latvian territorial waters in 2017, and further 19 possible cases in 2018.

We consider the following three cases of possible oil (and other pollutants) spills:

1) Possible oil spill in 2016.01.26 north of the harbor of Ventspils in Irbe strait.

2) Possible oil spill from ballast waters in 2017.05.14.

3) Possible pollutant (vegetable oil) spill in 2018.07.25.

Investigation of MRCC Riga sea pollution cases has revealed the following constraints and requirements for the FiMar oil drift model. First, the detected pollutant slicks are of size above 5 km2 and already of complex structure. This follows from the CleanSeaNet service detection capabilities and spill occurrence pattern; for example, dumping of ballast waters happens during the night, with spill being detected with the sunrise. Second, the pollutant slicks have a short lifespan and unknown chemical composition. Therefore, future development should focus on backtracking from a large target to a single most probable location in time-space.

The condition of nearly divergence-free flow is usually met in large-scale flows, but nonlinear changes in the properties of the oil are impossible to handle simply by reversing the direction of the wind field and the current field.

Method inspired by (Breivik, Bekkvik, Wettre and Ommundsen, 2011, BAKTRAK: Backtracking drifting objects using an iterative algorithm with a forward trajectory model.) was introduced into FiMar software and verified. This amends traditional reverse-time backtracking where a trajectory model is initialized and run in the forward direction, whereupon the individual ensemble particles that come within an acceptable time-space distance of the observation are used to initialize a new forward run. Unsuccessful particle trajectories thereafter are discarded. This procedure is then iterated until an acceptable number of trajectories end up within the target area (defined as a time-space radius around the location of the observation) is reached and time-space distribution of possible initial locations for the drifting object/ pollutant slick has been established.

The study was funded by Latvian Academy of Sciences, project lzp-2018/1-0162 DRIMO – Drift Modelling for pollution reduction and safety in the Baltic Sea, 2018-2021.