Assessing sensitivity of oil spill drift modeling.

Supervisors: Camilla Brekke, Martine M. Espeseth (The Arctic University of Norway).
With the rising awareness of ill effects of oil spills on the open waters and coastal environment, it has become crucial to timely launch effective cleanup and recovery actions. Oil drift modeling has become an essential part of this process to predict the horizontal and vertical movement of the slick. Although these models provide a lot of flexibility and control to the user by allowing parameters to be defined at the time of initialization of the simulation, it is important for the user to understand the effect of those settings on the results. OpenDrift framework developed at Norwegian Meteorological Institute (MET Norway) was used to perform simulations of the spilled oil. This study primarily focused on testing the sensitivity of the model to some selected parameters using Institute of Marine Research (IMR) NorKyst-800 model currents and winds derived from MET Norway Arome model. Data collected from multi-satellite Synthetic Aperture Radar (SAR) acquisitions (since 2011) of oils spilled during Oil-on-Water (OOW) exercises by Norwegian Clean Seas Association for Operating Companies (NOFO) in the North Sea were used for initializing simulations and for comparing the model output with the real SAR data. Based on the analysis of data from OOW2011, 2012, 2015 and 2016, it was observed that the model was sensitive to the number of seed points, time step and uncertainty radius of release position. However, very low sensitivity was observed for water fraction of the released oil. Modeling of oil spill drift largely depends on the accuracy of information provided on atmospheric wind and oceanic currents near the surface. A comparison of model currents and drifter currents based simulation was conducted for OOW 2011 and 2015 spills and showed that drifters improved the results significantly. For 2015 oil spills with high wind conditions, it was found that the IMR NorKyst model performed better than the MET Norway NorKyst model. Lastly, it was observed that the simulations with IMR NorKyst model data drifted the oil spill faster than the drift observed from SAR for less than two hour-long simulations from OOW 2011, 2012 and 2016 exercises.