Umuroa FPSO Oil spill incident New Zealand, October 2007

In the early hours on the 21st of October 2007, 23, 000 litres of oil was spilt from the Umuroa FPSO (floating, production, storage and offloading) vessel, offshore New Zealand's North Island. Approximately 60 hours following the spill, a report was received that a 10 km stretch of beach was contaminated in the Okato area ( 55 km northeast of the FPSO mooring). Following the incident, modelling was used to predict the likely path for the spilt oil using a purpose-developed trajectory particle and fates model, OILMAP. Three oil spill simulations were performed, using the same spatial wind data from the Global Forecast System (GFS), but with different current forcing. The first oil spill simulation was run using predicted tides generated for the region, with the model results showing the slick drifting immediately east from the release site and remaining 22 km off the coast. This indicates that the forcing of the tides (< 0.1 m/s) were insignificant in the deeper waters (121 m) at the incident site. Input data for the second simulation included satellite derived surface currents, which showed the oil spill meandering northeast from the release site, prior to reaching the shoreline by Tuesday evening. As this simulation resulted in the oil stranding immediately south of the actual contaminated beach, it is evident that the satellite derived current data was less accurate closer to the coast. The third oil spill simulation employed large scale forecast currents derived from the US ocean/coastal navy model (NCOM) as input. The OILMAP results showed th e oil spill initially drifting northeast from the release site for the first 24 hours, shifting eastward (similar to the second simulation) and stranding along the 10 km stretch of beach where stranding was observed by Tuesday evening. Hence, the findings from simulation 3 provided the highest degree of confidence of the likely oil spill path prior to stranding. This highlights the importance of deepwater currents in oil spill predictions and the need to access several datasets to reproduce the fate of an incident with confidence.