HYDROCARBON DISPERSION MODELLING
Numerical modelling of the dispersal and coastal impacts of accidental hydrocarbon spills from Site Tawhaki-1
prepared for OMV GSB
Version Revision Date Summary Reviewed by RevA 02/04/2019 Draft for internal review Weppe RevB 05/04/2019 Draft for internal review Berthot RevC 08/04/2019 Draft for client review Weppe RevD 22/04/2019 Updated draft for client review Weppe
Version Date Distribution
Document ID: P0441
MetOcean Solutions is a Division of Meteorological Services of New Zealand Ltd., MetraWeather (Australia) Pty Ltd [ACN 126 850 904], MetraWeather (UK) Ltd [No. 04833498]
and MetraWeather (Thailand) Ltd [No. 0105558115059] are wholly-owned subsidiaries of Meteorological Service of New Zealand Ltd (MetService).
The information contained in this report, including all intellectual property rights in it, is confidential and belongs to Meteorological Service of New Zealand Ltd. It may be used by the persons to which it is provided for the stated purpose for which it is provided and must not be disclosed to any third person without the prior written approval of Meteorological Service of New Zealand Ltd. Meteorological Service of New Zealand Ltd reserves all legal rights and remedies in relation to any infringement of its rights in respect of this report.
1. Introduction ... 5
2. Hydrocarbon dispersion modelling methodology... 7
2.1 Oil spill scenarios ... 7
2.2 Stochastic and deterministic approaches ... 7
2.3 Atmospheric and Oceanographic data sources ... 8
2.3.1 Wind data ... 8
2.3.2 Hydrodynamic data... 10
2.3.3 Wave data ... 10
2.4 Hydrocarbon dispersion modelling ... 10
2.5 Hydrocarbon weathering modelling ... 12
2.6 Post-processing of oil spill model outputs ... 13
2.6.1 Stochastic approach ... 13
2.6.2 Deterministic approach ... 14
3. Results 15 3.1 Stochastic modelling... 15
3.2 Deterministic modelling ... 20
4. Summary 26 5. References ... 27
6. Appendix A ... 29
List of Figures
Figure 1.1 Tawhaki-1 site location in Great South Basin of New Zealand. ... 6 Figure 2.1 The WRF model New Zealand domain at 12 km resolution. ... 9 Figure 2.2 Comparison of both CFSR data and a high-resolution WRF hindcast for Auckland Airport during a few days in January 2007 (left) and quantile-quantile plot of both CFSR (magenta) and the WRF hindcast (green) against the observations from Brother’s Island in the Cook Strait during 2007. ... 9 Figure 3.1 Time-independent mean surface oil distribution over the duration of the 200 oil spill events. The sum of all cell probabilities is 1. The probability density function was blanked below values of 1e-6 for plotting purposes (i.e. transparent). ... 16 Figure 3.2 Schematic representation of the main ocean circulation features off the east coast of New Zealand South Island. Main locations and oceanographic features of interest are highlighted (modified from Soutelino, R.G. and Beamsley, B. 2015).
Position of Tawhaki-1 is shown by the red dot on left panel. ... 17 Figure 3.8 Residual current rose at Tawhaki-1 at nearbed, mid-depth and surface levels.
Current direction are “going to”. ... 18 Figure 3.10 Wind rose at Tawhaki-1. Wind direction are “coming from”. ... 19 Figure 3.7 Time independent mean surface oil distribution averaged over the duration of worst-case spill scenario (starting 2010-06-20 17:47 ). The sum of all cell probabilities is 1. The probability density function was blanked below values of 1e-6 for plotting purposes (i.e. transparent). ... 21 Figure 3.8 Predicted daily oil spill positions throughout the simulation period from T0 to T0+7 days. ... 22 Figure 3.9 Predicted daily oil spill positions throughout the simulation period from T0+8 days to T0+15 days. ... 23 Figure 3.10 Predicted daily oil spill positions throughout the simulation period from T0+16 days to T0+23 days. ... 24 Figure 3.14 Oil mass budget over the duration of worst-case spill scenario (starting 2010-06-
20 17:47 ). The plot summarizes the amount and fate of oil discharged through the simulation. The amount of oil is expressed as mass on the left axis and fraction of total oil released on the right axis. Time between successive vertical grid lines is 10 days. ... 25
List of Tables
Table 1.1 Tawhaki-1 site position (WGS84). ... 6 Table 2.1 Oil spill scenario details following (Ranold 2019). Full oil characteristics used for weathering modelling are provided in Appendix A, Table 6.1. ... 8 Table 6.1 Oil properties of Tawhaki-1 used in the oil spill simulations... 29
OMV GSB Ltd. has commissioned MetOcean Solutions (MOS) to undertake an assessment of the potential dispersal and beaching of accidental hydrocarbon spills from the site Tawhaki-1 located in the Great South Basin of New Zealand (Figure 1.1).
The accidental discharge of hydrocarbons in the oceanic environment can occur at any time and therefore be subject to any combination of atmospheric and oceanographic forcings.
Additionally, the weathering characteristics of the hydrocarbons will be dependent on the chemical composition of the hydrocarbons, but also the oceanic conditions both at the time of release and during advection of the spill. Salient parameters that influence the weathering profile of a given hydrocarbon include salinity, temperature, wind speed and wave action; all of which will be spatially and temporally variable during the advection phase of the hydrocarbon dispersion. This means that any two spill events from a given source will have a unique weathering profile, with associated beachings. To account for this unpredictability, oil spill impact assessments must capture that weathering and forcing variability so that robust statistical outcomes can be derived.
For this study, a stochastic and deterministic modelling approach has been applied in order to ensure robust statistical outcomes for the dispersion of hydrocarbons are achieved. The study utilises a 10-year historical hindcast of atmospheric and oceanic conditions in which a large number of discrete hydrocarbon release events are simulated, each with random start-times and associated oceanic conditions during the advection and weathering phases. Post processing and merging the individual stochastic results provides a robust methodology that characterises the statistical probabilities related to oil dispersion, weathering and beaching patterns.
This report is structured as follows; Section 2 describes the general oil spill modelling approach and methods, including oceanographic data sources used, oil spill scenario considered and details of the oil spill model. Section 3 presents and interprets simulation results for both the stochastic and deterministic modelling, while a summary of the work is given in Section 4.
References cited are listed in Section 5.
Figure 1.1 Tawhaki-1 site location in Great South Basin of New Zealand.
Table 1.1 Tawhaki-1 site position (WGS84).
Sites Longitude [deg W] Latitude [deg N]
Tawhaki-1 171.6388 -46.9356
2. Hydrocarbon dispersion modelling methodology
2.1 Oil spill scenarios
Details of the oil spill scenarios modelled were defined from the statistical results presented in (Ranold 2019), which estimates most likely blowout release rates, durations as well as release depths. For the present case, the overall expected blowout duration is 21 days, with 10%
probability for a surface release and 90% probability for a seabed release and the total volume of oil released to the sea over that period is expected to be 136,612 stock tank barrels (STB hereafter).
The oil release rates were determined based on the volume spilled, the distribution of that volume within the surface and seabed layers and the duration of release. Oil chemical properties expected for the Tawhaki-1 site were provided by OMV GSB Ltd. (see Appendix A, Table 6.1 for details).
2.2 Stochastic and deterministic approaches
The stochastic modelling approach consists in simulating a large number of oil spill events with random start times over a time period of interest in order to capture the wide range of oceanic and atmospheric conditions expected at the site. The stochastic, i.e. probabilistic, approach is generally combined with a deterministic approach in which one or several individual worst-case spill events are identified and further analysed.
In the present study, the stochastic approach consisted of modelling 200 individual spill events at each site, with random start dates selected within a 10 year period (2004-2014) that spanned the available hindcast hydrodynamic and atmospheric datasets. The generated oil spill database was then post-processed to derive a range of statistical parameters (see Section 2.6).
The single worst case event within the 200 unique stochastic events was identified, and analysed further as part of the deterministic analysis. The worst case event was considered to be the event which resulted in the largest extent of shoreline impacted.
Table 2.1 Oil spill scenario details following (Ranold 2019). Full oil characteristics used for weathering modelling are provided in Appendix A, Table 6.1.
Tawhaki-1 Expected Blowout duration [days] 21
Oil volume [STB] 136,612
Oil volume [m3] 21721.3
Oil mass [kg] 17,920,079.1
fraction release surface [%] 10%
fraction release nearbed [%] 90%
release rate surface [m3/hour] 4.31 release rate nearbed [m3/hour] 38.79
Oil type Tawhaki-1
2.3 Atmospheric and Oceanographic data sources
Released oil dispersion and weathering profiles depend on atmospheric and hydrodynamic forcings. The datasets used for the oil spill simulations are described below.
2.3.1 Wind data
The near surface wind fields for the New Zealand region were primarily extracted from a 39- year (1979-2016) regional atmospheric hindcast carried out by MOS. The WRF (Weather Research and Forecasting) model was established over all of New Zealand at 12 km resolution (Figure 2.1). The WRF model boundaries were sourced from the CFSR (Climate Forecast System Reanalysis) dataset distributed by NOAA (Saha et al. 2010). These data span 39 years (1979- 2017) at hourly intervals and 0.31° by 0.31° resolution until December 2010 and 0.20° by 0.20°
resolution beyond January 2011. While the WRF hindcast produced atmospheric parameters at hourly intervals, including the near surface wind field (i.e. 10 minute mean at 10 m elevation).
The WRF model has been validated at numerous sites around NZ and has been shown to provide a significant improvement over global CFSR atmospheric datasets (i.e. Figure 2.2).
The recently atmospheric reanalysis ERA5 was used as secondary dataset in case oil particles moved out of the WRD NZ domain. More information on the ERA5 reanalysis is available in Hersbach et al. (2019) and ECMWF (2019).
Figure 2.1 The WRF model New Zealand domain at 12 km resolution.
Figure 2.2 Comparison of both CFSR data and a high-resolution WRF hindcast for Auckland Airport during a few days in January 2007 (left) and quantile-quantile plot of both CFSR (magenta) and the WRF hindcast (green) against the observations from Brother’s Island in the Cook Strait during 2007.
2.3.2 Hydrodynamic data
The primary hydrodynamic dataset for oil spill simulations consists of a 39-year NZ-scale hydrodynamic hindcast performed using the Regional Ocean Modeling System (ROMS model version 3.7) with an horizontal resolution of ~7 km that is described in (MOS 2019).
The Climate Forecast System Reanalysis CFSR product (Saha et al. 2010) from the National Center for Environmental Prediction (NCEP), which consists of a 0.5 degree global reanalysis with comprehensive data assimilation, was used as a secondary dataset when oil particles moved out of the ROMS NZ-scale domain.
2.3.3 Wave data
Wave conditions were defined from a suite of downscaled wave hindcasts covering a 39 year period. First, a global scale wave hindcast was produced by MOS using the WW3 (WAVEWATCH III) model with a resolution of 0.5° by 0.5°. The CFSR wind field was used for wind forcing and the physics of Tolman and Chalikov (1996) were applied in the model configuration. These global hindcast data were extracted at 3-hour intervals and used to prescribe spectral boundaries for a SWAN wave model domain covering New Zealand, with a 4 km resolution, forced with wind fields as described in Section 2.3.1.
2.4 Hydrocarbon dispersion modelling
Oil spill dispersion and weathering were simulated using the OpenOil3D oil drift model, which is part of the open-source, peer reviewed, dispersion modelling framework OpenDrift1 (Dagestad K.F et al. 2018).
The oil dispersion modelling consists of a trajectory tracking scheme applied to discretised oil particles in time and space-varying oceanic (current, wave) and atmospheric (wind) forcing fields, combined with “online” (i.e. concurrent) oil weathering. Key features of the model are outlined below. Full details are available in Dagestad K.F et al. (2018) and the model was applied in recent studies by Röhrs et al. (2019) and Hole et al. (2019).
In the horizontal plane, oil particles are assumed to drift with the oceanic currents. Surface oil particles are also subject to wind forces which move them at a fraction of the wind speed, in the wind direction. A recommended fraction of 3% of the wind speed was used in the present application (e.g. Zelenke, B. et al, 2012). Stokes drift, which can have a (secondary) minor effect on horizontal dispersion was not included in the modelling but is not expected to affect the salient outcomes of the dispersion modelling.
Horizontal diffusion is used to account for the motion of particles due to sub-grid scale turbulent processes, such as eddies, that are not explicitly resolved in the hydrodynamic model due to the spatial resolution. For this study, horizontal diffusion was parameterised by applying an uncertainty to the horizontal current magnitudes in the OpenDrift framework. Current uncertainty was estimated using the general diffusion distance equation (eqn 2.1) and assuming a generic value of 7.5 m2·s-1 (consistent with MOS, 2019).
∫𝑡𝑡+∆𝑡𝑢𝑡. 𝑑𝑡 = √6. 𝐾𝑢,𝑣. ∆𝑡 . 𝜃(−1,1) (2.1) where 𝜃(−1,1) is a random number from a uniform distribution between -1 and 1,
tis the time-step of the model in seconds (900 sec. used here) and Ku,v is the horizontal eddy diffusivity coefficient in m2·s-1.
The above horizontal drift components can lead to large gradients in drift patterns in the upper water column and it is important to adequately reproduce the vertical oil transport processes to correctly reproduce the oil spill behaviour.
Theoretically, vertical particle motions are controlled by vertical current velocities, entrainment in the water column due to wave breaking, oil droplets/particle buoyancy, as well as vertical turbulence (i.e. vertical diffusion).
Typically, vertical velocities have negligible effects on the vertical oil motions and were not used in the present simulations. The oil particle entrainment in the water column depends on the wind and wave conditions, as well as oil properties, and are parameterised following the formulations of (Li Z., Spaulding M.L., and French-McCay, D 2017). The buoyancy of oil particles (or “droplets”) is calculated according to empirical relationships and the Stokes law of Tkalich, P and Chan, E. S. (2002). The droplet size spectrum, which is important for the buoyancy process, is computed using the approach of Johansen M. R. et al. (2015), where a log-normal droplet spectrum is calculated explicitly based on wave height and oil properties, including viscosity, density, interfacial tension and surface film thickness.
Finally, the oil particles are subject to vertical turbulent motions through the water column (i.e.
diffusion). The process is parameterised in OpenDrift using a numerical scheme described in Visser A (1997). The present simulations used a constant vertical diffusion coefficient of 0.0001 m2·s-1.
For each of the 200 spill events simulated, the 21-day continuous release considered a total of 10,000 oil ‘particles’ released at near-bed and near-surface levels according to specifications.
The oil particles are then tracked in space and time over the duration of the simulation (i.e.
130 days), or until they become stranded on land, or move out of the hydrodynamic and wind model domains. The simulation time step was 900 seconds and oil particle positions were output every 3 hours.
2.5 Hydrocarbon weathering modelling
In addition to horizontal and vertical motions, the oil material is subject to weathering due to ambient oceanic and atmospheric conditions.
The OpenOil3D module is interfaced with the existing OilLibrary2 software developed by NOAA, which will eventually replace the original ADIOS23 oil library (Lehr et al. 2002). The OilLibrary contains a database of ~1000 oil types from around the world and is used by NOAA’s oil drift model PyGNOME4. User-defined oil types with specific properties can be prescribed.
In the absence of suitable oil records from the Great South Basin, the existing Tui-1 oil type was used within the OpenOil3D module as an appropriate analogue for the expected Tawhaki- 1 oil, in consideration of the geochemical similarity of the Upper Cretaceous coaly source rocks in the Great South Basin to those in the Taranaki Basin that sourced the Tui oil.
The OpenOil3D module includes state-of-the-art parameterisations of processes such as oil evaporation, emulsification and dispersion, adopted from NOAA PyGNOME model. These allow modelling the progressive oil mass reduction and dispersion over time and the production of oil budgets. All weathering parameterisations use time and space-varying oceanic and atmospheric conditions, as well as instantaneous oil properties (defined within the OilLibrary).
Dissolution is not currently included in the OpenOil3D module and thus not included in the present simulations. It is generally expected the dissolution process is responsible for only a small fraction of the overall oil losses (i.e. see National Research Council (US) Committee on Oil in the Sea: Inputs, Fates, and Effects., 2003; Afenyo M. et al., 2016) as it is in competition with oil evaporation which is generally much more efficient by one or more order of magnitudes. Further, most hydrocarbons, as considered here, generally contain relatively small or negligible proportions of soluble compounds therefore dissolution oil losses are not expected to be significant.
2.6 Post-processing of oil spill model outputs
2.6.1 Stochastic approach
Using the oceanic and atmospheric forcing fields, the oil spill simulations track discrete oil particle positions in space and time over the duration of the simulation (i.e. 130 days). The oil particle masses decrease concurrently over time due to weathering effects.
Results of each oil spill simulation were post-processed in order to form a database of possible oil spill dispersion (i.e. stochastic approach), which was then used to derive statistical metrics for both open-ocean dispersion and beaching potential.
The open-ocean dispersion patterns were estimated on grid ~800 km x 800 km with resolution of ~400 m, centred on the release site. For beaching analysis, the New Zealand shorelines were discretised into 1 km x 1 km cells.
Key parameters computed are outlined below:
• Mean surface slick dispersion patterns. The mean patterns are derived by combining the successive positions of oil particles found in the 10 m surface layer over the 200 runs and computing 2D oil mass histograms (i.e. box- counting), weighting each individual suspended particle according to the oil mass it carries.
• Probability of beaching. At each shoreline cell, the beaching probability is computed by dividing the number of spill events for which oil beaching, above given threshold occurred by the total number of runs (here, 200). Oil beaching fields are computed using a 2D beached oil histogram (i.e. box-counting), weighting each individual beached particle according to the oil mass it carries.
• Minimum time before beaching. At each shoreline cell, the minimum time before beaching is defined as the smallest time between the start of the simulation and first beaching, at concentrations above a given threshold, out of the 200 runs.
• Mean and maximum cumulative oil beaching. At each shoreline cell, the cumulative oil beaching is computed as the total oil mass reaching that cell over the duration of the simulation (i.e. 130 days). The cumulative oil beaching fields are computed for each run for subsequent statistical calculations.
• Mean and maximum length of shoreline exposed to oil beaching. The length of shoreline exposed is defined as the number of shoreline cells (1 km x 1 km) where beaching above the threshold occurred. The lengths of shoreline exposed are computed for each run to provide statistical mean and maximum exposures.
For the oil beaching analysis, a beaching event was defined as any time when the beached oil concentration in a shoreline cell was above a nominal threshold of 0.5 g.m-2. This equates to
an average thickness of ~0.5 m which, according to the Bonn Agreement Oil Appearance Code (Bonn Agreement, 2006), is described as a silvery to rainbow sheen in appearance.
Further, this thickness is considered the practical limit of observing oil in the marine environment (Australian Maritime Safety Authority, 2013) and the point at which the effectiveness of standard recovery systems would be at their limit. The 0.5 g.m-2 threshold is considered very low and conservative as it expected to be well below concentrations having adverse environmental impact. For example, French-McCay, D. P. (2009) and Owens, EH and Sergy, GA (1994) suggest threshold oil concentration having adverse impact on the nearshore aquatic wildlife (birds, invertebrates) are rather of order 10-100 g.m-2. Note the shoreline threshold is consistent, or smaller than previous oil spill study for the region (e.g. RPS APASA, 2015; Lebreton, LC and Franz, T. 2013).
2.6.2 Deterministic approach
The deterministic approach consists of selecting the worst-case spill event from the spill database (i.e. 200 runs) for further analysis.
The worst-case spill is normally defined as the spill event that contacted the longest stretch of coastline (above the 0.5 g.m-2 of beached oil concentration threshold). However, since the present set of simulations did not predict any oil beaching event above the considered threshold, the “worst-case” spill was qualitatively selected as the spill event for which the slick dispersion included most evident transport pathways towards the coast.
The section presents and interprets the results of the oil spill simulations. The stochastic results are presented first to outline the statistical oil dispersion and beaching patterns. Deterministic results for the worst-case scenario, i.e. the spill event for which the slick dispersion included most evident transport pathways towards the coast, are presented in a second section.
3.1 Stochastic modelling
The stochastic modelling approach allows definition of statistical metrics from the oil spill dispersion and beaching patterns, which capture the natural variability of the atmospheric and oceanic forcing fields within a given environment.
The time-independent mean surface oil distribution is show in Figure 3.1. The oil “distribution”
can also be interpreted as a probability of oil particle visitation in each cell. Note the considered oils have densities that are much lighter (~800 kg.m-3) than seawater (~1027 kg.m-
3), as such oil material, even if released near the seabed, will rise quickly to the surface layer, hence the focus on the “surface” oil dispersion pattern here.
The general oil dispersion footprint has an elliptic shape with a main north-south axis whose limits remain well off the New Zealand coastline (~100 km).
The oil beaching analysis indeed yields that no oil beaching over the considered threshold occurred for any of the simulated spill events. This outcome can be explained primarily by the local hydrodynamic and wind regimes at the site which are key drivers of the oil slick dispersion. The location of the Tawhaki-1 site is such that the discharged oil will be within the influence of the regional, northeast-directed Southland Current feature as shows on Figure 3.2 ( Soutelino, R.G. and Beamsley, B. 2015). Surface currents at the site are predominantly directed towards the east to northeast quadrant and therefore away from the coast (Figure 3.3). Visual assessment of oil dispersion pathways suggests the oil slick often connects with Southland Current further northwards and be advected along the coast rather than towards it. The oil transport away from the coast will be further reinforced by the wind action which is dominated by west to south-west directions (coming from) (Figure 3.4).
Figure 3.1 Time-independent mean surface oil distribution over the duration of the 200 oil spill events. The sum of all cell probabilities is 1. The probability density function was blanked below values of 1e-6 for plotting purposes (i.e. transparent).
Figure 3.2 Schematic representation of the main ocean circulation features off the east coast of New Zealand South Island. Main locations and oceanographic features of interest are highlighted (modified from Soutelino, R.G. and Beamsley, B. 2015). Position of Tawhaki-1 is shown by the red dot on left panel.
Figure 3.3 Residual current rose at Tawhaki-1 at nearbed, mid-depth and surface levels. Current direction are
Figure 3.4 Wind rose at Tawhaki-1. Wind direction are “coming from”.
3.2 Deterministic modelling
Although no event resulted in oil beaching above the 0.5g.m-2 threshold, some discrete events where subject to environmental conditions leading to the oil slick moving closer towards the coast. Based in visual assessment of time-independent surface dispersion footprint of all runs, the spill event starting on 2010-06-20 17:47 was selected as an example for further analysis.
Time independent mean patterns of surface oil dispersion over the simulation period (Figure 3.5) show a footprint with both northwest and northeast directed features.
The sequence of daily predicted oil plume positions over the initial 23-day simulation period (Figure 3.6 to Figure 3.8) show a clear initial north-directed pathway which veers north- eastwards off Oamaru as it connects with the regional Southland current feature (see similarity of patterns with Figure 3.2) and concurrently develops a southeast-directed dispersion components.
This example suggests that the regional Southland current will potentially shield the coast from any significant oil beaching by advecting the slick north-eastward and off the coast. This would allow oil weathering of the surface slick to continue over longer timer periods thereby further reducing the amount of oil in the system and potential for significant coastal impacts.
The oil mass budget over the duration of the event considered is shown in Figure 3.9. The amount of oil discharged expectedly increases over the initial 21 days (release period). Most of the oil is subject to dispersion (~82%) and the remaining oil mass (~18%) is evaporated (no oil beaching).
Figure 3.5 Time independent mean surface oil distribution averaged over the duration of worst-case spill scenario (starting 2010-06-20 17:47 ). The sum of all cell probabilities is 1. The probability density function was blanked below values of 1e-6 for plotting purposes (i.e. transparent).
Figure 3.6 Predicted daily oil spill positions throughout the simulation period from T0 to T0+7 days.
Figure 3.7 Predicted daily oil spill positions throughout the simulation period from T0+8 days to T0+15 days.
Figure 3.8 Predicted daily oil spill positions throughout the simulation period from T0+16 days to T0+23 days.
Figure 3.9 Oil mass budget over the duration of worst-case spill scenario (starting 2010-06-20 17:47 ). The plot summarizes the amount and fate of oil discharged through the simulation. The amount of oil is expressed as mass on the left axis and fraction of total oil released on the right axis. Time between successive vertical grid lines is 10 days.
The study characterises the oil dispersion and beaching patterns expected in the unlikely event of an accidental hydrocarbon spill in the oceanic environment during drilling operations at Tawhaki-1 site in the Great South Basin of New Zealand.
Since accidental discharge of hydrocarbon material can occur at any time and thus be subject to any combination of atmospheric and oceanographic forcings, the oil spill impact assessment must capture the natural forcing variability to derive robust statistical metrics on dispersion and beaching characteristics.
A stochastic oil spill modelling approach has been employed in the present study, which consisted of generating an extensive database of 200 oil spill trajectories, including oil weathering, with random start times within a 10-year period for which hydrodynamic, atmospheric and wave hindcast data was available. The oil spill modelling was undertaken using the open-source, peer-reviewed OpenDrift lagrangian modelling framework, which includes a dedicated oil spill trajectory and weathering module.
The general oil dispersion footprint has an elliptic shape with a main north-south axis whose limits remain well off the New Zealand coastline (~100 km). The oil beaching analysis indeed yields that no oil beaching over the considered threshold occurred for any of the simulated spill events. This outcome can be explained primarily by the local hydrodynamic and wind regimes at the site, which are key drivers of the oil slick dispersion. The location of the Tawhaki- 1 site is such that the discharged oil will be within the influence of the regional, northeast- directed Southland Current feature which limits oil dispersion pathways towards the coast. This is further reinforced by dominant west to south-west wind directions (i.e. away from the coast).
Afenyo M., Veitch B., and Khan F. 2016. “A State-of-the-Art Review of Fate and Transport of Oil Spills in Open and Ice-Covered Waters.” Ocean Engineering 119: 233–48.
Dagestad K.F, Röhrs J., Breivik O., and Ådlandsvik B. 2018. “OpenDrift v1.0: A Generic Framework for Trajectory Modelling.” Geoscientfic Model Development 11: 1405–1420.
ECMWF. 2019. “ERA5: Fifth Generation of ECMWF Atmospheric Reanalyses of the Global Climate.” https://cds.climate.copernicus.eu/cdsapp#!/home.
French-McCay, D. P. 2009. “State-of-the-Art and Research Needs for Oil Spill Impact Assessment Modelling.” In , 601–653. Ottawa.
Hersbach, et al. 2019. “Operational Global Reanalysis: Progress, Future Directions and Synergies with NWP- ERA5.” ERA Report Series 27. ERA Report.
Hole et al. 2019. “Revisiting the DeepWater Horizon Spill: High Resolution Model Simulations of Effects of Oil Droplet Size Distribution and River Fronts.” Ocean Science Discussion - in Review.
Johansen M. R., Reed M., and Bodsberg N. R. 2015. “Natural Dispersion Revisited.” Marine Pollution Bulletin 93 (1–2): 20–26.
Lebreton, LC, and Franz, T. 2013. “Trajectory Analysis of Deep Sea Oil Spill Scenarios in New Zealand Waters.” Prepared for Greenpeace New Zealand. Dumpark Ltd., Wellington.
Lehr, W., R. Jones, M Evans, D. Simecek-Beatty, and R Overstreet. 2002. “REvisions of the ADIOS Oil Spill Model.” Environmental Modelling and Software, 191–99.
Li Z., Spaulding M.L., and French-McCay, D. 2017. “An Algorithm for Modeling Entrainment and Naturally and Chemcally Dispersed Oil Droplet Size Distribution under Surface Breaking Wave Conditions.” Marine Pollution Bulletin, no. 119: 145–52.
MOS. 2019. “HYDROCARBON DISPERSION MODELLING - Numerical Modelling of the Dispersal and Coastal Impacts of Accidental Hydrocarbon Spills from Maui-8.” Report prepared for OMV New Zealand.
National Research Council (US) Committee on Oil in the Sea: Inputs, Fates, and Effects. 2003.
Oil in the Sea III: Inputs, Fates, and Effects. THE NATIONAL ACADEMIES PRESS.
Washington, D.C. https://www.ncbi.nlm.nih.gov/books/NBK220700/.
Owens, EH, and Sergy, GA. 1994. Field Guide to the Documentation and Description of Oiled Shorelines. Emergency Science Division. Canada: Environmental Technology Centre.
Ranold. 2019. “Blowout and Dynamic Wellkill Simulations Exploration Well Tawhaki-1.”
Röhrs et al. 2019. “The Effect of Vertical Mixing on the Horizontal Drift of Oil Spills.” Ocean Science 14: 1581–1601.
RPS APASA. 2015. “Financial Assurance Review: Oil Spill Modelling Study.” Prepared for Navigatus Consulting Q0344. RPS.
Saha, Suranjana, Shrinivas Moorthi, Hua-Lu Pan, Xingren Wu, Jiande Wang, Sudhir Nadiga, Patrick Tripp, et al. 2010. “The NCEP Climate Forecast System Reanalysis.” Bulletin of
the American Meteorological Society 91 (8): 1015–57.
Soutelino, R.G., and Beamsley, B. 2015. “The Influence of the Southland Current on Circulation Patterns within Pegasus Bay.” In , 7. Auckland, New Zealand.
Tkalich, P, and Chan, E. S. 2002. “Vertical Mixing of Oil Droplets by Breaking Waves.”
Marine Pollution Bulletin, no. 44: 1219–1229.
Tolman, H. L., and D. Chalikov. 1996. “Source Terms in a Third-Generation Wind Wave Model.”
Journal of Physical Oceanography 26 (11): 2497–2518.
Visser A. 1997. “Using Random Walk Models to Simulate the Vertical Dis-.” Marine Ecology- Progress Series, no. 158: 275–281.
Zelenke, B. et al, O’Connor, C. Barker, C.J. Beegle-Krause, and Eclipse L. 2012. “General NOAA Operational Modeling Environment (GNOME) Technical Documentation.” NOAA Technical Memorandum NOS OR&R 40. Seattle, WA: U.S. Dept. of Commerce,.
6. Appendix A
Table 6.1 Oil properties of Tawhaki-1 used in the oil spill simulations.
Oil Name Tawhaki-1
ADIOS Oil ID
Location NEW ZEALAND
Field Name Tui
Pour Point Min (K) 297.149994
Pour Point Max (K) 297.149994
Product Type crude
Asphaltene Content 0.006
Wax Content 0.178
Water Content Emulsion Emuls Constant Min Emuls Constant Max Flash Point Min (K) Flash Point Max (K)
Oil/Water Interfacial Tension (N/m) Oil/Water Interfacial Tension Ref Temp (K) Oil/Seawater Interfacial Tension (N/m) Oil/Seawater Interfacial Tension Ref Temp
(K) Density#1 (kg/m^3) 825
Density#1 Ref Temp (K) 288.149994
Density#1 Weathering 0
Density#2 (kg/m^3) 849
Density#2 Ref Temp (K) 288.149994
Density#2 Weathering 0.29
Density#3 (kg/m^3) 857
Density#3 Ref Temp (K) 288.149994
Density#3 Weathering 0.39
Density#4 (kg/m^3) 868
Density#4 Ref Temp (K) 288.149994
Density#4 Weathering 0.52
KVis#1 (m^2/s) 4.32E-06
KVis#1 Ref Temp (K) 303.149994
KVis#1 Weathering 0
KVis#2 (m^2/s) 2.46E-06
KVis#2 Ref Temp (K) 323.149994
KVis#3 (m^2/s) 1.74E-06
KVis#3 Ref Temp (K) 353.149994
KVis#3 Weathering KVis#4 (m^2/s) KVis#4 Ref Temp (K) KVis#4 Weathering KVis#5 (m^2/s) KVis#5 Ref Temp (K) KVis#5 Weathering KVis#6 (m^2/s) KVis#6 Ref Temp (K) KVis#6 Weathering DVis#1 (kg/ms) DVis#1 Ref Temp (K) DVis#1 Weathering DVis#2 (kg/ms) DVis#2 Ref Temp (K)
DVis#2 Weathering DVis#3 (kg/ms) DVis#3 Ref Temp (K) DVis#3 Weathering DVis#4 (kg/ms) DVis#4 Ref Temp (K) DVis#4 Weathering DVis#5 (kg/ms) DVis#5 Ref Temp (K) DVis#5 Weathering DVis#6 (kg/ms) DVis#6 Ref Temp (K) DVis#6 Weathering
Cut#1 Vapor Temp (K) 298.149994
Cut#1 Liquid Temp (K)
Cut#1 Fraction 0.027
Cut#2 Vapor Temp (K) 338.149994
Cut#2 Liquid Temp (K)
Cut#2 Fraction 0.07
Cut#3 Vapor Temp (K) 408.149994
Cut#3 Liquid Temp (K)
Cut#3 Fraction 0.213
Cut#4 Vapor Temp (K) 453.149994
Cut#4 Liquid Temp (K)
Cut#4 Fraction 0.31
Cut#5 Vapor Temp (K) 513.150024
Cut#5 Liquid Temp (K)
Cut#5 Fraction 0.429
Cut#6 Vapor Temp (K) 573.150024
Cut#6 Liquid Temp (K)
Cut#6 Fraction 0.569
Cut#7 Vapor Temp (K) 643.150024
Cut#7 Liquid Temp (K)
Cut#7 Fraction 0.714
Cut#8 Vapor Temp (K) 733.150024
Cut#8 Liquid Temp (K)
Cut#8 Fraction 0.886
Cut#9 Vapor Temp (K) 823.150024
Cut#9 Liquid Temp (K)
Cut#9 Fraction 0.948
Cut#10 Vapor Temp (K) Cut#10 Liquid Temp (K) Cut#10 Fraction
Cut#11 Vapor Temp (K) Cut#11 Liquid Temp (K) Cut#11 Fraction
Cut#12 Vapor Temp (K) Cut#12 Liquid Temp (K) Cut#12 Fraction
Cut#13 Vapor Temp (K) Cut#13 Liquid Temp (K) Cut#13 Fraction
Cut#14 Vapor Temp (K) Cut#14 Liquid Temp (K) Cut#14 Fraction
Cut#15 Vapor Temp (K) Cut#15 Liquid Temp (K) Cut#15 Fraction
Cut Units volume %
Oil Class group 1
Tox_EC(1)Species Tox_EC(1)24h Tox_EC(1)48h Tox_EC(1)96h Tox_EC(2)Species Tox_EC(2)24h Tox_EC(2)48h Tox_EC(2)96h Tox_EC(3)Species Tox_EC(3)24h Tox_EC(3)48h Tox_EC(3)96h Tox_LC(1)Species Tox_LC(1)24h Tox_LC(1)48h Tox_LC(1)96h Tox_LC(2)Species Tox_LC(2)24h Tox_LC(2)48h Tox_LC(2)96h Tox_LC(3)Species Tox_LC(3)24h Tox_LC(3)48h Tox_LC(3)96h Adhesion Benezene Naphthenes Paraffins Polars Resins
Reid Vapor Pressure Viscosity Multiplier
Conrandson Residuum Conrandson Crude Dispersability Temp (K) Preferred Oils