ABSTRACT

In ice covered waters, successful oil spill response requires predictions of where the oil and ice will travel. The International Association of Oil and Gas Producers (IOGP), Arctic Oil Spill Response Technology - Joint Industry Programme (JIP) funded research to improve oil spill response by leveraging new state-of-the-art ice forecasting into oil spill trajectory models. We present an overview of the systems and discuss how these advancements will provide responders with new information for spill preparedness and planning. The Nansen Environmental and Remote Sensing Center (NERSC) has developed two coupled ice-ocean models that cover the entire Arctic: TOPAZ4 and neXtSIM. TOPAZ4 uses both in situ ocean data and satellite data; the model also includes an ecosystem model. The neXtSIM model is a new high resolution (3km) coupled ice-ocean which uses daily sea ice thickness and concentration fields from satellites.

SINTEF’s Oil Spill Contingency and Response (OSCAR) model can now use output from both TOPAZ and neXtSIM. The OSCAR user can view the ice conditions with the spill, and the oil trajectory is modified by the time dependent ice coverages. Case studies will be discussed that test the implementation for different areas of the Arctic. Through these case studies, we provide new types of information for spill responders. The OSCAR model also includes information on oil weathering in ice from extensive laboratory and flume data for oils in water with and without ice.

  • Case Study 1: In the Beaufort Sea we compare observed ice drifter position time series with the ice drift calculated by the OSCAR model using input from the NERSC models. We then simulate a potential oil spill in the area.

  • Case Study 2: The 2009 Joint Industry Project included fieldwork and modeling for oil released in marginal ice zone in the Barents Sea. In May 2009, 7000 liters of fresh Troll oil was released into the marginal ice zone to study the oil weathering, spreading and overall oil trajectory.

Overall, the results of this study show that (1) using state-of-the art coupled ice-ocean forecast models is a key step in improving oil spill preparedness and response, and (2) using community data standards facilitate development of flexible operational systems.

INTRODUCTION

Oil spill planning, preparedness and response in ice covered waters are more complex than in temperate or tropical waters. The interaction between oil and sea ice is not simple. As shown in Error! Reference source not found., oil can reside on, in and under sea ice. In this project, we are considering the horizontal transport of oil in ice covered waters. In particular, the horizontal transport of surface oil which changes with varying ice concentrations. This JIP had two phases: (1) development of a new coupled ice-ocean model and (2) modification on the SINTEF OSCAR model to use the new coupled ice-ocean model output for simulating oil spill trajectories.

COUPLED ICE OCEAN MODELS

Numerical models of sea ice are commonly coupled ice-ocean models, using an atmospheric model or data as input. The ice and ocean dynamics in the coupled ice-ocean model leads to sea ice formation, movement and melting. Coupled ice-ocean models require a rheology (method for determining sea ice deformation and flow due to an applied stress) and a grid for discretization. Early models of sea ice used an “elastic-plastic” (EP) or “elastic-viscous-plastic” (EVP) model, such as the common Hibler (1977) sea ice model, which assumes that the interactions between ice grid cells increases as sea ice becomes thicker, and is still used today. The EVP rheology is in the NERSC TOPAZ (Towards an Operational Prediction system for the North Atlantic Zone) coupled ice-ocean model for the full Arctic, now in its fourth generation, TOPAZ4, (Sakov et al, 2012).

In the first phase of the JIP, NERSC initiated implementing the new neXtSIM sea ice model with an “elasto-brittle” (EB) ice rheology, intending to better represent the effects of increasing “damage” or numbers of cracks in sea ice (Girard et al, 2011). The EB model better simulates the long distances that stress can travel in high ice concentrations (>95%) and the heterogeneous nature of ice’s cracks and deformations. A novel moving finite element (triangular) model was also used for neXtSIM. The model triangles are reconfigured whenever the smallest angle of any triangle is smaller than 10° (Rampal et al, 2016), usually about once per hour.

The new ice rheology in neXtSIM has slowed down the ice flow, making ice simulations more realistic than the TOPAZ output fields, based comparison with ice drifters Girard et al. (2009); the neXtSIM rms errors for ice drift are 2.5 km/day in the U-component (East-West) and 2.3 km/day in the V component (North-South). This error accumulates – from OSCAR we estimated 138 km error over our full trajectories. These hindcast data sets of different than operationally produced data sets, e.g. for spill response, daily updated predictions that used the ice drifter data as an input should be examined. We expect that as in the DWH oil spill, utilizing (assimilating) more data in the area of the spill would improve forecasting, as occurred during the DWH as more observations became available for assimilation in the forecast cycles of the ocean circulation models.

Output from both TOPAZ and neXtSIM are available from NERSC. The neXtSIM output is interpolated onto the TOPAZ grid and formatted in network Common Data Format (netCDF), using the Climate and Forecast (CF) conventions. NetCDF is an open data standard that facilitates data sharing and development of general tools for data access and analysis. The CF conventions are agreed metadata conventions that facilitate data access and analysis (http://cfconventions.org/), i.e. a community of modelers has agreed to use “eastward_sea_water_velocity” and not “U” or “x_direction_currents”. This allows the model outputs to be shared for multiple purposes, e.g. many people may want to use the output for ocean currents and ice distributions beyond oil spills, e.g. for shipping, wind prediction, climate research, fisheries, etc. This also means OSCAR can now read any coupled ice-ocean model that provides output in netCDF CF compliant data. This is a key long-term view cost savings measure, which also encourages other potential coupled ice ocean data providers to also output netCDF compliant data.

WHY OPERATIONAL SEA ICE FORECASTING IS IMPORTANT

The Arctic Ocean is not the same from year to year. The Arctic-wide melt season has lengthened approximately by 1 day every 2 years since 1979 (Stroeve et al, 2014, see also Markus et al 2009). Open water absorbs more solar radiation than ice cover, so over the last decade, sea surface temperature (SST) in the Arctic has increased by 0.5–1.5°C, leading to delays in the timing of freeze-up initiation (Markus et al, 2009). This has led to changes in the overall ice thickness of Arctic sea ice1. Timing of migratory natural resources has changed with the sea ice, e.g. the autumn migration of Beluga whales has delayed an average of 4 days/year, based on data from 2008–2014 (Hauser et al, 2016). As the location migratory oil sensitive species changes, Environmental Sensitivity Index (ESI) maps may need to be revised with these resource timing shifts.

Though many people have heard of the El Niño / La Niña cycle in the Pacific, few are aware of similar interannual variability in the Arctic called the Arctic Oscillation (AO). Storm tracks change between the two AO phases leads to changes in winds, temperature and sea ice configurations. An interactive time series of the Arctic Oscillation Index can be found through U.S. National Oceanic and Atmospheric Administration (NOAA) National Weather Service / Climate Prediction Center – Arctic Oscillation2. Note that 2010 is the most extreme AO low since the record started in 1950. There has been a correlation between sea ice export through Fram Strait and the North Atlantic Oscillation, which is similar to the Arctic Oscillation. However, this relationship appears to be diminishing as sea ice thickness and amount has been decreasing in the 21st century (Jung and Hilmer, 2001). In order to appropriately sample variability, long (years) time series of environmental data are needed, and these may need to be updated based on significant transitions.

Access to highly peer-reviewed ocean circulation fields was a critical piece in the (NOAA) response to the Deepwater Horizon oil spill in 2010 (MacFadyen et al, 2011), and should be a goal for the Arctic. Over the entire Gulf of Mexico, there are different governing transport physics and scale: dynamics of freshwater entering the Gulf of Mexico from the Mississippi River; coastal circulation along varying coastal and continental shelf areas; and transport along the Louisiana-Texas shelf; the deep Gulf of Mexico circulation and the Loop Current. A freshwater lens overtop of seawater can act as a barrier to oil transport, but is at such a scale that operational circulation models do not simulate the phenomena. Access to some of models had previously been arranged, but NOAA standardizing with netCDF CF compliant output accessible via Thematic Real-time Environmental Distributed Data Services (THREDDS), new model output could quickly be added to the operational suite. NOAA trajectory analysts were able to switch among these models for inputs into the trajectory model (Beegle-Krause et al, 2007), so the input circulation model could be altered as the spill evolved and moved into different areas.

This JIP’s support of the netCDF CF standard for implementation of the coupled iceocean models means that similar flexibility is now available for trajectory modeling in the Arctic. Spill scenarios are each unique, and may require different inputs for different areas, e.g. near coastal river input, man-made structures (e.g. ice islands) or open waters, such as the Marginal Ice Zone. A short survey of coupled ice-ocean models yielded:

Development and improvement in Arctic sea ice model is continuing. The Forum for Arctic Modeling & Observational Synthesis (FAMOS3) project focuses on international collaboration toward inter-comparison and improvement in coupled ice-ocean modeling. Though focused on climate change time-scales, their work on improving forcing fields (atmosphere, ecosystem, land, ocean, rivers, sea ice) and process parameters (convection, eddies, ice formation, ice decay, ice ridging, mixing, tides, biological productivity) is applicable to all coupled ice-ocean models (Proshutinsky et al, 2011).

Leveraging these international efforts is important, as interest in the Northern Sea Route and the Northwest Passage is world-wide. These international efforts bring scientists of different countries together, leading to trust in the developed systems. During the DWH oil spill, NOAA had visitors from many Caribbean nations, who representatives wanted to understand why they should trust the trajectory forecasts.

MODELING THE TRAJECTORY OF OIL IN ICE

There has been a general “rule of thumb” for oil movement among ice floes since the 1970s with differing concentrations (or coverages) of sea ice, which was first published in El Tahan et al (1988):

  • For ice concentration of less than 3/10ths, the oil and ice drift independently,

  • For ice concentrations between 3/10ths and 6/10ths or 7/10ths, there is a transition with the ice drift increasingly controlling the oil drift, and

  • For ice concentrations above 6/10ths to 7/10ths, the ice completely controls the oil movement (references vary on this boundary).

The transition between 3/10ths to 7/10ths has been left ambiguous and has been assumed to be linear for applications to oil spills as discussed in theory by Ventatesh et al (1990) and El-Tahan et al (1988). There is no specific calibration for this algorithm, though one would expect increasing ice concentrations to increasingly control any oil on the ocean’s surface. Future research is needed on the transition between the dominance of waves in open water to ice covered waters, which also modifies the water column turbulence regime between wind/wave dominated and ice dominated.

SINTEF has previously organized and lead two large Joint Industry Projects focussed on experimental releases of oil in the marginal ice zone in the Barents Sea with modeling components. The first project, in 1993, included 3 releases of oil: 26 m3 at the surface in a broken ice field, and two releases of 1 m3 each under the ice cover (Sørstrøm et al, 1994). Numerical modelling was used to evaluate real time forecasting capabilities and the change in drift angle as a function of wind speed and ice thickness (Reed and Aamo, 1994). The project also resulted in improvements in oil weathering models, relating the effect of ice cover on evaporation, entrainment, and emulsification (Reed et al, 1995). The JIP in 2009 involved extensive laboratory and mesoscale work in preparation for the actual field experiment (Sørstrøm et al, 2010).

SINTEF also participated in the EU Arctic Climate Change, Economy and Society (ACCESS) project, comparing a long data set from the SINTEF SINMOD coupled ice-ocean model, which includes a biological model (Slagstad, 1981 and Ellingson et al, 2009). Selected oil spill scenarios were run under modern (2009–2013) and future (2050–2054) Arctic ice conditions (Nordam et al, 2016).

RESULTS

Case Study 1

Ten ice drifter tracks from the International Arctic Buoy Programme (IABP) were simulated using the EB model data in OSCAR. Trajectories all started on 1 December 2007. At that time, all drifters were located in the Beaufort Sea / Arctic Ocean (74–85°N, 96–165°W). Comparing the actual drifter track with the simulated track showed small deviations: after 150 days deviations were between 72 and 191 km (Figure 2). The mean difference between the OSCAR simulated trajectories using the EB model and NERSC simulated trajectories was 1.7 km with variance of 0.9 km.

Figure 1.

Season differences in oil / ice interactions.

Figure 1.

Season differences in oil / ice interactions.

Figure 2.

Observed 150 day drifter track (blue) overlaid on OSCAR drifter track (white) using the EB model output. This demonstrated that the OSCAR model is correctly reading the buoy data.

Figure 2.

Observed 150 day drifter track (blue) overlaid on OSCAR drifter track (white) using the EB model output. This demonstrated that the OSCAR model is correctly reading the buoy data.

Case Study 2

SINTEF has field data from the 2009 JIP which includes ice and oil drift information (Sørstrøm et al, 2010). The field experiment in May 2009 included ice drift data and the location of the oil within the ice (Daae et al, 2011). Seven cubic meters of Troll fresh crude was released between the ice floes in the marginal ice zone. Ice drift and ice field deformation data were recorded using a large number of GPS recorders in and around the oil slick. The original simulations were carried out with OSCAR Version 6.0, which did not include the observed movement of the ice. Thus ice and oil drift were computed by the model based on wind and current data alone.

Here we compare those results with results using ice drift computed by the NERSC model. Note that the overall drift of the ice was recorded as a compilation of GPS signals from an array of GPS receivers planted on the ice surface. In Figure 3, we show (a) the GPS positions of the ice camp, (b) OSCAR simulated drift using EB model implementation, and (c) OSCAR simulated drift using SINMOD. The tracks all have similar shape and extent. One note is that for the EB model, the start location was moved northward in order to find ice, as the area of the 2009 fieldwork in the EB was open water. We plan to go back to the original 2009 data to determine the dispersion of the ice floes from the mean path in order to estimate the path variance, to better compare our implementation with the NERSC data. Note that neXtSIM is an Arctic wide model, while SIMOD is SINTEF model that has been implemented, tuned and tested in the Barents Sea over many years.

Figure 3.

Evaluation of coupled ice-ocean models for simulation of the 2009 JIP ice field positions. (a) GPS positions of the drift of the field camp over the experimental period. (b) Northward displaced simulation of the 2009 JIP field camp positions using the NERSC data. (There is no ice in the field area in the TOPAZ data). (c) Simulation of the 2009 JIP field camp positions with the SINTEF SINMOD coupled ice ocean model.

Figure 3.

Evaluation of coupled ice-ocean models for simulation of the 2009 JIP ice field positions. (a) GPS positions of the drift of the field camp over the experimental period. (b) Northward displaced simulation of the 2009 JIP field camp positions using the NERSC data. (There is no ice in the field area in the TOPAZ data). (c) Simulation of the 2009 JIP field camp positions with the SINTEF SINMOD coupled ice ocean model.

CONCLUSION

Through this two-part JIP, oil spill trajectory modeling has been improved through development of an advanced coupled ice-ocean forecast model by NERSC and implementation of this model to control surface oil transport in the SINTEF OSCAR model.

In the first phase, NERSC developed an operational coupled ice-ocean forecasting system to improve upon the existing TOPAZ system through development of the neXtSIM coupled ice-ocean model. neXtSIM uses an EB ice rheology and moving triangular mesh grid. Compared to TOPAZ, the neXtSIM reduced ice drift rms errors for ice drift to 2.5 km/day in the U-component (East-West) and 2.3 km/day in the V component (North-South), by decreasing the currents to more realistic values. neXtSIM model output has been made available via netCDF CF compliant model output, and use of netCDF CF compliant coupled ice ocean model output has been implemented in the OSCAR model. OSCAR can now read data from any coupled ice ocean model that exports data in netCDF CF complaint format. The drifter comparison between the OSCAR model and NERSC simulations over the full simulations was 1.7 km with variance of 0.9 km. This is a significant step in improving operational oil spill trajectory forecasting for the Arctic and Antarctic, because the SINTEF OSCAR model can use TOPAZ, neXtSIM and add any other model similarly formatted.

The OSCAR model has been shown to properly read and display the neXtSIM output. We compare ice drifters and fieldwork from the 2009 Oil in Ice JIP, and show improvement by using neXtSIM. The traditional oil spill “rule of thumb” for ice control of oil movement has been implemented in OSCAR: below 3/10th ice cover, the oil and ice are independent, while at 6/10th to 7/10th ice cover, the ice completely controls the oil. Research on how increasing ice concentrations in the MIZ change dynamics nonlinearly to create this nonlinear rule of thumb would be the next step in improving oil spill trajectory forecasts using coupled ice ocean model output.

REFERENCES

REFERENCES
Ashik
and
Kulokov
(
2012
)
“Ice-ocean coupled model for operational predictions of sea ice and sea level conditions in the Arctic Ocean marginal seas”
,
poster4
.
Beegle-Krause
,
C.J.
,
C.
O’Connor
and
G.
Watabayashi
(
2007
)
“NOAA Safe Seas Exercise 2006: new data streams, data communication and forecasting capabilities for spill forecasting”
Proceedings of the 30th AMOP Technical Seminar
,
Edmonton, Alberta, Canada
,
June 5–7, 2007
.
Ottawa, Ont.
:
Environment Canada
.
Bertino
,
L.
K.A.
Kisæter
,
S.
Scient
(
2008
)
The TOPAZ monitoring and prediction system for the Atlantic and Arctic Oceans
.
Journal of Operational Oceanography
1
:
1
18
.
Daae
,
R.L.
,
L.G.
Faksness
,
I.
Durgut
,
P.J.
Brandvik
and
F.
Leirvik
(
2011
)
“Modelling of oil in ice with OSCAR”
.
Report no. 35 from the 2009 Oil in Ice JIP, SINTEF Report A 19804
,
17
pp
.
Ellingsen
,
I.H.
,
Slagstad
,
D.
,
Sundfjord
,
A.
(
2009
).
Modification of water masses in the Barents Sea and its coupling to ice dynamics: a model study
.
Ocean Dynamics
.
doi:10.1007/s10236-009-0230-5
.
El-Tahan
,
M.
,
G.
Comfort
, and
R.
Abdelnour
,
1988
.
“Development of a Methodology for Computing Oil Spill in Ice-Infested Waters.”
In
Atmospheric Environmental Service
,
90
+pp
.
Girard
,
L.
,
S.
Bouillon
,
J.
Weiss
,
D.
Amintrano
,
T.
Fichefet
,
V.
Legat
(
2011
)
“A new modeling framework for sea-ice mechanics based on elasto-brittle rheology
.
Annals of Glaciology
52
(
57
)
123
132
Hauser
,
D.D.W.
,
K.L.
Laidre
,
K.M.
Stafford
,
H.L.
Stern
,
R.S.
Suydam
,
P. R.
Richard
(
2016
)
“Decadal shifts in autumn migration timing of Pacific Arctic beluga whales are related to delayed annual sea ice formation.”
Global Change Biology
,
DOI 10.1117gcb.13564
.
Hibler
,
W.D.
1977
.
“A viscous sea ice law as a stochastic average of plasticity”
Journal of Geophysical Research
82
:
3932
3938
.
Hibler
,
WD.
1977
.
‘A viscous sea ice law as a stochastic average of plasticity’
,
Journal of Geophysical Research
,
82
:
3932
38
.
Jung
,
T.
, and
M.
Hilmer
(
2001
)
“The Link between the North Atlantic Oscillation and the Arctic Sea Ice Export through Fram Strait”
Journal of Climate
14
:
3932
3943
.
MacFadyen
,
A.
,
G.Y.
Watabayashi
,
C.H.
Barker
and
C.J.
Beegle-Krause
(
2011
)
“Tactical modeling of surface oil transport during the Deepwater Horizon spill response.”
In
Monitoring and Modeling the Deepwater Horizon Oil Spill: A Record Breaking Enterprise
.
American Geophysical Union
,
pages
167
178
.
Markus
,
T.
,
J. C.
Stroeve
and
J.
Miller
(
2009
)
“Recent changes in Arctic sea ice melt onset, freezeup, and melt season length”
Journal of Geophysical Research
,
DOI 10.1029/2009JC005436
.
Nordam
,
T.
,
D.
Donnebier
,
C.J.
Beegle-Krause
,
M.
Reed
,
D.
Slagstad
(
2016
)
“Impact of Climate Change and Seasonal Trends on the Fate of Arctic Oil Spills
.
Ambio
,
in press
.
Posey
,
P.G.
,
E.J.
Metzger
,
A.J.
Wallcraft
,
O.M.
Smedstad
,
M.W.
Phelps
(
2009
)
“Real-time Data Assimilation of Ice Concentration into a Coupled Ice/Ocean Forecast System”
IEEE OCEANS 2009 conference paper5
.
Proshutinski
,
A.
,
Y.
Aksenov
,
J. C.
Kinney
,
R.
Gerdes
,
E.
Golubeva
,
D.
Holland
,
G.
Holloway
,
A.
Jahn
,
M.
Johnson
,
E.
Popova
,
M.
Steele
,
E.
Watanabe
(
2011
).
“Recent Advances in Arctic Ocean Studies Employing Models from the Arctic Ocean Model Intercomparison Project”
Oceanography
24
(
3
)
doi 10.5670/oceanog.2011.61
.
Rampal
,
P.
,
D.
Bouillon
,
E.
Ólason
,
Mathieu
Morlighem
(
2016
)
“neXtSIM: a new Lagrangian Sea ice model”
The Cryosphere
10
:
1055
1073
.
Reed
,
M.
and
O. M.
Aamo
,
(
1994
).
“Real-time forecasting for an experimental oil spill in the Arctic marginal ice zone.”
Spill Science and Technology, Pergamon Press
(
1
)
1
:
69
-
77
.
Reed
,
M.
,
O.M.
Aamo
,
K.
Downing
,
P.S.
Daling
,
I.
Singsaas
,
T.
Strøm-Kristiansen
,
A.
Lewis
,
(
1995
)
“ESCOST task C1.2 Modelling of weathering of oil: laboratory, flume and field.”
IKU Report 41.5100.00/09/95. p
.
22
(
Restricted
)
Sakov
,
P.
,
F.
Counillon
,
L.
Berino
,
K.A.
Lisæter
,
P.R.
Oke
, and
A.
Korablev
(
2012
)
“TOPAZ4: an ocean-sea ice data assimilation system for the North Atlantic and Arctic”
Ocean Science
8
:
633
656
.
Slagstad
,
D.
(
1981
).
Modeling and simulation of physiology and population dynamics of copepods. Effects of physical and biological parameters
.
Modeling, Identification and Control
,
2
,
119
162
.
Sørstrøm
,
S.E.
,
Ø.
Johansen
,
S.
Vefsnmo
,
S.M.
Løyås
(
1994
).
Experimental oil spill in the Marginal Ice Zone, April 1993
.
SINTEF report 22.2120.00/02/94
.
Sørstrom
,
S.E.
,
P.J.
Brandvik
,
I.
Buist
,
P.
Daling
,
D.
Dickens
,
L.-G.
Faksness
,
S.
Potter
,
J.F.
Rasmussen
, and
I.
Singsas
(
2010
)
“Joint industry program on oil spill contingency for Arctic and ice-covered waters.”
SINTEF A14181
.
Stroeve
,
J.C.
,
M.
Thorsten
,
B.
Linette
,
J.
Miller
, and
A.
Barret
(
2014
)
“Changes in Arctic melt season and implications for sea ice loss.”
Geophysical Research Letters
41
:
1216
1225
.
Venkatesh
,
D.
, and
H.
El-Tahan
,
G.
Comfort
, and
R.
Abdelnour
(
1990
).
“Modeling the behavior of oil in ice infested waters.”
Atmosphere-Ocean
28
:
303
329
.
Wang
,
J.
,
H.
Hu
,
D.
Schwab
,
G.
Leshkevich
,
D.
Beletsky
,
N.
Hawley
,
A.
Clites
(
2010
)
“Development of the Great Lakes Ice-Circulation Model (GLIM): Application to Lake Erie in 2003–2004”
Journal of the Great Lakes
36
(
3
):
425
436
.
Yang
,
Q
,
Liu
J
,
Zhang
Z
,
Wu
H
,
Li
Q
and
Xing
J
(
2011
)
[A preliminary study of the Arctic sea ice numerical forecasting: coupled sea ice–ocean modelling experiments based on MITgcm]
.
Chinese J. Atmos. Sci.
,
35
(
3
),
473
482
[in Chinese]
.
Yang
,
Q
,
Li
C
,
Xing
J
,
Li
Q
,
Zhang
L
and
Li
M
(
2012
)
[Arctic sea ice forecasting experiments in the summer of 2010]
.
Chinese J. Polar Res.
,
24
(
1
),
87
94
[in Chinese]
.

1 Animations of changing sea ice export and ice thickness are available from NASA Goddard’s Scientific Visualization Studio (https://www.nasa.gov/feature/goddard/2016/arctic-sea-ice-is-losing-its-bulwark-against-warming-summers).

3 http://www.whoi.edu/projects/famos/ Formerly the Arctic Ocean Model Intercomparison Project (AOMIP). http://www.whoi.edu/projects/AOMIP/MIP