Implications of a fuel spill in the Bering and Chukchi Seas for two sentinel marine species: a simulation study
Data files
Jun 29, 2026 version files 58.09 GB
-
analysis_code.zip
147.39 MB
-
Model_Output.zip
14 MB
-
PolarBearRisk-TankerSpills_Final_Report.pdf
7.92 MB
-
README.md
11.67 KB
-
Scen1.zip
20.45 GB
-
Scen2.zip
2.01 GB
-
Scen3.zip
18.38 GB
-
Scen4.zip
1.30 GB
-
Scen5.zip
15.79 GB
Abstract
Sea-ice decline has opened Arctic shipping routes lengthening the maritime transit season. Along the Northern Sea Route (NSR), vessel traffic is projected to increase primarily from natural resource development-related traffic. The NSR traverses the range of the Chukchi Sea polar bear (Ursus maritimus) subpopulation, and Pacific walrus (Odobenus rosmarus divergens) haulouts dot Russian islands and mainland, the Alaska coast, and the Bering Strait along the NSR. Oil spills are a primary threat to marine Arctic wildlife from maritime traffic with the potential to affect many individuals. We estimated the number of polar bears and walruses potentially exposed to oil from a Chukchi Sea or Bering Strait ship-based spill. We simulated five large-scale boat-based spills in two NSR locations in August and October for crude and marine gas oil (MGO). We overlaid polar bear radiocollar locations onto trajectories to estimate the number exposed. If oil was ≤0.5 nm from a haulout, we presumed all walruses at the haulout were exposed. We report the most impactful scenario to help managers and emergency responders understand the upper levels of impact to these species and inform response planning. Crude oil remained on the surface for the entire 60-day simulations and >20% stranded onshore. Little surface MGO remained after 5 days or reached the shoreline. The Chukchi October crude spill exposed 24% of radiocollared polar bears to lethal and sublethal oil, or 711 (95% CI = 438–1,052) individuals as a proportion of the total population (n = 2,937). During the Chukchi August crude spill, ~183,702 walruses were exposed to lethal levels of surface oil. No polar bears or walrus haulouts were exposed to lethal levels of MGO from a spill. Our simulations suggest an oil spill would contaminate large areas of the marine environment and could have significant consequences for polar bears and walruses. The inability to conduct timely, effective cleanup underscores the importance of understanding and attempting to mitigate an Arctic oil spill. Spill preparation is expensive, time-consuming, and increasingly challenging geopolitically, yet a lack of preparation could have catastrophic consequences to marine life, regional subsistence users, and the fishing and tourism industries.
Dataset DOI: 10.5061/dryad.3n5tb2rz6
The data repository contains the following files needed for running the oil spill model or reviewing the results of the analysis.
analysis_code.zip
Model_Output.zip
Scen1.zip
Scen2.zip
Scen3.zip
Scen4.zip
Scen5.zip
Polar Bear Risk-Tanker Spills_Final_Report.pdf
Description of folder and file contents
Below we provide descriptions of the files contained in each zipped folder.
analysis_code.zip
This folder contains the following functions that are needed to run the spill simulation and analyze/summarize the results.
crawl_results.RDS
Estimated daily polar bear trajectories derived from 'movement_data.RDS" using the correlated random walk (CRAWL) model from Johnson et al. (2008) to accommodate location uncertainty in our oil spill analysis. It includes 25 realized paths for each polar bear in our data set.
- crawl_results.RDS is a list of 110 elements (1 for each bear) with 25 elements within each of the 110 representing that individual's set of 25 CRAWL realizations. Each realization data.frame object has the following column names:
- id: Represents the specific bear-year path
- mu.x: Represents the estimated x-coordinates from the CRAWL procedure. Coordinates are in NSIDC North Pole Stereographic projection.
- mu.y: Represents the estimated y-coordinates from the CRAWL procedure. Coordinates are in NSIDC North Pole Stereographic projection.
- datetime: Represents the specific date and time that the estimated locations correspond to.
movement_data.RDS
Raw polar bear data needed to develop "crawl" paths for use with the oil spill simulation model.
- movement_data.RDS is a data.frame object containing the raw location data used to develop the CRAWL paths used in the analysis. The data.frame contains the following columns:
- x: x-coordinate for observed polar bear locations used in CRAWL analysis. Coordinates are in NSIDC North Pole Stereographic projection
- y: y-coordinate for observed polar bear locations used in CRAWL analysis. Coordinates are in NSIDC North Pole Stereographic projection
- date: Date and time associated with a given observed polar bear location
- Loctype: Categorical classification of whether the location is an Argos or GPS-derived location
- lc94: Location quality index obtained for each Argos-based location
- Semi.major.axis: Parameter to define the error ellipse for Argos-based locations (i.e., length of the ellipse)
- Semi.minor.axis: Parameter to define the error ellipse for Argos-based locations (i.e., height of the ellipse)
- Ellipse.orientation: Parameter to define the error ellipse for Argos-based locations (i.e., orientation of ellipse in space)
haulout_use_months.csv
Walrus haulout data used to estimate impacts of simulated spills on walrus in the region.
- haulout_use_months.csv is a .csv file with information regarding each of the walrus haulouts considered in our analysis. The .csv file has the following columns:
- Name: Name of the walrus haulout
- start_date: Month the haulout typically forms
- end_date: Month the haulout typically disbands
- peak: Month that the peak number of walrus typically occurs
- start: The first date of the start month from 'start_date'
- end: The last date of the end month from 'end_date'
- max_occupancy: The maximum number of walrus expected at each haulout
- lon: Longitude of haulout in WGS 1984 projection
- lat: Latitude of haulout in WGS 1984 projection
- regular?: An indicator of whether the haulout is regularly used or just occasionally used
- include?: An indicator of whether the haulout should be used in our analysis or not
- comment: Additional information about each haulout (if available)
crawl_oilspill.R
Code used to develop CRAWL paths.
bear.oil.func.R
Function to interaction polar bear movement trajectories (i.e., CRAWL) with the spread of simulated oil spills to find the number and timing of bears coming within a given distance of simulated oil.
spill_analysis_code.R
Code to estimate the number of polar bears that potentially interact with simulated oil.
walrus_oil_code.R
Code to estimate the number of walrus at haulouts that could potentially interact with simulated oil.
oil.bootstrap.func.R
Function to summarize the outputs of the spill simulations interacted with polar bear movement data using bootstrapping.
bootstrap_summary_code.R
Code to summarize results using bootstrapping to ensure uncertainty introduced by variation in spill trajectories, animal movements, and across animals is properly propagated.
Model_Ouput.zip
Contains folders with output from "spill analysis" code for each of the five spill scenarios (e.g., 'Scen1') broken down into results for shoreline oil or oil on the ocean's surface (i.e., 'surface' oil). There is also a file folder containing the bootstrap summary results for each of the scenarios, saved as '.RDS' files. Finally, there is a folder for the output of the walrus haulout analysis (i.e., "walrus_oil_code.R") for each of the five spill scenarios.
- Each ScenX folder within Model_Output.zip contains 100 110 x 481 matrices, saved as .RDS files with 50 files for shore oil (e.g., Shore_Spillets_XX.RDS, where XX represents the spill iteration) and 50 files for surface oil (e.g., Spillets_XX.RDS, where XX represents the spill iteration), where each row represents one of the 110 polar bears with movement data in this study. Each column represents each 3-hour period for the full duration of the simulated spill. Cell values represent the greatest concentration of oil within 1 km of the polar bear at each time step.
- Within the 'Walrus_Output' folder, there are 5 .RDS files associated with the walrus outputs for each of the 5 spill scenarios. Within each scenario's output file, there is a list of 50 matrices (1 for each spill iteration) with rows associated with individual walrus haulouts, and columns representing each 3-hour period for the full duration of the simulated spill. Cell values represent the maximum oil concentration of surface oil within 0.5 km of a given haulout.
- Within the 'Summaries" folder, there are 5 .RDS files associated with each of the 5 spill scenarios. Within each .RDS file, there is a list object with 5 lists representing (in numeric order) the mean and median cumulative number of bears exposed to >= 1 g/m2 surface oil, >= 10 g/m2 surface oil, >= 100 g shoreline oil, any shoreline oil, and any oil, across all the 3-hour periods of the simulated spill.
Scen#.zip folders
Scen1.zip
A folder containing 50 .gpkg files for the 50 spill iterations for spill scenario 1; Crude oil released from a ship in August along the Chukotkan coast of Russia.
Scen2.zip
A folder containing 50 .gpkg files for the 50 spill iterations for spill scenario 2; Marine grade oil released from a ship in August along the Chukotkan coast of Russia.
Scen3.zip
A folder containing 50 .gpkg files for the 50 spill iterations for spill scenario 3; Crude oil released from a ship in October along the Chukotkan coast of Russia.
Scen4.zip
A folder containing 50 .gpkg files for the 50 spill iterations for spill scenario 4; Marine grade oil released from a ship in October along the Chukotkan coast of Russia.
Scen5.zip
A folder containing 50 .gpkg files for the 50 spill iterations for spill scenario 5; Crude oil released from a ship in October in the Bering Strait.
- Each ScenX.zip folder contains 50 realizations of the oil spill process. For each spill realization (i.e., files ending in '.gpkg'), there are three separate layers:
- Shore_Spillets: These are the modeled oil accumulations that end up on the coastline
- Each spill shoreline realization contains the following attributes. If labeled as NULL, it indicates that this variable was not relevant/needed for the current analysis:
- SpilletNum: NULL
- DateTime: Date and time (UTC) associated with the given spillet
- Type: An indicator for the type of oil. For shore spillets this always equals 3
- Mass_MT: Metric tons of oil represented by the spillet
- Density_gCM3: Density of oil represented by the spillet in grams per cubic cm
- Viscocity_cP: NULL
- WaterFraction: The proportion of water mixed into the spilled oil
- UnderWater_Hrs: NULL
- Diameter_M: NULL
- SumARMass_MT: NULL
- SumALMass_MT: NULL
- Each spill shoreline realization contains the following attributes. If labeled as NULL, it indicates that this variable was not relevant/needed for the current analysis:
- Spillets: These are the modeled oil accumulations that remain offshore in the water column or water surface
- Each offshore spill realization contains the following attributes:
- SpilletNum: A unique numeric value assigned to each spillet to track its status through time
- DateTime: Date and time (UTC) associated with the given spillet
- Type: An indicator for the type of oil. For shore spillets this always equals 0
- LocFlag: This is not a relevant parameter for this analysis, but is an artifact of the oil spill modelling framework as applied to other research topics.
- Radius_M: Radius (m) of the spillet
- Mass_MT: Metric tons of oil represented by the spillet
- Density_gCM3: Density of oil represented by the spillet in grams per cubic cm
- Viscosity_cP: Measure of oil viscosity in centipoise (cP)
- WaterFraction: The proportion of water mixed into the spilled oil
- Age_Hrs: How long the spillet is at a given time-step
- Depth_M: NULL
- FloatConc_mtM2: Concentration of floating oil (metric tons per meter squared)
- SubseaConc_mgM3: NULL
- WaterDepth_M: This is not a relevant parameter for this analysis, but is an artifact of the oil spill modelling framework as applied to other research topics.
- VertSpread_M: NULL
- DissolvedConc_mgM3: NULL
- Underwater_Hrs: NULL
- Diameter_M: NULL
- SumARMass_MT: NULL
- SumALMass_MT: NULL
- DemersalExp: NULL
- EnvCompartment: This is not a relevant parameter for this analysis, but is an artifact of the oil spill modelling framework as applied to other research topics.
- Each offshore spill realization contains the following attributes:
- Shore_Spillets: These are the modeled oil accumulations that end up on the coastline
- SpillPoint: Provides the location of where the simulated spill associated with the file occurred
- Each spill point has the following attributes:
- ConversionDate: Date at which the spill simulation was initially run
- Scenario: Label indicating which scenario was associated with this spill
- Lon: Longitude of spill initiation site (WGS 1984 projection)
- Lat: Latitude of spill initiation site (WGS 1984 projection)
- SpillDuration_Hrs: Duration (hours) that oil was actively being released into the environment
- Volume_bbl: Volume of oil (barrels) released into the environment associated with the spill
- StartDate: Date the spill was simulated to begin
- SimDuration_Days: Number of days the simulation tracked the spilled oil
- OilName: Type of oil simulated to have been spilled
- Each spill point has the following attributes:
Lastly, "PolarBearRisk-TankerSpills_Final_Report.pdf" is a .pdf that goes into more depth on the underlying oil spill model supporting our work.
***Please note, the file names above have a different numbering convention than how they are presented in the manuscript.
Scen1.zip is listed as Scenario 1 in the manuscript
Scen2.zip is listed as Scenario 3 in the manuscript
Scen3.zip is listed as Scenario 2 in the manuscript
Scen4.zip is listed as Scenario 4 in the manuscript
Scen5.zip is listed as Scenario 5 in the manuscript
