Contents lists available at ScienceDirect
Marine Pollution Bulletin
journal homepage: www.elsevier.com/locate/marpolbul
A standardised approach to the environmental risk assessment of potentiallypolluting wrecksFreya Goodsira,⁎, Jemma A. Lonsdalea, Peter J. Mitchella, Roxana Suehringb, Adrian Farcasa,Paul Whomersleya, Jan L. Branta, Charlotte Clarkea, Mark F. Kirbya, Matthew Skelhornc,Polly G. Hillca Cefas, Lowestoft Laboratory, Pakefield Road, Lowestoft, Suffolk NR33 0HT, UKbDepartment of Environmental Science and Analytical Chemistry (ACES), Stockholm University, SE-106 91 Stockholm, Swedenc Salvage and Marine Operations, Ash 2b #3212, MOD Abbey Wood, Bristol, BS34 8JH, UK
A R T I C L E I N F O
Keywords:Potentially polluting wrecksOilRisk assessment
A B S T R A C T
The potential risk to the marine environment of oil release from potentially polluting wrecks (PPW) is in-creasingly being acknowledged, and in some instances remediation actions have been required. However, wherea PPW has been identified, there remains a great deal of uncertainty around the environmental risk it may pose.Estimating the likelihood of a wreck to release oil and the threat to marine receptors remains a challenge. Inaddition, removing oil from wrecks is not always cost effective, so a proactive approach is recommended toidentify PPW that pose the greatest risk to sensitive marine ecosystems and local economies and communities.This paper presents a desk-based assessment approach which addresses PPW, and the risk they pose to en-vironmental and socio-economic marine receptors, using modelled scenarios and a framework and scoringsystem. This approach can be used to inform proactive management options for PPW and can be appliedworldwide.
1. Introduction
Oil can cause significant damage to the environment (Kingston,2002; Rogowska and Namieśnik, 2010) due to its toxic and persistentnature (Burns et al., 1994; Kingston, 2002; Peterson et al., 2003;Hailong and Boufadel, 2010). There are several sources from which oilcan enter the marine environment, including the natural slow release ofoil from the sea floor (seeps), oil and gas industry activities, shippingincidents and terrestrial based activities (i.e. run off and discharges).
There is an increasing concern over the pollution risk posed bypotentially polluting wrecks (PPW) (Michel et al., 2005; Faksness et al.,2015), as it is estimated that between 2.5 and 20.4 million metrictonnes (Michel et al., 2005) of oil may remain as bunker and/or cargooil within these wrecks. High profile incidents of oil release fromwrecks such as the S.S. Jacob Luckenbach (California, USA) and HMSRoyal Oak (Orkney, UK) have raised awareness of the potential of PPWas an oil polluter to the marine environment (Alcardo et al., 2007).With approximately 8500 identified wrecks worldwide, there isgrowing concern that these represent a significant source and risk ofpollution and thus pose a considerable level of threat to the marine
environment (Michel et al., 2005; Etkin et al., 2009).World War II wrecks have been submerged for over 70 years, so
pose increased potential of releasing oil due to the increased likelihoodof structural failure from corrosion of metal, compared to youngerwrecks. It is not possible to mitigate risk via intervention (e.g. removalof the oil) for all known wrecks, therefore there is a need to understandthe risk each of them pose. The first step is to understand the amountand composition of oil remaining on board and the likelihood of oilrelease to occur, to help prioritise efforts.
The oil types present in wrecks vary from heavy, persistent oils (e.g.bunker oils) to lighter, non-persistent oils (e.g. diesels or petrol). Heavyoils persist in the environment as they do not degrade readily (Law andMoffett, 2011), whereas light oils tend to degrade more quickly as theyare more bioavailable due to the small size of the compounds (ENSR,2006). Furthermore, there are a number of physical, biological andchemical factors which contribute to the weathering of oil (e.g. eva-poration, dispersion, sedimentation, microbial degradation) (Wang andFingas, 1995; Fingas, 1999; National Research Council, 2002; Plataet al., 2008), which can considerably alter the composition of oil in themarine environment (Liu et al., 2012). Oils contained in shipwrecks
https://doi.org/10.1016/j.marpolbul.2019.03.038Received 9 October 2018; Received in revised form 13 March 2019; Accepted 16 March 2019
⁎ Corresponding author at: Centre for Environment, Fisheries and Aquaculture Science, Pakefield Road, Lowestoft, NR33 0HT, UK.E-mail address: [email protected] (F. Goodsir).
Marine Pollution Bulletin 142 (2019) 290–302
0025-326X/ © 2019 Elsevier Ltd. All rights reserved.
T
(either for operational purposes and/or as cargo) have the potential tocause widespread negative effects to marine ecosystems (Peterson et al.,2003; Michel et al., 2013) and socio-economics activities (Surís-Regueiro et al., 2007; Smith et al., 2011) and can persist in marinesediments (Laane et al., 2013; Sühring et al., 2016).
The physical properties of oil are critical for understanding itsmovement through the water column and dispersion over time, but theexact composition is often unknown for historical wrecks. Therefore, itcan be necessary to make an estimation based on what is known aboutprevalent oils from the time and location. The type and scale of oil leakscan vary, occurring as sporadic events or continuous releases of dif-fering magnitudes (Etkin et al., 2009). As a result, predicting the po-tential quantity and rate of release of oil can be difficult. What can becertain is that, as the condition of a wreck deteriorates over time thelikelihood of remaining oil being released will increase (Monfils, 2005).
Historical records and survey data (i.e. acoustic and underwaterimagery) can provide information about a wreck's condition and in-ternal integrity, and therefore the likelihood of it releasing oil. In manyinstances there is limited information relating to the condition and oilleak history of wrecks, so proxy data or expert judgement is used to fillthe gaps, reducing confidence in the assessment.
Some national authorities have investigated the potential pollutionrisks posed by wrecks and have already started addressing and mana-ging PPW within their waters (NOAA, 2009, 2013; Bergstrøm, 2014).Depending on the perceived level of risk, mitigation measures can rangefrom doing nothing, initiating monitoring programmes, to removingremaining oil. Removing oil from wrecks can be extremely costly ran-ging from a few million to tens of millions of US dollars (Etkin et al.,2009), therefore, a proactive approach is recommended to identifywrecks that pose the greatest risk to local marine ecosystems, econo-mies and communities. This requires a robust and standardised risk
assessment approach incorporating the likelihood of an oil release andthe severity of the potential impacts. Such an approach enables directcomparison of PPW and the prioritisation of appropriate managementand funding.
Risk assessment methodologies which identify and evaluate the riskfrom wrecks are well documented (see for example Michel et al., 2005;Alcardo et al., 2007; ABP Marine Environmental Research Ltd., 2007;Etkin et al., 2009; NOAA, 2013; Landquist et al., 2013, 2016; Ventikoset al., 2016). However, existing assessments lack an evaluation ofquantitative risk to local receptors under multiple oil spill scenarios.This can lead to an under- or overestimation of threats to the marineenvironment and leads to incorrect prioritisation of PPW. A reliablemeans of prioritising PPW is essential to a management programmeseeking to reduce environmental risk to as low as reasonably possible(ALARP) whilst delivering value for money.
There is little standardisation in wreck risk assessment methodolo-gies at either national or international levels. For example, a standar-dised approach is useful to national authorities which have PPW dis-tributed across the globe as a means of prioritising PPW for remediationwork. Furthermore, a standardised approach could support aims set outwithin the Nairobi Convention 2007 such as determining the hazardposed by wrecks, the risk of a potential oil release into the environment,and the legal obligation for removal of PPW from the marine environ-ment.
This paper presents a standardised environmental desk based as-sessment (E-DBA) of risk posed by PPW in a multi-staged approach(Fig. 1). The E-DBA is the first of a multi-stage methodology for overallrisk assessment which builds on the strengths of previous risk assess-ments. Stage two incorporates on-site wreck integrity validation alongwith environmental measurements and sampling, and stage three in-volves intervention/remediation, should it be necessary. The
Fig. 1. Flow diagram illustrating the multi-staged methodology for overall risk assessment, and the stepped process for carrying out the environmental desk basedassessment (E-DBA).
F. Goodsir, et al. Marine Pollution Bulletin 142 (2019) 290–302
291
information collected during the stage one E-DBA and stage two on-sitewreck integrity and environmental surveys will support the decision-making process relating to stage three, remediation requirements, andassociated options at each wreck site.
The first stage of the approach reported here uses predeterminedmodelled scenarios, which can be applied to any wreck. Firstly, a long-term chronic release of oil from a wreck is modelled; and secondly, twoshort-term acute releases of oil from a wreck. Similar to previous as-sessments (NOAA, 2013), the modelled scenarios enable an estimate ofsea surface, water column, sediment and shoreline contamination froman oil release and inform an assessment of the potential impacts toecological and socio-economic marine receptors. Here the standardisedwreck assessment process is described. The approach has been devel-oped to ensure that the relative environmental risks of individualwrecks can be compared as an aid to environmental management, de-cision making and prioritisation for intervention/remediation purposes.
2. Risk assessment approach
The E-DBA for PPW builds on previous risk assessment approaches(i.e. NOAA, 2013; ABP Marine Environmental Research Ltd., 2007) andassesses the potential risk to marine receptors under the application ofnumerical models of both chronic and acute oil release scenarios. Theapproach is formed as a multi-staged approach (Fig. 1). Firstly, thelikelihood that a wreck will release oil, based on existing historicalinformation, surveys and dive reports. Secondly, the likelihood of ex-posure of ecological and socio-economic receptors to released oil usingspatially resolved modelled outputs of spill trajectory and fate. Lastly,the potential for oil exposure to harm marine receptors. The purpose ofthis scoring system is to allow comparison and ranking between dif-ferent wrecks around the world. A confidence assessment is also appliedat two stages of the assessment, the first is applied to the likelihood ofrelease and then separately to the severity of risk to marine receptors.
2.1. Likelihood of oil release
The likelihood that a wreck will release oil depends on severalfactors, of which eight are considered in this E-DBA (Tables 1 and 2).The volume and type of oil are assessed in the modelling scenarios inthe next assessment stage, and so are excluded from this stage to avoidthe same risk factor being assessed twice. Historical wreck informationis used to determine the likelihood of the release of fuel oil and oilstored as cargo. Each criterion is given a score of either low (1),medium (2), or high (3); as per Table 2. Weightings (1–3) are applied tocriteria according to their perceived influence on the likelihood of an oilrelease (Table 2).
Overall oil release likelihood scores are generated by multiplyingeach risk assessment criterion's score by its weighting and summing theresulting values. The minimum attainable oil release likelihood score is16 and the maximum score is 48. A confidence assessment is applied toindicate the suitability and accuracy of available information andhighlight where additional survey or investigative works could improvethe confidence in assessments. Confidence criteria are scored for eachcriterion as either low (1), medium (2), or high (3); as per Table 3. Theminimum obtainable confidence score is 8 which is attributed to thenumber of risk assessment criteria in Tables 2 and 3, and the maximumobtainable score is 24. The confidence score categories are as follows:low confidence (8–12) medium confidence, (13–19), high confidence(20–24).
2.2. Likelihood of exposure of marine receptors to oil
Three oil release scenarios are used to predict likelihood of en-vironmental impacts: one chronic and two acute. While it is recognisedthat there are other possible release scenarios, the scenarios here arechosen to demonstrate the range of possible outcomes through the riskassessment process.
Table 1Defined risk assessment criteria for the assessment of likelihood for wrecks to release oil (adapted from NOAA, 2013; ABP Marine Environmental Research Ltd.,2007).
Vessel depth Vessel depth influences the physical conditions that a wreck encounters. Wrecks in < 30 m of water potentially pose a higher riskbecause they are more likely to be affected by waves and storm events. Wrecks at shallower depths are therefore expected to breakapart more rapidly, increasing the chance of an oil release. Wrecks in > 100 m are less influenced by hydrodynamics.Where vessel depth information is not available, water depth is taken from admiralty charts of the last know location of a wrecksto give a best estimate. Where location is unknown, literature or historical records can be used to provide an approximatelocation.
History of leaks Historical incident, dive or observation reports may contain information on whether there have been reported leaks since thevessel sunk or whether oil from an unknown source has been observed near a wreck. This information can tell us something aboutthe structure of the vessel and whether tanks may have been ruptured in the incident, however the amount of oil leaking from thevessel will remain unknown.
Integrity of wrecks If a wreck is broken into several pieces, tanks may have been ruptured and pipes and vessels may have been broken contributingto the release of oil. However, in some cases wrecks that have been broken into several pieces still run the risk of containing someoil which may be released. Where applicable, and evidence is available, post-sinking damage i.e. from salvage or natural eventssuch as storms, will be considered.
Age of vessel at time of sinking Older vessels pose an increased potential of releasing oil due to the increased likelihood of structural failure from corrosion ofmetal, compared to younger wrecks. This has been kept separate from the ‘length of time the vessel has been submerged’ as thesetwo criteria can affect a wreck's integrity differently as vessels are likely to accumulate wear and tear during their working life,such as rust or changes to the structural integrity due to modifications. This criterion also functions as a proxy for condition ofvessel prior to sinking.
Length of time the vessel has beensubmerged
Over long periods of submergence, wreck integrity will deteriorate as it is exposed to environmental factors that may acceleratecorrosion and collapse. Therefore, the longer a wreck has been submerged, the more likely it is to structurally fail leading to arelease event.
Method of storage The method of storage reflects how securely the oil is stored and how resistant the vessel may be to release the oil. Bunker tanksare more resistant to weathering than drums stored on deck, which are more exposed to external elements and therefore have ahigher potential to break free and rupture, though these contain smaller volumes of oil.
Type of incident The type of incident relates to the severity of structural damage that occurred at the time of sinking. This can provide an idea ofthe internal damage sustained, indicating whether oil is likely to have been released at the time of sinking. More severe types ofincidents have been classified as low, as a significant portion of any oils on-board are likely to have been released at the time ofsinking.
Stability of seabed The stability of the seabed is important for assessing whether conditions are likely to change around to a wreck, potentiallycausing wreck movement and accelerating breakup. Areas of stable seabed are less likely to cause breakup of a wreck than areaswith a high degree of movement and instability.
F. Goodsir, et al. Marine Pollution Bulletin 142 (2019) 290–302
292
The chronic release scenario is a continuous release of 50 kg oil perday and is fixed between wrecks for direct comparison. This value isreached using previous information for the sunken warship HMS RoyalOak, for which the oil release was considered “noticeable” when therate of oil release exceeded 100 kg per day (Michel et al., 2005). HMSRoyal Oakwas chosen as a representative reference because this methodwas initially developed for the assessment of nearshore World War I andII wrecks assumed to contain large amounts of oil. If information isavailable regarding the expected release rate, the release rate in thechronic scenario should be adjusted accordingly, however, such data isunlikely to be available.
The “worst-case” acute release scenario is based on the total esti-mated volume of oil being released within a 24-h period. The volume ofoil remaining on board is estimated based on best available evidencefrom historical reports, information such as how much oil the vesselwas carrying at the time of incident, and loss of oil and internal damagesustained during the incident.
It is unlikely that all the oil would be released at once, even fol-lowing an extreme failure of wreck integrity. This is particularly thecase for large vessels where the oil is stored in multiple tanks to preventsuch an event. Therefore, a “most probable” acute scenario simulates asingle tank failure within a 24-h period. Tank volumes are determinedfrom ships plans found in historical reports. Where limited or no in-formation is available for scenario development, proxy information isused from similar ships, ideally sister ships built by the same company.If no proxy information is available, 10% of total oil volume is used.
The properties of oil, such as composition, density and viscosity, arecritical for predicting movement through the water column and dis-persion over time. The exact composition and physical-chemical prop-erties of oil from historical wrecks is often unknown and can vary de-pending on when and where they were refuelled. Oil can also varybetween tanks on the same wreck due to different fuels being loaded(ship vs aviation fuel) and inconsistent rates of biodegradation betweentanks. Pre-WWII Admiralty specifications of oil are very vague re-garding the type and origin of oils, and could include residuals oils,shale distillates and various types of crude oils, but in general specifyviscosities at low temperature (32 °F or 0 °C) and require that the fuelsare suitable to use at low ambient temperatures (Brown, 2003). Forscenario development, and where oil type wasn't specified in archivedreports, oil types are chosen from the literature and/or the OSCAR Oildatabase (SINTEF, 2013) based on prevalent oils used on vessels at thetime of sinking, the location of loading fuel/cargo and the propulsiontype of the vessel.
For the presented work, the chronic scenario is modelled using theDose-related Risk and Effects Assessment Model (DREAM) and theacute scenarios are modelled using the Oil Spill Contingency andResponse (OSCAR) component of the Marine Environmental ModellingWorkbench (MEMW; SINTEF). The proposed E-DBA is a modular as-sessment method. Therefore, other environmental fate and transportmodels are able to predict long-term low volumes as well as short-termhigh volume sub-sea releases and distribution of oil can be used. Thescenarios explore sea surface, water column, sediment and shorelineimpacts. Table 4 provides model parameters for the different scenariotypes, including oil release description, model domain and grid re-solution based on the SS Baku Standard. the model domain should ac-commodate the transport of oil particles throughout the simulation,while the grid resolution (the size of grid cells, which determines thenumber of cells within the model domain) should ensure a good balancebetween the spatial resolution of the model outputs and computationalcost. Fig. 2 presents the oil deposition in sediments (kg/m2) from thethree scenarios using information based on the SS Baku Standard.
2.2.1. Chronic modelling scenarioDREAM is a chemical dispersion model that uses 2D surface wind
and 3D current data to model how a discharge or spill disperses in thewater column (SINTEF, 2013). DREAM simulates how the componentsTa
ble2
Risk
asse
ssm
entc
rite
ria,
wei
ghtin
gan
dri
sksc
ore
cate
gori
es.C
rite
ria
are
give
na
wei
ghtin
gof
eith
er1,
2or
3,w
ith3
bein
gth
em
osts
igni
fican
t.Th
ew
eigh
ting
scor
esre
flect
that
cert
ain
crite
ria
are
stro
nger
indi
cato
rsof
oilr
elea
seth
anot
hers
.For
each
ofth
e8
crite
ria,
wre
cks
are
assi
gned
ari
sksc
ore
of1
(low
),2
(med
ium
)or
3(h
igh)
.(Cr
iteri
ase
lect
edfr
omN
OA
A,2
013;
ABP
Mar
ine
Envi
ronm
enta
lRes
earc
hLt
d.,2
007)
.
Risk
asse
ssm
entc
rite
ria
Wei
ghtin
gof
crite
ria
Low
(sco
reof
1)M
ediu
m(s
core
of2)
Hig
h(s
core
of3)
Vess
elde
pth
2>
100
m30
–100
m<
30m
His
tory
ofle
aks
3N
okn
own
leak
sU
nkno
wn
oran
ecdo
tale
vide
nce
Doc
umen
ted
hist
ory
ofle
aks
Inte
grity
ofw
reck
s2
Brok
enin
tom
ore
than
thre
epi
eces
Brok
enin
totw
oor
thre
epi
eces
Inta
ct,i
non
epi
ece
orun
know
nA
geof
vess
elat
time
ofsi
nkin
g1
<10
year
s10
–30
year
s>
30ye
ars
Leng
thof
time
vess
elha
sbe
ensu
bmer
ged
3<
50ye
ars
50–9
0ye
ars
>90
year
s
Met
hod
ofSt
orag
e2
Spec
ific
bunk
erta
nkIn
hold
On
deck
,dru
ms,
cont
aine
rs,c
rate
sTy
peof
inci
dent
caus
ing
sink
ing
1M
ultip
leto
rped
ode
tona
tions
,mul
tiple
min
es,
seve
reex
plos
ion
Sing
leto
rped
o,sh
ellfi
re,s
ingl
em
ine,
rupt
ure
ofhu
ll,br
eaki
ngin
half,
grou
ndin
gon
rock
ysh
orel
ine
orun
know
nFo
ulw
eath
er,g
roun
ding
onso
ftbo
ttom
,co
llisi
onSt
abili
tyof
seab
ed2
Know
nto
best
able
seab
edRe
lativ
ely
stab
leor
not
know
nU
nsta
ble
and/
orhi
ghde
gree
ofm
ovem
ent
F. Goodsir, et al. Marine Pollution Bulletin 142 (2019) 290–302
293
of a discharge behave based on physicochemical properties, biode-gradability, and hydrodynamic properties of the release. With theseparameters, DREAM calculates the Predicted Environmental Con-centration (PEC) of a discharge as concentrations of contaminant“particles” in a 3D grid around the wreck. The addition of appropriatePredicted No-Effect Concentration (PNEC) values for the discharge al-lows DREAM to establish PEC/PNEC ratios throughout a modelled vo-lume and calculates the risk to sensitive species. In the presented study,DREAM was used to establish the predicted environmental risk ofchronic oil release from PPW. However, other models could be applied(or adapted) to predict the PEC/PNEC ratio. To generate a PEC/PNEC,local predicted oil concentrations are divided by the PNEC that is basedon available ecotoxicological literature data.
In DREAM, a PEC/PNEC value of > 1 represents concentrations atwhich 5% or more of the most sensitive species are likely to be affected.The species may or may not be present in the vicinity of the release site,and are not specifically identified, but rather they represent thosespecies for which the PEC is equal to or exceeds their PNEC. The modeluses acute toxicity data for marine algae, crustaceans and fish to cal-culate the PNEC of the investigated components. In most cases, thereare few data relating to chronic toxicity or sub-lethal effects such ascarcinogenic effects and endocrine disruption, and so these are not fullyconsidered by the model.
If a wreck is within a seasonally or partially stratified water column,the chronic model is run under both winter and summer conditions.Where seasonal variation in model results is observed, risk assessmentis based on the worst case of the two model outputs. If no seasonalvariation in model results is observed, or a wreck is within a perma-nently mixed water column, only one set of model outputs is reported.
The PEC/PNEC approach is an internationally accepted standard forenvironmental risk assessment and routinely applied in both theEuropean regulation for the Registration, Evaluation, Authorisation andRestriction of Chemicals (REACH) and Stockholm Convention assess-ments. The DREAM model is routinely used in the risk assessment forproduced water and oil releases from offshore oil and gas production inaddition to accidental spill modelling (SINTEF, 2013).
These initial chronic risk predictions are used in the first assessmentof PPW that are intended to inform decisions on the prioritisation offurther, on-site assessment and survey work. Therefore, the scenariosare based on general, easily available ecotoxicological data. The pre-dictions can be further refined by conducting ecotoxicological experi-ments using species typically living in wreck habitats.
2.2.2. Acute modelling scenariosOSCAR consists of a trajectory module for transport processes (ad-
vection and dispersion) coupled to an oil weathering module for the
fate processes (evaporation, dissolution, emulsification, stranding,biodegradation, and interaction with the sediments) (Spaulding, 2017).The horizontal transport of oil in the marine environment is driven bywind, currents and, to a lesser extent, waves (Guo and Wang, 2009).Thus, it is dependent on the meteorological and hydrodynamic condi-tions occurring from the initial moment of acute oil release until thefinal fate of the oil has been determined (Spaulding, 1988, 2017; Reedet al., 1999). The fate of the released oil is typically determined overdays (e.g. light oils or when the environmental conditions favour rapidonshore stranding) or weeks (e.g. heavy oils release, far from shore)(Guo and Wang, 2009; Spaulding, 1988, 2017; Reed et al., 1999).
Single model outputs can show a great variability, which is due notonly to the variability of the initial and driving environmental condi-tions, but also due to the uncertainty in parameter values used to set upthe model (parametric uncertainty), as well as uncertainty related tothe design of the equations used to model the physical processes and theway they are solved computationally (structural uncertainty).Structural uncertainty can be addressed through a multi-model ap-proach (e.g. using different models for the fate and trajectory of spilledoil), while the parameter uncertainty can be explored by varying theparameter values. Arguably, one of the most uncertain and importantparameters is the volume of released oil, and although here two pos-sible scenarios (the most likely and the worst case volumes) are de-scribed, these scenarios can still inform our estimation of risk to marinereceptors (which can broadly be categorised as low/medium/high),especially if the two scenarios result in similar patterns of oil exposureprobabilities, only with different magnitudes. However, the mainvariability of the model outputs arises from the variability of thedriving environmental conditions, which is more thoroughly exploredby the running of multiple trajectories.
The consistency of outcomes from multiple simulations determinesour estimates of probability. While a high probability (e.g. of impact ofa certain receptor) indicates an agreement between a large fraction ofthe model runs, this in not necessarily a proof of prediction robustnesson its own, unless it could be argued that the underlying driving pro-cesses for this particular outcome are well known and well representedin the model. In this case, it is often the prevailing winds and currentsthat are broadly responsible for the high probability predictions ofimpact, and arguably these are reasonably represented by the drivingdataset and implemented in the model.
While detailed data (e.g. hydrographic, tidal currents), which oftenis not available, would be necessary for accurate predictions at smallscale or in complex topographic coastal areas, it should be also notedthat many of the receptors span large areas, often tens of kilometresacross, which together with the broad categorisation of the risk to thesereceptors, allows for a meaningful estimation of the risk.
Table 3Confidence scores for the likelihood of oil release criteria based on underlying data are applied to each risk assessment criterion, assessed as 1 (low) to 3 (high).
Confidence Score Definition
High 1 The data used are timely, the best available, robust and the outputs are well supported by evidence. Majority of experts agree.Medium 2 The data are limited and/or proxy information. There is a majority agreement between experts; however, evidence is inconsistent and there are differing
views between experts.Low 3 The data are limited and not well supported by evidence. Experts do not agree.
Table 4Basic model set-up parameters used for the three oil spill modelling scenarios based on the SS Baku Standard.
Parameter Chronic Most probable acute Worst-case acute
Release amount 50 kg/day 848 t 5057 tOil type Texas crudeOil release position 56°48.5′ N, 002°12.9′ EOil release height 2 m above seabedModel domain 70 km × 70 km × 100 m 500 km × 500 km × 100 m 500 km × 500 km × 100 mGrid resolution 150 m × 150 m × 10 m 1 km × 1 km × 10 m 1 km × 1 km x 10 m
F. Goodsir, et al. Marine Pollution Bulletin 142 (2019) 290–302
294
The variability of the environmental conditions (both short-termand seasonal), determines the predicted behaviour and fate of the oil(Fig. 3a). This variability is included by running multiple (200) tra-jectories beginning at different times over a one-year period for eachscenario, namely a stochastic simulation. The trajectories are combinedto produce a statistical output. The results of individual trajectories mayvary widely in the direction and shape of the spread of oil depending onthe release time. The results of any single trajectory may thus not re-present the worst case or be representative of the actual risk posed tothe marine receptors. Therefore, all results presented are based on thestatistical analysis of the 200 model trajectories and do not depict theextent of oil contamination from a single trajectory (Fig. 3b).
Thresholds are applied to exclude model output data that are notdeemed environmentally harmful. Each stochastic simulation scenariois set up to include a threshold value at which an impact could occur atthe sea surface, in the water column and on the shoreline (Table 5).Therefore, the post simulation statistical analysis calculates the prob-ability of oil exceeding the minimum threshold in each model grid cellat some point during the simulation period. The threshold exceedanceduration cannot be specified as a model parameter and thus can be asshort as a single time-step (e.g. 20 min) or as long as the full simulationduration (e.g. 100 days). The actual threshold exceedance time at agiven location will vary between each individual model trajectory. Thesensitivity to oil varies dependant on the receptors assessed, for ex-ample surface oil sublethal effects on marine mammals, birds and seaturtles could occur at around 1.0 g/m2, whereas commercial and re-creational fishing could be prohibited at 0.01 g/m2 (French McCay,2016). Standard thresholds were applied (Table 5) for a rapid risk as-sessment to identify priority to wrecks for further investigation.
For each of the three environmental compartments (shoreline, seasurface and water column), the probability of contamination is calcu-lated by mapping out the percentage of model trajectories where agiven cell is hit by the contaminant in exceedance of the specifiedthreshold (Fig. 3b). For computational reasons, stochastic simulationsdo not include the sedimentation of oil as this is reflected as oil thatexits the model domain. It is therefore necessary to compute sedimentcontamination probabilities using manually set up trajectories ratherthan the OSCAR stochastic simulation tool.
It is important to note that the model outputs predict the extent ofan oil release incident with no remediation actions undertaken. It ispossible that salvage efforts to remove the oil before a potential releaseinto the environment could significantly reduce the extent and magni-tude of environmental impacts.
2.3. Severity of risk to marine receptors
Determining the level of risk an oil spill poses to marine receptorsrequires interpretation of the probability and magnitude of oil ex-posure, whilst considering the sensitivity of each receptor. Marine re-ceptors are described in this paper as ecological (marine and coastaldesignated areas, marine mammals, reptiles, seabirds, fish, and pro-tected benthic species); or socio-economic (shipping, tourism, fishingand infrastructure). Thresholds have been defined for ecological andsocio-economic receptors as the minimum level of oil contamination atwhich an impact might occur at the sea surface, water column, shore-line and in the sediments, for both chronic and acute modelling sce-narios (Tables 6, 7, 8 and 9). It must be noted that national prioritieschange considerably by region, for example mangroves, seagrass beds
Fig. 2. Modelled oil release scenarios: a) chronic 50 kg/day, b) most probable acute release of 843 t and c) worst cast acute release of 5057 t based on the wreck SSBaku Standard.
F. Goodsir, et al. Marine Pollution Bulletin 142 (2019) 290–302
295
and coral reefs are commonly of high priority for protection in tropicalregions, whereas the protection of marine mammals and seabirds maybe prioritised in more northern latitudes. Therefore, it is possible thatecological and socioeconomic receptors and their associated thresholdscan be refined dependant on the local priorities as required.
Risk to each marine receptor is evaluated and characterised as ei-ther low (1), medium (2) or high (3). Model outputs from the threescenarios are mapped along with marine receptor spatial distributiondata to assess the potential risk.
Fig. 4 illustrates mapped outputs showing environmental and socio-economic risk associated with an acute oil release from the SS BakuStandard, a wreck off the east coast of Scotland. Risk is expressed asprobability of contamination, based on the proportion of trajectoriesreaching a receptor during the modelled acute oil spill simulations. Thisbasic method cannot give an exact probability of contamination for a
receptor that spans more than one model grid cell, it can only give aminimum probability, but further analysis of the trajectory outputs inconjunction with receptor shapefiles can provide these figures.
Where a risk is identified for a marine receptor, expert judgementalong with additional sources (e.g. tourist board information, citizenscience) is used to verify risk. For example, under an acute scenario alarge area of the sediment is shown to be at high risk of oil con-tamination. In the absence of spatial information on benthic features,e.g. cold coral reefs, it is difficult to determine whether the feature ispresent and therefore overlaps with the area at risk. However, recordsof coral reefs have been reported in or close to the area at high risk.Therefore, a high risk score is given based on the large spatial extent athigh risk of oil contamination under an acute scenario, suggesting thatthis feature is likely to be exposed to high concentrations of oil.
A confidence assessment is also applied to evaluate the level of
Fig. 3. a): Single model outputs: single model trajectories of oil released in different months, variations in slick spreading is primarily due to different windconditions. 3b) Stochastic model outputs: a probability of 50% translates to 100 of the 200 individual trajectories resulting in contamination in a location for at leastthe minimum time step during each 30-day simulation. Values shown are the maxima across the 200 trajectories. The extent of any single model trajectory is notdepicted in this figure.
Table 5Threshold values applied to the shoreline, sea surface and water column for acute release scenarios.
Shoreline 50 kg/km which corresponds to 1 g/m2 (or the equivalent of 2 tar balls of approximately 1 cm in diameter per m2), under the assumption the shore is 50 m wide.Sea surface 0.1 t/km2 or 0.1 g/m2, equivalent to a 0.1 μm thick layer, where sheens will start to be visible.Water column 50 ppb, corresponds to the PNEC (predicted no effect concentration) of the most toxic components of oil, in this study C25 is used which was the main component
by mass fraction of the fuel oils investigated. The value will vary on the type of oil, and the contribution to toxicity made by the main components of that oil.
F. Goodsir, et al. Marine Pollution Bulletin 142 (2019) 290–302
296
accuracy of spatial data and suitability of underpinning information foreach receptor. Confidence criteria are scored for each criterion as eitherlow (1), medium (2), or high (3); as per Table 3. For example, in theabsence of spatial data and evidence on extent of the receptor, con-fidence is scored as low, whereas if spatial data is present and recent,confidence is score high.
2.4. Overall risk score
The overall risk posed by a wreck is assessed by generating anoverall risk score. The overall risk score is calculated by multiplying thelikelihood of release score by the severity of risk to marine receptorsscore. Ecological and socio-economic risks are kept separate so as not toassume an equal weighting.
The minimum score that can be achieved for ecological receptors is96 based on a score of 1 for each of the 6 receptors (total score 6) andthe lowest likelihood of oil release score (16). The maximum score is864 based on a score of 3 for the 6 receptors (total score 18) and thehighest likelihood of release score (48). Ecological risk is defined as low(< 252), medium (253–384) or high (385–864).
The minimum score that can be achieved for socio-economic re-ceptors is 64, based on a score of 1 for each of the four receptors (totalscore 4) multiplied by the lowest likelihood of oil release score (16).The maximum score is 576 based on a score of 3 for each of the re-ceptors (total score 12) and the highest likelihood of release score (48).Socio-economic risk is defined as low (< 168), medium (169–256) orhigh (257–576).
A confidence score is given based on 1) the robustness of dataavailable and 2) the level of agreement among a range of assessors onthe risk score assigned. This is used to identify limitations, uncertaintyand gaps in the available data. Confidence scores are assigned for eachof the assessed receptors as low (1), medium, or high (3). Scores aresummed then divided by the number of assessed receptors and multi-plied by 100 to gain a percentage score. The confidence score is used todescribe the overall level of confidence as low (< 50%), medium(50–80%) or high (> 80%).
A worked example is available in the supplementary material.
3. Discussion
This standardised risk assessment approach has been developed toassess the likelihood of oil release from a PPW and the potential of thatoil to cause harm to ecological and socio-economic marine receptors.The approach uses a combination of numerical modelling, geo-spatialdata and expert scientific judgement to quantify risks and generateconfidence scores to highlight areas of uncertainty. It attempts to ad-dress some of the current shortfalls in environmental risk assessments inrelation to PPW and builds on the strengths of previous assessmentstowards a robust approach. The aim is to assist managers in identifyingthose PPW which pose the greatest risk, to highlight where additionalsurveys or investigative works could improve the confidence of theseassessments, and to facilitate the development of appropriate man-agement and remediation strategies.
The described method uses three predetermined scenarios, in-cluding both chronic and acute oil releases to demonstrate risk of ex-posure and the magnitude of risk to key marine receptors. Multiplescenarios are essential, as the volume and duration of a release is fun-damental to understanding the movement of oil and therefore the risk itmay pose (NOAA, 2013). Large volume acute releases have the poten-tial to cover a wider area, potentially effecting many receptors, whereaslong duration chronic releases tend to be more localised but still havethe potential to impact receptors such as mariculture or benthic species.By understanding the history of a wreck, such as the incident thatcaused its sinking, and current structural condition, these scenarios canbe tailored to reflect the most likely oil spill scenario. If a wreck isknown to be leaking, the modelled scenario can be changed to reflectTa
ble6
Risk
ofch
roni
coi
lrel
ease
harm
ing
ecol
ogic
alm
arin
ere
cept
ors
issc
ored
aslo
w(1
),m
ediu
m(2
)or
high
(3).
Ther
eis
deem
edto
bea
risk
whe
nth
ePr
edic
ted
Effec
tCon
cent
ratio
n(P
EC)
exce
eds
the
conc
entr
atio
nof
cont
amin
atio
nat
whi
ch‘n
oeff
ect’
isob
serv
ed(P
redi
cted
No
Effec
tCo
ncen
trat
ion
(PN
EC),
i.e.P
EC/P
NEC
>1)
.
Risk
asse
ssm
entc
rite
ria
Rele
vant
oils
pill
mod
elou
tput
Low
(sco
reas
1)M
ediu
m(s
core
as2)
Hig
h(s
core
as3)
Mar
ine
and
coas
tald
esig
nate
dar
eas
Wat
erco
lum
n,se
dim
ent,
sea
surf
ace
and
shor
elin
e
<0.
002
PEC/
PNEC
risk
ofov
erla
pw
ithan
ypr
otec
ted
area
.0.
002–
0.2
PEC/
PNEC
risk
ofov
erla
pw
ithan
ypr
otec
ted
area
.>
0.2
PEC/
PNEC
risk
ofov
erla
pw
ithan
ypr
otec
ted
area
.
Spec
ies
and
feat
ures
ofco
nser
vatio
nin
tere
stM
arin
em
amm
als
(pin
nipe
ds)
Wat
erco
lum
n,se
asu
rfac
ean
dsh
orel
ine
<0.
2PE
C/PN
ECri
skof
over
lap
with
any
pinn
iped
haul
-out
orpu
ppin
glo
catio
n.0.
2–1
PEC/
PNEC
risk
ofov
erla
pw
ithan
ypi
nnip
edha
ul-
out
orpu
ppin
glo
catio
n.>
1PE
C/PN
ECri
skof
over
lap
with
any
pinn
iped
haul
-out
orpu
ppin
glo
catio
n.M
arin
em
amm
als
(cet
acea
nsan
dsi
reni
ans)
Wat
erco
lum
nan
dse
asu
rfac
e<
0.2
PEC/
PNEC
risk
ofov
erla
pw
ithan
impo
rtan
tkn
own
ceta
cean
orsi
reni
anha
bita
t.0.
2–1
PEC/
PNEC
risk
ofov
erla
pw
ithan
impo
rtan
tkno
wn
ceta
cean
orsi
reni
anha
bita
t.>
1PE
C/PN
ECri
skof
over
lap
with
anim
port
ant
know
nce
tace
anor
sire
nian
habi
tat.
Mar
ine
rept
iles
(tur
tles)
Wat
erco
lum
n,se
asu
rfac
ean
dsh
orel
ine
<0.
2PE
C/PN
ECri
skof
over
lap
with
any
turt
lene
stin
gsi
te.
0.2–
1PE
C/PN
ECri
skof
over
lap
with
any
turt
lene
stin
gsi
te.
>1
PEC/
PNEC
risk
ofov
erla
pw
ithan
ytu
rtle
nest
ing
site
.Se
abir
dsSe
asu
rfac
ean
dsh
orel
ine
<0.
002
PEC/
PNEC
risk
ofov
erla
pw
ithan
yIm
port
ant
Bird
Are
a(I
BA).
0.00
2–0.
2PE
C/PN
ECri
skof
over
lap
with
any
IBA
.>
0.2
PEC/
PNEC
risk
ofov
erla
pw
ithan
yIB
A.
Bent
hic
feat
ures
(e.g
.ree
fs)
and
spec
ies
incl
udin
gde
sign
ated
shel
lfish
grou
nds
Wat
erco
lum
nan
dse
dim
ent
Pred
icte
dto
talf
ootp
rint
ofoi
ldep
ositi
onon
sedi
men
t<
100
km2
and
<0.
002
PEC/
PNEC
risk
ofov
erla
pw
ithpr
otec
ted
bent
hic
feat
ures
and
spec
ies.
Pred
icte
dto
talf
ootp
rint
ofoi
ldep
ositi
onon
sedi
men
tbe
twee
n10
0an
d10
00km
2or
0.00
2–0.
2PE
C/PN
ECri
skof
over
lap
with
prot
ecte
dbe
nthi
cfe
atur
esan
dsp
ecie
s.
Pred
icte
dto
talf
ootp
rint
ofoi
ldep
ositi
onon
sedi
men
t>
1000
km2
or>
0.2
PEC/
PNEC
risk
ofov
erla
pw
ithpr
otec
ted
bent
hic
feat
ures
and
spec
ies.
Fish
spaw
ning
and
nurs
ery
area
sW
ater
colu
mn
and
sedi
men
tN
okn
own
spaw
ning
ornu
rser
yar
eas.
Oil
spill
inte
ract
sw
ithkn
own
disc
rete
area
sus
edfo
rsp
awni
ngan
d/or
nurs
ery
area
.O
ilsp
illin
tera
cts
with
high
inte
nsity
spaw
ning
and/
ornu
rser
yar
eas.
Fish
(sen
sitiv
eor
char
ism
atic
spec
ies)
Wat
erco
lum
nan
dse
dim
ent
No
know
nsp
ecie
s.O
ilsp
illin
tera
cts
with
know
ndi
scre
tear
eas
used
byse
nsiti
veor
char
ism
atic
spec
ies.
Oil
spill
inte
ract
sw
ithar
eaus
edby
larg
enu
mbe
rsof
sens
itive
orch
aris
mat
icsp
ecie
s.
F. Goodsir, et al. Marine Pollution Bulletin 142 (2019) 290–302
297
Table7
Risk
ofch
rom
icoi
lrel
ease
harm
ing
soci
o-ec
onom
icm
arin
ere
cept
orss
core
das
low
(1),
med
ium
(2)o
rhig
h(3
).Ri
skis
asse
ssed
asan
yov
erla
pbe
twee
nth
eso
cio-
econ
omic
rece
ptor
and
mod
elle
doi
lcon
cent
ratio
nsab
ove
the
defin
edth
resh
olds
whe
rean
impa
ctha
sth
epo
tent
ialt
ooc
cur
(see
Tabl
e5)
.
Risk
asse
ssm
entc
rite
ria
Rele
vant
oils
pill
mod
elou
tput
sLo
w(s
core
as1)
Med
ium
(sco
reas
2)H
igh
(sco
reas
3)
Curr
ent
and
plan
ned
infr
astr
uctu
reO
ffsho
rew
ind
farm
sSe
asu
rfac
eN
oov
erla
pw
ithan
yw
indf
arm
.Se
ason
alov
erla
pfo
r>
5%of
aw
indf
arm
leas
ear
ea.
Year
-rou
ndov
erla
pfo
r>
5%of
aw
indf
arm
leas
ear
ea.
Offs
hore
oila
ndga
sin
stal
latio
nsSe
asu
rfac
eN
oov
erla
pw
ithan
yin
stal
latio
n.Se
ason
alov
erla
pw
ithan
yin
stal
latio
n.Ye
ar-r
ound
over
lap
with
any
inst
alla
tion.
Indu
stri
alw
ater
inta
kes
Shor
elin
eN
oov
erla
pw
ithan
yin
dust
rial
wat
erin
take
.Se
ason
alov
erla
pw
ithan
yin
dust
rial
wat
erin
take
.Ye
ar-r
ound
over
lap
with
any
indu
stri
alw
ater
inta
ke.
Aqu
acul
ture
Wat
erco
lum
nan
dse
asu
rfac
eN
oov
erla
pw
ithan
yaq
uacu
lture
faci
lity.
Seas
onal
over
lap
with
any
aqua
cultu
refa
cilit
y.Ye
ar-r
ound
over
lap
with
any
aqua
cultu
refa
cilit
y.
Tour
ism
and
leis
ure
area
sTo
uris
m(C
oast
alto
wns
,bea
chfr
onts
,and
beac
hre
sort
sar
epo
pula
rho
liday
and
recr
eatio
nala
reas
supp
ortin
ga
rang
eof
busi
ness
esan
dco
mm
uniti
es)
Shor
elin
eN
oov
erla
pw
ithan
ykn
own
tour
ist
area
s.Se
ason
alov
erla
pw
ithan
ykn
own
tour
ist
area
s.Ye
ar-r
ound
over
lap
with
any
know
nto
uris
tar
eas.
Hig
hus
ear
eas
(mon
itore
dbe
ache
s,po
pula
rdi
ving
loca
tions
,rec
reat
iona
lmar
inas
and
boat
ing
area
san
dto
uris
tre
sort
s)Sh
orel
ine
No
over
lap
with
any
high
use
area
s.Se
ason
alov
erla
pw
ithan
yhi
ghus
ear
eas.
Year
-rou
ndov
erla
pw
ithan
yhi
ghus
ear
eas.
Fish
ing
grou
nds
Dem
ersa
lSe
dim
enta
ndse
asu
rfac
e<
180
days
offis
hing
effor
tim
pact
ed.
180–
365
days
offis
hing
effor
tim
pact
ed.
>36
5da
ysof
fishi
ngeff
orti
mpa
cted
.
Pela
gic
Wat
erco
lum
nan
dse
asu
rfac
e<
180
days
offis
hing
effor
tim
pact
ed.
180–
365
days
offis
hing
effor
tim
pact
ed.
>36
5da
ysof
fishi
ngeff
orti
mpa
cted
.
Crus
tace
anSe
dim
enta
ndse
asu
rfac
e<
180
days
offis
hing
effor
tim
pact
ed.
180–
365
days
offis
hing
effor
tim
pact
ed.
>36
5da
ysof
fishi
ngeff
orti
mpa
cted
.
Ship
ping
Impo
rtan
tsh
ippi
ngla
nes
Sea
surf
ace
No
over
lap
with
any
impo
rtan
tsh
ippi
ngla
nes.
Tem
pora
ryov
erla
pw
ithan
yim
port
ant
ship
ping
lane
s.Ye
ar-r
ound
over
lap
with
any
impo
rtan
tsh
ippi
ngla
nes.
Port
sSh
orel
ine
No
over
lap
with
any
port
s.Te
mpo
rary
over
lap
with
any
port
s.Ye
ar-r
ound
over
lap
with
any
port
s.
F. Goodsir, et al. Marine Pollution Bulletin 142 (2019) 290–302
298
the actual rate of release.Risk assessments rely on the availability of relevant and robust data.
The presented approach has been specifically designed with theknowledge that there is often limited data to support a comprehensiverisk assessment (Whittington et al., 2017). This methodology in-corporates information about the current and historical state of awreck, environmental variables such as currents, tides and winds usedin oil release models, and marine receptor geo-spatial and sensitivitydata. The data intensive nature of this methodology can be problematic.The availability of data is highly variable between regions and dataproviders. Furthermore, where data do exist they may be outdated, orhave licence restrictions on their use. This creates difficulties whenidentifying the number and location of potentially impacted marinereceptors, which may limit the confidence of the assessment. Poor dataabout vessel condition or sensitive receptors can lead to an under- oroverestimation of threats to the marine environment, which has thepotential to undermine the quality of management decisions. However,by scoring confidence in the data, as well as risk, this approach iden-tifies which criteria are lacking information. This provides managerswith two options, either take steps to improve confidence or take aprecautionary approach based on risk and confidences scores. Con-fidence can be improved by purchasing existing sensitivity and/or sa-tellite data and accessing dive site reports. It can be improved furtherby collecting data through high resolution multibeam echosounderand/or Remotely Operated Vehicle (ROV) surveys, which indicate thestructural integrity of the wreck and hull plating damage (i.e. howmany oil tanks remain intact). These additional data can be used tomodel more probable oil release scenarios.
An environmental risk assessment such as the one proposed hereprovides a snap shot in time, it is based on the present understanding ofthe condition of the wreck and the environmental conditions at the timeof the assessment. If no monitoring or remediation is undertaken, riskassessments may need to be updated on a periodic basis to take accountof changes to wreck integrity and the surrounding environment as wellas including new information, as and when it becomes available.
Further refinement could include a risk assessment procedure forlong-term effects of an oil release, taking account of longer-term issuessuch as accumulation in sediments. Long-term oil accumulation in se-diments represents an environmental risk as a local contaminationsource for benthic organisms, as well as higher trophic levels throughbiomagnification, and can potentially impact communities near a pol-luting wreck over several decades. Environmental surveys and numer-ical modelling could predict the effects of a long-term oil release andhelp determine the level of risk posed to the local area.
An on-site environmental survey requires robust design and im-plementation to deliver the necessary information regarding site char-acterisation and a baseline assessment of contaminants present. Modeloutputs generated in the stage one E-DBA can be used to inform siteselection and sample collection, such as probable locations of oil de-position in the sediment. Verification of any previously conductedwreck integrity assessments (whether it is structurally intact) willprovide a more complete understanding of the wreck in question at thetime of survey.
National authorities are under political pressure and moral obliga-tion to avoid environmental incidents, which can be done by taking aproactive approach to PPW management (Whittington et al., 2017).Proactive PPW management will not only limit environmental damage,but also avoid the reputational damage and clean-up costs associatedwith marine oil spills. Management decisions may include monitor andevaluate, hot tap of wreck hull and removal of oil, or partial or com-plete wreck removal. The quantitative risk assessment approach pre-sented here is a management tool for prioritising PPW, ensuring thatresources are focussed on those PPW which pose the greatest risk tomarine receptors.
Table8
Risk
ofac
ute
oilr
elea
seha
rmin
gec
olog
ical
mar
ine
rece
ptor
ssc
ored
aslo
w(1
),m
ediu
m(2
)or
high
(3).
The
prob
abili
tyof
cont
amin
atio
n,ex
pres
sed
asa
perc
enta
ge,r
eflec
tsth
efr
actio
nof
the
200
traj
ecto
ries
whi
chre
sulte
din
cont
amin
atio
nab
ove
the
thre
shol
dat
the
loca
tion
ofth
ere
cept
or.T
here
isde
emed
tobe
ari
skw
hen
the
prob
abili
tyof
cont
amin
atio
nex
ceed
sth
epr
e-de
fined
thre
shol
dfo
rea
chsp
ecifi
cre
cept
or.
Risk
asse
ssm
entc
rite
ria
Rele
vant
oils
pill
mod
elou
tput
sLo
w(s
core
as1)
Med
ium
(sco
reas
2)H
igh
(sco
reas
3)
Mar
ine
and
coas
tald
esig
nate
dar
eas
Wat
erco
lum
n,se
dim
ent,
sea
surf
ace
and
shor
elin
e<
5%pr
obab
ility
ofov
erla
pw
ithan
ypr
otec
ted
area
.5–
50%
prob
abili
tyof
over
lap
with
any
prot
ecte
dar
ea.
>50
%pr
obab
ility
ofov
erla
pw
ithan
ypr
otec
ted
area
.
Spec
ies
and
feat
ures
ofco
nser
vatio
nin
tere
stM
arin
em
amm
als
(pin
nipe
ds)
Wat
erco
lum
n,se
asu
rfac
ean
dsh
orel
ine
<5%
prob
abili
tyof
over
lap
with
pinn
iped
haul
-out
orpu
ppin
glo
catio
n.5–
50%
prob
abili
tyof
over
lap
with
pinn
iped
haul
-out
orpu
ppin
glo
catio
n.>
50%
prob
abili
tyof
over
lap
with
any
pinn
iped
haul
-out
orpu
ppin
glo
catio
n.M
arin
em
amm
als
(cet
acea
nsan
dsi
reni
ans)
Wat
erco
lum
nan
dse
asu
rfac
e<
5%pr
obab
ility
ofov
erla
pw
ithan
impo
rtan
tce
tace
anor
sire
nian
habi
tat.
5–50
%pr
obab
ility
ofov
erla
pw
ithan
impo
rtan
tce
tace
anor
sire
nian
habi
tat.
>50
%pr
obab
ility
ofov
erla
pw
ithan
impo
rtan
tcet
acea
nor
sire
nian
habi
tat.
Mar
ine
rept
iles
(tur
tles)
Wat
erco
lum
n,se
asu
rfac
ean
dsh
orel
ine
<5%
prob
abili
tyof
over
lap
with
any
turt
lene
stin
gsi
te.
5–50
%pr
obab
ility
ofov
erla
ptu
rtle
nest
ing
site
.>
50%
prob
abili
tyof
over
lap
with
any
turt
lene
stin
gsi
te.
Seab
irds
Sea
surf
ace
and
shor
elin
e<
5%pr
obab
ility
ofov
erla
pw
ithan
yIB
A.
5–50
%pr
obab
ility
ofov
erla
pw
ithan
yIB
A.
>50
%pr
obab
ility
ofov
erla
pw
ithan
yIB
A.
Bent
hic
feat
ures
(e.g
.ree
fs)
and
spec
ies
athi
ghpr
obab
ility
ofov
erla
pin
clud
ing
desi
gnat
edsh
ellfi
shgr
ound
s
Wat
erco
lum
nan
dse
dim
ent
Pred
icte
dto
talf
ootp
rint
ofoi
ldep
ositi
onon
sedi
men
t<
100
km2
and
noov
erla
pw
ithpr
otec
ted
bent
hic
spec
ies.
Pred
icte
dto
talf
ootp
rint
ofoi
ldep
ositi
onon
sedi
men
tbe
twee
n10
0an
d10
00km
2an
dno
over
lap
with
prot
ecte
dbe
nthi
csp
ecie
s.
Pred
icte
dto
talf
ootp
rint
ofoi
lde
posi
tion
onse
dim
ent
>10
00km
2
Fish
spaw
ning
and
nurs
ery
area
sW
ater
colu
mn
and
sedi
men
tN
oov
erla
pw
ithan
ysp
awni
ngor
nurs
ery
area
s.O
verl
apw
ithdi
scre
tear
eas
used
for
spaw
ning
and/
ornu
rser
yar
ea.
Ove
rlap
with
high
inte
nsity
spaw
ning
and/
ornu
rser
yar
eas.
Fish
(sen
sitiv
eor
char
ism
atic
spec
ies)
Wat
erco
lum
nan
dse
dim
ent
No
over
lap
with
any
spec
ies.
Ove
rlap
with
disc
rete
area
sus
edby
sens
itive
orch
aris
mat
icsp
ecie
s.O
verl
apw
ithar
eaus
edby
larg
enu
mbe
rsof
sens
itive
orch
aris
mat
icsp
ecie
s.
F. Goodsir, et al. Marine Pollution Bulletin 142 (2019) 290–302
299
Table9
Risk
ofac
ute
oilr
elea
seha
rmin
gso
cio-
econ
omic
mar
ine
rece
ptor
ssco
red
aslo
w(1
),m
ediu
m(2
)orh
igh
(3).
The
prob
abili
tyof
cont
amin
atio
n,ex
pres
sed
asa
perc
enta
ge,r
eflec
tsth
efr
actio
nof
the
200
traj
ecto
ries
whi
chre
sulte
din
cont
amin
atio
nab
ove
the
thre
shol
dat
the
loca
tion
ofth
ere
cept
or.T
here
isde
emed
tobe
ari
skw
hen
the
prob
abili
tyof
cont
amin
atio
nex
ceed
sth
epr
e-de
fined
thre
shol
dfo
rea
chsp
ecifi
cre
cept
or.
Risk
asse
ssm
entc
rite
ria
Rele
vant
oils
pill
mod
elou
tput
sLo
w(s
core
as1)
Med
ium
(sco
reas
2)H
igh
(sco
reas
3)
Curr
ent
and
plan
ned
infr
astr
uctu
reO
ffsho
rew
ind
farm
sSe
asu
rfac
e<
5%pr
obab
ility
ofov
erla
pof
oilw
ithan
yw
indf
arm
.5–
50%
prob
abili
tyof
over
lap
ofoi
labo
ve5%
ofa
win
dfar
mle
ase
area
.>
50%
prob
abili
tyof
over
lap
ofab
ove
5%of
aw
indf
arm
leas
ear
ea.
Offs
hore
oila
ndga
sin
stal
latio
nsSe
asu
rfac
e>
5%pr
obab
ility
ofov
erla
pof
anin
stal
latio
n.5–
50%
prob
abili
tyof
over
lap
ofan
inst
alla
tion.
>50
%pr
obab
ility
ofan
yov
erla
pof
anin
stal
latio
n.
Indu
stri
alw
ater
inta
kes
Shor
elin
e>
5%pr
obab
ility
ofov
erla
pw
ithan
ypo
wer
stat
ion.
5–50
%pr
obab
ility
ofov
erla
pw
ithan
ypo
wer
stat
ion.
5–50
%pr
obab
ility
ofan
yov
erla
pw
ithan
ypo
wer
stat
ion.
Aqu
acul
ture
Wat
erco
lum
nan
dse
asu
rfac
e<
5%pr
obab
ility
ofov
erla
pw
ithan
yaq
uacu
lture
faci
lity.
5–50
%pr
obab
ility
ofov
erla
pw
ithan
yaq
uacu
lture
faci
lity.
>50
%pr
obab
ility
ofov
erla
pw
ithan
yaq
uacu
lture
faci
lity.
Tour
ism
and
leis
ure
area
sTo
uris
mSh
orel
ine
<5%
prob
abili
tyof
any
shor
elin
eim
pact
.5–
50%
prob
abili
tyof
any
shor
elin
eim
pact
.>
50%
prob
abili
tyof
any
shor
elin
eim
pact
.H
igh
use
area
sSh
orel
ine
<5%
prob
abili
tyof
over
lap
with
any
high
use
area
.5–
50%
prob
abili
tyof
over
lap
with
any
high
use
area
.>
50%
prob
abili
tyof
over
lap
with
any
high
use
area
.
Fish
ing
grou
nds
Dem
ersa
lSe
dim
enta
ndse
asu
rfac
e<
180
days
offis
hing
effor
tim
pact
edin
area
of50
–100
%pr
obab
ility
ofoi
lcon
tam
inat
ion.
180–
365
days
offis
hing
effor
tim
pact
edin
area
of50
–100
%pr
obab
ility
ofoi
lcon
tam
inat
ion.
>36
5da
ysof
fishi
ngeff
orti
mpa
cted
inar
eaof
50–1
00%
prob
abili
tyof
oilc
onta
min
atio
n.Pe
lagi
cW
ater
colu
mn
and
sea
surf
ace
<18
0da
ysof
fishi
ngeff
orti
mpa
cted
inar
eaof
50–1
00%
prob
abili
tyof
oilc
onta
min
atio
n.18
0–36
5da
ysof
fishi
ngeff
orti
mpa
cted
inar
eaof
50–1
00%
prob
abili
tyof
oilc
onta
min
atio
n.>
365
days
offis
hing
effor
tim
pact
edin
area
of50
–100
%pr
obab
ility
ofoi
lcon
tam
inat
ion.
Crus
tace
anSe
dim
enta
ndse
asu
rfac
e<
180
days
offis
hing
effor
tim
pact
edin
area
of50
–100
%pr
obab
ility
ofoi
lcon
tam
inat
ion.
180–
365
days
offis
hing
effor
tim
pact
edin
area
of50
–100
%pr
obab
ility
ofoi
lcon
tam
inat
ion.
>36
5da
ysof
fishi
ngeff
orti
mpa
cted
inar
eaof
50–1
00%
prob
abili
tyof
oilc
onta
min
atio
n.
Ship
ping
Impo
rtan
tsh
ippi
ngla
nes
Sea
surf
ace
<5%
prob
abili
tyof
over
lap
with
impo
rtan
tshi
ppin
gla
nes.
5–50
%pr
obab
ility
ofov
erla
pw
ithim
port
ant
ship
ping
lane
s.>
50%
prob
abili
tyof
over
lap
with
impo
rtan
tsh
ippi
ngla
nes.
Port
sSh
orel
ine
<5%
prob
abili
tyof
over
lap
with
apo
rt.
5–50
%pr
obab
ility
ofov
erla
pw
itha
port
.>
50%
prob
abili
tyof
over
lap
with
apo
rt.
F. Goodsir, et al. Marine Pollution Bulletin 142 (2019) 290–302
300
Acknowledgments
This work is funded by and is part of the ongoing research projectC6107 Environmental Desk Based Assessments for the Ministry ofDefence (MoD). The MoD has defined a programme of work to under-take standardised E-DBAs taking account of oil pollution from wrecksfor which they are responsible.
Appendix A. Worked wreck example
Supplementary data to this article can be found online at https://doi.org/10.1016/j.marpolbul.2019.03.038.
References
ABP Marine Environmental Research Ltd, 2007. Research Project 567- Potentially
Polluting Wrecks in the UK Pollution Control Zone Phase 1 and 2. ABP MarineEnvironmental Research Ltd (148 pp).
Alcardo, L., Amato, E., Cabioch, F., Farchi, C., Gouriou, V., 2007. DEEPP Project,Development of European Guidelines for Potentially Polluting Shipwrecks. ICRAM,Instituto Centrale per la Ricerca scientifica e tecnologica Applicata al Mare, CEDRE,Centre de Documentation de Recherché et d'Epérimentations surles pollutions acci-dentelles des eaux.
Bergstrøm, R., 2014. 2014. Lessons learned from offloading oil from potentially pollutingship wrecks from world war II in Norwegian waters. International Oil SpillConference Proceedings 2014 (1), 804–813 May.
Brown, W.M., 2003. The Royal Navy's Fuel Supplies 1898–1939: The Transition from Coalto Oil. University of London (329 pp).
Burns, K.A., Garrity, S.D., Jorissen, D., MacPherson, J., Stoelting, M., Tierney, J., Yelle-Simmons, L., 1994. The galeta oil spill. II. unexpected persistence of oil trapped inmangrove sediments. Estuar. Coast. Shelf Sci. 34 (4), 349–364.
ENSR Corporation, 2006. Pipeline Risk Assessment and Environmental ConsequenceAnalysis. Document Number 10623–004. (52 pp).
Etkin, D.S., van Rooij, H., McCay French, D., 2009. Risk assessment modeling approachfor the prioritization of oil removal operations from sunken wrecks. In: Effects of Oilon Wildlife, (Tallin, Estonia. 10 pp).
Faksness, L., Daling, P., Altin, D., Dolva, H., Fosbæk, B., Bergstrøm, R., 2015. Relative
Fig. 4. Risk to marine and coastal receptors from an acute oil release off the east coast of Scotland. Top: probability of contamination of marine and coastaldesignated areas (> 5%) Marine and coastal designated areas at risk include Marine Protected Areas (MPAs), Special Protected Areas (SPAs), Special Area ofConservation (SAC), Site of Special Scientific Interest (SSSI), and Ramsar sites. Bottom: probability of contaminations of ports.
F. Goodsir, et al. Marine Pollution Bulletin 142 (2019) 290–302
301
bioavailability and toxicity of fuel oils leaking from world war II shipwrecks. Mar.Pollut. Bull. 94, 123–130.
Fingas, M., 1999. 1999. The evaporation of oil spills: development and implementation ofnew prediction methodology. International oil spill conference proceedings 1999 (1),281–287 March.
French McCay, D., 2016. Potential effects thresholds for oil spill risk assessments. In:Proceedings of the 39th AMOP Technical Seminar on Environmental Contaminationand Response, pp. 285–303.
Guo, W.J., Wang, Y.X., 2009. A numerical oil spill model based on a hybrid method. Mar.Pollut. Bull. 58 (5), 726–734.
Hailong, L., Boufadel, M.C., 2010. Long-term persistence of oil from the Exxon Valdezspill in two-layer beaches. Nat. Geosci. 3, 96–99.
Kingston, P.F., 2002. Long-term environmental impact of oil spills. Spill Sci. Technol.Bull. 7 (1–2), 53–61.
Laane, R.W.P.M., Vethaak, A.D., Gandrass, J., Vorkamp, K., Köhler, A., Larsen, M.M.,Strand, J., 2013. Chemical contaminants in the Wadden Sea: sources, transport, fateand effects. J. Sea Res. 82, 10–53. https://doi.org/10.1016/j.seares.2013.03.004.
Landquist, H., Hassellöv, I.-M., Rosén, L., Lindgren, J.F., Dahllöf, I., 2013. Evaluating theneeds of risk assessment methods of potentially polluting shipwrecks. J. Environ.Manag. 119, 85–92.
Landquist, H., Rosén, L., Lindhe, A., Hassellöv, I.-M., 2016. VRAKA—A probabilistic riskassessment method for potentially polluting shipwrecks. Front. Environ. Sci. 4, 49.https://doi.org/10.3389/fenvs.2016.00049.
Law, R.J., Moffett, C.F., 2011. The Braer Oil Spill, 1993. Oil Spill Science and Technology.vol. 36. pp. 1119–1126.
Liu, Zhanfei, Liu, Jiqing, Zhu, Qingzhi, Wu, Wei, 2012. The weathering of oil after theDeepwater Horizon oil spill: insights from the chemical composition of the oil fromthe sea surface, salt marshes and sediments. Environ. Res. Lett. 7, 035302.
Michel, J., Etkin, D.S., Gilbert, T., Urban, R., Waldron, J., Blocksidge, C.T., 2005.Potentially Polluting Wrecks in Marine Waters. An Issue Paper Prepared for the 2005International Oil Spill Conference. (40 pp).
Michel, J., Owens, E.H., Zengal, S., Graham, A., Nixon, Z., Allard, T., Holton, W., et al.2013. Extent and degree of shoreline oiling: deepwater horizon oil spill, Gulf ofMexico, USA. PLoS One 8(6):e65087. doi:https://doi.org/10.1371/journal.pone.0065087.
Monfils, R., 2005. The global risk of marine pollution from WWII shipwrecks: examplesfrom the seven seas. In: International Oil Spill Conference. vol. 2005 IOSC, MiamiBeach, FL.
National Oceanic and Atmospheric Administration (NOAA), 2009. Wreck Oil RemovalProgram (WORP). 2009. Demonstration Project Overview. National Oceanic andAtmospheric Administration.
National Oceanic and Atmospheric Administration (NOAA). 2013. Risk assessment forpotentially polluting wrecks in U.S. waters. (195 pp).
National Research Council, 2002. Oil in the Sea III: Inputs, Fates, and Effects. NationalAcademy Press, Washington, DC.
Peterson, C.H., Rice, S.D., Short, J.W., Esler, D., Bodkin, J.L., Ballachey, B.E., Irons, D.B.,2003. Long-term ecosystem response to the Exxon Valdez oil spill. Science 302(5653), 2082–2086.
Plata, D.L., Sharpless, C.M., Reddy, C.M., 2008. Photochemical degradation of polycyclicaromatic hydrocarbons in oil films. Environ. Sci. Technol. 42, 2432–2438.
Reed, M., Johansen, Ø., Brandvik, P.J., Daling, P., Lewis, A., Fiocco, R., Mackay, D.,Prentki, R., 1999. Oil spill modeling towards the close of the 20th century: overviewof the state of the art. Spill Sci. Technol. Bull. 5 (1), 3–16.
Rogowska, J., Namieśnik, J., 2010. Environmental implications of oil spills from shippingaccidents. In: Reviews of Environmental Contamination and Toxicology. vol. 206.Springer, New York, pp. 95e114.
SINTEF, 2013. MEMW 6.5 user guide: SINTEF marine environmental technology. MEMW6.5 user's manual. Retrieved from: www.sintef.no.
Smith, L.C., Smith, M., Ashcroft, P. 2011. Analysis of environmental and economic da-mages from British petroleum's deepwater horizon oil spill. Albany Law Review,74(1): 563–585. Available at SSRN: https://ssrn.com/abstract=1653078 ordoi:https://doi.org/10.2139/ssrn.1653078.
Spaulding, M.L., 1988. A state-of-the-art review of oil spill trajectory and fate modelling.Oil Chem. Pollut. 4 (1), 39–55.
Spaulding, M.L., 2017. State of the art review and future directions in oil spill modelling.Mar. Pollut. Bull. 115, 7–19.
Sühring, R., Busch, F., Fricke, N., Kötke, D., Wolschke, H., Ebinghaus, R., 2016.Distribution of brominated flame retardants and dechloranes between sediments andbenthic fish – a comparison of a freshwater and marine habitat. Sci. Total Environ.542, 578–585. https://doi.org/10.1016/j.scitotenv.2015.10.085.
Surís-Regueiro, J.C., Garza-Gil, M.D., Varela-Lafuente, M.M., 2007. The prestige oil spilland its economic impact on the Galician fishing sector. Disasters 31 (2), 201–215.
Ventikos, N.P., Louzis, K., Koimtzoglou, A., Delikanidis, P., 2016. Enhanced decisionmaking through probabilistic shipwreck risk assessment: focusing on the situation inGreece. Front. Mar. Sci. 3, 97.
Wang, Z.D., Fingas, M., 1995. Study of the effects of weathering on the chemical com-position of a light crude oil using GC/MS and GC/FID. J. Microcolumn Sep. 7,617–639.
Whittington, M., Zhang, A., Campion, D., 2017. To remove or not to remove? Dealingwith pollution risks from ship wrecks. In: 2017 International Oil Spill Conference, pp.1–21.
F. Goodsir, et al. Marine Pollution Bulletin 142 (2019) 290–302
302