A fate and transport ocean model for persistent organic ......A fate and transport ocean model for...

s 63 (2006) 1–19www.elsevier.com/locate/jmarsys

Journal of Marine System

A fate and transport ocean model for persistent organicpollutants and its application to the North Sea

T. Ilyina a,⁎, T. Pohlmann a, G. Lammel b, J. Sündermann a

a Institute of Oceanography, University of Hamburg, Bundesstr. 53, 20146, Hamburg, Germanyb Max Planck Institute for Meteorology, Bundesstr. 53, 20146, Hamburg, Germany

Received 23 November 2005; received in revised form 19 April 2006; accepted 20 April 2006Available online 15 June 2006

Abstract

An ocean model (FANTOM) has been developed to investigate the fate of selected persistent organic pollutants (POPs) in the NorthSea. The main focus of the model is on quantifying the distribution of POPs and their aquatic pathways within the North Sea. Keyprocesses are three-dimensional transport of POPswith ocean currents, diffusive air–sea exchange,wet and dry atmospheric depositions,phase partitioning, degradation, and net sedimentation in bottom sediments. This is the first time that a spatially resolved, measurements-based ocean transport model has been used to study POP-like substances, at least on the regional scale. The model was applied for thesouthern North Sea and tested by studying γ-HCH behaviour in sea water in the years 1995 to 2001. Themodel's structure and processesare described in details. Concentrations of γ-HCH and its fluxes between upper sediment, sea and atmosphere were modelled, based ondischarge and emission estimates available through various monitoring programmes.Model results are evaluated against measurements.Modelled concentrations of γ-HCH in sea water are in good agreement with the observations; the spatial distribution and the downwardtrend in the entire North Sea are reproduced during the simulation period. The correlation between themodel results andmeasurements isbetter during warmer seasons suggesting the importance of the temperature dependency of the air–sea exchange.© 2006 Elsevier B.V. All rights reserved.

Keywords: Persistent organic pollutants; Transport; Fate; Modelling; Lindane; North Sea

1. Introduction

Persistent organic pollutants (POPs) are synthetic or-ganic chemical compounds which are environmentallypersistent, bioaccumulative and toxic. They can be trans-ported over long distances in the atmosphere and oceans.Included in the POPs are organochlorine compounds, e.g.hexachlorocyclohexane (HCH) which is the most abun-dant organochlorine pesticide in both the atmosphere andoceans. POPs are distributed throughout the oceans as a

⁎ Corresponding author. Tel.: +49 40 428387489; fax: +49 40428387488.

E-mail address: [email protected] (T. Ilyina).

0924-7963/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.jmarsys.2006.04.007

consequence of atmospheric deposition and direct intro-duction into aquatic systems. Already in the 1970s, it wasbeing predicted that the oceans may be recipients of mostof the persistent pesticides used globally (Goldberg,1975). This was confirmed by observations (Bidlemanet al., 1995) and global budget calculations (Strand andHov, 1996) showing that the oceans are a major storagemedium of HCH. Oceans are traditionally thought to be aglobal reservoir and ultimate sink of many POPs (Iwataet al., 1993; Dachs et al., 2002) and may be a slow butsignificant medium for their long-range transport (Waniaand Mackay, 1999; UNEP, 2003). Upon long-termongoing uptake of HCH the ocean may achieveequilibrium with respect to HCH partitioning between

mailto:[email protected]

http://dx.doi.org/10.1016/j.jmarsys.2006.04.007

Fig. 1. Key processes included in FANTOM affecting the fate of POPs.

2 T. Ilyina et al. / Journal of Marine Systems 63 (2006) 1–19

air and sea, which has been observed for α-HCH(Bidleman et al., 1995; Lakaschus et al., 2002).

The North Sea is surrounded by highly industrialisedcountries and receives large amounts of POPs from atmos-pheric deposition and river discharge. Monitoring cam-paigns show that nearly all known POP-like chemicalshave been detected in the North Sea (OSPAR, 2000;Weigel et al., 2002) as a consequence of their persistenceand continued input from contaminated sites and distantsources. Although concentrations of some POPs in theNorth Sea have been diminishing since the early 1990s,even the reduced levels can still be harmful to the envi-ronment (OSPAR, 2000). Concentrations of some POPs insea water are often five orders of magnitude higher thanthose in the air (Wania and Mackay, 1999). This willcontribute to the capacity of sea water to act as a source byreleasing POPs to the atmosphere via the volatilisationprocess (Hornbuckle et al., 1993; Bidleman et al., 1995;Ridal et al., 1996; Sahsuvar et al., 2003). Furthermore, inshelf areas the sinking particles carry POPs down to thebottom sediments; they may then enter the water columnagain via re-suspension and return back to the sea surface.Therefore, as primary pollutant sources are reduced, remo-bilisation from previous repositories, such as water bodies,can act as secondary sources to the atmosphere.

A model to investigate the environmental fate of POPsis a unique tool for testing hypotheses about the processesaffecting POPs cycling, especially those whose character-istics are not accessible by experimental approaches, e.g.for calculating single episodes of pollutant release andtransport. With regard to availability of comprehensivedata on POPs, the North Sea is comparatively well served.The datasets have been accumulated throughout nationaland international monitoring programmes (OSPAR, 2000)and research projects (Sündermann, 1994), though thedata coverage is better for Lindane (γ-HCH) than for othercompounds. The environmental fate of α- and γ-HCH inthe Baltic Sea has been studied using box models withinthe POPCYCLING-Baltic project during 1996–1999(Pacyna, 1999). Besides that, there have been campaign-based studies, limited in both time and space, whichaddressed the occurrence and distribution of POPs in openocean and shelf seas (Iwata et al., 1993; Schulz-Bull et al.,1998; Lakaschus et al., 2002; Jaward et al., 2004).

The fate of POPs in sea water depends on a number ofmechanical (transport with moving flows), chemical (che-mical decay, amalgamation with other chemicals, transferto gaseous state, etc.), physical (transfer to another aggre-gative state, adsorption) and biological (pollutant accumu-lations and transport by biota) processes. These processescan only be fully taken into account with a three-dimensional, hydrodynamic ocean model. Therefore, the

objective of this study was to design and evaluate such amodel and use it to investigate the cycling of γ-HCH in theNorth Sea. It is a first study of this kind.

Section 2 describes the architecture of the fate andtransport ocean model (FANTOM). Results from the6.5 years simulations of the fate of γ-HCH in the NorthSea, model evaluation and its sensitivity towards individualprocesses are presented in Section 3. Section 4 gives theconclusions and an outlook for future developments.

2. Model description

FANTOM is a three-dimensional Eulerian model de-signed for describing the long-term fate of POP-like con-taminants in the coastal and shelf aquatic environment.FANTOM is aimed at tracing substances released frompoint or diffuse sources. The tracers can enter the modeldomain via rivers, adjacent seas or atmospheric deposi-tion. The key processes are described in Fig. 1. They fallinto four broad categories.

(a) Transport with ocean currents. In sea water thepollutant is transported by ocean currents via ad-vection and turbulent diffusion (Section 2.1).

(b) Air–sea exchange. Air–sea exchange (Section2.2) is represented by three mechanisms: revers-ible gaseous exchange (Appendix A), dry particledeposition (Appendix B) and wet deposition(Appendix C).

(c) Phase distribution. The pollutant in the model iseither dissolved or bound to suspended particulatematter (SPM) present in seawater (Section 2.3). Theparticle-bound fraction of the pollutant (AppendixE) follows the SPM dynamics (Appendix D). It issubject to gravitational sinking and deposition to thebottom sediments. Redistribution of particles takesplace in the sediment due to the disturbance of

3T. Ilyina et al. / Journal of Marine Systems 63 (2006) 1–19

sediment layers by biological activity (bioturbation).It can also be re-mobilised back to the water columnwhen disturbed by erosion processes.

(d) Degradation in sea water. Both tracer fractions aresubject to degradation in sea water (Section 2.4).

2.1. Transport with ocean currents

Evolution of the total (dissolved and particle-bound)concentration of the pollutant C at a fixed locationresults from the sum of sources, sinks, and mechanicaltransport of a flow field. The latter has a key role inshaping the pollutant's field in sea water. It has twocomponents: transport governed by the averaged currentvelocity field (advection) and transport due to thepresence of the random chaotic component in thevelocity field (diffusion). Turbulence in the ocean isdetermined by the current velocity gradients, surfaceand deep perturbations, and sea water stratification. Itplays an important role in the intensity of the diffusionprocesses and thus the pollutant's spatial distribution(Baumert et al., 2005). Eulerian transport of tracers dueto advection and turbulence is calculated in FANTOM ina similar way to that used by Pohlmann (1987) whereonly passive transport of conservative and dissolvedtracers in the North Sea is considered. Additionally,tracers in FANTOMmay undergo other processes in seawater (Fig. 1) acting as sources (QC) or sinks (RC) in themodel, leading to the following formulation:

∂C∂t

¼ ∂∂x

Kx∂C∂x

� �þ ∂∂y

Ky∂C∂y

� �þ ∂∂z

Kv∂C∂z

� �

� u∂C∂x

þ v∂C∂y

þ w∂C∂z

� �

þ QCðt; x; y; zÞ � RCðt; x; y; zÞ ð1ÞHorizontal flow field components in eastern and nor-

thern directions (u and ν) at everymodel grid point are inputparameters required for calculating the horizontal advectionof the tracer concentration. These parameters are providedby the ocean circulation model (Section 2.5.2), with thevertical component of the flow field w being calculatedfrom u, ν using the continuity equation. In addition thehorizontal and vertical turbulent diffusion is calculatedusing the horizontal diffusion coefficients (Kx and Ky) andthe vertical diffusion coefficient (Kv) which are also pro-vided by the ocean circulation model (Section 2.5.2).

2.2. Air–sea exchange

Air–sea exchange is believed to be the major pathwayfor atmospheric input to oceans and seas for many orga-

nochlorine contaminants (Iwata et al., 1993; Bidleman etal., 1995; Wania and Mackay, 1999; Lakaschus et al.,2002). Atmospheric deposition can occur as dry gaseousor dry particle deposition, or as wet deposition of gasesand particles incorporated in rain droplets or snow.Hence,the net flux of a pollutant to the sea surface from theatmosphereFsurf (ngm

−2 s−1) is represented in FANTOMby the net gaseous air–sea flux Fa–w (Appendix A), thedry particle deposition fluxFdry (Appendix B) and the wetdeposition flux Fwet(Appendix C):

Fsurf ¼ Fwet þ Fdry þ Fa−w ð2Þ

Previous studies have shown that gaseous air–seaexchange dominates over wet and dry particle deposi-tions of organochlorine compounds. Exceptions occur inregions and seasons with intensive precipitation andareas with a high concentration of atmospheric aerosolparticles. The relative importance of the different me-chanisms of the air–sea exchange is still in debate.

2.3. Phase distribution

Contaminants may be present in sea water eitherfreely dissolved or bound to the suspended particulatematter (SPM). Redistribution between the dissolved andparticulate phases essentially affects the dynamics of thetracer concentration distribution in the marine environ-ment. The dissolved tracer fraction follows the path ofthe water masses, while the particles bound fractionquickly sedimentises and remains in areas where sedi-mentation is promoted.

Particulate organic carbon (POC), an organic carbonfraction of SPM, is calculated in FANTOM (Appendix D)and is used as a sorbing matrix for POPs. Such anapproach is commonly used in modelling POPs accumu-lation in biota (Skoglund and Swackhamer, 1999) andtheir export to the deep sea (Scheringer et al., 2004).

The substance fraction bound to POC, fPOC is cal-culated (Skoglund and Swackhamer, 1999; Scheringeret al., 2004) as:

fPOC ¼ Kocd CPOC

Kocd CPOC þ 1ð3Þ

The organic carbon–water equilibrium partitioncoefficient Koc(l kg

−1) is compound specific (AppendixE) and CPOC (kg l−1) is the concentration of POC in thesolution (Appendix D).

Eq. (3) implies that the transfer of POPs from theparticulate phase to the dissolved phase and thus also theirtransport behaviour are controlled by the abundance of

Table 1Parameters used in FANTOM

Parameter Symbol Value Units Reference

Intercept of the temperature dependent b 10.14±0.55 – Sahsuvar et al., 2003Henry's law constant 7.54±0.54 – Kucklick et al., 1991Slope of the temperature dependent m −3208±161 K Sahsuvar et al., 2003Henry's law constant −2382±160 K Kucklick et al., 1991Dry particle deposition velocity νdep 2×10−5 m s−1 McMahon and Denison, 1979; Slinn, 1983Specific aerosol surface θ 1.5×10−4 m2 m−3 Pekar et al., 1998Adsorption constant s 0.17 Pa m Junge, 1977Octanol–water partitioning coefficient Kow 3.98×103 – Klöpffer and Schmidt, 2001Settling velocity of SPM νset 3×10−4 m s−1 Pohlmann and Puls, 1994Threshold shear velocity for erosion of SPM ν⁎

cr,e 0.028 m s−1 Pohlmann and Puls, 1994Threshold shear velocity for deposition of SPM ν⁎

cr,d 0.01 m s−1 Pohlmann and Puls, 1994Degradation rate in ocean water at 298 K kdeg 2.3×10−8 s−1 Klöpffer and Schmidt, 2001


POC. This implies that increasing POC content willtransfer the chemical from the dissolved to the particulatestate. Higher POC concentrations result in lower con-centrations of POPs (in ng kg−1) in the particulate phase.Most of the POC present in sea water is in the form ofparticles (Appendix D), which sink to the sea bed bygravitational settling with a sinking velocity νset (m s−1).Correspondingly, the flux of chemicals bound to POC thatis removed from the upper sea layers togetherwith sinkingparticles Fset(μg m−2 s−1) is calculated as:

Fset ¼ msetd fPOCd C ð4Þ

Many organic compounds are lipophilic (fat soluble)and hydrophobic. Lipophilic compounds are characte-rised by high values of octanol–water partitioning co-efficient, Kow(N10

5) (Appendix E). These are mostlybound to POC and tend to disperse and accumulate inthe sediment rather than in the water column.

2.4. Degradation in sea water

Combined abiotic (due to photolysis and hydrolysis)and biotic degradation in sea water is represented in themodel by a first order rate decay coefficient, kdeg (s

−1),with the higher order kinetics being neglected. It is assu-med that degradation is linearly dependent on thecompound total concentration dC /dt=−kdeg·C. No mea-surements of degradation in sea water exist. There is noexperimental evidence suggesting that anthropogenicorganic compounds bound to marine SPM are moreresistant to degradation processes than those dissolved insea water. Consequently the kdeg value for γ-HCH(Table 1) has been chosen on the basis of a thoroughcompilation of physical–chemical properties of selectedPOPs (Klöpffer and Schmidt, 2001; Lammel et al., 2001).

It is a recommended value for fresh water divided by afactor of 10 to account for reduced biotic degradation insea water relative to fresh water. Following the expertestimates (EU TGD, 1996) the kdeg value is assumed todouble per 10 K temperature increase. Degradation in thesediment is neglected.

2.5. Model setup

2.5.1. Model areaThe model covers the southern and central North Sea

up to 57.1°N (Fig. 2), a shallow region with mean depthof 50 m and a maximum depth of 160 m. The horizontalresolution of the model is 1.5′ by 2.5′ (corresponding to2.5–3 km) and there are 21 vertical layers of varyingdepth, i.e. 5 m in the upper 50 m, and 10 m in lowerlayers. The number of vertical layers is different fromone grid point to another depending on the local depth.

In the North Sea the distribution and mixing of watermasses is largely subject to tidal currents, meteorologicalconditions, and run-off from rivers and the Baltic Sea.Westerly winds prevail over the North Sea. The pre-dominant circulation driven by winds and tides is anti-clockwise along the North Sea coast causing shortflushing times (time needed for the water mass to beexchanged). That means that the existing climate happensto be favourable for the North Sea ecosystem. However,this circulation pattern can regionally be reversed whentransient prevailing easterly winds cause an extension ofwater mass flushing times (Sündermann et al., 2002).

The flushing time of water, calculated by the inflowsand outflows, is estimated to be about 1 year in the entireNorth Sea (OSPAR, 2000). However the flushing timevaries in different subregions of the North Sea: it rangesfrom 28 days in the northern part to 40 days in thecentral North Sea (Lenhart and Pohlmann, 1997). The


prevailing currents cause polluted coastal waters to havehigh residence time (time it takes for a substance toleave the water body, e.g. the North Sea) and be trans-ferred along the coastline. This aspect is of key im-portance, because lying between land and sea coastalhabitats are subject to a range of influences and areparticularly sensitive to anthropogenic pressure.

2.5.2. Ocean circulationThe transport processes in FANTOMwhich are driven

by ocean currents are calculated from the distributions ofthe flow field (Section 2.1) available from ocean cir-culation models. In this study the HAMburg Shelf OceanModel (HAMSOM) was coupled with FANTOM (Fig. 3)for the entire simulation period. HAMSOM is a baro-clinic, primitive equation circulation model based on asemi-implicit numerical scheme described elsewhere(Backhaus, 1985; Pohlmann, 1996). HAMSOM coversthe same domain and uses the same spatial resolution of1.5′ by 2.5′ as FANTOM (Fig. 2). Simulations werecarried out with a time step of 10 min. Atmosphericforcing for HAMSOM is calculated using the ECMWFERA-40 data at a 6 h time step and 1.125° spatialresolution (ECMWF, 2005). Boundary conditions for the

Fig. 2. FANTOM domain and bathymetry (depth in m) on a grid of 1.5′ by 2.5observational time series were available for the model evaluation are labelleOceanographic Data Centre (DOD, 2005), with data for points P, N and O colocations of three stations (Westerland, Kollumerwaard and Lista) where meshown (Lista lies outside of the modelling domain). River mouths are indicatNordzeekanaal, Nieuwe Waterweg and Haringvliet). A dashed line along 3°

open ocean are obtained from HAMSOM applied to theentire North Sea and a part of the north-eastern Atlantic(Pohlmann, 1996).

The results of the circulation model (e.g. flow fieldsand their variances, vertical diffusion coefficients, seatemperature and salinity distribution and sea surfaceheight— see Fig. 3) are stored as daily means averagedover two periods of the predominant semidiurnal lunartide M2 (Bartels, 1957). Such a coupling approach hasbeen used previously (Pohlmann, 1987; Luff and Pohl-mann, 1995).

2.5.3. Compound selectionLindane is one of the most widely used pesticides

globally and it has been used for a long time. It was chosenfor this study as it is a semi-volatile organic compoundwhich is mostly observed in the dissolved phase in water.It possesses a relatively high water–air partitioning coef-ficient and a low octanol–water partitioning coefficient(Table 1).

Lindane consists of about 98%ofγ-HCH, an isomer ofthe technical HCH which contains five stable isomers,namely α (60–70%), β (10–12%), γ (6–10%), δ (3–4%)and ε (3–4%) (Willet et al., 1998). The last two of these

′ (corresponding to 2.5–3 km). Geographical locations of points whered with letters. Data for points A to M were provided by the Germanming from Dutch database Waterbase (DONAR, 2005). Geographicalasurements of γ-HCH atmospheric concentrations were available areed (the rivers Rhine and Meuse drain into the North Sea through Ijssel,E shows section referred to in Fig. 9.

Fig. 3. The input–output diagram for the 1995–2001 integration of HAMSOM and FANTOM.


compounds are not routinely found in environmentalsamples. Technical HCHwas banned in the 1980 becauseof its toxic effects in the environment. γ-HCH is still usedin many countries as an insecticide in agriculture, as awood and building preservative, and as a biocide tocombat lice and scabies. Its use in the countries adjacent tothe North Sea is restricted (SRU, 2004). But despiterestrictions and bans, α-β- and γ-HCHs are still widelyfound in the North Sea. A large proportion enters thewater through flooding, run-off from treated areas, andincorrect disposal of left-over mixtures into farm drainsand sewage systems. Atmospheric depositions can alsocontain γ-HCH transported from remote regions (Seme-ena and Lammel, 2005).

Inventories and knowledge about HCH isomers ingeneral and γ-HCH in particular are relatively complete.γ-HCH is subject to regular monitoring in the seathrough national (Bund–Länder monitoring programmein Germany) and international monitoring programmes(OSPAR, HELCOM).

2.5.4. Initial and boundary conditionsPhysical–chemical properties of a chemical, environ-

mental parameters and release data are required input datafor the model (Fig. 3). Initial and boundary conditionsinclude initial distribution of the selected chemical and its

values on the boundaries throughout the simulation pe-riod. For the studied domain (Fig. 2), the followingboundaries are considered in the model: air–sea interfaceand boundaries with the neighbouring water bodies (i.e.Atlantic Ocean, Baltic Sea, English Channel) and theinflowing rivers that drain into the North Sea.

Daily mean distributions of flow components, theirvariances, sea surface height, diffusion coefficients, sali-nity and sea temperature were obtained from the cor-responded HAMSOM simulations (Section 2.5.2).

The initial distribution of γ-HCH concentrations(Fig. 4) and its values on the sea boundaries were in-terpolated using measured non-filtered samples (Theo-bald et al., 1996) and therefore represent the totalconcentration. Initial conditions for the concentrationare defined by the function C0 as C(x,y,z,0)=C0(x,y,z).Boundary conditions for the water boundaries B areprescribed:

Cjðx;y;zÞaB ¼ C Bðx; y; z; tÞ ð5Þ

whereCB is the concentration at the water boundary. Theboundary condition at the coast is ∂C /∂n|(x,y,z)∈Γ=0. Atthe bottom, with the bottom depthD, the tracer flux to thebottom is calculated according to Eq. (4) F|z =D=−Fset.

Fig. 4. Initial distribution of γ-HCH concentration (ng l−1) in the North Sea in the summer, 1995 interpolated from measurements indicated bynumbers.


When the re-suspension occurs (Appendix D) the tracerflux at the sea bottom is directed upwards.

Because the North Sea is a tidally influenced shelf sea,the inflowing rivers have estuarine characteristics, i.e. the

Fig. 5. a) γ-HCH concentrations in the atmosphere at the EMEP stations anlocations of the measurement sites are shown on the map (Fig. 2).

tide and river flow interact. The phenomenon of mixingwatermasses is taken into account by usingmeasurementsat the last tide-free gauge station of each river (Lenhartand Pätsch, 2004). The river loads of the selected

d b) γ-HCH loads (Qriv×Criv) in the continental rivers. Geographical

Table 2List of stations used for the model evaluation, their geographical location, number of measurements, months of sampling, mean and range ofmeasurements and the model results, and correlation coefficient

Symbol onthe map

Lat. Long. Number ofmeasurements

Months ofsampling

Measurements: mean andrange (ng l−1)

Model results: mean andrange (ng l−1)

Correlationcoefficient

A 54.00° N 8.00° E 36 5–9 1.50 (0.37 – 3.00) 1.48 (0.58 – 3.03) 0.94B 53.72° N 7.45° E 14 11–2 0.70 (0.10 – 3.00) 1.41 (0.81 – 2.24) 0.35C 53.85° N 8.06° E 14 11–2 0.83 (0.10 – 3.00) 1.27 (0.68 – 1.88) 0.40D 53.86° N 8.13° E 14 11–2 0.80 (0.10 – 2.00) 1.32 (0.69 – 1.32) 0.30E 53.98° N 8.22° E 26 11–3, 5, 7–9 1.86 (0.10 – 3.00) 1.32 (0.67 – 2.47) 0.74F 54.05° N 7.86° E 13 11–2 0.82 (0.10 – 2.00) 1.08 (0.62 – 1.52) 0.23G 54.22° N 8.38° E 41 1, 2, 5–9 1.65 (0.43 – 3.95) 1.77 (0.86 – 3.00) 0.77H 54.25° N 7.50° E 25 5–9 1.63 (0.42 – 2.55) 1.19 (0.68 – 1.76) 0.93I 54.67° N 6.33° E 20 5–9 1.45 (0.30 – 3.49) 0.95 (0.38 – 1.31) 0.77J 55.00° N 6.25° E 12 5–8 0.71 (0.23 – 1.32) 0.67 (0.35 – 1.23) 0.84K 55.00° N 8.25° E 21 5–9 1.25 (0.42 – 2.75) 1.50 (0.74 – 2.33) 0.86L 55.50° N 4.17° E 18 5, 9, 12 0.58 (0.17 – 0.95) 0.58 (0.22 – 0.91) 0.84M 56.00° N 3.00° E 4 7 0.42 (0.28 – 0.70) 0.49 (0.38 – 0.59) 0.45N 52.58° N 3.53° E 21 2, 3, 5, 8, 11, 12 0.79 (0.20 – 1.80) 1.00 (0.63 – 1.51) 0.83O 51.96° N 2.68° E 6 2, 5, 8, 11 1.25 (0.50 – 0.70) 1.08 (0.96 – 1.15) 0.68P 54.42° N 4.04° E 6 8 0.72 (0.10 – 1.20) 0.67 (0.32 – 1.07) 0.94


chemical in FANTOM from the European continentalrivers including the Elbe, the Weser, the Ems, the Rhineand Meuse, and the Scheldt are calculated as a product ofthe daily fresh water dischargeQriv (m

3 s−1) (Lenhart andPätsch, 2004) and the concentrations of the compoundCriv (ng l−1) (Fig. 5b). Data for Criv from the Germanrivers were provided by the various EnvironmentalAgencies (ARGE-ELBE, 2005; NLOE, 2005). Thepollutant concentrations in the Dutch rivers are availablefrom the Dutch online database Waterbase (DONAR,2005). The loads of POPs reported toOSPAR (2000)wereused for the Thames and the Humber (Fig. 2). Values ofCriv are available with different temporal resolutions (e.g.monthly means) and were interpolated using splineinterpolation to fit the temporal resolution of the model.

Boundary conditions at the air–sea interface are basedonmeasuredmonthly averaged total (gaseous and particlesorbed) concentration of γ-HCH in the air and its totalconcentration in precipitation Cpr (Fig. 5a) available fromthe EMEP online database (EMEP, 2005). The measuredvalues from several EMEP stations (Fig. 2) were inter-polated to the whole domain. The wind speed and direc-tion, air temperature and precipitation rateP are calculatedusing the ECMWF 6 hourly ERA-40 data (ECMWF,2005). The tracer's flux on the sea surface is determinedaccording to Eq. (2): F|z = 0=Fsurf.

Compound specific physical–chemical properties (e.g.Kow and kdeg) are listed in Table 1. The Henry's lawconstants were calculated as a function of temperatureusing values recently given by Sahsuvar et al. (2003) forall model simulationswith the exception of the experimentwhere values reported byKucklick et al. (1991)were used.

3. Results and discussion

The model simulations were carried out for theperiod from July 1995 to December 2001 with a timestep of 10 min. Model results were stored as daily ave-raged values.

3.1. Model evaluation

Model evaluation consisted of two stages.

(a) calculated γ-HCH concentrations in sea waterwere compared with the measured ones (Section3.1.1).

(b) model sensitivity to input parameters was ana-lysed (Section 3.1.2).

Time series for observed γ-HCH concentrations in seawater covering the simulation period were available in 16locations (Fig. 2). Most of the data were provided by theGerman Oceanographic Data Centre (DOD, 2005) cove-ring only the German exclusive economic zone of theNorth Sea. Time series at locations N, O and P are fromDutch database Waterbase (DONAR, 2005).

3.1.1. Comparison between modelled and measured seawater concentrations

Daily averaged concentrations in the uppermost sealayer were compared with the individual measurementsand plotted for six of these points (Fig. 6) located indifferent subregions of the North Sea and with sufficientnumber of measurements. A clear seasonal pattern with

Fig. 6. Observed (dots) and FANTOM (solid line) γ-HCH concentrations (ng l−1) in the surface layer at the locations A, G, K, L, N, P. Geographical locations of points are given in Table 2 and areshown in Fig. 2. 9

T.Ilyina

etal.

/Journal

ofMarine

Systems63

(2006)1–19

Fig. 7. Observed versus FANTOM γ-HCH concentrations (ng l−1) in the surface layer at the locations A, G, K, L, N, P. Geographical locations ofpoints are given in Table 2 and are shown in Fig. 2.


higher concentrations in summer is predicted for mostyears in all locations. Similar patterns are also found inthe atmospheric concentrations (Fig. 5a) and in thecontinental river loads of γ-HCH (Fig. 5b). The highestconcentrations, up to 3.95 ng l−1 (Table 2), were found atlocations near the coast, e.g. A (Fig. 6a), G andK (Fig. 6band c); the lowest ones, of less than 0.2 ng l−1 (Table 2),were found in the open North Sea at locations L and P(Fig. 6d and f). Concentration of γ-HCH had decreasedin all locations throughout the simulation period. Thebiggest drop in γ-HCH concentrations, 6–12 times, was

found in the German Bight and near the southern NorthSea coast, with the smallest, 4–5 times, in the centralNorth Sea. Both observed and modelled concentrations insea water at all locations increased during the first half ofthe simulation period, with highest concentrations in1997–1998. During the second half of the simulation, inthe years 1998–2001, there are negative trends in con-centrations at all locations, both in observations and pre-dictions by FANTOM. Also atmospheric concentrationsof γ-HCH reflect such pattern with the highest concentra-tions of γ-HCH detected in precipitation during spring

Fig. 8. Observed (dots) and FANTOM γ-HCH concentrations (ng l−1) in the surface layer at the location A, under different model setup: solid line—main model simulation; dashed line—when γ-HCH concentrations inflowing via the lateral boundaries were equal zero; dash–dot line—when air–sea exchange was calculated using Henry's law constants given by Kucklick et al. (1991). Geographical locations are given in Table 2 and are shownin Fig. 2.


and summer of 1997 and decreased concentrations during1998–2001 (Fig. 5a).

At most of the locations, namely at A, E, G to P,measurements were made during warm seasons (May–September) corresponding to the predicted maxima. Atlocations B, C, D and F measurements were conductedbetween December and March.

In order to evaluate the model results, measured γ-HCH concentrations are plotted against calculated onesin scatter plots (Fig. 7) for the same locations as in Fig. 6.Correlation coefficients for all locations where timeseries were available are presented in Table 2. Goodcorrelation was found for locations A, H and P withcorrelation coefficients of 0.94, 0.93 and 0.94 respec-tively. Lower correlation coefficients between 0.23 and0.40 were found for the locations F, D and B where thetime series consist of the measurements conducted bet-ween December and March. The number of measure-ments and their temporal resolution, in particular at thelocations P, O and M, make further statistical analysisinappropriate.

Time series of γ-HCH concentration in the southernNorth Sea surface water reported elsewhere (SRU, 2004)show a gradual decrease since the 1980s due to therestriction in technical HCH application in the countriesadjacent the North Sea. This decrease is captured by themodel during the studied period. Furthermore, model

results suggest that γ-HCH concentrations declined inthe entire modelling domain.

3.1.2. Uncertainty and sensitivity analysisAlthough γ-HCH is a rather well studied contami-

nant and subject to national and international monitoringprogrammes, data availability remains an importantconstraint for constructing an observation based, spa-tially resolved model. Unlike regular monitoring in theatmosphere, where monthly or even weekly mea-surements of γ-HCH are available for the Europeanregion (e.g. EMEP, 2005), observations of sea waterconcentrations are discrete in time and space due tooperational constraints. Many model input parametersare of limited availability and have to be estimated byextrapolation and interpolation methods. Those valuesthat are available for a given parameter often exhibitlarge variability and uncertainties. Also there are un-certainties introduced by the poor temporal and spatialresolution in measured data, both in the sea water and inthe atmosphere.

Experiments aimed at explaining the deviations inmodelled and observed γ-HCH concentrations wereperformed (Fig. 8). In the first experiment it was assu-med that the concentration of γ-HCH in water flowinginto the North Sea from the English Channel and fromthe northern boundary of the modelling domain is zero,

Fig. 9. Vertically averaged annual mean concentrations of γ-HCH (ng l−1) in the North Sea calculated by FANTOM for each year from 1996 to 2001.


i.e. CB=0 (Eq. (5)). The resulting time series of theconcentration was plotted for location A in the GermanBight where the impact of inflow from these boundarieswas expected to be less pronounced than elsewhere. Inthe second experiment the Henry's law constanttemperature dependency was calculated using the valuesobtained by Kucklick et al. (1991) instead of those bySahsuvar et al. (2003) used in other model runs (Table1). Daily averaged concentrations in the uppermost sealayer resulting from the main simulation and both expe-

riments were compared with the individual measure-ments at the location A (Fig. 8).

No regular measurements of γ-HCH concentrationsin the English Channel and on the northern sea boundarywere available for this study. Since concentrations in thewhole North Sea decreased they have probably alsodecreased in the English Channel and on the northernsea boundary without being captured by the model. Thisoffers an explanation for the overestimation in calculat-ed concentrations during the years 2000–2001. The


overestimation is most pronounced at location N, lyingin the passage of the English Channel water (Fig. 6e).

Correlation coefficients calculated for locations B, C,D and F (Table 2) are lower, even though they aresituated only within a few model grid points from loca-tions A, E, G and H, where correlations are much better.Calculated concentrations at locations B, C, D and F arehigher than observed (Table 2). One explanation for thisdiscrepancy could be that during winter the strongerwinds which occur over the North Sea induce highercurrent speeds enhancing stronger water mass flushing.Consequently there is an increase in the amount of lesspolluted waters from the open lateral boundaries beingtransported into the German Bight. As mentioned abovethe decrease in γ-HCH concentrations in the EnglishChannel and on the northern boundary might not becorrectly captured by the model leading to overesti-mated winter concentrations. This hypothesis is verifiedby the modelling experiment when γ-HCH concentra-tions in the English Channel and on the northern boun-dary of the modelling domain were set to zero for theentire simulation period. Fig. 8 indicates that concen-tration without inflow from the lateral boundaries was asignificant component of the γ-HCH burden until thesecond half of the year 2000. Furthermore, ongoingcycling of earlier emitted γ-HCH leakages from inap-propriate disposal may play a role.

Atmospheric concentrations of γ-HCH used for thisstudy were measured at coastal stations. Therefore there isno information on possible concentration gradients overthe whole North Sea and in the subregions remote from thepolluted coastal areas. Concentrations of γ-HCH inprecipitation measured in Lista, Westerland and Kollu-merwaard exhibit significant differences (Fig. 5a). Forsome pollutants, e.g. for polychlorinated biphenyls(PCBs), air concentrations over the coastal waters areexpected to be higher than those over the open oceanwaters because of their vicinity to sources. However, thereare indications that air measured in Lista originate from theNorth Sea about 2/3 of the time (Haugen et al., 1998).Since the measured air concentrations were available onlyfromLista theywere used for thewholemodelling domain.

The seasonal pattern predicted by the model withhigher γ-HCH concentrations during summer thanduring winter emphasises the importance of temperaturedependency of the air–sea exchange. For the GreatLakes of North America and for the Arctic Ocean it hasbeen found that during summer, the air–sea gas transferis net volatilisational, whereas during colder period it isdepositional (Ridal et al., 1996; Jantunen and Bidleman,1996). Measurements suggest (Sahsuvar et al., 2003)that temperature dependency of Hc has a profound effect

on gas exchange predictions. This hypothesis is illus-trated by Fig. 8 where γ-HCH concentrations resultingfrom model calculations using two different temperaturedependency coefficients of Hc are shown. Recent expe-rimental data (Sahsuvar et al., 2003) used in this studylead to higher (10–50%) sea water concentrations.Therefore as parameterisation of the air–sea exchangeseems to play a key role, its uncertainty should bereduced. Discrepancies in the experimentally deter-mined values of Hc and possibly physical processes notincluded in the model (see Appendix A) contribute tothis uncertainty.

3.2. Horizontal and vertical distributions in sea water

Measurements in the North Sea's uppermost layer of5 m thickness compiled during summer 1995 andinterpolated on the modelling domain of FANTOM(Fig. 4) show clear gradients of γ-HCH with the highestconcentrations in the southern regions of the North Sea,in the estuaries of rivers and in the English Channel. Thehighest concentration of 8 ng l−1 was found close to themouth of the River Elbe.

Annual mean γ-HCH concentrations averaged ver-tically for the whole modelling domain in the years1996–2001 were calculated (Fig. 9). These results showpositive gradients of concentration towards the coasts; asimilar pattern was found in the initial distribution(Fig. 4). During 1995–1997 total concentration of γ-HCH in the continental coastal water was above 1 ng l−1

with a concentration of more than 25 ng l−1 in the RiverElbe estuary. As with the γ-HCH concentration at theselected locations (Fig. 6), concentrations in the wholemodelling domain decreased. This effect is morepronounced in the southern regions of the North Seawhere initial concentrations had been highest.

Observed spatial distribution of γ-HCH total con-centration with decreasing concentrations towards thenorth-western part of the North Sea, similar to thoseshowed by this modelling study (Fig. 9), have beenreported in the scientific literature (Gaul, 1988; Theo-bald et al., 1996; Hühnerfuss et al., 1997, Lakaschus etal., 2002) and by the OSPAR Commission (OSPAR,2000). Low concentrations in the north-western NorthSea (Fig. 9) are due to the inflow of cleaner Atlanticwater. Along the coastline from southern Britain toDenmark close to the estuaries there are high concentra-tions; this indicates the importance of the river inflowwhich modulates the local circulation. In fact, measure-ments in coastal waters close to large estuaries suggestthat river input is a significant source for γ-HCH in theNorth Sea (Sündermann, 1994; Hühnerfuss et al., 1997).

Fig. 10. Vertical sections from south to north along 3° E (see Fig. 2) of sea temperature (colours, 0C) and salinity (lines) in (a) January and (c) August1997 calculated by HAMSOM; (b) and (d) the corresponding total concentrations of γ-HCH (ng l−1) calculated by FANTOM.


Another indication for that is the jagged pattern in theconcentration's time series in the German Bight (Figs.6a, b and 8) which is due to variability of the fresh waterinflow (Fig. 5b). As opposed to the German Bight, in theopen North Sea (Fig. 6d, f) remote from the pollutedestuaries, time series are smoother and sea water con-centrations seem to be balanced by the volatilisation–deposition process. This offers an explanation for theslower decrease in γ-HCH concentrations in the centralNorth Sea than in the southern German Bight. This alsoimplies that in coastal areas, waterborne inputs of γ-HCHmay be dominant, whereas atmospheric depositionis more important in the open sea and can exhibit thesame order of the magnitude as river inputs (Hühnerfusset al., 1997). A correlation between γ-HCH concentra-tions in water (Fig. 6) and its concentration in pre-cipitation and in the air (Fig. 5a) is found.

The mean anti-clockwise currents pattern commonlyobserved in the North Sea (Section 2.5.1) favours shortflushing times (stronger transport) ofwatermass along thesouth eastern coast of about 11 days and longer ones ofabout 40 days in the central North Sea (Lenhart andPohlmann, 1997; OSPAR, 2000). Weaker transports (or

longer flushing times) in some subregions of the NorthSea suggest that the concentrations of semi-volatile com-pounds, such as γ-HCH, should be more sensitive tovertical exchange rates and less dependent on thehorizontal circulation. This also implies that a relativelywater soluble compound, e.g. γ-HCH, which ends up inthese areas of the North Sea may not be subject to trans-port with sea currents to any significant extent. On theother hand, γ-HCH in the subregions with strongertransports, e.g. in the German Bight where the concentra-tions are also higher, will be transported northwardswithin the coastal current. Typical speeds of sea currentsare 0.01 m s−1 which are 2–3 orders of magnitude slowerthan in the atmosphere. This, together with the variationsin the water mass flushing times (see Section 2.5.1),implies that only those substanceswith a residence time insea water of up to a year or longer may undergo transportwith sea currents over long distances in the North Sea.

Vertical distribution of sea temperature and salinity inJanuary and August 1997 and corresponded γ-HCHdistribution from south to north along the 3° E are plottedin Fig. 10. In winter the southern North Sea is in generalwellmixed to the bottom (Fig. 10a). This is also reflected in


the distribution of γ-HCH total concentrations (Fig. 10b).During warm seasons a stable thermal stratification isformed by summer heating (Fig. 10c) enhancing verticalgradients in γ-HCH total concentrations in the northernpart of the domain (Fig. 10d). High values of about 0.6 ngl−1 are found near the sea surface, and low values of lessthan 0.2 ng l−1 below, where the Atlantic water inflows.The high γ-HCH concentrations of more than 1.2 ng l−1

were found in the continental river estuaries both in Januaryand August. As already illustrated by the time series(Fig. 6),γ-HCH concentrations inAugust were higher thanin January. Concentrations beneath the thermocline, ifpresent, do not show seasonal variations.There were nomeasurements available on the vertical distribution of γ-HCH concentrations in the North Sea. Model results sug-gest that the vertical structure of totalγ-HCH concentration(Fig. 10b, d) seems to be shaped by water column strati-fication and its seasonal variability (Fig. 10a, c).

4. Conclusions and outlook

The fate and transport ocean model FANTOM wasdeveloped to describe the oceanic fate of POPs in theNorth Sea. The processes considered are transport with seacurrents, air–sea exchange, partitioning of particulate or-ganic carbon, net sedimentation and degradation in the sea.

The model was applied to study the aquatic fate of theinsecticide Lindane (γ-HCH) in the years 1995–2001.This is the first time that such model has been applied onthe regional scale. Predicted γ-HCH concentrations inthe surface water were compared with measured ones.The model is able to reproduce the spatial distribution ofγ-HCH concentrations and capture its decrease duringthe simulation period.While there is excellent agreementfor warm seasons, FANTOM overestimates winter con-centrations of γ-HCH.

Model uncertainties and sensitivities were explored.From these it is concluded that particular data needs existwith regard to off-shore atmospheric environment and thedescription of the air–sea exchange. Concentrations in theatmosphere correlate with concentrations in sea water,indicating the significance of the atmospheric input of γ-HCH in the North Sea. Overestimations in the modelledconcentrations in 2000–2001 seem to be due to thelacking data for the open sea boundaries. There were nomeasurements on the vertical concentration structure.Therefore model performance with regard to sediment-water exchange cannot be assessed.

Although FANTOM is capable of reproducing thespatial and temporal distribution of a fairly water solublecontaminant such as γ-HCH, it remains a future task toinvestigate the fate of compounds with different phy-

sical–chemical properties and hence environmental be-haviour. FANTOM can be used to study the contributionof different sources and sinks to the mass budget. Alsothe sensitivity of POPs behaviour in the aquatic enviro-nment to the ambient processes under present and futureclimate conditions can be studied.

Acknowledgements

This study was supported by the International MaxPlanck Research School for Maritime Affairs at theUniversity of Hamburg. The authors wish to thank Dr.Nies and Dr. Theobald of the German Federal Hydro-graphic Agency (BSH) for providing unpublished data.We are grateful to Dr. Walter Puls of the GKSS ResearchCentre for his advice during the model development andvaluable discussions.

Appendix A. Gaseous air–sea exchange

The gaseous air–sea transfer of organochlorines canbe treated as a diffusion of the trace gases through spa-tially and temporally varying thin boundary layers inboth media whose thicknesses are a function of near-surface turbulence and molecular diffusivity (Schwar-zenbach et al., 1993). The kinetics of air–sea exchange inthe open ocean is driven by near-surface turbulence withwind stress being the major controlling factor. In a winddriven ocean-atmosphere system, turbulence is generat-ed due to shear, buoyancy, and large- and micro-scalewave breaking. The gas transfer dependency on windspeed over the ocean is often non-linear (Wanninkhof,1992). In low wind speeds, when buoyancy may domi-nate in generating the turbulence and there are organicfilms, this dependency is weaker. Additionally, variabil-ity is introduced by the presence of small-scale wavesand rain. Furthermore, many organochlorines are semi-volatile so they can occur in the atmosphere in bothgaseous and condensed states under ambient tempera-tures. Therefore temperature may also play a central rolein the air–sea transfer of POPs.

FANTOM uses a description of the gaseous air–sea gasexchange based on the stagnant two-film theory formulatedbyWhitman (1923) and restated by Liss and Slater (1974),with the adoption of the fugacity formulation as describedin Mackay (2001). Accordingly, the net mass transferacross the air–sea interface is expressed as a product of akinetic parameter representing the resistance to interfacialtransfer and a term representing the deviation from thechemical equilibrium between air and water as a drivingforce for interfacial transfer. The chemical equilibriumbetween the two compartments is controlled by the ambient


parameters (e.g. temperature and wind speed), physical–chemical properties of the compound and its abundance inthe environment.

The two mass transfer coefficients, u1 and u2 (m s−1)for the stagnant (unstirred) atmospheric boundary layerand for the stagnant water layer close to the air–waterinterface respectively are calculated as a function ofwind speed WS (m s−1) at 10 m above the surface usingrelationships (according to Schwarzenbach et al., 1993)u1 = 6.5·10− 4 ·(6.1 + 0.63·WS ) 0 . 5 ·WS and u2 =1.75·10−6·(6.1+0.63·WS)0.5·WS.

Fugacity capacities of air and water Za and Zw (molm−3 Pa−1) at air temperature Ta (K) and sea surfacetemperature Tw (K) are calculated as Za=1 /R·Ta andZw=1 /Hc(Tw) respectively. The ideal gas constant R(R=8.314, Pa m3 mol−1 K−1) and the Henry's lawconstant Hc (Pa m

3 mol−1) at Tw are used to describe theequilibrium partitioning of trace gases between air andwater. Experimentally derived relationships for Hc arecalculated from the temperature dependent equation(Kucklick et al., 1991; Paasivirta et al., 1999; Sahsuvaret al., 2003) using slope m (K) and intercept b:

logHc ¼ bþ m=Tw ðA1ÞThe overall exchange rate constant Dwa for volatilisa-

tion from seawater (mol Pa−1 s−1) is calculated accordingto Mackay (2001) and Wania et al. (2000) Dwa=Aw/ (1 /u1·Za+1/u2·Zw), where Aw (m2) is the surface area of thewater compartment. The gaseous exchange rate constantfor the dry gaseous deposition is calculated similar tovolatilisational one. Transfers from the atmosphere to thesea surface by dry particle and wet depositions arecalculated separately (Appendix B and Appendix C).

The net mass transfer rate (mol s−1) is calculated forthe tracer gaseous concentrations in the air Ca anddissolved in sea water Cw expressed in mol m−3:

dma−w

dt¼ Dwad ðCa=Za−Cw=ZwÞ¼ Dwad ðCad Rd Ta−Cwd HcðTwÞ

�ðA2Þ

The air–sea flux Fa–w (ng m−2 s−1) to the surface isthen re-calculated from Eq. (A2) using the tracer's molarmass M (g mol−1). The direction of the flux Fa–w isdetermined by its sign (i.e. positive values ofFa–w indicategaseous deposition, and negative values indicate volati-lisation from the sea surface).

Appendix B. Dry particle deposition

Organic contaminants sorbed to atmospheric aerosolparticles can settle to the sea surface by dry particle de-

position. The dry deposition flux from the atmosphere tothe sea surface Fdry (ng m

−2 s−1) is expressed as a productof the contaminant particle-bound concentration in air Cap

(ng m−3) and a dry deposition velocity νdep (m s−1):Fdry=Cap·νdep.

Dry deposition velocities depend on particle size,underlying surface properties, and meteorological para-meters (e.g. wind speed). For this study a uniform value ofνdep=2·10

−5 m s−1 was used. This was based on anempirical relationship between νdep and the mass mediandiameter (an average value used to describe aerosolsparticles) for oceanic conditions (McMahon and Denison,1979; Slinn, 1983) and the assumption that the γ-HCHdistribution follows the air particles size distributionwhich peaks in the accumulation mode.

Atmospheric concentrations reported by themonitoringprogrammes often represent the total (gaseous and particlesorbed) compound concentration in air. The fraction fap ofthe total tracer concentration in air sorbed by aerosolparticles is needed to estimate the dry particle depositionflux. It can be calculated based on an empiric relationwhich assumes that the equilibrium between the gaseousand aerosol particles bound fractions is determined by thesubstance vapour pressure and is independent of theparticles chemical properties (Junge, 1977; Pankow, 1987)fap=s·θ / (Pol+s·θ), wherePol is the temperature dependentsaturated vapour pressure for supercooled liquid (Pa),calculated in a way similar to Eq. (A1). The specificaerosol surface θ (m2 m−3) and adsorption constant s (Pam) used in this study (Table 1) are constants and representthe North Sea conditions (Pekar et al., 1998).

Appendix C. Wet deposition

Due to their semi-volatility POPs are episodicallyscavenged from the atmosphere by precipitation in boththe gas and particulate phases (Pankow, 1987, Bidleman,1988). During wet periods the removal of gaseous andparticle sorbed compounds dominate other depositionalprocesses. Because precipitation is intermittent and a localphenomenon, it is crucial to consider its spatial andtemporal variability.

The wet deposition flux Fwet (ng m−2 s−1) iscalculated in FANTOM as a product of the tracer con-centration in precipitation Cpr (ng l−1) which includesboth the dissolved and particulate phases and precipita-tion rate P (m s−1) Fwet=Cpr·P.

Spatial and temporal distributions of Cpr and P arebased on measurements as described in Section 2.5.4.In the present model configuration no distinction ismade between precipitation scavenging of vapours andparticles.


Appendix D. Particulate organic carbon content

The total SPM in sea water consists of inorganic andorganic portions, namely microflocs of mineral particlesand organic matter. The fine sediment (mud) or particlessmaller than 20 μmmake up to 85% of SPM in the NorthSea (Eisma and Kalf, 1987). The composition of SPM insea water is controlled by a number of factors such as therate of primary productivity, the amount of lithogenousinput to the sea, and the sinking rate. Biogeochemicalprocesses leading to release and/or uptake of elementsfrom sea water during the horizontal or vertical fluxes ofSPM are also responsible for its composition. The shelfseas are normally more productive than the open ocean.SPM concentrations in the North Sea, for instance, are inthe order of 0.1–100 mg l−1 (Puls et al., 1994). Much ofit is biogenic consisting of detritus and planktonic algae(Eisma and Kalf, 1987). In winter the fraction of organicmatter is about 20% of the total SPM, whereas in otherseasons the appearance of SPM may be dominated byphytoplankton (Puls et al., 1994).

The concentration of POCCPOC (kg l−1) in FANTOM is

a composite of the concentrations of biogenic organic carbonCbio (kg l−1) and sediment organic carbon Csed (kg l−1):CPOC=Cbio+Csed. The biogenic concentration Cbio repre-sents the POC of phytoplankton, zooplankton, bacteria, andsmall and large detritus suspended in the water column.These concentrations are calculated by an ecosystemmodel,described elsewhere (Pätsch et al., 2002).

Sediment organic carbon concentration Csed is derivedfrom plant and animal detritus, bacteria or planktonformed in situ, or derived from natural and anthropogenicsources in river catchments. In the shallow regions understormy conditions the bottom sediment can reach back tothe sea surface. Measurements (Van der Zee and Chou,2005) suggest that this sediment is an importantcontribution to the POC burden in the North Sea, espe-cially in winter when sea currents are stronger and stormevents are more frequent.

The sediment organic carbon is calculated in FAN-TOM as a portion of the total fine sediment distributed inthe upper 2 cm of the sediment bed, a layer where nearlyall the benthic biomass is found (Pohlmann and Puls,1994). The POC content pPOC of the bottom fine sedimentpmud (in % of dry mass) is calculated based on themeasurements reported by Wiesner et al. (1990) as

pPOC ¼ 5−1:4logðpmudÞ2:6

ifpmudb50%pmudN50%

�.

The bottom sediment enters the near-bottomwater layerof themodel due to erosion. It is diffused to the upper waterlayers and may be returned to the bottom sediment viadeposition by settling. In FANTOMdeposition of SPMand

erosion of bottom sediment are controlled by the bed shearvelocity. The latter is a characteristic of the bed shear stressthat depends on the wind and density driven currents, tidalcurrents andwaves. The shear stress is calculated accordingto the formulation given by Soulsby (1997). Wave inducedshear stress is dominant in shallow waters, such as theNorth Sea (Puls et al., 1994). Therefore, only bed shearvelocity due to waves ν⁎ is considered in the model. Thethreshold shear velocity for erosion ν⁎

cr,e is 0.028 m s−1 andthe one for deposition ν⁎

cr,d is 0.01 m s−1 (Pohlmann andPuls, 1994). Thus, if ν⁎Nν⁎

cr,e, sediments from the disturbedsea bottom enter the water column and are distributeduniformly in the bottomwater layer (Sündermann and Puls,1990). The amount of eroded SPM depends on the fractionof fine sediment at the bottom. Eroded SPM is then diffusedthrough the water column. SPM in the water column issubject to gravitational sinking with the uniform settlingvelocity νset of about 25 m day−1. Further, if ν⁎bν⁎

cr,e, aportion of SPM in the bottom water layer is deposited backto the bottom sediments.

In the sediment the deposited SPM is re-distributedvertically by benthic organisms constantly disturbing thesediment by burrowing and feeding.Vertical bioturbation isdescribed in the model as a diffusive transport process(Pohlmann and Puls, 1994).

In the water column SPM settles due to gravitationalsinking. In the deep sea this process is the ultimate sink forSPM, whereas in the shallow regions re-suspension (ero-sion of previously deposited SPM) caused by currents andwaves carries it back to the water column.

Finally, knowing the mass of the bottom sediment inthe water column at a specific time and its density allowsCsed is calculated.

The approach described here provides first orderrealistic spatial and temporal POC distributions. The fullSPM dynamics calculations are described elsewhere(Sündermann and Puls, 1990; Pohlmann and Puls,1994; Puls et al., 1994).

Appendix E. Fraction bound to particulate organiccarbon present in sea water

The calculation of the particle-bound fraction of POPsis based on the assumption that the equilibrium betweenthe tracer's concentrations in dissolved and particulatephase is established instantaneously (Skoglund et al.,1996; Skoglund and Swackhamer, 1999). As equilibra-tion time is neglected the following relationship isfulfilled Cp /Cw=Koc·CPOC, where Cp (ng l−1) and Cw

(ng l−1) are the concentrations of POPs associated withPOC and in the dissolved phase respectively; and CPOC

(kg l− 1) is the concentration of POC in solution


(Appendix D). The organic carbon–water equilibriumpartition coefficient Koc (l kg

−1) is commonly employedin modelling organic chemicals (Koziol and Pudykie-wicz, 2001; Malanichev et al., 2004) to calculate theirpartitioning in different aquatic particulate matrices. TheKoc value is compound specific and related to thedimensionless octanol–water partition coefficient Kow

according to Karickhoff (1981) as Koc=0.411·Kow. Theoctanol–water partition coefficient is the ratio of theconcentration of a chemical in octanol and in water atequilibrium at a specified temperature (Karickhoff,1981). Octanol is an organic solvent that is used as asurrogate for natural organic matter. This approach todescribe the phase partitioning in water is appropriate forthe fast uptake into the organic matrix. The uptakeprocess may be far from the equilibrium during higherbiological activity (i.e. during fast phytoplankton gro-wth). Then a kinetic description would be more adequate(Skoglund et al., 1996; Axelman et al., 1997).

References

ARGE-ELBE, 2005. Annual Report. Arbeitsgemeinschaft fr dieReinhaltung der Elbe, 2005. http://www.arge-elbe.de/.

Axelman, J., Broman, D., Näf, C., 1997. Field measurements of PCBbetween water and planktonic organisms: influence of growth, particlesize, and solute–solvent interactions. Environ. Sci. Technol. 31,665–669.

Backhaus, J.O., 1985. A three-dimensional model for the simulation ofshelf sea dynamics. Dtsch. Hydrogr. Z. 38, 167–187.

Bartels, J., 1957. Gezeitenkräfte. Handbuch der Physik, Band XLVIII,Geophysik, II.

Baumert, H., Simpson, J., Sndermann, J. (Eds.), 2005. MarineTurbulence — Theories, Observations and Models. CambridgeUniversity Press, Cambridge.

Bidleman, T.F., 1988. Atmospheric process. Environ. Sci. Technol. 22,361–367.

Bidleman, T.F., Jantunen, L.M., Falconer, R.L., Barrie, L.A., Fellin, P.,1995. Decline of hexachlorocyclohexane in the arctic atmosphere andreversal of air–sea gas exchange. Geophys. Res. Lett. 22, 219–222.

Dachs, J., Lohmann, R., Ockenden,W.A.,Mejanelle, L., Eisenreich, S.J.,Jones, K.C., 2002. Oceanic biogeochemical controls on globaldynamics of persistent organic pollutants. Environ. Sci. Technol. 36,4229–4237.

DOD, 2005. German Oceanographic Data Centre, online dataset,http://www.bsh.de/.

DONAR, 2005. Dutch online database Waterbase, http://www.water-base.nl/.

ECMWF, 2005. European Centre for Medium-Range Weather Forecasts.Re-Analysis ERA-40 online dataset, http://www.ecmwf.int/.

Eisma, D., Kalf, J., 1987. Distribution, organic content and particle size ofsuspended matter in the North Sea, Netherlands. J. Sea Res. 21,265–285.

EMEP, 2005. EMEP data online, http://www.emep.int/.EU TGD, 1996. Technical Guidance Document in support of Commission

Directive 93/67/EEC on Risk Assessment for new notified substances,Commission Regulation (EC) No 1488/94 on Risk Assessment for

existing substances and Directive 98/8/EC of the European Parliamentand of the Council concerning the placing of biocidal products on themarket. European Chemicals Bureau, ISPRA.

Gaul, H., 1988. Der Eintrag von Organohalogenverbindungen ber dieAtmosphäre in die Nordsee. Dtsch. Hydrogr. Z. 36, 191–212.

Goldberg, E.D., 1975. Synthetic organohalides in the sea. Proc. R. Soc.Lond., Ser. B 189, 277–289.

Haugen, J.E., Wania, F., Ritter, N., Schlabach, M., 1998. Hexachlorocy-clohexanes in air in Southern Norway. Temporal variation, sourceallocation, and temperature dependence. Environ. Sci. Technol. 32,217–224.

Hornbuckle, K.C., Achman, D.R., Eisenreich, S.J., 1993. Over-waterand over-land polychlorinated biphenyls in Green Bay, LakeMichigan. Environ. Sci. Technol. 27, 87–98.

Hühnerfuss, H., Bester, K., Landgraff, O., Pohlmann, T., Selke, K., 1997.Annual balances of hexachlorocyclohexanes, polychlorinated biphe-nyls, and triazines in theGermanBight.Mar. Pollut. Bull. 34, 419–426.

Iwata, H., Tanabe, S., Sakai, N., Tatsukawa, R., 1993. Distribution ofpersistent organochlorines in the oceanic air and surface seawaterand the role of ocean on their global transport and fate. Environ.Sci. Technol. 27, 1080–1098.

Jantunen, L.M., Bidleman, T.F., 1996. Air–water gas exchange ofhexachlorocyclohexanes (HCHs) and the enantiomers of α-HCH inArctic regions. J. Geophys. Res. 101, 28837–28846.

Jaward, F.M., Barbera, J.L., Booijb, K., Jones, K.C., 2004. Spatialdistribution of atmospheric PAHs and PCNs along a north–southAtlantic transect. Environ. Pollut. 132, 173–181.

Junge, C.E., 1977. Basic considerations about trace constituent in theatmosphere is related to the fate of global pollutant. In: Suffet, I.H.(Ed.), Fate of Pollutants in the Air and Water Environment. JohnWiley & Sons, New York, pp. 7–26.

Karickhoff, S.W., 1981. Semi-empirical estimation of sorption ofhydrophobic pollutants on natural sediments and soils. Chemo-sphere 10, 833–846.

Klöpffer, W., Schmidt, E., 2001. A multimedia load model for theBaltic Sea. Environ. Sci. Pollut. Res. Int. 8, 180–188.

Koziol, A., Pudykiewicz, J., 2001. Global-scale environmental transport ofpersistent organic pollutants. Chemosphere 45, 1181–1200.

Kucklick, J.R., Hinckley, D.A., Bidleman, T.F., 1991. Determination ofHenry's law constants for hexachlorocyclohexanes in distilled waterand artificial seawater as a function of temperature. Mar. Chem. 34,197–209.

Lakaschus, S., Weber, K., Wania, F., Bruhn, R., Schrems, O., 2002.The air–sea equilibrium and time trend of hexachlorocyclohexanesin the Atlantic ocean between the Arctic and Antarctica. Environ.Sci. Technol. 36, 138–145.

Lammel, G., Feichter, J., Leip, A., 2001. Long-range transport andmultimedia partitioning of semivolatile organic compounds: a casestudy on two modern agrochemicals. Report No. 324 MPI for Meteo-rology, Hamburg.

Lenhart, H., Pätsch, J., 2004. Daily loads of nutrients, total alkalinity,dissolved inorganic carbon and dissolved organic carbon of theEuropean continental rivers for the years 1977–2002. Berichte ausdem Zentrum fr Meeres-und Klimaforschung, Reihe B: Ozeano-graphie, vol. 48, p. 160.

Lenhart, H.J., Pohlmann, T., 1997. The ICES-boxes approach in relation toresults of a North Sea circulation model. Tellus 49A, 139–160.

Liss, P.S., Slater, P.G., 1974. Flux of gases across the air–sea interface.Nature 247, 181–184.

Luff, R., Pohlmann, T., 1995. Calculation of water exchange times inthe ICES-boxes with a Eulerian dispersion model using a half-lifetime approach. Dtsch. Hydrogr. Z. 47, 287–299.

http://www.argelbe.de/

http://www.bsh.de/

http://www.waterbase.nl/

http://www.waterbase.nl/

http://www.ecmwf.int/

http://www.emep.int/


Mackay, D., 2001. Multimedia Environmental Models: The FugacityApproach, Second ed. Lewis Publishers, Boca Raton.

McMahon, Denison, 1979. Empirical atmospheric deposition para-meters — a survey. Atmos. Environ. 13, 571–585.

Malanichev, A., Mantseva, E., Shatalov, V., Strukov, B., Vulykh, N.,2004. Numerical evaluation of the PCB transport over the Northernhemisphere. Environ. Pollut. 128, 279–289.

NLOE: Niedersächsisches Landesamt fr Ökologie, http://www.nloe.de/, 2005.

OSPAR Commission for the protection of the marine environment ofthe North–East Atlantic, 2000. Quality status report. Region II —Greater North Sea. OSPAR Commission, London.

Paasivirta, J., Sinkkonen, S., Mikkelson, M., Raantio, T., Wania, F.,1999. Estimation of vapor pressures, solubilities and Henry's lawconstants of selected persistent organic pollutants as functions oftemperature. Chemosphere 39, 811–832.

Pacyna, J.M., 1999. Final report for Project POPCYCLING-Baltic. EUDGXII, Environment and Climate Program ENV4-CT96-0214,CD-rom, Norwegian Inst. Air Research., Kjeller, Norway.

Pankow, J.F., 1987. Review and comparative analysis of the theorieson partitioning between the gas and aerosol particulate phases inthe atmosphere. Atmos. Environ. 21, 2275–2283.

Pätsch, J., Kühn, W., Radach, G., Santana Casiano, J.M., GonzalezDavila,M., Neuer, S., Freudenthal, T., Llinas, O., 2002. Interannualvariability of carbon fluxes at the North Atlantic station ESTOC.Deep-Sea Res., Part 2, Top. Stud. Oceanogr. 49, 253–288.

Pekar, M., Gusev, A., Pavlova, N., Strukov, B., Erdman, L., Ilyin, I.,Dutchak, S., 1998. Long-range transport of selected persistentorganic pollutants. Development of transport models for lindane,polychlorinated biphenyls, benzo(a)pyrene. Part I. EMEP/MSC-EReport 2/98. EMEP/MSC-E, Moscow, Russia.

Pohlmann, T., 1987. Validation of a Three Dimensional Dispersion Modelfor 137Cs in the North European Shelf Sea. ICES. C.M. 1987/C:34.

Pohlmann, T., 1996. Predicting the thermocline in a circulationmodel of theNorth Sea. Part I: model description, calibration, and verification. Cont.Shelf Res. 7, 131–146.

Pohlmann, T., Puls, W., 1994. Currents and transport in water. In:Sndermann, J. (Ed.), Circulation and contaminant fluxes in theNorth Sea. Springer, Berlin, pp. 345–402.

Puls, W., Doerffer, R., Sndermann, J., 1994. Numerical simulation andsatellite observations of suspended matter in the North Sea. IEEE J.Oceanic Eng. 19, 3–9.

Ridal, J.J., Kerman, B., Durham, L., Fox, M.E., 1996. Seasonality ofair–water fluxes of hexachlorocyclohexanes in lake Ontario.Envron. Sci. Technol. 30, 852–858.

Sahsuvar, L., Helm, P., Jantunen, L., Bidleman, T.F., 2003. Henry'slaw constants for α-, β-, and γ-hexachlorocyclohexanes (HCHs) asa function of temperature and revised estimates of gas exchange inArctic regions. Atmos. Environ. 37, 983–992.

Scheringer, M., Stroebe, M., Wania, F., Wegmann, F., Hungerbhler, K.,2004. The effect of export to the deep sea on the long-range transportpotential of persistent organic pollutants. Environ. Sci. Pollut. Res. Int.11, 41–48.

Schulz-Bull, D.E., Petrick, G., Bruhn, R., Duinker, J.C., 1998. Chlor-obiphenyls (PCB) and PAHs in water masses of the northern NorthAtlantic. Mar. Chem. 61, 101–114.

Schwarzenbach, R.E., Gschwend, P.M., Imboden, D.M., 1993.Environmental Organic Chemistry. Wiley, New York, p. 681.

Semeena, V.S., Lammel, G., 2005. The significance of the grasshoppereffect on the atmospheric distribution of persistent organic substances.Ge op h ys . Re s. Le tt . 3 2, L0 78 04 , doi:10.1029/2004GL022229.

Skoglund, R.S., Swackhamer, D.L., 1999. Evidence for the use of organiccarbon as the sorbingmatrix in themodelling of PCBaccumulation inphytoplankton. Environ. Sci. Technol. 33, 1516–1519.

Skoglund, R.S., Stange, K., Swackhamer, D.L., 1996. A kinetic modelfor predicting the accumulation of PCBs in phytoplankton.Environ. Sci. Technol. 30, 2113–2120.

Slinn,W.G.N., 1983.Apotpourri of deposition and resuspension questions.In: Pruppacher, H.R., Semonin, R.G., Slinn, W.G.N. (Eds.), Precipi-tation Scavenging, Dry Deposition and Resuspension. Elsevier, NewYork, pp. 1361–1416.

Soulsby, R., 1997. Dynamics of Marine Sands. Tomas Telford, London.SRU, 2004. The German Advisory Council on the Environment: Marine

Environment Protection for the North and Baltic Seas. Special Report.NOMOS, Baden-Baden, Germany.

Strand, A., Hov, O., 1996. A model strategy for the simulation ofchlorinated hydrocarbon distribution in the global environment. WaterAir Soil Pollut. 86, 283–316.

Sündermann, J. (Ed.), 1994. Circulation and Contaminant Fluxes in theNorth Sea. Springer, Berlin.

Sündermann, J., Puls, W., 1990. Modelling of suspended sedimentdispersion and related transport of lead in the North Sea. Mitt.Geol.-Paleontol. Inst. Univ. Hamb. 69, 143–155.

Sündermann, J., Beddig, S., Radach, G., Schlnzen, H., 2002. TheNorth Sea — Problems and Research Needs, Report. Centre forMarine and Climate Research, University of Hamburg.

Theobald, N., Gaul, H., Ziebarth, U., 1996. Verteilung vonorganischen Schadstoffen in der Nordsee und angrenzendenSeegebieten. Dtsch. Hydrogr. Z. 6, 81–93 (Suppl.).

UNEP, 2003. Regionally based assessment of persistent toxicsubstances. Global report. UNEP Chemicals, Geneva, Switzerland.

Van der Zee, C., Chou, L., 2005. Seasonal cycling of phosphorus in thesouthern bight of the North Sea. Biogeosciences 2, 27–42.

Wania, F., Mackay, D., 1999. Global chemical fate of alpha-hexachlor-ocyclohexane. 2.Use of a global distributionmodel formass balancing,source apportionment, and trend predictions. Environ. Toxicol. Chem.18, 1400–1407.

Wania, F., Persson, J., di Guardo, A., McLachlan, M.S., 2000. ThePOPCYCLING-Baltic Model. A non-steady state multicompart-ment mass balance model of the fate of persistent organicpollutants in the Baltic Sea environment. Report Norwegian Inst.Air Research OR 10/2000, U-96069, Kjeller, Norway.

Wanninkhof, R., 1992. Relationship between wind speed and gasexchange over the ocean. J. Geophys. Res. 97, 7373–7382.

Weigel, S., Kuhlmann, J., Hühnerfuss, H., 2002. Drugs and personalcare products as ubiquitous pollutants: occurrence and distributionof chlofibric acid, caffeine and DEET in the North Sea. Sci. TotalEnviron. 295, 131–141.

Whitman, W.G., 1923. The two-film theory of gas absorption. Chem.Metall. Eng. 29, 146–150.

Wiesner, M.G., Haake, B., Wirth, H., 1990. Organic facies of surfacesediments in the North Sea. Org. Geochem. 15, 419–432.

Willet, K.L., Ulrich, E.M., Hites, R.A., 1998. Differential toxicity andenvironmental fates of hexachlorocyclohexane isomers. Environ.Sci. Technol. 32, 2197–2207.

http://www.nloe.de/

http://www.nloe.de/

http://dx.doi.org/10.1029/2004GL022229

A fate and transport ocean model for persistent organic ......A fate and transport ocean model for...

Documents

Transcript of A fate and transport ocean model for persistent organic ......A fate and transport ocean model for...