Cycling of DDT in the global environment 1950–2002:

World ocean returns the pollutant

Irene Stemmler1,2 and Gerhard Lammel3,4

Received 13 October 2009; revised 13 October 2009; accepted 23 November 2009; published 31 December 2009.

[1] The global distribution and fate of the insecticide DDTwas modeled for the first time using a spatially resolvedglobal multicompartment chemistry-transport modelcomprising a 3D coupled atmosphere and ocean GCM,coupled to 2D vegetation surfaces and top soils. DDT entersthe model environment as a pesticide in agriculture only.Final sinks of DDT in the total environment are degradationin air (hydroxyl radical reaction), on vegetation surfaces, inocean sediments and soils. The process resolution of theocean compartment, i.e., either a fixed or variable size andsinking velocity of suspended particles, has almost no effecton the large-scale cycling and fate of DDT. The residencetimes in various ocean basins were declining but variedregionally. The global ocean absorbed until 1977 and sincethen has been losing DDT, while large sea areas are stillaccumulating the pollutant. The main sink is volatilizationto the atmosphere. In 1990, the year when emissions ceased,292 kt of DDT were deposited to the global ocean, 301 ktwere volatilized, and 41 kt were exported from the surfacelayer to the deeper levels. The sea region that has beenrepresenting the most significant (secondary) DDT source isthe western N Atlantic (Gulf stream and N Atlantic Driftregions). It has been a source since approximately 1970.Also large parts of the tropical ocean and the southern mid-latitude ocean have turned net volatilizational (i.e.,volatilization flux > deposition flux) during the 1980s.Despite the emissions migrating southward as a consequenceof substance ban in mid latitudes, the geographic distributionof the contaminant (and, hence, environmental exposure) hasbeen migrating steadily northward since the 1960s.Citation: Stemmler, I., and G. Lammel (2009), Cycling of

DDT in the global environment 1950–2002: World ocean returns

the pollutant, Geophys. Res. Lett., 36, L24602, doi:10.1029/

2009GL041340.

1. Introduction

[2] Apart from environmental conditions (includingocean dynamics and the status of the marine biosphere),the capacity of the ocean surface layer to store organic tracecontaminants is regulated by their water and – because ofsuspended particulate matter - organic solubilities, partialpressure and the level of pre-contamination of the seawater.Organochlorides have been uptaken by the world ocean for

decades. Saturation and subsequently reversal of the air-seagas-exchange has been observed in the open ocean forhexachlorocyclohexane (HCH) in northern mid and highlatitudes [Bidleman et al., 1995; Lakaschus et al., 2002], butso far not for any other organic contaminant with exclu-sively anthropogenic sources.[3] DDT is an insecticide which was first applied in the

late 1940s and until the 1970s in large amounts in agricul-ture worldwide, approximately 1.5 Tg [United NationsEnvironment Programme (UNEP), 2003]. After its persis-tence and toxicity to wildlife (upon bioaccumulation alongfood chains) had been discovered [Carson, 1962], it wassubstituted in mid-latitude and later in low-latitude agricul-tures. Accordingly, the substance has been reported todecline in air and soils, but is thought to still accumulatein the ocean and in sediments [UNEP, 2003]. Due to itspersistence DDT continues to cycle between compartmentsand also to be primary emitted, as an impurity of a substitutepesticide, dicofol [Qiu et al., 2005], and since a number ofAsian and African countries reserve the right of its appli-cation for malaria control.[4] Transport in ocean and atmosphere and fluxes be-

tween these compartments are essential to understand globalcycling of contaminants. It was the aim of this modelingstudy to follow the global fate and both the geographic andcompartmental distributions of a high-volume productionchemical from the early years upon first release until farbeyond the peak application years.

2. Methods

2.1. Model Description

[5] The multicompartment chemistry-transport modelMPI-MCTM is based on the three-dimensional coupledatmosphere-ocean general circulation model ECHAM5-HAM/MPIOM-HAMOCC [Roeckner et al., 2003;Marslandet al., 2003]. It includes two-dimensional top soils andvegetation surfaces. An aerosol module (HAM [Stier et al.,2005]) is embedded in the atmosphere and a biogeochemis-try module (HAMOCC5 [Maier-Reimer et al., 2005]) isembedded in the ocean. HAMOCC uses a nutrient-phyto-plankton-zooplankton-detritus ecosystem model [Six andMaier-Reimer, 1996; Maier-Reimer et al., 2005], and acarbon chemistry formulation following Maier-Reimer[1993]. The model offers two ways of defining gravitationalsinking of detritus, i.e., constant sinking with 5 m d�1, andparticle-size dependent sinking. In the latter case aggregationof marine snow from phytoplankton and detritus produces atemporally and spatially varying size distribution andcorresponding sinking velocities [Kriest, 2002; Maier-Reimer et al., 2005].

GEOPHYSICAL RESEARCH LETTERS, VOL. 36, L24602, doi:10.1029/2009GL041340, 2009ClickHere

for

FullArticle

1Max Planck Institute for Meteorology, Hamburg, Germany.2Now at Max Planck Institute for Chemistry, Mainz, Germany.3Max Planck Institute for Chemistry, Mainz, Germany.4Research Centre for Environmental Chemistry and Ecotoxicology,

Masaryk University, Brno, Czech Republic.

Copyright 2009 by the American Geophysical Union.0094-8276/09/2009GL041340$05.00

L24602 1 of 5

[6] Chemicals are cycling in atmosphere (gas, water andparticle phase), ocean (dissolved, colloidal, particulatephase), and soil and vegetation surfaces. Cycling includeschemical degradation and mass exchange processes, whichhave been described previously [Semeena et al., 2006;Guglielmo et al., 2009]. In the ocean, in addition toadvection and diffusion, chemicals in the particulate phaseare subject to gravitational settling at the same velocity asdetritus.

2.2. Emissions

[7] DDT application estimates for this study were adap-ted from a compilation of global agricultural usage 1950–1990 [Semeena and Lammel, 2003]. The emissions werecompiled from data reported by member states to the UNFood and Agricultural Organisation (FAO), scaled with a1� � 1� cropland distribution. The total inventory of p,p0-DDT is approximately 1.2 � 106 t (see Figures 1 and S11

for annual application rates). Other DDT usage, such asmalaria control or illegal use are ignored (due to dataavailability and/or quality). The emissions do not resolveany diel or seasonal cycles.

2.3. Model Experiments

[8] A resolution of about 5� (T21) and 19 vertical levelsfrom 1000–10 hPa in the atmosphere with a time step of 40min, and a nominal resolution of 3� in the ocean with a timestep of 144 min were used. The ocean model uses anorthogonal curvilinear C-grid with one pole located overGreenland and the other over Antarctica. The model oceanhas 40 vertical levels with level thickness increasing withdepth. Eight levels are within the upper 90 m and 20 levelsare within the upper 600 m. The coupling time step betweenatmosphere and ocean was 1 day. In the experiment, DDTwas introduced following the agricultural applications dur-ing 1950–1990 using a split upon entry of 20% to soil and80% to vegetation. No primary emissions in the period1991–2002 were considered.

[9] The chemical DDT is characterized by a low vaporpressure and water solubility and a medium lipophilicity(Table S1). Degradation of the DDT in the environmentalcompartments is notoriously slow and is set to zero inseawater. DDT removal from the model environment is bydegradation in soil only, represented as a first-order process(4.05 � 10�9 s�1 at 298 K [Hornsby et al., 1996]) andassumed to double per 10 K temperature increase.

3. Results and Discussion

[10] The temporal evolution of the DDT burden in allenvironmental compartments follows the pattern of theapplications (Figure S1), partly with significant delay(Table 1). Offsets from the application peak reflect theresidence time of DDT in the compartments, which isshortest in the atmosphere (7.3 days), and longest in theocean (23.7 years). The global ocean loses mass since 1977.Sinks for DDT in the model ocean are volatilization anddegradation in sediment. Volatilization was found to be thedominant sink process: after 30 and 40 years, respectively,loss in the sediment still accounts for only 3.8 and 5.5% oftotal loss, respectively.[11] The distribution of DDT in seawater shows a spatio-

temporal pattern, which is far from reflecting the applicationpattern in the neighbouring continental regions. Obviously,sea currents and other features influencing residence time inseawater are involved. Thereby, neighboring regions, e.g., inthe Atlantic Ocean, behave differently (Figure 2a), obviouslyin response to the environmental conditions, such as surfacetemperature, currents and deep water formation, and atmo-spheric deposition patterns. The Equatorial Ocean, as well asparts of the Arctic Ocean, and the Gulf Stream (including theNorth Atlantic Drift) region do not reach a maximum inburden until 1990, but still continue to accumulate DDT.Due to low mean sea surface temperatures volatilization isreduced in the Arctic Ocean, and DDT sources (depositionand inflow) dominate. The vertical stratification of DDT isdefined by export with sinking particulate matter, and watermovements (including deep water formation and upwelling).Both are spatially and temporally highly variable, but ingeneral downward water movements dominate the removalof DDT from surface waters [Guglielmo et al., 2009].Globally, in the ocean below 100 m DDT mass is increasingcontinuously. This process significantly slowed down sincethe 1970s (Figure 2b). 76.0, 101.1, and 101.2 kt of DDT hadbeen stored below 100 m of depth since the years 1970, 1980and 1990, respectively. The surface ocean reaches a maxi-mum DDT load in the late 1970s and early 1980s. Asignificant amount of DDT is not exported into the deepocean, but returned to the atmosphere.[12] The DDT volatilization flux from the global ocean

was 2472 kt/year in 1970 but 301 kt/year in 1990. The totalamount returned from the global ocean since the peak year

Figure 1. DDT emissions: accumulated application inagriculture (a) 1950–1990 and (b) year of peak application.

Table 1. DDT Dynamics in Environmental Compartmentsa

Emissions Atmosphere Ocean Soil Vegetation

Year of peak burden 1960 1961 1977 1973 1966Residence time [a] - 0.02 23.70 14.93 1.20

aGlobal means were used.1Auxiliary materials are available in the HTML. doi:10.1029/

2009GL041340.

L24602 STEMMLER AND LAMMEL: WORLD OCEAN RETURNS DDT L24602

2 of 5

1977 was 13.4 kt until 1990. In 1990 292 kt were depositedto the ocean, 301 kt were volatilized, and 41 kt wereexported from the ocean surface layer (i.e., �90 m depth)to the deeper levels. Outgassing of DDT started in theAtlantic ocean in the 1970s (Figure 3b), and was strongestin the extension of the Gulf Stream close to the mouth of theSt Lawrence river. The Pacific ocean starts to return DDT tothe atmosphere only 20 years after the Atlantic ocean, in the1990s. The reason for the delay between the oceans is themuch earlier application peak on American soils upwindwith regard to the oceanic regions (Figure 1) in the 1950s, incontrast to the application peak in Asia in the 1980s (notshown). Large parts of the tropical ocean and the southernmid-latitude (so-called ‘roaring fourties’) turned net-vola-tilizational, i.e., volatilization flux exceeded deposition flux,in the 1980s. In general, in remote areas, regions with strongmean wind speed turn net-volatilizational earlier thanregions with lower mean wind speeds (Figure 3b).[13] The spatial pattern of net deposition in 2002 (Figure 3a)

is obviously determined by prevailing environmental con-ditions, whereas the application pattern (Figure 1a) is notdominant anymore. This was different in 1990, when the net

deposition pattern over land was still dominated by thespatial distribution of the applications in the 1980s. Inregions of high DDT usage, such as India, large parts ofEurope, and western Africa, the surface fluxes were netvolatilizational. The model identifies the highest net vola-tilization fluxes in the year 2002 in the Northern Atlanticocean, whereas over India, Europe and West Africa, depo-sition and volatilization are near equilibrium. Expectedly,ignored primary sources (health programs/malaria control,leakage from improper storage, illegal use) cause a modifiedspatial pattern in recent years.[14] The residence time of DDT was derived for several

oceanic regions from storage, volatilization, degradation insediment, and advection out of the region. It was found tobe strongly spatially and temporally variable (Table 2).Longest residence times are found in the Arctic, (1–7 a)and Southern Ocean (0.13–2.4 a), where deposition isfavoured over volatilization due to cold mean surfacetemperatures. Short residence times are found in the IndianOcean (0.04–1.0 a). These values are much lower than theresidence time of DDT in the total compartment (23.7 a,Table 1), because the individual sea regions are smallcompared to the global ocean (Figure S2), and advectioncontributes to DDT turnover. Driven mainly by the decreas-ing deposition flux, the long-term trend is that the residencetime of DDT is decreasing over time.[15] DDT distribution in the total environment has been

migrating slowly northward. The centre of gravity (COG,corresponding to the 50th percentile of the cumulativedistribution function in zonal direction [Leip and Lammel,2004]) of the total environmental burden of DDT waslocated at 47.4, 48.5, 50.0 and 53.0�N in 1960, 1970,1980 and 1990, respectively. This is not explained by the

Figure 2. DDT in the global environment: (a) year of peak burden and (b) vertical distribution in the global oceanintegrated from the sea bottom upward.

Figure 3. Net deposition: (a) annual mean in 2002 and(b) year in which the ocean turns net-volatilizational. Greyindicates no net-volatilization until 2002.

Table 2. Residence Times of DDT in Various Ocean Regions

1950–1990a

Region 1950 1960 1970 1980 1990

Labrador Sea, Baffin Bay, Hudson Bay 477 269 259 194 129Norwegian Sea, Barents Sea 1090 525 491 251 156Arctic Ocean 2592 1241 579 504 386North Atlantic 172 103 86 63 50Central Atlantic 84 37 39 47 62South Atlantic 317 113 68 51 40Indian Ocean 151 54 39 29 21Pacific Ocean 229 73 57 46 37Southern Ocean 893 165 98 70 50

aResidence times are in [d].

L24602 STEMMLER AND LAMMEL: WORLD OCEAN RETURNS DDT L24602

3 of 5

emissions which were located more southerly and, incontrary, moved further southward (COG at 43.1 and42.7, 29.8 and 28.8�N in 1960, 1970, 1980 and 1990,respectively). Instead, it is obviously a consequence of thesubstance mobility, which is driven by atmospheric andoceanic transports and by repeated cycles through theatmosphere before final removal by degradation or transferto the deep sea, the so-called grasshopper effect [Wania andMackay, 1993; Semeena and Lammel, 2005; Jurado andDachs, 2008; Guglielmo et al., 2009].[16] Due to lack of monitoring data before the early

1990s, validation of these model results is hardly possible.No observational data have been collected in most openocean regions since decades. Oversaturation and reversal ofthe air-sea exchange had been observed previously for another organochlorine pesticide, HCH, with respect to theisomer a-HCH in the Bering and Chukchi Sea between1988 and 1993 [Bidleman et al., 1995], in the NorthAtlantic off the Norwegian coast and in the Southern Oceanby 1999/2000 [Lakaschus et al., 2002] and with respect tothe isomer g-HCH in the North Atlantic off the Norwegiancoast [Lakaschus et al., 2002], but not for polychlorinatedbiphenyls in the North Atlantic and Arctic Oceans [Gioia etal., 2008]. The historic usage of technical HCH, PCB andDDT was by and large in phase [Lohmann et al., 2007].Both DDT and PCB are less water-soluble, more lipophilic(and, therefore, do partition more efficient to organic phasessuspended in seawater) and less volatile than HCH, suchthat substantial differences of the cycling in the marineenvironment and the time reversal of air-sea exchange isexpected.

4. Conclusions

[17] The results of a global long-term simulation of DDTcycling using historical application data suggest that untilthe 1970s the ocean acted as a global sink of DDT. Butvertical export of DDT into the deep ocean was notsufficiently effective to compensate for high atmosphericdeposition as a consequence of intense usage of the pesti-cide. From the end of the 1970s onwards, as a consequenceof declining emissions following the ban of DDT in thedifferent countries and, hence, depositions, parts of theocean surface layer became oversaturated, and the net air-sea flux was reversed. This was not indicated by any studybefore. The decline of atmospheric levels over the oceanfollowing the ban of DDT across countries and continentswas hence superimposed by re-emissions from the ocean.To some extent, even atmospheric concentrations over theEuropean continent may have been affected by DDTreturned from the ocean via long-range transport in theatmosphere.[18] A strong poleward migration of the contaminant’s

geographic distribution (and, hence, environmental expo-sure) is found in the northern hemisphere in agreement withprevious findings, which, however, were based on simpli-fied (non-transient) emission scenarios [Guglielmo et al.,2009] or neglected oceanic transports [Semeena et al.,2006].[19] Direct evidence of DDT distribution and model

validation is hampered by lack of monitoring data in themarine environment. The global cycling of DDT, due to

re-entry in malaria and other vector-bourne diseases con-trol and as impurity in a registered pesticide (dicofol), butalso due to the chemodynamics in ocean and terrestrialcompartments, deserves more scientific attention. Measure-ment campaigns should address potential source areas inthe open ocean, such as the North Atlantic, and possiblepast influence on continental monitoring stations in Europeand the Arctic by advection from these sea regions shouldbe studied.

[20] Acknowledgments. IS was supported by the International MaxPlanck Research School for Maritime Affairs and the Max BuchnerResearch Foundation.

ReferencesBidleman, T. F., L. M. Jantunen, R. L. Falconer, L. A. Barrie, and P. Fellin(1995), Decline of hexachlorocyclohexane in the Arctic atmosphere andreversal of the air-sea gas-exchange, Geophys. Res. Lett., 22, 219–222,doi:10.1029/94GL02990.

Carson, R. (1962), Silent Spring, 363 pp., Houghton Mifflin, New York.Gioia, R., R. Lohmann, J. Dachs, C. Temme, S. Lakaschus, D. Schulz-Bull,I. Hand, and K. C. Jones (2008), Polychlorinated biphenyls in air andwater of the North Atlantic and Arctic Ocean, J. Geophys. Res., 113,D19302, doi:10.1029/2007JD009750.

Guglielmo, F., G. Lammel, and E. Maier-Reimer (2009), Global environ-mental cycling of DDT and g-HCH in the 1980s: A study using a coupledatmosphere and ocean general circulation model, Chemosphere, 76,1509–1517, doi:10.1016/j.chemosphere.2009.06.024.

Hornsby, A. G., D. R. Wauchope, and A. E. Herner (1996), PesticideProperties in the Environment, Springer, New York.

Jurado, E., and J. Dachs (2008), Seasonality in the ‘‘grasshopping’’ andatmospheric residence times of persistent organic pollutants over theoceans, Geophys. Res. Lett., 35, L17805, doi:10.1029/2008GL034698.

Kriest, I. (2002), Different parameterizations of marine snow in a 1D-modeland their influence on representation of marine snow, nitrogen budget andsedimentation, Deep Sea Res., Part I, 49, 2133–2162, doi:10.1016/S0967-0637(02)00127-9.

Lakaschus, S., K. Weber, F. Wania, and O. Schrems (2002), The air-seaequilibrium and time trend of hexachlorocyclohexanes in the AtlanticOcean between the Arctic and Antarctica, Environ. Sci. Technol., 36,138–145, doi:10.1021/es010211j.

Leip, A., and G. Lammel (2004), Indicators for persistence and long-rangetransport potential as derived from multicompartment chemistry-transportmodelling, Environ. Pollut., 128, 205 – 221, doi:10.1016/j.envpol.2003.08.035.

Lohmann, R., K. Breivik, J. Dachs, and D. Muir (2007), Global fate ofPOPs: Current and future research directions, Environ. Pollut., 150,150–165, doi:10.1016/j.envpol.2007.06.051.

Maier-Reimer, E. (1993), Geochemical cycles in an ocean circulationmodel: Preindustrial tracer distributions, Global Biogeochem. Cycles,7, 645–677, doi:10.1029/93GB01355.

Maier-Reimer, E., I. Kriest, J. Segschneider, and P. Wetzel (2005), TheHAMburg Ocean Carbon Cycle Model HAMOCC5.1—Technical De-scription Release 1.1, Rep. Earth Syst. Sci., vol. 14, 57 pp., Max PlanckInst. for Meteorol., Hamburg, Germany.

Marsland, S. J., H. Haak, J. H. Jungclaus, M. Latif, and F. Roske (2003),The Max Planck Institute global ocean-sea ice model with orthogonalcurvelinear coordinates, Ocean Modell., 5, 91–127, doi:10.1016/S1463-5003(02)00015-X.

Qiu, X., T. Zhu, B. Yao, J. Hu, and S. Hu (2005), Contribution of dicofolto the current DDT pollution in China, Environ. Sci. Technol., 39,4385–4390, doi:10.1021/es050342a.

Roeckner, E., et al. (2003), The atmospheric general circulation modelECHAM5 part 1: Model description, Rep. 349, 127 pp., Max PlanckInst. for Meteorol., Hamburg, Germany.

Semeena, V. S., and G. Lammel (2003), Effects of various scenarios ofentry of DDT and g-HCH on the global environmental fate as predictedby a multicompartment chemistry-transport model, Fresenius Environ.Bull., 12, 925–939.

Semeena, V. S., and G. Lammel (2005), Significance of the grasshoppereffect on the atmospheric distribution of persistent organic substances,Geophys. Res. Lett., 32, L07804, doi:10.1029/2004GL022229.

Semeena, V. S., J. Feichter, and G. Lammel (2006), Significance of regionalclimate and substance properties on the fate and atmospheric long-rangetransport of persistent organic pollutants: Examples of DDT and g-HCH,Atmos. Chem. Phys., 6, 1231–1248.

L24602 STEMMLER AND LAMMEL: WORLD OCEAN RETURNS DDT L24602

4 of 5

Six, K., and E. Maier-Reimer (1996), Effects of phytoplankton dynamics onseasonal carbon fluxes in an ocean general circulation model, GlobalBiogeochem. Cycles, 10(4), 559–583, doi:10.1029/96GB02561.

Stier, P., et al. (2005), The aerosol-climate model ECHAM5-HAM, Atmos.Chem. Phys., 5, 1125–1156.

United Nations Environment Programme (UNEP) (2003), Regionally BasedAssessment of Persistent Toxic Substances, Global Rep. 2003, 211 pp.,Geneva, Switzerland.

Wania, F., and D. Mackay (1993), Global fractionation and cold condensa-tion of low volatility organochlorine compounds in polar regions, Ambio,22, 10–18.

�����������������������G. Lammel and I. Stemmler, Max Planck Institute for Chemistry, J. J.

Becher Weg 27, D-55128 Mainz, Germany. (g.lammel@mpic.de)

L24602 STEMMLER AND LAMMEL: WORLD OCEAN RETURNS DDT L24602

5 of 5

Auxiliary Material for Paper 2009XXXXXX Cycling of DDT in the global oceans 1950-90: World ocean returns the pollutant Irene Stemmler 1, Gerhard Lammel 2,3 (1 = Max Planck Institute for Meteorology, Hamburg, Germany 2 = Max Planck Institute for Chemistry, Mainz, Germany 3 = Masaryk University, Research Centre for Environmental Chemistry and Ecotoxicology, Brno, Czech Republic) Geophys. Res. Lett., xx, doi:10.1029/2009GLXXXXXX, 2009 Introduction This material contains two tables, two text files, and seven figures. 1. 2009isXXXXXX-ts01.txt Physico-chemical properties and degradation rates of DDT 1.1 Column "Physico-chemical property" of DDT 1.2 Column "Unit" SI-unit of the property 1.3 Column "Value (at temperature)" value for p,p'-DDT 1.4 Column "Reference" Literature reference or estimate 2. 2009isXXXXXX-ts02.txt Sensitivity of fate and distribution towards vertical particle flux parameterizations 3.2009isXXXXXX-ts03.txt Validation of DDT model predictions 4. 2009isXXXXXX-ts04.txt Validation of atmospheric concentrations using data from Iwata 1993 4.1 Column "Location" of the observations 4.2 Column "Observation [pg m-3]" DDT in air from Iwata 1993 4.3 Column "Model [pg m-3]" DDT in air model predictions 4. 2009isXXXXXX-fs01.tif Annual global DDT applications 1950-2002 5. 2009isXXXXXX-fs02.tif Map of ocean regions for which DDT residence times were derived 6. 2009isXXXXXX-fs03.tif Global DDT compartmental burden time series and DDT applications 7. 2009isXXXXXX-fs04.tif Validation of modeled soil burden 8. 2009isXXXXXX-fs05.tif Modeled and observed vertical profiles of DDT in the ocean 9. 2009isXXXXXX-fs06.tif Modeled and observed DDT concentrations in air Kosetice 10. 2009isXXXXXX-fs07.tif Modeled and observed DDT concentrations in air Zeppelinfjell

The chemical DDT is characterized by a very low vapor pressure (2.5e-5 Pa at 293 K) and water solubility (3.4e-3 mg/L at 298 K) and a medium lipophilicity (Kow = 6.19). Substance degradation in soil is assumed to be represented by a first-order process (4.05e-9 s-1 at 298 K [Hornsby et al., 1996]). Degradation on vegetation surfaces, In lack of data no discrimination between isomers present in technical DDT is possible. Temperature dependencies of these properties are accounted for by usage of published data or estimated. Thus, a doubling of degradation rate in soil is assumed per 10 K temperature increase. Physico-chemical property Unit Value (at temperature) Reference Water solubility mg/L 3.4e-3 (298 K) 1 Saturation vapor pressure Pa 2.5e-5 (293 K) 2 Octanol-air partitioning log 10.09 coefficient Octanol-water partitioning log 6.19 3 coefficient Enthalpy of vaporisation kJ/mol 118.00 estimated Enthalpy of solution kJ/mol 27.00 estimated Degradation rate in soil 1/s 4.05e-9 4 Degradation rate in seawater 1/s 0.00 Degradation rate in 1/s 4.05e-9 4 ocean sediment OH radical rate coefficient cm^3/(molec s) 1.3e-13 estimated References: [1] Biggar J.W., G.R. Dutt and R.L. Riggs (1967), Predicting and measuring the solubility of p,p'-DDT in water, Bulletin Environ. Contam. Toxicol., 2, 90-100. [2] Rippen, G. (2008), Umweltchemikalien, CD-ROM edition, Ecomed, Landsberg, Germany. [3] O'Brian R.D. (1975), Nonenzymic effects of pesticides on membranes, in: Environmental dynamics of pesticides, Plenum Press, New York, USA, 331-342. [4] Hornsby A.G., D.R. Wauchope, A.E. Herner (1996), Pesticide properties in the environment, Springer, New York, USA.

The sensitivity of the fate and distribution towards representations of the vertical flux in the ocean was determined: One experiment used 5 m/d sinking velocity, and the other aggregation of marine snow. In these experiments the resolution of the atmosphere model was T63L19 with a time step of 20 min, and the ocean used a nominal resolution of 1.5┬░. The ocean grid cell size varies gradually between 15 km in the Arctic and approximately 184 km in the Tropics. As in the 3┬░ resolution it has 40 vertical levels with level thickness increasing with depth. The time step of the ocean model was 72 min and the coupling time step was 1 day. In the experiments the distribution of DDT was simulated based on 1980 applications. In total 7665 t were applied annually to soil (20 %) and vegetation (80 %). It was found that overall persistence deviate <4% and long-range transport indicators (e.g. plume displacement; [Leip and Lammel, 2004]) deviates <1% as a consequence of different representations of the vertical flux in the ocean. Reference: Leip, A., and G. Lammel (2004), Indicators for persistence and long-range transport potential as derived from multicompartment chemistry-transport modelling, Environ. Poll. 128, 205-221.

Validation of DDT model predictions Soil Soil and vegetation are only represented as single layer (topsoil) surfaces in the MPI-MCTM, hence their contamination is expressed as a mass per surface area. Soil burdens were converted into concentrations by dividing them by soil dry bulk density and a fixed soil depth of 10 cm. The average DDT concentration in soil between 40 N and 60 N was compared to measured soil and sediment concentrations from Northern North America and Great Britain [Dimond (1996) [1], Meijer (2001) [2], and others compiled by Schenker 2008 [3] ]. For intercomparison reasons only relative soil concentrations are compared to observational data. Each set of observations was normalized to its 1990 value. Observations for DDT levels in US and European soils show that 1965 values of concentrations are about 2 to 6 greater than those in 1990( Fig. 2009isXXXXXX-fs04.tif). Model data from MPI-MCTM show a peak in the concentrations around 1972 which are 2 times higher than 1990 values. Modeled and observed concentrations are decreasing from then on. For the second half of the simulation the model results are well within the observed range of relative concentrations, whereas for the first half model results are at the lower boundary of the range spanned by the observations. Ocean Vertical profiles from individual locations, measured in 1976 [ Tanabe and Tatsukawa (1983) [4]], were used to validate the modeled distribution of DDT in the water column. Vertical DDT profiles were compared with observations (Fig.2009isXXXXXX-fs05.tif) at four locations in the Pacific, Indian and Antarctic oceans. Measurements in the Pacific were conducted at two locations west of Japan in July and August 1976. At location A probes were taken at 0 m, 50 m, 200 m, 500 m, 1000 m, and 1500 m. Modeled and observed concentrations show a similar profile decreasing from the surface down to 200 m. Model results decrease from 1.5 ng/L to approximately 0.5 ng/L, whereas observations range from 0.75 ng/L to 0.25 ng/L. From 500 m to 700 m model results first increase and than decrease moderately down to 1000 m after reaching a maximum. The observations fail to capture this profile, because of resolution, i.e. no samples were taken between 500 m and 1000 m. Below 1000 m model results show a rapid decrease to 0 ng/L. Observations are increasing from 1000 m (0.58 ng/L) to 1500 m (0.67 ng/L). At location B the modeled profile shows a similar pattern like at location A. Both, model results and observations decrease from 1 ng/L (observations), 1.5 ng/L (model data) to approximately 0.4 ng/L at 100 m. As for location A the profiles cannot be compared directly because observations capture 7 depths between 0 and 2500 m. Due to the limited model resolution and consequential deficiencies in the model bathymetry the model ocean is shallower than the deepest depth of 2500 m in the observations at location B.

In the Indian ocean southwest of Indonesia, model results and observations show a different profile. At the surface observational data are higher than model results, decreasing down to 200 m, and showing a slight increase at 600 m. The model results increase down to about 200 m and decrease afterwards. In the Antarctic ocean the model results exceed the observations for all depths. Both profiles show a subsurface increase down to approximately 100 m followed by a constant decrease. Maximum model results are 0.5 ng/L, whereas maximum observed concentrations are 0.027 ng/L. Atmosphere Atmospheric concentrations predicted by the model were validated against observations from two stations of the EMEP monitoring network (at Zeppelinfjell (Fig. 2009isXXXXXX-fs06.tif) and Kosetice (Fig. 2009isXXXXXX-fs07.tif M) [7]), and against observation over seawater published in Iwata 1993 [5] (Table 2009isXXXXXX-ts04.txt). Both observations and model results show a decrease in concentration from lower to higher latitude. For DDT, modeled concentrations, ranging between 6 and 700 pg m-3, are higher than observed ones (Table Table 2009isXXXXXX-ts04.txt)), except in the Caribbean Sea and the South China Sea. Observed concentrations range between 0.9 and 250 pg m-3. In the Bay of Bengal model results show a mean concentration of 712.5 pg m-3, whereas the observed mean value is 250 pg m-3. The model prediction lies within the range of the observations ranging from 42-1000 pg m-3. A better agreement is found in lower latitudes (e.g. the Gulf of Mexico) where model exceeds observations less than two times. Best agreements between model and observations are found in the Caribbean sea, the Gulf of Mexico, the Southern ocean and the South China Sea. In high latitudes of the northern hemisphere modeled concentrations exceed observed (weekly) values by 1, maximum 2 orders of magnitude. The discrepancies of a factor of 10 and larger between modeled and observed concntrations is also evident from the comparison with data at Zeppelinfjell and Kosetice. Location Observation [pg m-3] Model [pg m-3] Gulf of Alaska 3.9 30.0 Bering Sea 3.6 37.6 Chukchi Sea 5.8 37.5 Gulf of Mexico 48.0 58.8 East China Sea 19.0 187.3 Caribbean Sea 13.0 9.1 North Atlantic 8.7 34.5 North Pacific 12.0 34.2 South China Sea 54.0 41.3 Bay of Bengal 250.0 712.5 Souther:n Ocean 2.4 6.3 Among the areas sampled 1989-90 (Iwata et al., 1993) reversal of air-sea exchange is predicted by the model for the Eastern Indian and Southern Ocean. The concentration levels reported (Iwata et al., 1993) suggest that DDT was close to saturation in the Eastern Indian Ocean (i.e. about equal fugacities from seawater and air), but not in the Southern Ocean (fugacity from seawater << fugacity from air). In general, discrepancies between model results and observed data could be

due to incorrect model results and uncertainties of the observations. Incorrect model results can be caused by uncertainties in the emission data and chemical properties of the compounds, and by incompletely reproduced or lacking relevant environmental processes. The emission inventory for DDT was built on agricultural usage data reported to the Food and Agriculture Organisation (FAO), and does not include DDT usage for forestry, usage in health programs (e.g. for indoor residual spraying), or storage and disposal as waste. Only a limited number of countries report to FAO, as reporting to FAO is voluntary [Voldner(1995) [6]]. For these reasons the emission inventory is expected to underestimate the actual sources of DDT. However, no general trend of a underestimation of environmental concentrations in the model is found based on the comparison with observational data. Furthermore, the latitudinal distribution of the emissions is probably less subject to errors than the absolute value. Hence, deficiencies of the emission inventory are not the major cause of deviations between model data and observations. Due to the complexity of the model the prospect to identify a single process responsible for discrepancies between model results and observations is low. And a comprehensive sensitivity analysis is not feasible due to computing costs. The possible sensitivity to selected input parameters was addressed by MBM simulations (not shown here). They show that some substance properties, i.e. the photochemical degradation rate and solubility of DDT, have the potential of reducing the erroneous latitudinal gradients in ocean and atmosphere. Also the implementation of processes not yet captured, like degradation of DDT in the particle phase in air or degradation of DDT in the ocean could contribute to better model results. Metabolites of DDT, DDE and DDD, have not been included in this study. In seawater and soil DDE levels are often higher than DDT levels. Validation of model results could be improved with DDE included in model simulations, as it would imply a larger observational data base, besides other. Metabolites of DDT, DDE and DDD, have not been included in this study. In seawater and soil DDE levels are often higher than DDT levels. Validation of model results could be improved with DDE included in model simulations, as it would imply a larger observational data base, besides other. References: [1] Dimond JB and Owen RB (1996), Long-term residue of DDT compounds in forest soils in Maine, Environ. Poll., 92, 227-230. [2] Meijer SN, CJ Halsall, T Harner, A Peters, WA Ockenden, AE Johnston, KC Jones (2001), Organochlorine pesticide residues in archived UK soil, Environ. Sci Technol., 35, 1989-1995 [3] Schenker U, M Scheringer, K Hungerbohler (2008), Investigating the global fate of DDT: Model evaluation and estimation of future trends, Environ. Sci. Technol., 42, 1178-1184 [4] Tanabe S and R Tatsukawa (1983), Vertical transport and residence time of chlorinated hydrocarbons in the open ocean water column, J. Ocean. Soc. Japan, 39, 53-62 [5] Iwata H, S Tanabe, R Tatsukawa (1993), Distribution of perstitent organochlorines in the oceanic air and surface seawater and the role of ocean on their global transport and fate, Environ. Sci. Technol., 27, 1080-1098

[6] Voldner EC, YF Li (1995), Global usage of selected perstitent organochlorines, Sci. Tot. Environ. 160/161, 201-210 [7] EMEP, http://tarantula.nilu.no/projects/ccc/