Atmospheric Input of Lead Into the German Bight- a High Resolution Measurement and Model Case Study

MARINE ECOLOGY PROGRESS SERIES Vol. 156: 299-309,1997 Published September 25 Mar Ecol Prog Ser

Atmospheric input of lead into the German Bight- a high resolution measurement and model case study

K. Heinke S~hliinzen'~',Thomas Stahlschmidt2, Andreas ~ebers~,Ulrike Niemeierl, Michael Kriews3, Walter an neck er^

'Zentrum fiir Meeres- und Klimaforschung der Universitat Hamburg, Meteorologisches Institut. Bundesstr. 55, D-20146 Hamburg, Germany 21nstitut fiir Anorganische und Angewandte Chemie, Universitat Hamburg, Martin-Luther-King-Platz 6, D-20146 Hamburg, Germany 3~lfred-~egener-~nstitutfiir Polar- und Meeresforschung, Postfach 120161, D-27515 Bremerhaven, Germany

ABSTRACT: The atmospheric input of lead into the German Bight is calculated from measured atmospheric concentrations, from modelled as well as measurement-derived deposition velocities, and from measured wet deposition values, with a temporal resolution between 6 and 24 h. The measurements were taken at several coastal sites and on a ship in the German Bight during a 1 wk drift experiment in April 1991. The applied model is a 3-dimensional nonhydrostatic atmospheric mesoscale model. The calculated input data depend on deposition velocities and integration time. Due to variation in dry deposition the atmospheric input values can differ by a factor of 3. Total deposition values and thus atmospheric input data can differ by a factor of 10. Calculated backward trajectories show that the observed time-lag in the occurrence of concentration maxima can be explained by the travel time of the advected air masses between the measurement sites. By use of backward trajectories and results of the mesoscale model the measured values are associated with source areas for a day with very high atmospheric lead concentrations. It is shown that up to 50% of the contaminants can reach the German Bight via long-range transport. Mesoscale atmospheric phenomena also influence the wind field and the concentrations.

KEYWORDS: North Sea. German Bight. Atmospheric input. Dry deposition. Wet deposition . Atmos- pheric lead concentration . Mesoscale model

INTRODUCTION for the Second International North Sea Conference (INC 1987). The input via the atmosphere amounts to The pollution of marginal seas is caused by inputs more than two thirds of the total for cadmium and lead not only from rivers, coastal industries, and dumping and to more than one third for zinc and copper. The but also from the atmosphere in considerable amounts. data for the German Bight compiled in Fig. lb are This is especially true for the North Sea, which is situ- based on the input values for the Wadden Sea area ated close to major industrial centres and highly popu- (Schliinzen 1994a) recalculated for the German Bight lated areas. The input quantities have been widely dis- (area: 24000 km2). Due to the proximity of the river cussed in earlier papers (e.g. INC 1987, Warmenhoven estuaries, the proportion of the atmospheric input is et al. 1989, QSR 1993) and thus are only briefly sum- lower for the more coastal area compared to the whole marized here. Fig. la shows the relative contributions North Sea. However, it still accounts for a considerable of the atmosphere and other paths for the whole North amount of lead and cadmium as well as of some Sea area. The values are based on the data compiled organic contaminants which are not regarded here. The relative contributions given in Fig. 1 correspond to annual means. These are strongly influenced by

0 Inter-Research 1997 Resale of fullarticle not permitted 300 Mar Ecol Prog Ser 156: 299-309, 1997

1m North Sea ]b German B~ght l

Fig. 1. Fractional inputs (Oh) into (a)the North Sea (data from INC 1987) and (b) the German Bight (data from Schliinzen 1994a)

short-term deposition events, as has been derived from Kersten et al. 1992, Kriews 1992, Krause et al. 1994). In long-term investigations (Dannecker et al. 1994, Ker- addition, the atmospheric transport and deposition of sten et al. 1994). Measurements and model results other heavy metals like cadmium is similar to that for show that the atmospheric input can change consider- lead, which makes lead a very good example for study- ably within 1 d. This can be caused by changes in con- ing the atmospheric input. taminant concentration in air due to a wind shift, as Since the atmospheric contribution to the mean input shown in this paper (factor of 50: section 'Measured is high for lead (see Fig. 1) some effort has to be made concentrations' below), and by changes in deposition to estimate the atmospheric input of lead into the sea at velocity due to different atmospheric stratification and a resolution matching the measurements in the water. wind velocity (factor of 2: Schliinzen & Krell 1994). Since it is not possible to measure the dry deposition Additionally, wet deposition can enhance dry deposi- directly with a high temporal resolution, the values tion by up to a factor of 10 within a few hours (section presented in this paper are calculated from high reso- 'Calculated atmospheric input ...' below; Spokes et al. lution concentration measurements (section 'Measur- 1993). For this reason, studies of transfer processes in ing atmospheric concentrations') and model results the ecosystem as described by Sundermann (1997 in (section 'Modelling atmospheric concentrations and this volume) need input data with a temporal resolu- deposition'). The resulting deposition values and tion of a few hours. derived input data are given in the section 'Calculated To investigate the transfer processes in the German atmospheric input.. .'. Bight ecosystem, lead was used as a key contaminant in the German research project PRISMA. Lead is released into the marine environment by river dis- MEASURING ATMOSPHERIC CONCENTRATIONS charges, atmospheric deposition, relllobilization iron1 sediment and removal from bacterio-, phyto-, zoo- The atmospheric concentrations of lead and other plankton and other biota it might be incorporated heavy metals were measured at 5 sites in the area of within or attached to. The accumulation factors for the German Bight during the first PRISMA experiment. phytoplankton can be much higher for lead than they It was performed in April 1991 and included 3 drift- are for other heavy metals (Karbe et al. 1994). Most experiment periods. Details on the experiment are lead, however, is bound to the body surface and only a given by Brockmann et al. (1997 in this volume). little is incorporated. Therefore, lead is a less crucial Below, atmospheric concentrations and the atmos- heavy metal for the plankton than e.g. mercury is. pheric input are presented for the third drift period Nevertheless, lead is suitable for studying temporal (April 23 to 30, 1991). Measurements and model results concentration changes in the ecosystem, because it is for other compartments of the ecosystem are given by found and can reliably be measured in all compart- Konig et al. (1997), Moll (1997) and Raabe et al. (1997) ments of the ecosystem (including the atmosphere; (all in this volume). Schliinzen et al.: PRISMA-Atmospheric input of lead 301

Measurement strategy were carried out under clean room conditions. Sampling and analytical techniques are described in detail by Durlng the experiment, bulk aerosol samples were Kriews (1992) and Schulz (1993). taken on board the RV 'Valdivia', from which measurements in the water column were also performed. For interpretation of the measured atmospheric con- Measured concentrations centrations, additional samples were taken at the Ger- man research platform Nordsee (FP 'Nordsee'), at the At the beginning of the drift experiment (April 23) isle of Helgoland, and at 2 coastal measurement sites, the observed atmospheric lead concentrations were namely Westerhever on the Eiderstedt peninsula and low (Fig. 2), because comparatively unpolluted air Neuharlingersiel on the East Frisian coast (see Fig 8 masses were advected from the northwest. Intense for the location of the sites). The sampling intervals precipitation events preceding the measuring cam- were scheduled on board RV 'Valdivia' depending on paign additionally reduced the concentrations in air. the local wind direction and in consistency with other During April 24 the lead concentrations distinctly meteorological measurements, e.g. the rawinsonde increased (factor of 50) with values above 30 ng m-3 at ascents (Schrum et al. 1997 in this volume). During RV 'Valdivia'. Backward trajectories were calculated to periods of easterly wind directions, when air masses of explain concentration trends during the drift experi- continental origin were advected, the temporal resolu- ment and to associate the measured concentrations tion of the measurements was increased. The atmos- with the origin of the air masses. pheric aerosol was accum.ulated for 6 h, thus the mea- Fig. 3 shows backward trajectories for April 24 calcu- surements were 6-hourly integrations. For westerly lated for the measurement sites RV 'Valdivia' (Fig. 3a) wind directions, when the advected air was very clean, and Westerhever (Fig. 3b). From 00:OO to 06:OO h UTC the intervals were extended to 24 h. The measured alr air masses from the northern North Sea reached RV concentrations confirmed this sampling strategy, since 'Valdivia' carrying low contaminant concentrations. they clearly changed for different wind directions dur- During the day, the wind turned from northerly via ing the drift experiment. northeasterly to easterly directions, and the air masses Aerosol was collected at a fixed inlet velocity on pre- passed Denmark; in the afternoon they were of conti- washed cellulose nitrate filters under non-isokinetic con- nental origin. The concentrations measured at Wester- ditions. Under these conditions, the upper size limit of hever were high most of the day. The corresponding the collected particles depends on the wind velocity trajectories (Fig. 3a) show the influence of continental (Zebel1979).The maximum aerodynamic diameters lie air masses for April 24 and northerly wind directions at about 30 pm for no-wind conditions (Kriews 1992) and for the day before (not shown here). With an average are lower for higher wind velocities. Trace element wind velocity of 3 m S-' air masses that passed Wester- analysis was performed with Graphite Furnace Atomic hever during the night hours could have arrived at RV Absorption Spectrometry (GF-AAS) after a wet oxidative 'Valdivia' on the morning of April 24, about 10 h later. acid digestion procedure. Rain samples were collected Thus the high concentrations observed at Westerhever with a specially prepared funnel on an event basis on and the more westerly sites FP 'Nordsee' and RV 'Val- board RV 'Valdivia' and with automatic rain collectors or divia' might originate at the same sources, with the bulk samplers at Westerhever and Neuharlingersiel. On time-lag in occurrence to be explained by the travel board RV 'Valdivia' sample preparation and handling time of the air. The sources were possibly in Jutland or

RVVALDMA :>$TtT FP NO~S~F

NEUHARIINOERSIEL HELGOLAM)

Fig. 2. Atmospheric lead concentrations (Cp,,)measured at (a) RV 'Valdivia' and Westerhever, and (b) other sampling sites 24 26 28 30 24 26 28 3 0 between April 23 and 30, 1991 April 199 1 April 199 1 Mar Ecol Prog Ser 156: 299-309, 1997

Fig. 4 shows backward trajectories for April 25 to 27, 1991, calculated for RV 'Valdivia'. During this time interval, air masses with high concentration levels reached RV 'Valdivia' as well as the other measurement sites. As can be seen from the backward trajectories, the sampled air masses had somewhat different origlns during the 2 days. On April 25 the air crossed Schleswig- Holstein and was possibly of continental origin again. Thus, highly polluted air was advected. During April 26 the wind changed slightly to northeasterly directions, and air masses transported over the Baltic 1 Ea&wardlramclr*W Sukewmd Dsstlnencm.FS Wine' I Emlrrreloa: I Sea, passing the southern region of Sweden, reached the sampling site. The concentrations decreased during April 26. Rain events which were observed over land can only partly explain the reduction; it was mainly caused by the wind shift and thus differences in the source areas. In the following days the measured atmospheric concentrations were around 10 ng m-3 at RV 'Val- divia', and the probed air masses at the measurement sites were mainly advected from northerly and northeasterly directions. The concentrations were at times reduced by precipitation which occurred over land. over the German Bight, and around the ship. On the 28th RV 'Valdivia' was in a precipitation area and wet deposition samples could be taken at Fig. 3. Backward wind trajectories for (a) RV 'Valdivia' and (b) Westerhever for April 24, 00:OO to 24:OO h UTC this site. over the continent, but cot close tc the German Bight. MODELLING ATMOSPHERIC CONCENTRATIONS On the same day, the concentrations were still low at AND DEPOSITION the more southerly measurement site Helgoland, whereas the East Frisian coastal measurement site The model calculations were performed using the 3- Neuharlingersiel showed higher concentrations (Fig. dimensional, nonhydrostatic atmospheric mesoscale 2b). The difference can be explained by the mesoscale transport and flow model METRAS. The model was flow pattern (see Fig. 6) which shows that different air developed at the Meteorological Institute, University masses are advected to these measurement sites. Pos- of Hamburg, for the calculation of atmospheric flows in sibly, these 2 air masses have different sources situated the mesoscale-7 and mesoscale-P range (Schliinzen close to the German Bight. These hypotheses are con- 1990, Wu & Schlunzen 1992, Niemeier & Schliinzen firmed by the mesoscale model results, as will be 1993) and for pollution transport studies especially in shown below ('Modelling atmospheric concentrations coastal areas (Schliinzen & Pahl 1992, Schliinzen and deposition-Model results'). 1994a, Schlunzen & Krell 1994). In the present paper Schlunzen et al.: PRISMA-Atrnosphenc input of lead 303

the model 1s used to calculate the deposition veloclty and is applied in a 3-dimensional case study of lead deposition

Description of the atmospheric transport model

An expllclt description of the model is given in Schlunzen (1988, 1990) and Schlunzen et a1 (1996) The main features are typical for nonhydrostatic mesoscale models, as can be seen from Schlunzen (1994b) Here, only those qualities of the model are descnbed in detail whlch Ba~hwaldIrdK'WlCS Surtace wlM Desl~nalfonFS Valohla Erpbral!ons are Important for the slnlulation of 25~~1991 39~~UTCICZ~AD~I lWl C~OOUTC Number of lralecrnrres 25 MaxIf~llme 60h Trne~nr-n I r hleanwo*jue\ocly 29mr Plo't-3 lraiecrores 25 contaminant transport over coastal areas and for modelling dry deposl- tion b sw~/ul A- u.--.&d ~d7 /S 58 rill The model is based on the funda- JY JY mental principles of S ,l conservation 4 fluid dynamics, namely those of c/- mass, momentum and energy Wind i \ B field, temperature, humidity, cloud i d and ralnwater content, as well as tracer concentrations are calculated from prognostic equations, pressure and denslty from diagnostic ones The subgnd-scale turbulent fluxes important for the dispersion of emit- ted contarmnants are parameterlzed ' I2 with a flrst-order closure scheme The determlnatlon of exchange coef- ficients dependent on wlnd shear BaCkwa d lrBp totar Sudacw wld ks11ra.k-r FS Valbvln and atmospheric stratification fol- 26Av1' '391 09W U-C IU Z:A~I1-1 W uu UTC lows Dunst (1982). A surface energy budget equation is applied to corn- Fig. 4 Backward wind trajectories for RV 'Valdlv~a'for (a)Apnl 25, 09:OO h UTC, pute the time-dependent surface to April 26, 09:OO h UTC, and (b) Apnl 26, 09:OO h UTC, to Apnl 27, 09:OO h UTC temperature over land and mudflats. A corresponding equation is used for humidity. Dry deposition is calculated as a function of water: Garratt & Hicks 1973). These are also consid- deposition velocity Vd and concentration C at the low- ered in the values for the surface resistances, which est grid point above the surface, following the resis- are taken from Walcek et al. (1986).The deposition ve- tance model concept (e.g. Chang et al. 1987): locity of lead is determined differently, since lead is transported on aerosols of a particle size between 0.1 Vd.C = (r,+rm+rs)-'.C and 1 pm (Seinfeld 1986). Following Voldner et al. The atmospheric resistance r, depends on the calcu- (1986), the surface resistances rs are taken to be zero lated wind velocity and atmospheric stability. For for particles. The values for the sublayer resistances r,, gases, the sublayer resistance rm and the surface resis- depend on land use characteristics and short wave ra- tance rs both have to be taken into account. The sub- diation. Over water, the sublayer resistance is zero, re- layer resistances are calculated with respect to differ- sulting in higher deposition velocities over water com- ent land use characteristics (urban areas: Brutsaert pared to the values over land. Neglecting the sublayer 1975; vegetation, mudflats: Weseley & Hicks 1977; resistance r, over water is equivalent to neglecting Mar EcolProg Ser 156: 299-309, 1997

smaller particles In this area. In the present study only (Voldner et al. 1986) and above the values (by a factor hygroscopic aerosols (e.g, s.ulphate) of a particle size of 2 to 10) calculated from impactor measurements above 0.8 pm are considered in the modelled deposi- (Steiger et al. 1989, Kriews 1992, Dannecker et al. tion velocities. Thus, the deposition velocities calcu- 1994). The discrepancy can be explained by the lated here are in the high end of the possible range. aerosol spectrum, for which different assumptions are With the described features, the atmospheric meso- made when calculating deposition velocities in the scale model can be applied to simulate a realistic diur- model and from the impactor measurements. In the nal cycle of the atmospheric boundary layer height and aerosol spectrum applied to measurements, the parti- of vertical mixing, as well as of wind, temperature, cle growth due to the high humidity over water sur- humidity, and precipitation fields. In addition, the faces and thus the increase in aerosol mass is not con- tracer transport and dry and wet deposition can be cal- sidered. In contrast, the METRAS model results are culated at a high temporal and spatial resolution. Thus based on the dssumption that lead is transported over the deposition velocity and the atmospheric concentra- water on sulphate aerosol which has grown (particle tion can be calculated at the same temporal resolution size above 0.8 pm). Both calculations neglect the bi- as is necessary for comparison with and interpretation modal particle size distribution found over the sea of measurements. (Steiger et al. 1989, Schulz et al. 1994). For this reason, the modelled deposition velocities characterise the upper limit of the deposition velocity values, whereas Model results the deposition velocities calculated from the impactor measurements mark the lower limit for Vd over water. In the case study presented In this paper, cloud Very high concentrations were measured at the development and wet deposition processes have not northern sampling sites on April 24. For this day back- been included, because wet deposition took place ward trajectories show that long-range transport might directly at the measurement site RV 'Valdivia' only for have caused the high concentrations (section 'Mea- a few hours (see the section 'Measured concentrations' sured concentrations' above). To confirm this hypothe- above). sis, the 3-dimensional version of the METRAS model The l-dimensiona.1 version of the mesoscale model was appljed. The model, was initialized with mean pro- METRAS was applied to calculate the dry deposition files derived from the rawins0nd.e data taken from RV velocity corresponding to the measurement intervals 'Gauss' (April 23: 10:OO h UTC; April 24: 10:00, 16:00, given in Fig. 2. The model was initialized for a neutral 22:OO h UTC), which was close to RV 'Valdivia' in the atmospheric stratification and with the wind velocities German Bight. The mean stratification in the atmos- measured at RV 'Valdivia' (Schrum et al. 1997). The phere generally was slightly stable with an inversion at resulting deposition velocities are given in Fig. 5 Most an altitude of 2000 m. The wind velocity close to th.e of the time the simulated deposition velocities are ocean surface was 4.2 m s ' from the northeast an.d the around 0.6 cm S-'. On April 27 the values are highest water temperature was 7.2"C. For details on the mete- (l.? cm S-'), because the wind velocities were above orological situation see Schrum et al. (1997). The lead 10 m S-' for some hours. The modelled values are in the emission data for Germany were taken from Munch & range given by measurements over the Great Lakes Axenfeld (1990); they correspond to emission values for 1987. Because of the emission redu.ctlons carried out in Germany in the meantime, the values used in the model for Germany lie above the emiss~onsfor 1991 The sources in the model areas representing Denmark and the Netherlands were appropriately estimated, but point sources were neglected. In Fig. 6 the simulated wind and temperature fields are given for a height of 10 m, above the ground for 08:OO and 14:OO h UTC. In the morning the fie1d.s are quite homogeneous-the air temperature over the land is about 2°C higher than over the German Bight. The wind comes from easterly directions and the wind 23 24 25 26 27 28 29 30 velocity is slightly higher over the German Bight (3 to April 199 1 4 m SS') than over land (l to 2 5 m S-').During the day, Fig. 5. Simulated deposition velocities for lead, dependent on the temperature rises up to ab0u.t 12°C over the land measured wind velocities at RV 'Valdivia' for the measure- surfaces (Fig 6b). Since the temperature over water ment intervals given in Fig. 2a does not change correspondingly, the resulting tem- Schlunzen et a1 PRISMA-Atrnosphenc input of lead

Fig. 6. Simulated wind and temperature f~eldfor the area of the German Bight for Apnl 24 (a) 08 00 h UTC, (b) 14:00 h UTC Temperature increment 1s 1°C perature difference at the coast causes a convergence The wind flelds influence the deposition field as can zone with a wind parallel to the coast. At the East be seen from Fig. 7 The hourly dry deposition values Frisian coast the modelled wind velocity is highest (up are between 200 and 300 ng m-2 h ' at 08:OO h UTC. to 6 m S-') whereas it is lower in the area of the German Durlng the day the boundary layer height increases, Bight (3 to 6 m S-'), at the North Frisian coast and over and the concentrations (not shown here) close to the land (both 1 to 3 m S-'). The simulated wind and tem- surface decrease over the German Bight and at the perature fields correspond to the observed wind direc- North Frisian coast. They increase at the East Frisian tions and temperatures over land, but show a differ- coast. These model findings are consistent with the ence in wind direction of 30" relative to the directions measurements (Fig. 2). Lower concentrations were measured at RV 'Valdivia' measured at Westerhever and Helgoland during the Mar Ecol Prog Ser 156: 299-309, 1997

afternoon, whereas the concentrations were higher at sion data. While these were too high for Germany, the Neuharlingersiel. The concentrations measured at emission values estimated for the southern part of Den- these sites probably originate from sources close to the mark were too low and did not include industrial point German Bight. However, large-scale transports were sources. However, only the very southern part of Den- also ~mportant;they influenced the concentration level mark is included in the model calculations and most of and caused the increase in concentration at FP 'Nord- the sources there are included. The third reason again see' and RV 'Valdivia' (see the section 'Measured con- focuses on missing emission data. These might not be centrations' above). missing in the model area itself, but fluxes across the The deposition velocity remains about the same fur- lateral boundaries- which can include influences of ther offshore during the day, but decreases in the con- remote sources-might be underestimated. In the sim- vergence zone at the North Frisian coast (1.2 mm s-'). ulations they are assumed to produce background con- For this reason and due to the reduced concentra- centration values outside the model area. If long-range tions, the atmospheric input is mostly lower in the transport were included in the model calculations, afternoon compared to 08:OO h UTC, and minimum these values would have to be considerably higher. It input values are to be found close to the North Frislan can be concluded from mesoscale model results and coast. Only at the East Frisian coastline, where higher backward trajectories that during the first measure- deposition velocities and concentrations are modelled, ment interval on April 24 at least 50% of the concen- is the atmospheric input higher at 14:OOh than at trations measured at RV 'Valdivia' had sources outside 08:OO h UTC. the model area and that long-range transport was The simulated lead concentrations are around 13 ng responsible for the increase in concentration. Sources m-3 at RV 'Valdivia'. This value is lower by up to a fac- at a distance of up to 100 km may have caused concen- tor of 3 than the concentrations measured there on trations of 13 ng m-3 for the day examined. April 24 (32, 17 and 13 ng m-3; Fig. 2a). There are 3 possible explanations. The simulated wind field gives a wind direction for the drift area which is 30" too far to CALCULATED ATMOSPHERIC INPUT - the north, and thus some sources in Schleswig-Hol- MERGED MEASUREMENT AND MODEL RESULTS stein might not have contributed to the concentrations modelled for the measurement site RV 'Valdivia'. Fig. 8 shows lead deposition values calculated with 3 However, this effect seems to be of minor importance different methods at the 5 sampling sites for the period since the backward trajectories presented in Fig. 3 of the drift experiment. The dry deposition rates, Fdlr indicate that the contaminants have been transported are based on the concentration measurements and an over Denmark. The second reason might be the ernis- average deposition velocity for lead of 0.17 cm S-',

1 Neuharhgersiel 2 FP NORDSEE 3 Westerheversand

5 RV VALDMA

Fig. 8. Mean daily dry deposition values of lead (pg m-? d-l) calculated 10 with deposition velocities based on impactor 201Lo measurements. Fdlrand with modelled deposition Fdl Fw velocities, FdZ;and mean wet deposition values, F,. Results are integral data for the whole drift experiment period (6 d). No wet deposition oc- 10 curred at Helgoland and 20 FP 'Nordsee' Displayed 0 M are mean values and 10" Fd~ F~ standard deviations Schlunzen et al.. PRISMA-Atmosphenc input of lead 307

derived from impactor measurements. These mark the iment underlines the importance of long-term mea- lower limit for dry deposition values (see the section surements when mean atmospheric input values are to 'Model results' above). The upper limit for dry deposi- be estimated. At the same time, it shows the impor- tion rates is given by Fd2, which is calculated from tance of high resolution measurements when causes of measured concentrations and modelled deposition high or low concentrations in seawater are to be inves- velocities. The wet deposition rates, F,, are calculated tigated. from sampled precipitation amounts and lead concen- Assuming that the atmospheric input of lead com- trations measured in the rain samples. pounds (3 to 30 pg m-' d-') was completely absorbed In general, dry deposition seems to be representa- by the sea water, the average daily increase of lead in tively measured, because deposition values are at the upper meter of the water column was between 3 about 3 1-19 n1r2 d-' for Fdl and at about 10 pg m-2 d-' for and 30 ng kg-' d-'. This value is 1 to 100% of the F,, for all measurement sites. Wet deposition values observed low lead concentrations in the water and is differ. They lie between minimum and maximum dry far beyond or in the range of observed changes in con- deposition values at the measurement sites RV 'Val- centrations. For the measurement site RV 'Valdivia', divia' and Westerhever. Based on these measure- where the measurements in the water column were ments, about half of the atmospheric input into the performed, the mean total deposition was about 9 ng German Bight was caused by dry and half by wet kg-' d-l. However, for individual days the total deposi- deposition during the drift expenment. However, the tion was considerably higher. greater proportion of wet deposition at Neuharlinger- Fig. 9a shows the lead concentration in air and the siel (about 70% of the total) shows the uncertainty of dry deposition values calculated with the modelled this conclusion when it is applied to the whole German deposition velocity of lead. Individual lead dry deposi- Bight. The enhanced wet deposition at Neuharlinger- tion values range from 1 to 25 pg m-' d-' and are thus siel is caused by the more frequent rain events in that up to a factor of about 3 higher than the mean value of area. The low rain amounts at Helgoland (below 9 pg m-, d-l. Furthermore, the maximum dry deposi- 0.1 mm for the drift period) result in negligible wet tion is only about 10% of the total deposition (Fig. 9b). deposition values here. The different results show that The difference was caused by 1 rain event which an integration period of 6 d is too short to calculate lasted for 5 h only, but considerably influenced the area-representative wet deposition values, due to the total deposition. With a total deposition of 220 yg m-, random occurrence of precipitation. d-' the daily increase of lead concentration in the During the drift experiment, the atmospheric lead upper meter of the water column was about 200 ng input into the German Bight was low in the area of kg-' d-l), a value well in the range of measured sea Helgoland, whereas it was a factor of 10 higher at the water concentrations and their changes. East Frisian coastline. Deriving yearly input data based on these results for the whole German Bight, the values are between 25 t yr-l (based on Helgoland data, CONCLUSIONS Fdl) and 250 t yr-' (based on Neuharlingersiel data, F,, plus F,). Yearly means calculated from long term mea- The results presented on atmospheric input into the surements show a lower range (110 to 240 t yr-'; recal- German Bight elucidate possible causes of high con- culated values based on Schlunzen 1994a). The centrations and show the range of input data. The inte- enlarged range resulting from the present 1 wk exper- gral dry deposition values were within the same order

Fig. 9. (a) Atmospheric lead concentrations, C,, (ng m-3), and lead dry deposition rates, F,(Pb) (pg m-2 d-' ), calculated with the modelled deposition velocity for lead. (b) Dry deposition rates, F,(Pb), compared with total deposition rates, F,iPbi:,\ ,. note difference in scales April 1991 April 1991 308 Mar Ecol Prog Ser 1

of magnitude at all sites in the measurement area. In Acknowledgcn~mts. We thank Ursula Bredthauer, Holger the case of precipitation, wet deposition could become Gerwiy, dnd Gotz Stcinhoff at the Institut fur Anorganische the main contributor to the total deposition, and the und Angewandte Chemie, Universitat Hamburg, for the chemical analysis, XIartind Fdlke (now at the Urnweltbehorde values dlffered considerably between the measure- Hamburg) for taking the meteorological measurements on ment sites (up to a factor 10). A sampling time of 1 wk board RV 'Gauss', and Volker Reinhardt at the Meteorolog- did not result in similar deposition values over an area isches Institut, Universitat Hamburg, for calculating the back- of about 120 by 120 km'. This shou1.d be considered ward trajectorics. We thank the German Weather Serv~cefor kindly maklng the meteorological data available to us. This when comparing results from large-scale models with work was funded by the Universitat Hamburg and by the single measurements. It can be assumed that for longer Minister fiir Forschung und Technologie, Bonn, under grant time periods the probability that similar rain amounts number 03F0558A1 The authors are responsible for the con- will be measured at all sltes increases and measure- tents of thls publication. ments and model results will become more compara- ble. Due to the relevance of wet deposition for the total LITERATURE CITED deposition, wet and dry deposition values should be determined separately, in measurements as well as in Brockmann UH, Raabe T, Nagel K, Haarich M (1997) Mea- models. These data allotv comparisons even for short surement strategy of PRISIMA; design and realisation. Mar Ecol Prog Ser 156-245-254 time intervals, as shown in the present paper (section Brutsaert W (1975) The roughness length for water vapour, 'Model results'). sensible heat, and other scalars. J Atmos Sci 32:2028-2031 The differences in the deposition velocities calcu- Chang JS, Brost RA, Isaksen ISA, Madronich S, Middelton P. lated from impactor measurements and in the model Stockwell WR, Walcek CJ (1987) A three dimensional have been explained by the used aerosol slze distribu- Eulerian and deposition model. physical concepts and for- mulation. J Geophys Res 92:14681-14700 tions ('Model results'). The 2 values for dry deposition Dannecker W. Hinzpeter H, Kirzel HJ, Luthardt H, Kriews M, data presented in this paper mark the lower and upper Naumann K, Schulz M, Schwikowsky M, Steiger M, limits. For future model studies and evaluations of Terzenbach U (1994) Atmospheric transport of contami- measurement data, the bimodal structure of the nants, their ambient concentration and input into the North Sea. In: Sundermann J (ed) Circulation and contam- aerosol size spectrum and the growing of hygroscopic inant fluxes in the North Sea. Springer-Verlag, Berlin, particles over water due to humidity should be taken p 138-189 into account. Dunst M (1982) On the vertical structure of the eddy diffusion Models have been shown to be a helpful tool for the coefficient in the PBL. Atmos Environ 16:2071-2074 interpretation of measurements and they can help to Garratt JR, Hicks BB (1973) Momentum, heat and water vapour transfer to and from natural and artificial surfaces. pinpoint main source areas. In the work described Q J R Meteor01 Soc 99:680-687 here, 2 models were applied for different scales (back- INC (1987) Der Qualitatszustand der Nordsec-Zusammen- ward trajectories, mesoscale model). The model results fassung. Report by Wissenschaftlich-Technische Arbeits- underline the importance of long-range transport, gruppe zur zwelten Internationalen Nordseekonferenz. Her Majesty's Stationery Office, London which can be responsible for up to 50% of the mea- Karbe L, Aletsee L, Diirselen CD, Heyer K, Kammann U, sured concentrations. Mesoscale phenomena were Krause M, Rick HJ, Steinhardt H (1994) Bioaccumulation also found to influence the concentrations. The present and effects of plankton and benthos on the fate of contam- measurement and model case study verifies again that inants. In: Sundermann J (ed) Circulation and contaml- sources in the area of the German Bight as well as nant fluxes in the North Sea. Springer-Verlag, Berlin, p 555-597 remote sources contribute substantially to the atmos- Kersten M, Forstner U, Kriews M, Krause P, Dannecker W, pheric lead input. Garbe CD, Hock M, Terzenbach U, Grass1 H (1992) Pollu- The range of high-resolution total deposition data tion source reconnaissance using stable lead isotope ratios and the resultant concentration changes in the water ('""'?"~b) In: Vernet JP (ed] Trace metals in the environment Impact of heavy metals on the environment. Else- column (sectlon 'Calculated atmospheric input.. .') vier. Amsterdam, p 311-325 underline the importance of high-resolution data when Kersten M, Kriews M, Kuhn W, Rick HJ (1994) Combined investigating changes in the water column. The 6 d effects of abiotic and biotic factors on heavy metal fluxes. mean value for the drift experiment (9 ng kg-' d-l) was In: Siindermann J (ed)Circulation and contaminant fluxes not sufficient to reflect possible minima and maxima in in the North Sea. Springer-Verlag, Berlin, p 598-619 Konig P, Schrum C (1997) Hydrographic observations and atmospheric input. These lay an order of magnitude model results from a PRISMA drift experiment. Mar Ecol below or a factor of 25 above the mean values. Since Prog Ser 156:255-261 they are crucial parameters when explaining transfer Krause P. Kriews M. Dannecker W, Garbe-Schonberg CD, processes in the water column, they should be mea- Kersten M (1994) Determination of '0""U7Pbisotope ratios by ICP-MS in particulate matter from the North Sea envi- sured and modelled with a high temporal and spatial ronment. Fresenius J Analyt Chem 347:324-329 resolution, at least in investigations which aim to clar- Knews M (1992) Charakterisierung mariner Aerosole in der ify transfer processes in the ocean. Deutschen Bucht sowie ProzeDstudien zum Verhalten Schlunzen et al.: PRISMA-Atmospheric input of lead 309

von Spurenmetallen beim Ubergang Atmosphare/Meer- Schrum C, Konig P, Michaelsen K, Niemeier U, Pohlmann T wasser. Schriftenreihe Angewandte Analytik, Institut fur (1997) Meteorological and oceanographic situation in the Anorganische und Angewandte Chemie, Univers~tyof German Blght from 23 to 29 April 1991. Mar Ecol Prog Ser Hamburg 15 156.263-273 Moll A (1997) Phosphate and plankton dynamics during a Schulz M (1993) Raumliche und zeitliche Verteilung atmo- drift experiment in the German Bight: simulation of phos- sphanscher Eintrage von Spurenelementen in die Nord- phorus-related plankton production. Mar Ecol Prog Ser see. Schriftenreihe Angewandte Analytik, lnstitut fur 156:289-297 Anorganlsche und Angewandte Chemie, University of Munch G, Axenfeld F (1990) Emissions-Datenbasis fiir die Hamburg 17 Untersuchung und Bewertung des Pb-Eintrags uber die Schulz M, Stahlschmidt T, Maenhaut W, Francois F, Larsen Atmosphare in Nord- und Ostsee mit Hilfe eines meso- SE (1994) The change of aerosol size distributions mea- skaligen Ausbreitungsmodells (GESIMA). In: Eppel DP, sured in a Lagrangian-type-experiment to study deposi- Kapitza H (eds) Simulation des Transports und der Depo- tion and transport processes in the marine atmosphere. In: sition atmospharischen Bleis in der norddeutschen Proceedings of the EUROTRAC Symposium. April 1994, Kustenregion. AbschluDbericht, Forschungsprojekt UFA- Garmisch-Partenkirchen, Germany. EUROTRAC Office. FB 92-104 02 583. GKSS-Forschungszentrum, Geesthacht Garmisch-Partenkirchen, p 702-706 Niemeier U, Schlunzen KH (1993) Modelling steep terrain Seinfeld JH (1986) Atmospheric chemistry and physics of air influences on flow patterns at the isle of Helgoland. Beitr pollution. Wiley & Sons, New York Phys Atmosph 66~45-62 Spokes L, Jickells T. Rendell A, Schulz M, Rebers A, Dan- QSR (1993) Quality status report of the North Sea 1993. necker W, Kriiger 0, Leermakers M, Baeyens W (1993) Report on sub-region ?a. Salchow R (ed) Bundesamt fur High atmospheric nitrogen deposition events over the Seeschiffahrt und Hydrographie, Hamburg North Sea. Mar Pollut Bull 26:698-703 Raabe TU, Brockmann UH, Durselen CD, Krause M, Rlck HJ Steiger M Schulz M, Schwikowski M, Naumann K, Dan- (1997) Nutrient and plankton dynamics during a spring necker W (1989) Variability of aerosol size distributions drift experiment in the German Right. Mar Ecol Prog Ser above the North Sea and its implication to dry deposition 156:275-288 estimates. J Aerosol Sci 20:1229-1232 Schliinzen KH (1988) Das mesoskalige Modell 'METRAS'- Sundermann J (1997) The PRISMA project: an investigation Grundlagen, Validierung, Anwendung. Hamburger Geo- of processes controlling contaminant fluxes in the German phys Einzelschr A88 Bight. Mar Ecol Prog Ser 156:239-243 Schlunzen KH (1990) Numerical studies on the inland pene- Voldner EC, Barrie LA, Sirois A (1986) A literature review of tration of sea breeze fronts at a coastline wlth tidally dry deposition of oxides of sulphur and nitrogen with flooded mudflats. Beitr Phys Atmosph 63:243-256 emphasis on long-range transport modelling In North Schlunzen KH (1994a) Atmospharischer Eintrag von Nahr- America. Atmos Environ 20:2101-2123 und Schadstoffen. In: Lozan JL, Lenz W, Rachor E, Reise Walcek CJ, Brost RA, Chang JS (1986) SOz,sulfate and HNO, K, von Westenhagen H (eds) Warnsignale aus dem Wat- deposition velocities computed using regional landuse tenmeer-wissenschaftliche Fakten. Blackwell Wissen- and meteorological data. Atmos Environ 20:949-964 schafts-Verlag, Berlin, p 45-48 Warmenhoven JP, Duiser JA, de Leu LTh. Veldt C (1989)The Schlunzen KH (1994b) Mesoscale modelling-overview on contribution of the input from the atmosphere to the cont- the German nonhydrostatic models. Beitr Phys Atmosph amination of the North Sea and the Dutch Wadden Sea. 67:243-253 Rep No. R89/349A, TNO, Apeldorn Schlunzen KH, Bigalke K, Niemeier U, Lenz CJ, Lupkes C, Weseley ML. Hicks BB (1977) Some factors that affect the von Salzen K (1996) Concept and realisation of the meso- deposition rate of sulfur dioxide and similar gases on veg- scale transport and fluid-model 'METRAS' METRAS Tech etation. J Air Pollut Control Assoc 27:1110-1116 Rep 5, Meteorologisches Institut, University of Hamburg Wu Z, Schliinzen KH (1992) Numerical studies on the local Schlunzen KH, Krell U (1994) Mean and local transport in air. wind structures forced by the complex terrain of Qingdao In: Sundermann J (ed) Circulation and contaminant fluxes area. Acta Meteor01 Sin 6:355-366 in the North Sea. Springer Verlag, Berlin, p 317-344 Zebel G (1979) Modellrechnungen uber den Einflun des Schlunzen KH, Pahl S (1992) Modification of dry deposition in Windes auf die Einsaugkoeffizienten von Staubteilchen a developing sea-breeze circulation-a numerical case fiir zwei verschiedene Probenahmekopfe. Staub Reinhal- study. Atmos .Environ 26A: 51-61 tung Luft 39:349-362

This article was submitted to the editor Manuscript received: January 2, 1997 Revised version accepted: June 2, 1997