Interaction between nitrogen and sulfur cycles in the polluted marine boundary layer

. Simultaneous measurements are reported of the nitrate radical (NO3), nitrogen dioxide (NO2), ozone (O3), and dimethylsulfide (DMS) in the nighttime marine boundary layer over Biscayne Bay in South Florida. These field observations are analyzed and used to initialize a boundary layer box model which examines the relative importance of the various sinks for NO,(cid:127) in the marine boundary layer. The results show that the observed lifetime of NO3 (<6 min.) is probably controlled both by the loss of nitrogen pentoxide (N205) to reaction with water vapor and aerosols and by the reaction between NO3 and DMS. The model is then extended to investigate the loss of nitrogen oxides from an air parcel that remains in the boundary layer with a constant sea-to-air DMS flux for several days. The principal conclusions are (1) that DMS is a much more important sink for NO3 at lower NO2 levels and (2) that the reaction between NO3 and DMS is an important sink for DMS in the marine boundary layer and could exceed that of the daytime removal by OH.


Introduction
Dimethylsulfide (DMS) is the most important source of biogcnic sulfur in the atmosphere [Saltzman and Cooper, 1989;Andreae, 1990;Bates et al., 1992].It is produced in the upper ocean by phytoplankton and evades into the marine boundary layer (MBL), where its mixing ratio ranges between about 40 and 300 parts per trillion by volume (pptv) [Saltzman and Cooper, 1989;Andreae, 1990;Bates et al., 1992].It is oxidized by the hydroxyl radical (OH) during the daytime and by the nitrate radical (NO3) at night.In both cases the eventual reaction products containing sulfur are concentration only becomes significant (i.e., greater than 0.1 pptv) during the night [Wayne et al., 1991].It has been shown that N205 adsorbs readily onto wet aerosol particles to form aqueous HNO3, with an accommodation coefficient of up to 0.05 [DeMore et al., 1990;Dentener and Crutzen, 1993].N205 may also react very slowly with gas-phase H20 to form HNO3. Since N205 is coupled to NO3 through the equilibrium with NO2, these loss pathways for N205 cause the indirect loss of NO3.It should be noted that although NO3 reacts very rapidly with a variety of unsaturated hydrocarbons [Atkinson, 1991;Wayne et al., 1991 ], the concentrations of these species in the MBL are unlikely to be large enough for their reaction with NO3 to compete with the other loss processes for NO3 discussed above [Russell et al., 1986].Last, the unimolecular thermal decomposition of NO3 to NO and 02 has been suggested as a significant loss pathway in the atmosphere [Davidson et al., 1990].However, a recent photodissociation study of NO3 [Davis et al., 1993] has shown that the potential energy barrier for this process is (198 __ 3) kJ mol '•, so the rate of this process will be negligible at tropospheric temperatures.This is in accord with the hour-long lifetime of the radical that has been observed in the boundary layer at low relative humidities [Platt et al., 1984].
Here we report the results of a field and modeling study of the relative importance of these loss pathways of NxOy in the nighttime MBL.
The results are then generalized to investigate the relative importance of these nighttime losses of NxOy compared to the daytime loss, which occurs principally through the recombination of OH and NO2 to form HNO3 [Finlayson-Pitts and Pitts, 1986].
A related problem is then to examine the diel variation in the rate of oxidation of DMS.The major sink for DMS in the MBL is thought to be through its reaction with OH.Given a constant sea-to-air flux of DMS and this sink, the expected diel cycle for atmospheric DMS should show a predawn maximum due to nighttime accumulation and a late afternoon minimum due to daytime oxidation by OH.This cycle has been seen in data collected in the remote MBL [¾von et. al., 1996] where NO, is low.The reaction between DMS and 1379 NOs may be a significant nighttime sink for DMS in relatively polluted environments [Andreae et al., 1985;Plane and Nien, 1991 ].This nighttime sink could affect the amplitude and, if it is large enough, the sense of the DMS diel cycle.Tenax is then heated to 80øC to release the DMS, which is then injected onto a Chromosil 330 packed column.The DMS is separated isothermally at 50øC.DMS elutes at approximately 6 min and is detected by using a flame photometric detector.This method has a detection limit of 3 pptv.
Wind speed and direction, and ancillary measurements of 03 and NOx, were made by Dade County Environmental Resources Management (DERM), which has an air quality monitoring station in the rooftop laboratory at RSMAS.

Field Measurements
Most data were recorded during the months May-July and October-November 1989.It became apparent that the nighttime NO3 levels were generally less than 2 pptv in clean marine air from the prevailing southeasterly wind.However, this situation altered when urban emissions were mixed into the marine air. Figure 2a illustrates the wind pattem during the night of July 18, 1989.For the first 6 hours afler 1800 EST the wind was from the city or else stagnant, and then it swung round to the SSE and blew steadily from the ocean for the rest of the night.Figure 2b shows the NO3 and DMS concentrations during the night.Figure 2c   The chemistry of this model is similar to that described by Thompson and Lenschow [ 1984] with the addition of some reactions involving O(3p) and some reactions affecting the concentrations of peroxyacetyl nitrate (PAN).A complete list of the daytime and nighttime NOx reactions used in the model is shown in Table 1.This set of reactions emphasizes the importance of NO3 as a nighttime oxidant of many reduced species, as well as its role in coupling NOx to HNO3 and PAN.
The model employs fourth-order Runge-Kutta integration [Press et al., 1986] to solve the coupled ordinary differential equations describing the time-resolved variation of the minor species.

Reaction
in the higher NOx case (case 2) is due to the shift in the equilibrium between NO3 and N2Os toward N2Os at higher NO2 concentrations.This shift also significantly reduces the importance of the NO3 + DMS reaction during the first half of the case 2 model ran.It should be mentioned here that the rate constant for the reaction of N205 with water vapor is an upper limit, and this reaction may be very much slower.However, the shift in the equilibrium between NO3 and N205 toward N205 at higher NO2 concentrations would still occur,

Figure 1 .
Figure 1.A map of the location of the field study of NO3 and DMS in the marine boundary layer off Miami.
is a comparison of NO2, 03, and NO3.NO3 appeared at about 2030 EST, which was just after local sunset.It then tracked the changing concentrations of NO2 and 03, peaking when both NO2 and 03 were high at midnight.This finding is expected, since the rate of production of NO3 is proportional to the product of NO2 and 03.NO3 then dropped away rapidly as the wind switched to the SSE, and finally it disappeared at dawn.Inspection of Figure 2b indicates a striking anticorrelation between NO3 and DMS.Although one might be tempted to conclude that this was due to the rapid reaction between NO3 and DMS, it should also be noted that high DMS levels are a reliable indicator of air of marine origin, which may have lower NOx levels than continental air.Either of these last two factors will also tend to decrease the NO3 concentration.

Figure 3 Figure 3 .
Figure3shows a study over the nights of November 20 and 2 l, 1989, during which a weather front passed through South Florida.On the night of November 20 (Figure3a), frontal precipitation occurred after 2000 EST, and both NO2 and NO3 were substantially washed out in the following 2 hours.During the following night (Figure3b) the wind blew steadily from the north to NNE, conditions typical immediately after the passage of a cold front.The DMS concentration was highly variable, indicating significant mixing of urban and marine air, which is to be expected, since the wind was blowing parallel to the coast.Under these conditions the NO2 remained high most of the night.We began recording nighttime spectra about 30 min after sunset, when the NO3 was already up to 14 pptv.Thereafter it tracked the variations in NO2 and 03 and disappeared at dawn.The NO3 concentrations recorded on this night were the highest during 45 nights of measurements between May and December 1989.The NO3 lifetime can be estimated from the steady state relationship shown below: --•NO + NO2 + 02 k = 8.2 x 10 '14 Exp(-1480/T) NO:) + DMS -•HNO:) + products k = 1.1 x 10 '12 NO3 + CH3CHO --•HNO3 + CH3CO3 k = 1.4 x 10 '12 Exp(-1900/T) H2S + NO:) -=) products k tropospheric concentration of 03, DMS, NO, and NO2 are assumed to be 30 ppbv [Routbier et al., 1980; Carroll et al., 1990], 0 pptv [Andreae et al., 1988; Berresheim et al., 1990], 13 pptv, and 36 pptv [Carroll et al., 1990], respectively.The concentrations of organics that are used in the model simulations are typical of a ppbv (Figure 3b).In this case the DMS flux is the same as that used in case 1 and maintains a mean nighttime DMS concentration of•50 pptv in this case as well.Case 3 is designed to examine the effect that DMS has on NO3 concentrations.This case uses the same conditions as case 1 with the exception of the DMS flux, which was set to zero.model runs when the source of NO is turned on and a constant average NO, concentration is maintained (days 2-6 in Table 3), NO2 + OH is the most important sink for N, Oy in case 1, while in the higher NO, case (case 2) entrainment of NO and NO2 and the reaction of N205 with water and aerosols also become important.The importance of the N205 reactions

Figure 4 .Figure 5 .Figure 6 .
Figure 2. It can been seen from the top panels in this figure that the nighttime NO2 was sustained at approximately 2 ppbv until 150 hours into the ran, when the NO source was turned off.After the NO source was turned off, the NO and NO2 concentrations fall off rapidly.NO• concentrations fall below 40 pptv within 3 days of moving offshore.The middle panels in this figure show how the NO3 and N205 concentrations vary with time during this model ran.The N205 concentration becomes insignificant once the NO2 concentration falls below 100 pptv.The nighttime NO3 concentrations shown here are in good agreement with those measured during the latter half of the night shown in Figure 2, and they too fall off rapidly after the air mass is advected off shore.The results from case 2 are shown in Figure 5.This case was designed to represent the more polluted night shown in Figure 3b.The starting NO• concentrations in this case are an order of magnitude higher than they were in case 1, and within 6 days the NOx concentrations fall below 40 pptv.This is only twice as long as in case 1.As shown in the middle panels of Figure 5, both the NO3 and N2Os concentrations are much higher in this case than in case 1.The shift in the NO3[N2Os equilibrium mentioned earlier is apparent during the first part of the model mn with the N2Os concentrations being 5 times higher than the NO3 concentrations, while in case 1, NO3 is approximately 1.5 times greater than N205.The NO3 concentrations predicted for the first part of the case 2 model mn are in good agreement with the concentrations measured during the first half of the night shown in Figure 3b.The agreement becomes worse for the second half of that night, because the drop in the measured 03 concentrations during this period is not accounted for in the model nm.We also examined the effect that NO3 can have on the oxidation rate of DMS.Because the flux of DMS into the boundary layer and the entrainment velocity into the free troposphere are constant, changes in the diel cycle of DMS are