Chemical signatures of aged Pacific marine air: Mixed layer and free troposphere as measured during PEM-West A

. The Pacific Ocean is one of the few remaining regions of the northern hemisphere that is relatively free of direct anthropogenic emissions. However, long-range transport of air pollutants is beginning to have a significant impact on the atmosphere over the Pacific. In September and October 1991, NASA conducted the Pacific Exploratory Mission-West A expedition to study the atmospheric chemistry and background budgets of key atmospheric trace species. Aircraft sampling centered on the northern Pacific, 0 ø to 40øN and 115 ø to 180øE. The paper summarizes the chemical signature of relatively well-aged Pacific marine air (residence time >-10 days over the ocean). The chemical signatures show that marine air is not always devoid of continental influences. Aged marine air which circulates around the semipermanent subtropical anticyclone located off the Asian continent is influenced by infusion of continental air with anthropogenic emissions. The infusion occurs as the result of Asian outflow swept off the continent behind eastward moving cold fronts. When compared to aged marine air with a more southerly pathway, this infusion results in enhancements in the mixing ratio of many anthropogenic/continental species and typically those with lifetimes of weeks in the free troposphere. Less enhancement is seen for the short-lived species with lifetimes of a few days as infused continental emissions are depleted during transport (about a week) around the semipermanent subtropical high.


Introduction
The Pacific Ocean is one of the few remaining regions of the northern hemisphere that is relatively free of direct anthropogenic emissions. In remote regions of the Pacific some distance from the continents and under selected meteorological conditions, biogeochemical cycles can be studied in relatively unperturbed (limited anthropogenic influence) or background conditions. For example, measurements under downslope conditions at the Mauna Loa Observatory (Hawaii) are being used to address remote Pacific region questions including our understanding of photochemistry in low-NO:• free-tropospheric marine environments [Ridley and Robinson, 1992]. However, there is little doubt that long-range transport of air pollutants from Asia and, to a lesser extent, North America is beginning to have a significant impact on the atmosphere with the result that background Pacific marine environments are not so pristine as, for example, a decade ago. In fact, one may For the meteorological conditions and flight locations of PEM West A, air masses residing over the ocean for periods as brief as 1 day to periods exceeding 10 days were frequently sampled. The focus of this paper is to summarize the chemical signature of relatively well-aged Pacific marine air sampled by the NASA aircraft during PEM-West A. In the analyses, only air (mixed layer and free troposphere) which has resided (based on back trajectory analyses) over the ocean for at least 10 days is considered. The term "marine air" used throughout the paper refers to this class of air. About 20% of approximately 120 aircraft flight hours met this aged condition. The chemical signatures presented herein represent a background against which special case studies (from companion papers) of  Table 2 list the trace species measured on the aircraft along with sample frequency and limits of detection (LAD). All species listed were considered in our analyses, and results for most species are reported in the tables. However, for our marine classification, some measurements were routinely reported as at LaD. These are not given in our tabular results and include (LaDs are given in parentheses) isoprene (3 parts per trillion by volume (pptv)), N-hexane (3 pptv), pyruvic acid (10 pptv), oxalic acid (2 pptv), methyl sulfonic acid (1 pptv), phosphate (20 pptv), potassium (17 pptv), magnesium (6 pptv), calcium (14 pptv), aluminum (75 pptv), and vanadium (0.05 pptv).

Meteorology
Numerous PEM meteorological products are available and most are discussed by Bachmeier et al. [this issue]. A brief summary of the meteorology as it applies to conditions important to aged marine air is given below and is referenced to the mean winds derived from 12 hourly griddcd National Meteorological Center analyses during the PEM-W½st A time period (Figure 1). Figures la through ld depict the mean surface winds, 700-hPa, 500-hPa, and 300-hPa mean winds, respectively. These mean winds suggest two prominent transport pathways (north and south in the figure) for aged marine air into the PEM-West A study area (shown as a box in the figure). The major synoptic-scale features affecting long-range transport into the study area are (1) migratory cyclones (low pressure) with associated cold fronts, (2) the semistationary subtropical anticyclone (high pressure), and (3) the Intertropical Convergence Zone and tropical cyclones.
During PEM-West A, migratory midlatitude cyclones would frequently form and intensify over China/Mongolia or along the east coast of Asia. As these systems moved eastward, their trailing cold fronts formed the boundary between continental air of midlatitude origin and marine air of either tropical or midlatitude origin. Similar cyclones often formed over the central North Pacific. The mean positions of the cold fronts associated with these cyclones are depicted by the thick lines in the figure, with the southernmost frontal location denoted by the dotted lines.

During this time of year the semipermanent subtropical anticyclone is very evident over the western and central North
Pacific, centered near 35øN, 167øE at the surface and near 30øN, 162øE at 500 hPa, as illustrated by "A" in Figures la and lc. This feature is primarily responsible for the north transport pathway of aged marine air. While the 10-day backward trajectories for such air masses indicated no passage over land, infusion of continental air swept off the continent behind the eastward moving fronts did affect the chemical composition of the air. Any scavenging of these anthropogenic/continental emissions by cloudiness or precipitation would be minimized in the stable, subsiding flow around the anticyclone. Aged marine air arriving along the north pathway would, at times, exhibit higher levels of anthropogenic/continental emissions than would be expected for air aged 10 days in the marine environment.
The Intertropical Convergence Zone (ITCZ) was situated near 10øN, the normal September-October position. As a result, easterly trade winds originating from both the northern and the southern hemisphere often entered the southern portion of the study region. These deep tropical easterlies were responsible for the south transport pathway of aged marine air. Because of the general presence of these easterlies across the entire tropical Pacific, it is likely that prior to the 10-day back trajectories, some of the aged marine south air had significantly longer residence time over the Pacific and probably came from a region less populated or industrialized than eastern Asia. Once established, the easterlies would be perturbed by the frequent development of tropical cyclones (commonly referred to as "tropical storms" or "typhoons"). Of the six tropical cyclones which developed during the PEM-West A period, five moved through some portion of the study region).

PEM-West Classification of Air Masses
For purposes of defining source areas and previous emission history, air masses are classified by several methods: (1) back trajectory analyses, (2) aircraft lidar measurements of O 3 and aerosols, and (3) chemical ratios. For the most part, these classification schemes are discussed in four papers of this issue, and classifications developed in one or more of the papers are used by authors of companion papers for their analyses of chemical budgets, processes, and trends. Below is a brief sum-  issue], which describes the source of the gridded meteorological data, significant assumptions made in the analyses, and the relative reliability of trajectories under the different environmental conditions encountered. Briefly, isentropic back trajectories were calculated postmission for aircraft sampling positions (times) identified as important flight events, e.g., start and end of constant-altitude flight legs, regions of elevated pollution identified from the aircraft data, and regions of expected thermodynamical or dynamical changes identified from aircraft and/or predicted meteorological data. Typically, 15 to 20 trajectories were calculated for each flight. These trajectories represent approximately 60 to 70% of the regions and types of air masses sampled during PEM-West A and include altitudes from within the mixed layer to above 10 km. The goal of each trajectory analyses was to calculate backward trajectories for 10 days at 12-hour intervals. However, as the result of the trajectories leaving the domain (horizontally or vertically) for isentropic analyses, some trajectories end at about 7 or 8 days. Trajectories for mixed layer air generally end after only a few days. The first time step for each trajectory is less than 12 hours and projects the air backward to either 0000 or 1200 UTC. Subsequent time steps are for 12-hour periods ending at either 0000 or 1200 UTC.
Using the trajectory analyses supplied by Merdll [this issue], the data from flights 6 to 19 were classified as to measurements of either continental or marine air. Data for which the back trajectories showed the air mass sampled by the aircraft had passed over land within 5 days of the sampling time were classified as "continental." Data for which trajectories showed no land passage for 10 days were labeled as "marine." Islands were treated as land masses. Only data measured during constant-altitude flight are included in the analyses and each constant-altitude flight leg was considered separately in assigning a trajectory path. Thus data from the same flight are found in both databases as some flights sampled both continental and marine air. Where the trajectory did not extend to 10 days and ended, for example, at 8 days, a judgment was made as to whether the air would have passed over land during the addi-tional 2 days of backward trajectory. This judgment was based upon the location of the air at the last calculated back trajectory location, the direction/magnitude of synoptic winds, and the direction/magnitude of aircraft-measured winds at the flight altitude of interest. When there were insufficient data or some uncertainty as to land passage, the data were excluded from both databases. Included in the continental and marine databases are measurements from seven and nine flights, respectively. For some analyses, the marine data are further subdivided according to trajectory region and flight altitude. For each subgroup, the chemical data for samples within the group were examined to eliminate any singular data or data from a single flight which were inconsistent with other values for the data grouping. The few exceptions were limited to data measured in the vicinity of Typhoon Mireille (mostly free tropospheric and about 10% of the data initially classified as free tropospheric). For these exceptions the chemical data were inspected, particularly PAN and dimethyl sulfide, as suggested by the results of Newell et al. [this issue], and a judgment made as to the representativeness of data in the vicinity of Typhoon Mireille.

Measurement Results
Before discussing the marine results, it is important to place the marine data into perspective in relationship to the continental data. As previously mentioned, the companion paper by Talbot et al. [this issue] discusses continental chemical signatures as a function of air mass age (1 to 4 days from the continent); thus only a brief discussion of continental air is presented herein and is limited to observations important to the aged marine air discussions. The databases used for these discussions are those defined above as continental and marine.

The continental database (number of samples) may be slightly different than that used by Talbot et al. [this issue] which
divides continental air into north and south Asian cases; however, results are similar.
In comparing the continental and marine chemical data, it is noted that the continental data show substantial variation in mixing ratios for many species. This is not unexpected since our continental classification does not distinguish between the age of the air and includes air aged 1 to 5 days from the continent. However, after 10 days over the ocean, even the marine data for some species and at times show sizable variations in mixing ratio. The range of variations often includes mixing ratios that are at values typical of clean background air as well as mixing ratios sufficiently high to suggest an anthropogenic influence. For example, for the marine CO data approximately 50% of the observations in the 7-to 9-km altitude band (see Figure 2c, altitude code F) are at mixing ratios <75 parts per billion by volume (ppbv), while 25% are > 100 ppbv.  1. NOy (particularly the GT measurement) and nitric acid clearly indicate a larger supply of reservoir nitrogen for north marine air as compared to south marine air. A portion of this north/south difference may be attributed to processes not solely dependent on the trajectory classification, some of which are latitude dependent. For example, the increase in stratospheric influences on tropospheric air for the northern regions of the PEM sampling area contributes to the reactive nitrogen reservoir via decomposition of PAN during transport to low altitudes [Singh et al., 1992]  troposphere was impacted by stratospherically influenced air. At latitudes of 0 ø to 20øN this percentage decreased to -6%. 2. Processed species (intermediate species produced via cloud processing, photochemistry, and/or heterogeneous chemistry) in some cases show sizable differences (e.g., nitric and formic acid) between the groupings, while in other cases no differences are observed (e.g., acetic acid and the peroxides). For the first cases the differences (asterisks) of Table 1 may not be entirely the result of the trajectory classification and may also be influenced by the magnitude of previous (during transport) cloud processes, heterogeneous chemistry, and/or wet deposition (rain) which are not considered in the grouping of the data. Similarly for the latter cases, statistical differences which might be present as the result of the trajectory classification may be overshadowed by differences in local production rates (marine environment), production rates during transport (OH chemistry and cloud processing) in the lower troposphere (ML to few kilometers), and loss processes during transport (via wet deposition) on the various flight days. The relatively large interquartile ranges for acetic acid and the peroxides tend to indicate the importance of multiple processes and suggest the need for a database which includes additional observational days (perhaps with sufficient data to include classifications for some of the above processes) before we can be confident that the observed differences (or lack of) for these species are purely the result of trajectory classification. Heikes  tabase apply to both CO2 and N20. The difference shown for CO2 is barely significant, and while it is consistent with the known vegetative uptake of CO2 over the continent [Tans et al., 1990] and an eastern tropical Pacific CO2 source [Wong et al., 1993], a larger database is desirable. On the basis of the above long-lived theory, one expects N20 to exhibit a statistically significant difference between the data classifications of Table  1    As noted in Figures 7b and 7c, CO and ethane show an increasing gradient with altitude (no maxima in the 4-to 8-km altitude range). While this may be partly due to both the sparsity and the grouping of the data, we believe that the data suggest (most prominent in the ethane data of Figure 7c) that additional anthropogenic/continental outflow is occurring at altitudes as high as 12-14 km. This outflow results in elevated mixing ratios, especially for the longer-lived hydrocarbon species, and tends to mask any maxima that might be present in the 4-to 8-km altitude range. Propane data are similar to the ethane results of Figure 7 and the ratio of propane/ethane (not given in the figure) increases from 0.8 (8-to 10-km band) to -1.2 (12-to 14-km band and limited data). While our data and analyses do not permit a rigorous assessment, it is highly probable that the anthropogenic/continental emissions present at these higher altitudes are of multiple-continental sources (Asia, Europe, North America) and were advected aloft at a relatively young age and then transported globally into the study area. Hydrocarbons advected into the upper troposphere have relatively long lifetimes compared to those which are transported at lower tropospheric altitudes (e.g., propane lifetime at surface is -6 days as compared to -40 days at 10 km [Greenberg et al., 1990]). It is also probable that some of these high-altitude emissions may have been advected aloft (via clouds) as the air left the Asian coast. While the focus of our analyses is not this upper altitude outflow, it should not be assumed that this outflow is unimportant. For specific days and meteorological conditions (e.g., strong subsidence or storm activity) the higher-altitude outflow could be very important to the chemical signature of air classified in this study as north marine.
The south marine data when subdivided into altitude bands does not contain sufficient samples (some altitude bands include only data from flight 15) to rigorously compare with the north data. However, south data are shown for 03 and SO2 ("a" panels and shifted +0.25-km altitude units) and imply that similar enhancements in the 4-km to 8-km altitude range are not prominent in the south data. Where sufficient data are available, enhancements were not observed for most species for the south data.  Table 1 with ML data of Table 2 clearly shows the ML (north and south) is enhanced in di-   The ocean is a well-documented source of atmospheric DMS [Bates et al., 1987[Bates et al., , 1992Andreae, 1990;Spiro et al., 1992]. With a DMS lifetime in the FT of only a few hours, mixing ratios decrease rapidly with altitude, as shown in Figure 8a. Little difference is seen between the north and the south (all altitudes) results.
As shown in Figure 8b, CH3OOH (as did total organic) exhibits a rapid decrease with altitude, but based on the south data (north data not available), the decrease does not occur until above -2 km altitude or well above the marine mixing layer. Again, little difference is observed between the north and the south results. H202 (Figure 8c, north and south)  As shown in Figure 9 and Table 2, ML nitric and formic acid mixing ratios are notably higher for the north region than the south, and acetic acid values are similar for both mixing layers.
Other results of Table 2 show statistical differences for some species (asterisks of Table 2). We briefly comment on these observations noting the desire for a larger database from which to rigorously discuss ML observations. First, for those species which showed statistical differences between the north/south FT databases of  Table 2 (GT data only) is only about 1 pptv and is well within the precision of the measurement and the statistical representation of one and two flights for the south and north ML data, respectively. For 03 the differences observed within the ML between the north and south data are not the result of changes in ML photochemical loss/production rates associated with the different latitude of the measurements. Davis et al. [this issue] calculate the ozone photochemical tendency along the aircraft flight path within the ML; their Table 5 shows values for daily net ozone loss of about 2.5 x 105 molecules/cm3/s (-0.8 ppbv/ 24-hour day) for both the ML at 00-20 ø and 20ø-40øN. These net loss rates are lower than calculated for the remote marine ML over the Atlantic Ocean northeast of Brazil (5øN to 5øS latitude) of -6 ppb/d [Anderson et al., 1993;Davis et al., 1993]. Assuming similar loss rates of <1 ppbv/d within the marine north and south mixing layers, the 4 ppbv difference (south lower) in 03 shown in Table 2 is not the result of differences in ozone photochemistry but is probably the result of the higher reservoir FT supply of 03 for the north marine air (Table 1,

versus 22 ppbv). Calculations of Davis et al. [this issue]
show that for altitudes above the ML, ozone photochemical tendency increases toward more production and less loss as both altitude and north latitude increase. Table 2 can be used to illustrate our earlier concern for a larger database to represent ML results. As shown in Table 2, CS2 is about a factor of 10 (5.7 versus 0.6 pptv) higher in the north ML than in the south ML. No differences were observed for the FT data. However, the CS2 results for north ML data originated from only two flights. The median value of CS2 for the two flights were 6.5 and 1.4 pptv. This is within the range observed for the eight flights included in the north FT grouping in which two flights showed medians around 6 pptv, while the other six flight median values were about 1 pptv. For the north data, the ML (flight 8) and FT (flights 6 and 8) sampling in which CS2 mixing ratios were the highest have back trajectories which suggest strong and persistent continental outflow into the Pacific Ocean relative to other flights. Thornton et al. [this issue (b)] uses CS2 mixing ratios of <1 and >2 pptv to represent clean background marine air and anthropogenically influenced air, respectively.

CS2 results of
We conclude that for the marine ML in the PEM-West study area and for the long-lived species (1) the FT is a reservoir source of these gases for both the north and the south marine ML and (2) both mixing layers are net sinks for these gases (via chemistry and/or deposition). Further, we postulate that the tendency for the long-lived species to exhibit higher mixing ratios (CO2 lower) in the north marine ML as compared to the south ML is the result of the higher reservoir of these species in the north FT (subsequent transport into the ML) as opposed to any sizable differences in the magnitude of deposition and/or chemical loss processes occurring within the respective ML. In arriving at these conclusions, we have assumed that mixing processes between the FT and ML on the experiment days (regardless of latitude) were not drastically different for the reasonably clear flight days. Short-lived species showed no statistical differences between the north and the south ML as none existed in the FT reservoir supply of these species to the ML.

Modeling Results
As part of the PEM-West program, atmospheric chemical modeling was incorporated to aid in the interpretation of the measurement results. The modeling included premission, in the field, and postmission analyses with photochemical threedimensional and box models. Modeling results are included as both stand-alone papers in this special issue and in papers which focus on trace species or chemical families. In these papers some of the modeling results are processed to provide a simulation of the chemical signature of aged marine air. In many cases it was not possible (due to model mechanics and/or differences in time and spatial scales of the models compared to measurements) to provide simulations that are exactly comparable to measurement products; however, in general, results or trends suggested by the models are consistent with the aged marine observations and conclusions. We comment briefly on results discussed in two of the modeling papers but direct the reader to these and other companion papers for in-depth discussions of models and their application to the aged-marine environment.

Davis et al. [this issue
] use a box model to calculate photochemical O3 tendency (production, loss) along the flight track of the DC-8 aircraft. The calculations incorporated diagnostic modeling techniques which utilized some real-time aircraft measurements of critical O3-related photochemical parameters and meteorological data to calculate 03 tendencies along the flight track. In addition, mixing ratios of other key trace species important to 03 photochemistry are calculated, e.g., OH, nitric acid, and H202. In general, their results compare favorably to the marine observations. Selecting O3 as an example, the model shows that the marine ML is a net sink for ozone with a loss rate, when diurnally averaged, of the same magnitude at 00-20 ø (south marine) and 20 ø to 40øN latitude (north). This agrees with our marine observations in which O3 ML mixing ratios are consistently lower than FT values and provides supporting evidence for our conclusion that the 4 ppbv higher median O3 for the north ML (compared to south) is not the result of ML photochemistry but is due to a higher north FT reservoir supply of 03. The model also shows that above the ML, 03 photochemistry tendency moves toward more production (less loss) as flight altitude and north latitude increase and thus supports the marine observations of a higher FT reservoir of 03 for the north data. This move to more 03 production within the FT results in a model-calculated 03 column (surface to 12 km altitude) changing from a net loss of about 11 x 10 •ø (0ø-20øN) to a net production of 5 x 10 •ø molecules/cm2/s (20ø-40øN)

Concluding Discussion
The chemical signature for aged marine air which has been over the Pacific Ocean for at least 10 days shows that for some trajectories the air is not always devoid of continental/ anthropogenic influences. Analyses of mean winds suggest two distinct pathways for transport of aged marine air into the PEM-West A study area (0ø-40øN, 120ø-160øE). The major synoptic-scale features important to long-range transport of marine air into the study area are (1) migratory low-pressure cyclones, (2) the semistationary subtropical high-pressure anticyclone, and (3) the Intertropical Convergence Zone and tropical cyclones (storms, typhoons). Marine air (resided over the ocean for at least 10 days) which circulates around the semipermanent subtropical anticyclone located off the Asian continent (e.g., 30øN, 165øE) is influenced by infusion of continental air with anthropogenic emissions. This infusion is the result of persistent Asian continental outflow which is swept off the continent behind cold fronts accompanying eastward moving cyclones. This marine air, classified in the paper as north marine air, shows enhancements in some continental/ anthropogenic source species as compared to aged marine air with a more southerly pathway (classified as south marine). The south marine air originates from the eastern and more southerly regions of the Pacific (may have a southern hemisphere component) and is transported into the PEM study region by easterlies.
The enhancement observed for the north marine air occurs mainly for continental/anthropogenic species with lifetimes in the free troposphere (FT) of weeks to months as compared to species with lifetimes of only a few days. Short-lived species are depleted in concentration as the infused continental/ anthropogenic emissions are transported (about a week) around the semipermanent subtropical high. The enhancements for the north air occur in the altitude range of a few kilometers to about 8 km, consistent with expected altitudes of outflow from the Asian continent. With the exception of species with sizable oceanic sources (e.g., dimethyl sulfide), differences in mixed layer (ML) concentrations between north and south aged marine air are, for the most part, the result of differences in the FT reservoir supply for the specie rather than major differences in ML chemistry, photochemistry, or surface deposition. For processed or intermediate species like H202, CH3OOH, and nitric acid, differences between ML and FT mixing ratios shown for north and south aged marine air may not solely be the result of the trajectory classification but may also depend upon the degree of cloud processing, wet/dry deposition, and heterogenous processing which occurred during transport.