Large-scale air mass characteristics observed over Western Pacific during summertime

. Remote and in situ measurements of gases and aerosols were made with airborne instrumentation to investigate the sources and sinks of tropospheric gases and aerosols over the western Pacific during the NASA Global Tropospheric Experiment (GTE)/Pacific Exploratory Mission-West A (PEM-West A) conducted in September-October 1991. This paper discusses the general characteristics of the air masses encountered during this experiment using an airborne lidar system for measurements of the large-scale variations in ozone (03) and aerosol distributions across the troposphere and airborne in situ instrumentation for comprehensive measurements of air mass composition. In low latitudes of the western Pacific the airflow was generally from the east, and under these conditions the air was observed to have low aerosol loading and low ozone levels throughout the troposphere. Ozone was found to be below 10 parts per billion volume (ppbv) near the surface to 40-50 ppbv in the middle to upper troposphere. In the middle and high latitudes the airflow was mostly westerly, and the background 03 was generally less than 55 ppbv. On 60% of the PEM-West A flights, 03 was observed to exceed these levels in regions that were determined to be associated with stratospheric intrusions. In convective outflows from typhoons, near-surface air with low ozone (<25 ppbv) was transported into the upper troposphere (> 10 km). Several cases of continental plumes from Asia were observed over the Pacific during westerly flow conditions. These plumes were found in the lower troposphere with ozone levels in the 60-80 ppbv range and enhanced aerosol scattering. At low latitudes over the central Pacific the troposphere primarily contained air with background or low ozone levels; however, stratospherically influenced air with enhanced ozone (40-60 ppbv) was observed several times in the lower troposphere. The frequency of observation of the air masses and their average chemical composition are also discussed in this paper. in measurements 03 the at latitudes.


Introduction
The primary objectives of the NASA Global Tropospheric Experiment (GTE)/Pacific Exploratory Mission-West A (PEM-West A) field experiments were to investigate the atmospheric chemistry of ozone (03) and its precursors over the western Pacific, to examine the natural budgets of these species, and to assess the atmospheric impact of anthropogenic emissions [

H20 and O(ID) and that odd nitrogen is deposited as nitric acid
(HNO3) and particulate nitrate (NO3-) [Liu et al., 1983]. One of the key findings from Project GAMETAG was the unusually large 0 3 mixing ratios found in the troposphere between San Francisco and Hawaii, which led to further understanding of stratospheric injections of 03 into the troposphere [Danielsen, 1980;Danielsen and Hipskind, 1980]. The composition and chemistry of the marine boundary layer over the Pacific was studied as part of the Third Soviet-American

Gas and Aerosol Experiment (SAGA 3) conducted in February-
March 1990 over the western tropical Pacific aboard the former Soviet ship Akademik Korolev [Johnson et al., 1993], and aerosols were studied off the coast of China on survey cruises from 1985 to 1987 by Zhou et al. [1992]. These ship measurements have provided extensive data sets on trace gases and aerosols in the marine boundary layer over the Pacific.
To study the budget of 03 across the troposphere, we must be able to examine the sources and sinks of 0 3 as they are related to different air masses that are observed over the western Pacific.
In addition to the airborne DIAL system the DC-8 had instrumentation for in situ measurements of 03, aerosol size distribution, aerosol number density, NOx, NOy, CO, CH 4, N20, SO2, SO4, HNO3, PAN, NMHCs, CFCs, H202,7Be, other trace species, and meteorological parameters such as temperature, dew point, and winds. A general description of these systems and their measurements is given by Hoell et al. [this issue], and more details are provided in companion papers in this issue.

Data Results and Discussion
Eighteen missions were conducted as part of the PEM-West A field experiment between September 16 and October 21,1991. A list of the PEM-West A missions, the mission objectives, and the latitude and longitude ranges of the missions are given in Table 1, and the flight tracks for these missions are shown in Figure 1. During this field experiment, eight flights were survey missions between NASA ARC, Anchorage, Alaska, Yokota, Japan, Okinawa, Hong Kong, Guam, Wake Island, Hawaii, and NASA ARC; and ten flights were "local" missions for intensive measurements over the Pacific from the bases in Japan, Hong Kong, Guam, and Hawaii.
Ozone and aerosol distributions were measured remotely with the DIAL system on all flights during PEM-West A. These measurements provided nearly complete altitude coverage of 03 and aerosol distributions from near the surface to the tropopause region along the aircraft flight track.
The air masses observed during PEM-West A were broadly divided into six main categories: background air in the free troposphere, near-surface (mixed layer) air, stratospherically influenced air, convective outflows, plumes, and clean Pacific air. Examples are presented below for the 03 and aerosol distributions that were observed during PEM-West A (PWA). Characteristics are discussed for the different types of air masses, and an analysis is given of the extent to which the air masses were observed in various regions of the Pacific in different altitude ranges. The average 03 profile and chemical composition of major air mass types are also presented.

Continental Outflow
The survey flight from Anchorage, Alaska, to Yokota, Japan, on September 18, 1991 (PWA flight 5), revealed the presence of extensive stratospheric intrusions that extended down into the lower troposphere (<5 km) at midlatitudes over the western Pacific. Plate 1 shows the aerosol and 03 distribution observed east of Japan on that flight. An air mass with relatively low aerosol scattering (<200) and high 03 mixing ratios (>60 ppbv) can be seen in the center of Plate 1. This air mass is bounded by air with enhanced aerosol scattering (>400) that sometimes exceeds the scattering in the clean regions by an order of magnitude (>2000) and is optically thick in some locations (note attenuation of lidar signals in those regions). There are also enhanced 0 3 levels (>50 ppbv) in association with the enhanced aerosol regions below ~6 km. The upper level (>6 km) aerosol layers are clouds, and they are associated with intermediate levels of 03 (30-50 ppbv).
The in situ measurements on the DC-8 indicated that we were flying in the lower stratosphere at 11 km along this flight track. The highest values of 0 3 (262 ppbv) were obtained at 0111 UT, and at that time the dew point (DPT) was -66øC; CO was 52 ppbv; CH 4 was 1670 ppbv; N20 was 300 ppbv; SO2 was 226 pptv (parts per trillion by volume); SO4 was 208 pptv; NO was 30 pptv; NOy was 593 pptv; 7Be was 2653 fCi/m 3' and PAN was 116 pptv at 0108 UT. The levels of SO2 and SO4 were elevated due to the injection of gases and aerosols into the lower stratosphere from the Mount Pinatubo eruption that occurred in the Philippines in June 1991. The zenith DIAL measurements showed that 03 continued to increase above the aircraft, further confirming that we were in the lower stratosphere. The air mass below us with the elevated 03 and the relatively low aerosol scattering (compared to the tropospheric aerosol loading) appeared to be an extension of the air mass we were flying in at 11 km. A cross section of potential vorticity (PV) derived from the European Center for Medium-Range Weather Forecasting (ECMWF) meteorological analysis for a portion of our flight track and within 2 hours of our time is shown in Figure 2. The PV analysis confirmed the intrusion of air with enhanced PV levels (>4xlO -7 deg m 2 (kg s) -l) extending down to -3 km near 40øN and the sampling of the aircraft in the lower stratosphere at -250 hPa (PV~20). The vertical and horizontal extent of the intrusion was clearly evident in the airborne DIAL data south of 50øN, and there was good correlation between the 03 enhanced regions and the PV analysis. Large-scale stratospheric intrusions have been observed previously with an airborne lidar at midlatitudes in the spring [Browell et al., 1987] and at high latitudes in the summer [Browell et al., 1992[Browell et al., , 1994; however, this was the first observation of this large-scale transport process at midlatitudes in the summer.
An air mass backtrajectory was calculated to determine the origin of the air mass with enhanced aerosols and enhanced 03 below --6 km. The wind analysis along the flight track on September 18, 1991, indicated that the air mass with enhanced aerosols below ~6 km (Plate 1) was coming off the Asian continent, and the enhanced 03 was thought to be due to photochemical 03 production from precursor gases released over Asia. Air mass backtrajectories were computed for four altitudes    , the lower levels of 03 that are near the surface are also transported into this region. This topic will be discussed in more depth in a section below. A continental outflow from Asia was investigated east of Japan on September 24, 1991 (PWA flight 7). Plate 3 shows the aerosol and 03 distributions observed over the western Pacific on that mission. Significant amounts of aerosols were observed in the outflow region below 8 km (see left side of Plate 3). Except for the near-surface air (<2 km) the 03 generally exceeded -50 ppbv in the enhanced aerosol air masses, and in some areas, 03 exceeded 70 ppbv. There was large variability in the aerosol scattering and 03 levels near the surface with values ranging from 30 to 75 ppbv. In regions above 6 km the amount of aerosol scattering was reduced from the aerosol loading in the lower troposphere; however, the 03 levels were as high or higher than in the aerosol-enhanced layers. The zenith measurements are shown with a more sensitive aerosol-scattering scale, and while the level of aerosol scattering is reduced, there are some aerosols clearly present below 11 km where 03 exceeds 100 ppbv in some regions. The enhanced 0 3 with low aerosol loading above -12 km is associated with stratospheric air. The tropopause (as indicated by the O3=100 ppbv level) decreases in altitude from 15.5 km at -35øN to 12 km at -42øN with enhanced 0 3 below the tropopause indicating a broad transition region of mixing between the stratosphere and the troposphere. Evidence of cirrus clouds with lower 0 3 (<40 ppbv) can also be seen between 9.5 and 12.5 km. A backtrajectory analysis is presented in Figure 4 for selected altitudes at 0450 UT in this flight. This analysis clearly shows that the flow was from the west at all altitudes bringing gases and aerosols over the Pacific from China and Japan.
The aerosol and 03 distributions shown in Plate 4 were obtained on October 6, 1991, near Taiwan. The aerosol distribution below 6 km indicated that some portion of the air resulted from a continental outflow; however, the 03 distribution only showed enhanced 0 3 associated with the heavier aerosol loading region below about 3.5 km. Low 0 3 (<20 ppbv) was observed in part of the light aerosol scattering region from the 3.5-to the 5.5-km region, and above -5.5 km to the tropopause at -16.5 km the 0 3 ranged from 40 to >100 ppbv in what appeared to be relatively clean air (low aerosol loading). The backtrajectory analysis done for this case ( Figure 5) showed that the low-altitude air (-3 km) traveled over portions of the ocean and land areas before arriving at our measurement location. The low 03 component of the air resulted from the marine portion of the trajectory, and the enhanced aerosols and 03 came from inputs from either Taiwan and/or from the coast of China. The clean air with elevated 0 3 came from over China with the air at -6 km descending from -9 km during the trajectory. While the low aerosol loading and enhanced 03 levels of these air masses are consistent with the characteristics of stratospheric intrusions discussed above, the PV analysis did not show evidence for a stratospheric intrusion that would explain the magnitude of the 03 enhancement that was observed. Clouds associated with deep convection (> 12 km) over China were observed from the aircraft during this flight, and in situ measurements of air mass composition at -9 km confirmed that the air was not of stratos_pheric origin. Large aerosols can be washed out of the air during the deep convective events [Gatz, 1977], while insoluble gases will pass into the outflow region in the middle to upper troposphere to photochemically produce 0 3 if the precursor gases are present [Pickering et al., 1989[Pickering et al., , 1991. This discussion has presented examples of the major processes that were observed to increase 0 3 in the troposphere over the western Pacific during PEM-West A. These processes include stratospheric intrusions which mix elevated concentrations of 03 with the background tropospheric air to produce what we call stratospherically influenced air, and continental outflows which can increase 03 if the precursor gases are present in the air that is directly advected over the Pacific in aerosol-laden plumes or in outflows from convection over the continent.

Low Ozone Air and Convective Transport over Pacific
The first observation of an extensive region of air containing low 03 (<30 ppbv) was on October 1, 1991, on a survey flight that was orignally planned to go from Yokota, Japan, to Hong Kong, but because of a problem with the weather radar on the DC-8 and the presence of Typhoon Nat over the South China Sea was diverted to Okinawa. Plate 5 presents the aerosol and 03 distributions observed east of the northern tip of the Philippines on this flight. Except for the aerosols contained predominantly in the marine boundary layer below -1.5 km and some cloud activity seen on the right of the cross section, the atmosphere was clean of aerosols below -11.5 km. The region of low 03 (<30 ppbv) extended from the marine boundary layer to just

Tropospheric Occurrence of Different Air Masses
To assess the relative impact of various air mass types on determining the composition of the troposphere over the Pacific, the nadir and zenith DIAL 03 and aerosol data were used to categorize the air mass types observed on all flights during PEM-West A. Prior to establishing the criteria for the different air mass types, a reference or background 0 3 profile had to be defined. The reference 03 profile shown in Figure 8 is an approximation to the average of a series of 0 3 profiles that were selected to be representative of air that was reasonably unaffected by recent sources/sinks of 03 and aerosols. While the reference 0 3 profile may be close to the average western Pacific midlatitude background 0 3 profile, it is really only a model for use as a discriminator in this analysis. A better estimate of the average background 03 profile will come from the air mass analysis to be discussed below.
Nine air mass types were identified during this field experiment, and they were categorized using the following criteria in each altitude range: (1)   ::::::::::::::::::::::::::::::::::  and in vicinity of cirrus clouds; (9) convective outflows from convection over the continent (CO-C), 03 more than 120% of reference 0 3 profile, low aerosol scattering, and no evidence of stratospheric intrusions in PV analysis. The results of the air mass analysis are presented in Plate 7. At high latitudes (40ø-60øN) over the Pacific the average tropopause was at an altitude of-8 km (>70% of SINF were above this altitude). This is also confirmed by the average 03 profile for the SINF for this region (Figure 9) where the average 03 at 8 km excedded 80 ppbv. A significant percentage of the troposphere above 6 km (>20%) consisted of SINF and the balance of the troposphere was mainly composed of plumes with enhanced 03 from continental outflows. The average 03 profile for these plumes was 17-20 ppbv (-45%) greater than the reference 03 profile (Figure 9). Less than 25% of the troposphere consisted of plumes with background 03 levels. In this latitude range the main flow was from the Asian continent over the Pacific throughout the troposphere.
At midlatitudes (20ø-40øN) over the western Pacific the average tropopause was at -16 km, and the background air was the predominant air mass type throughout the free troposphere, representing about 40% of the troposphere from 3 to 16 km. This provides confirmation that the 03 reference profile used in the air mass classifications does represent the average air mass type in this region. In addition, the average 03 profile for this air mass type was very close to the reference 0 3 profile chosen for the air mass discrimination (Figure 10). The SINF and convective outflows (marine) also contributed greatly to the composition in the upper troposphere. In the midtroposphere, many different air mass types were observed, and in the lower troposphere, plumes from continental outflows and convective outflows from the continent were the dominant air masses. Figure 10 shows the average 0 3 profiles for these air masses in the midlatitude region. The high 03 plumes had up to 100% higher 03 than that found in the background air, and in the upper troposphere the 0 3 in the SINF was -40% higher than the background average. The 03 levels in the convective outflows (marine) were comparable to 03 levels 5 to 10 km lower in the troposphere.
The results for the western Pacific low-latitude region show that the tropopause was typically above 16 km and that outflows from convection over the ocean were present a large percentage of the time in the upper troposphere. The low 03 associated with the clean Pacific air and convective outflows dominated the air masses in the southwestern Pacific. Figure 11 shows the average 0 3 profiles for the these two dominant air mass types. The 03 in the clean Pacific air near the surface was less than 9 ppbv, which is more than 50% lower than the 03 reference profile, and in the middle to upper troposphere, the low 03 air in outflows was typically <33 ppbv, which is about 50% lower than the reference profile. Only a low percentage of plume cases were observed in this region, and they were mostly below 5 km. Background air mass conditions were found in a low percentage of observations Tables 2-4 present summaries of the regional and altitude dependence of several major air mass types observed during Table 2. The latitude dependence of the tropopause levels and the frequency of occurrence of enhanced 03 due to strataspheric intrusions are clearly seen in this table. A significant tropospheric extent of SINF at midlatitudes over the Pacific was observed with generally increasing frequency with altitude and covering an average of more than 39% of the troposphere above 9 km. The amount of SINF in the upper troposphere (>6 km) at high latitudes and in the lower troposphere (<5 km) at low latitudes over the central Pacific is given in this table. Table 3 gives the total percentage of the troposphere that contains all plume types. The amount of plumes below 8 km increased with increasing latitude over the western Pacific. This was predominantly the result of the advection of boundary layer continental air over the Pacific as a result of the climatalogical high-pressure system that is present in the southwestern Pacific during the summer. Some plumes in the lower troposphere were also observed in the central Pacific as a result of long-range transport in the 2-to 4-km range over thousands of kilometers into the central Pacific. The low 03 air, as represented by the clean Pacific air and convective outflow (marine) air (Table 4), consisted of 55-95% of the entire troposphere from 2-15 km at low latitudes in the western Pacific to 35-66% in the same altitude range in the central Pacific. At midlatitudes the extent of low 0 3 air was reduced to 18-38% in the 4-to 14-km range, and there was negligible evidence of low 03 air in the high-latitude region.

PEM-West A. The distribution of SINF is shown in
The average percentage of the troposphere that contains the different air mass types in the different locations is given in Ozone, ppbv The in situ measurements of background air (BKG) measurements were grouped primarily in the midtroposphere (5.2-8.6 km), while the plumes (BPLU, LPLU, and HPLU) were sampled over most of the lower troposphere (1.4-8.6 km). The clear cases of in situ measurements of stratospherically influenced air (SINF) were found in the mid to upper troposphere (8.6-11.0 km) at middle to high latitudes over the western Pacific and in the low to middle troposphere (3.4-4.6 km) at low latitudes over the central Pacific.
The convective outflow-continental (CO-C) air was measured in the midtroposphere (7-8.6 km), and the clean Pacific (CP) air was measured in the low to midtroposphere (3.7-8.6 km). The convective outflow (CO) marine air was sampled in the upper troposphere (8.3-11.3 km).
Chemical Characteristics of Different Air Masses 80 1 O0 While the altitudes of these in situ measurements were generally consistent with the regions where the air masses were most often observed in the low to mid troposphere (<10 kin), they cannot necessarily be considered to be representative of the same air mass type observed remotely by the DIAL system in the upper troposphere (-10-16 km). This is particularly true for SINF which has a chemical composition that is determined by significantly diluting air from the lower stratosphere with highly variable tropospheric air. In general, many compounds follow the trends that might be expected for the different air masses. The chemical composition was similar for continental plumes with high 03 levels (HPLU) and for continent convective outflow air (CO-C). Both had ele- These compounds are generally elevated in air from continental surface sources that have some pollution, and the backtrajectories also indicate that the observed air was from a continental source.
The C2H2/CO and C3H8/C2H 6 ratios in these air masses were also elevated (1.72-2.10 and 0.097-0.167, respectively) indicating that these air mass types have had a short processing time from their source over the continent. While the insoluble gas composition was similar in these air mass types, the levels of H202 (348 pptv) and CH3OOH (146 pptv) were greatly reduced in the continental convective outflow (CO-C) air (444 and 141 pptv, respectively) A detailed chemical characterization of the various air mass types encountered during PEM-West A was made using the comprehensive in situ measurement capabilities on the DC-8 [Hoell et al., this issue]. Missions and time periods were identified that corresponded to cases where in situ sampling occurred within each air mass type characterized by the DIAL system (Table 6). These measurement periods primarily corresponded to level flight legs that had either been overflown or underflown in the same geographic region during wall-type missions. Table 6 also shows the altitudes and geographic locations of the measurements. Only one case of low-O 3 plumes (LPLU) and two cases of background-0 3 plumes (BPLU) and near surface air (NS) could be clearly identified as having been in situ sampled. While the DIAL classification scheme could readily identify these air mass types, the limited in situ sampling of these air masses must be recognized in the interpretation of their chemical characteristics. Table 7 gives a summary of the combined chemical signatures of each of the air mass types based on the in situ sampled cases. The average chemical composition derived for each air mass type represents the first attempt to relate the large-scale classification capabilities of the airborne DIAL system with the chemical signatures of these air masses.   't' 150øE -160øW region.

Conclusions
The troposphere over the Pacific during the late summer/early fall was found to be composed of a complex combination of different air masses. The distribution of 0 3 and aerosols in these air masses has been used to differentiate between different air mass types. Backtrajectory analyses were used to identify the probable source (continental or marine) of the tropospheric air, and inference of stratosphere-troposphere exchange was made through PV analyses. The tropospheric composition was found to be greatly dependent on the region of the Pacific and the altitude range of the air masses. Nine types of large-scale air masses were characterized during PEM-West A. The spatial distribution, frequency of observation, and composition of these air masses were discussed above.
Significant enhancements in 03 levels (typically >40% greater than reference 03 profile) were found at high latitudes (40ø-60øN) over the Pacific during the summer/fall due to photochemical 03 production in continental plumes in the low to midtroposphere and to air that has been previously involved in stratospheric intrusions (SINF). Enhanced PV levels were associated with the SINF, and backtrajectory analysis confirmed the origin of the plumes over the Asian continent. The high-O 3 continental plumes and SINF represented 64% and 27%, respectively, of the troposphere in this high-latitude region.     BKGmbackground or reference air; NSmnear surface air; BPLU•plumes with background 03 in the free troposphere; LPLU•plumes with low 03 in the free troposphere; HPLU--plumes with high 03 in the free troposphere; SINF--stratospherically influenced air; CO-C•onvective outflow from convection over the continent; CP•lean Pacific air; CO•onvective outflow from convection over the ocean.

*Average values.
At midlatitudes (20ø-40øN) over the western Pacific the tropo-(<10 ppbv) below 2 km. In the central Pacific (150øE-160øW) at sphere was dominated by background 0 3 and aerosol conditions low latitudes the average 03 levels were not so low as in the (31%); however, 03 was enhanced due to stratospheric influences western Pacific. Background 03 was observed over 46% of the in 15-51% of the troposphere between 8 and 16 km. Plumes were troposphere, and clean Pacific air and convective outflows were frequently observed (over 37% of the time) in the lower troposphere below 5 km, and as in the high-latitude results, the majority of plumes had enhanced levels of photochemically produced 0 3 . These were due to advection of air from urban sources or biomass-burning regions. Low 03 was found in association with cirrus clouds in up to 30% of the cases in 8-to 15-km region. This air was transported from near the surface into the upper troposphere as a result of convective storm activity over the Pacific.
At low latitudes (0ø-20øN) in the western Pacific, low-O 3 air (up to 50% lower than reference 03 profile) was found in 68% of found 37% of the time. Observations of enhanced 0 3 in the lower troposphere (below 6 km) were unexpected. The low aerosol loading and enhanced PV values associated with this air confirmed that the enhanced 0 3 was due to previous stratospheric intrusions. The SINF had probably resulted from stratospheric intrusions that occurred at midlatitudes over eastern Asia, and this air subsequently descended into the lower troposphere of the central Pacific around the anticyclonic circulation in the western South Pacific. Plumes with enhanced aerosols and enhanced 0 3 (40-60 ppbv) were also observed in the lower troposphere over the central Pacific. These plumes were just above the marine the troposphere. The extensive vertical distribution of the low-O 3 boundary layer, and they were determined to have originated from air was attributed to convective storm activity associated with the distant fires in Borneo based on backtrajectory analyses and many typhoons in this region of the Pacific during the late reports of extensive burning on the island. summer. Photochemical destruction of 0 3 in the moist tropical The fractional contribution of the various air masses to the marine atmosphere was responsible for the low 03<30 ppbv average 0 3 profiles observed in the different regions was throughout most of the troposphere and the very low 03 levels generally similar to the fractional occurrence of the air mass types in that region; however, the higher 0 3 levels in the SINF and high-O 3 plume air masses resulted in these air mass types having an enhanced contribution to the average 03 profile at the expense of air mass types that have low 0 3 levels, such as the clean Pacific and marine convective outflow air. At high latitudes, air from the lower stratosphere dominated the 03 profile above 8 km, and high 0 3 associated with high aerosols in continental outflow plumes dominated the 0 3 profile in the low to midtroposphere.
At midlatitudes in the western Pacific, SINF contributed over 35% of the 0 3 above 6 km, and this is significant since the tropopause is above 16 km in this region. Many different types of air masses contributed to the average 0 3 profile below 6 km; however, the high-O 3 plume air made the most significant (40-65%) influence on the 0 3 level. With the predominant easterly flow at the low latitudes the average 03 profile was low (typically <30 ppbv), and the clean Pacific and convective outflow air mass types determined the 0 3 profile above 7 km, and in the lower troposphere the contribution of the clean Pacific air decreased with altitude to less than 35% at 2 km. Below 5 km, plumes from various continental sources (enhanced aerosols and 0 3) combine to influence the average 03 profile. A very surprising result was found in the central Pacific at low latitudes.
While the background and clean Pacific air mass types dominated the average 0 3 profile above 6 km, the air associated with SINF was found to have contributed 27 to 40% of the average 0 3 below 6 km. This shows the unexpected importance of this enhanced 03 source to the 03 budget in this remote region of the Pacific.
In general, the average chemical composition of all the air mass types followed trends that either reflected anthropogenic influences from continental sources, stratospheric influences or long processing times over the tropical marine environment. Convective transport processes over the land and the ocean tended to decrease soluble gases and washout aerosols; however, the insoluble species remained to reflect the near-surface source of the air. Thus the continental convective outflow air resembled the insoluble composition of the continental plume air, and the marine convective outflow air resembled the insoluble composition of the clean Pacific air. With the exception of the stratospherically influenced air the average 0 3 levels in the different air mass types were generally inversely related to the amount of processing of the air as defined by the C2H2/CO ratio. The amount of air mass processing (i.e., photochemistry plus mixing) significantly influenced the composition of the observed air masses [Liu et al., this issue; $myth et al., this issue]. As was done for 0 3 , the contribution of various air mass types to the tropospheric budget of other compounds can be determined from the chemical composition and fractional occurrence information presented in this paper for each air mass type.