An assessment of ozone photochemistry in the central/eastern North Pacific as determined from multiyear airborne field studies



[1] In this study, the photochemistry of ozone is examined for the central/eastern North Pacific (CENP), a region bounded by 180°–120°W and 0°–45°N. Measurements of ozone precursors have been made in this region during 10 previous airborne studies covering the last two decades. The two seasons for which the most extensive analysis has been possible are spring (March–May) and fall (September–November). Box model results have been displayed in the form of latitudinal/altitudinal plots for ozone formation, destruction, and net tendency. They indicate that for both seasons of the year, significant net destruction is found for altitudes in the 0–4 km range; whereas, at higher altitudes, the ozone tendency calculations lead to small values of net production. For both seasons, however, the total column integrated ozone trend is one of net destruction. The largest difference between seasons was in the value for ozone destruction. During springtime, this value was significantly larger than for fall. This trend coincides with higher average springtime ozone levels. The trends in CENP ozone were also compared to those for the western Pacific using measurements recorded during PEM-West A (fall 1991) and PEM-West B (spring 1994). While the PEM-West A results revealed a neutral ozone column tendency, those from PEM-West B showed net production at all altitudes. Both results are in sharp contrast to those reported here for the CENP region showing net destruction. This suggests that while lightning and surface emissions of NOx from the northwestern Pacific Rim have a strong influence on the ozone tendency in the surrounding region, due to the combined effects of dispersion and chemical loss most of this NOx does not reach the CENP region. Hence, ozone in the central Pacific is typically destroyed by photochemistry.

1. Introduction

[2] A knowledge of tropospheric ozone is essential to having a comprehensive understanding of some of the chemical controlling species in the troposphere. Serving as a major source of HOx radicals (OH and HO2), ozone is among the most critical agents that indirectly regulate the atmospheric sinks of many reduced compounds released from the biosphere [Logan et al., 1981; Thompson and Cicerone, 1986; Berresheim et al., 1995]. An understanding of the processes controlling ozone levels is therefore important to our understanding of the lifetimes, distributions, and future trends of many tropospheric species.

[3] Early studies of tropospheric ozone led to the conclusion that this gas was controlled primarily by stratospheric intrusions, with deposition to the surface defining the main sink [Fabian and Pruchniewicz, 1977]. Recognition of the key role played by free radicals in the troposphere led to the examination of photochemistry as yet a second major factor [Levy, 1971; Chameides and Walker, 1973; Fishman and Crutzen, 1978; Liu et al., 1980]. At the same time the highly variable lifetime of ozone (1–13 weeks) also resulted in the recognition that transport can play an important role in controlling local ozone levels. In fact, over vast regions of the ocean, the net effect of photochemical processes is often controlled by long range transport of NOx generated over continents [Davis et al., 1996; Thompson et al., 1996; Crawford et al., 1997b; Schultz et al., 1999; Jaegle et al., 2000]. Given the current interest in air quality in North America, the possible impact from outflow from Eurasia is of particular interest [Jacob et al., 1999; Jaffe et al., 1999; Newell and Evans, 2001].

[4] This study was designed to expand on previous efforts to understand North Pacific ozone, an undertaking seriously begun in the early 1990s by NASA's GTE program as well as by other research groups. Some of the earliest NASA studies focused on making airborne observations in the northwestern Pacific, e.g., the Pacific Exploratory Mission West A (PEM-West A) [Davis et al., 1996] and PEM-West B [Crawford et al., 1997b]. More recently, however, this program has shifted its focus to the central/eastern North Pacific (CENP), a region downwind from the Pacific rim as well as the continental landmass of Eurasia. Thus, we are here looking at air masses that have experienced even longer chemical processing times from their point of release. The effects from this chemical processing, as well as that from further mixing, will be assessed in this study, both by comparing the respective levels of ozone photochemical precursors for both regions [Davis et al., 1996; Crawford et al., 1997b; DiNunno et al., 2002], and by contrasting the resulting photochemical ozone formation and destruction rates.

2. Observational Database

[5] The observational database used in this study is that reported by DiNunno et al. [2002] and consists of data assembled from 10 aircraft missions in the CENP region whose time history spans 20 years. Five of these missions took place in the spring (defined here as March, April, May) and five were scheduled in the fall (defined as September, October, and November). (Details concerning the methods used to create this database, as well as the justification of these methods can be found in DiNunno et al. [2002].) The geographical boundaries of the data used in this study are 0° to 45°N in latitude and 180° to 120°W in longitude. The CENP region has been further subdivided into 5 degree latitude and 1 km altitude bins. A median value for each photochemical species has also been estimated for each bin. Finally, these median values were used as input to a photochemical box model for purposes of generating ozone formation, destruction, and tendency rates. Some evaluations of precursors as a function of longitude were also carried out to demonstrate the absence of major longitudinal gradients in the database.

[6] To further facilitate the readability of this text, a short summary of the precursor observational data reported by DiNunno et al. [2002] is given below. The distribution of ozone was found to shift from one season to another, the highest levels being observed during spring (∼75 ppbv). Both seasons experienced minimum values at low altitudes in the tropics (5–20 ppbv), and maximum values at midlatitudes (30°–45°N) and high altitudes. Ozone in the transition zone (15°–30°N) resembled that found at midlatitudes during spring (40–70 ppbv), while during fall it was more similar to that found in the tropics (20–40 ppbv). Carbon monoxide also maximized during spring (150+ ppbv), showing elevated values relative to the fall time period at all latitudes. Some of the highest levels, however, occurred at midlatitudes. Sharp springtime latitudinal gradients were also observed, turning into more modest gradients during the fall.

[7] For the most part, observed levels of NOx (NO + NO2) were below 15 pptv for both spring and fall seasons, though sporadic elevated concentrations can be found in a few grid boxes. In the latter case median values of 30–50 pptv have been reached. These events can be understood in terms of either plumes from Asia or local/regional lightning. In general, the low NOx values in the CENP region were found to be in sharp contrast to those recorded in the western Pacific [Singh et al., 1996; Kondo et al., 1996, 1997; Davis et al., 1996; Crawford et al., 1997a]. This is particularly true at low altitudes where during PEM-West B values of 20–40 pptv of NO were observed as far out as 2000 km from the Asian mainland [Crawford et al., 1997b].

[8] Water concentration trends in the CENP region revealed a strong decrease with altitude, and typically were found to drop an order of magnitude from the surface to 5–9 km. Fall concentrations exceeded those for spring, particularly at midlatitudes. Nonmethane hydrocarbons (NMHC) followed a trend similar to that for CO, showing maximum values at midlatitudes and for the spring season. Concentrations of the more reactive hydrocarbons during spring exceeded those during fall by factors of 1.5 to 5.

3. Model and Method Description

[9] The formation, destruction, and net tendency for O3, (i.e., F(O3), D(O3), and P(O3) respectively), have been defined in previous work [Chameides et al., 1987; Davis et al., 1996; Crawford et al., 1997b]. Key reactions are shown in Table 1. The model calculated quantities F(O3), D(O3), and P(O3) are defined as:

equation image
equation image
equation image

These definitions are appropriate under the assumption of remote marine NOx conditions, wherein the photolysis of NO2(J1) greatly dominates over its conversion to HNO3 [Chameides et al., 1987; Crawford et al., 1997b].

Table 1. Key Reactions That Define the Formation and Destruction of Ozone for Remote Marine Tropospheric Conditions
R1NO + HO2 → NO2 + OH
R2NO + CH3O2 → NO2 + CH3O
R3NO + RO2 → NO2 + RO
J1NO2 + hν → NO + O3P
R4O3P + O2 + M → O3 + M*
J2O3 + hν → O1D + O2
R5O1D + H2O → 2 OH
R6O3 + HO2 → O2 + OH
R7O3 + OH → O2 + HO2

[10] Ozone photochemical trends provided here are diurnal average values based on the output from a time-dependent box model (TDM). This model has been used in several previous studies involving marine photochemistry and all relevant reactions have been listed in the work of Crawford et al. [1999]. Modeled chemical species are integrated in time until their diurnal cycles have reached a stable state, with results matching the previous day's pattern. Rate constants for the HOx–NOx–CH4 chemistry were taken from DeMore et al. [1997] with the exception of the ozone photolysis rate which was taken from Talukdar et al. [1998]. The NMHC oxidation mechanism is that based on the condensed version of the Lurmann et al. [1986] model with updates in rate coefficients as reflected in the work of Atkinson et al. [1992]. Other modifications include explicit chemistry for some previously “lumped” family hydrocarbons (e.g., acetone, propane, and benzene), and added reactions for remote low-NOx environments (e.g., formation of organic peroxides). An evaluation of the error in various model products was generated using Monte Carlo techniques by combining the uncertainties in the reaction rate and photochemical constants. This has been determined to be less than 30% for key species such as OH and HO2 [Davis et al., 1993].

[11] Operationally, model runs are constrained using observational data for the critical inputs such as O3, NO, dew point, CO, and nonmethane hydrocarbons. To minimize the impact resulting from diurnal variations, the NO observations were filtered such that only those corresponding to solar zenith angles of less than 70° were processed. NO observational data were used to generate a total “short-lived” nitrogen budget (e.g., NO, NO2, NO3, N2O5, HONO, and HO2NO2) the concentration partitioning of which was defined by each model run as has been previously discussed by Crawford et al. [1999]. (Note, earlier studies established that the assumption of photochemical equilibrium for the NO–NO2 system is valid for remote conditions [Bradshaw et al., 1999]). Ozone column densities were generated using multiyear data derived from the TOMS instrument for the three month seasonal time periods defined in this study. These composite column densities are shown in Figure 1. From here it can be seen that while the two seasons are similar in the tropics, the springtime ozone column densities for the subtropics and midlatitudes tend to be significantly higher than those for fall.

Figure 1.

Latitudinal trends in ozone column density for spring (March, April, and May) and fall (September, October, November) seasons averaged from 1979–1991, 180°–120°W, as derived from TOMS data.

[12] Photolysis rates were estimated using the aforementioned ozone column densities in conjunction with the NCAR Tropospheric Ultraviolet-Visible (TUV) radiative transfer code (S. Madronich, personal communication). The median solar flux for each season was that defined by 15 April for the spring model runs, and 15 October for the fall. Direct observations of the UV solar intensity from PEM-Tropics A and PEM-Tropics B were used to synthesize an average cloud correction factor for the composite database. The latter two missions were chosen because of their geographic coverage and their similarity in instrumentation. This assessment of the cloud impact on J was performed by comparing the J-values based on in situ observations of spectrally resolved actinic flux [Lefer et al., 2001] to those generated from “clear-sky” model calculations. The ratio of these two values was then used to define our cloud correction factor. However, due to the large variation in the data within a season, as well as the absence of a simple latitudinal trend, it was decided to synthesize just two altitudinal dependent cloud correction factors. For the boundary layer (BL), the correction was a 4% reduction, while in the free troposphere it was estimated that a 6% increase in all J-values was necessary. For purposes of the current model runs, the above cited reduction was applied to the lowest 1 km in the troposphere, and the enhancement was applied for all altitudes over 1 km.

4. Results and Discussion

[13] In our examination of the photochemical trends in ozone for the CENP region, two major comparisons are explored. The first of these involves the influence of seasonal changes on photochemical ozone. Here we examine latitudinal and altitudinal patterns in O3 formation, destruction, and net production as well as variations in column integrated values for these same quantities. The photochemical column destruction and formation rates are then compared to the estimated stratospheric flux for O3, thus attempting to define the relative importance of each source.

[14] The second comparison will be that between model estimated midlatitude CENP ozone and the corresponding values published earlier for the western North Pacific (WNP) as reported by Davis et al. [1996] and Crawford et al. [1997b]. The latter comparison provides further insight about the efficiency with which NOx emissions from the Pacific Rim are transported across the North Pacific, and hence, their potential for influencing ozone levels within continental North America.

4.1. CENP Seasonal Photochemical Ozone Trends

[15] As previously noted, the data from the ten aircraft missions used in this study fall broadly into two major seasons, spring and fall. Modeling results using each of these seasonal databases are presented below.

4.1.1. Spring: Formation, Destruction, and Net Tendency

[16] The meteorological patterns for the CENP region were defined by DiNunno et al. [2002] in terms of three distinct latitudinal zones: tropical, 0°–15°N; transitional, 15°–30°N; and midlatitude, 30°–45°N. The selection of these three zones was based on the prevailing climatological flow patterns for this region. The tropical zone was found to be dominated by easterlies, the midlatitudes by westerlies, and the transitional zone was controlled by a mixture of wind patterns that varied with altitude and season. Since DiNunno et al. presented the ozone photochemical precursor profiles based on these meteorological divisions, ozone formation, destruction, and tendency will be similarly presented here.

[17] As shown in Figure 2a, ozone formation during spring is generally quite low. Although difficult to discern on the plate due to the low levels involved, tropical formation rarely exceeds 2.0 × 105 molecules/cm3/s. Formation rates in this zone above 8 km are typically 50% smaller than at lower altitudes. Within the transition zone, somewhat higher values of formation can be found (e.g., 2.0–4.0 × 105 molecules/cm3/s), reflecting both higher NOx and HOx levels. The model estimated elevated levels of HOx can largely be understood in terms of the higher levels of O3 in this zone. At midlatitudes, where both the solar flux and water levels are lower, HOx is also lower and it follows that F(O3) rates decrease relative to the transition zone. This decrease is further accentuated by the lower NOx levels in this zone. Hot spots, however, are seen in the springtime F(O3) profile (e.g., 1–2 km), reaching values of 6.0–8.0 × 105 molecules/cm3/s. These “hot spots” reflect highly elevated pollution events, the source of which has been traced back to Asia [Clarke et al., 2001]. They therefore represent one of the clearest cases recorded of continental plumes in the North Pacific that have spanned most of the Pacific Ocean.

Figure 2.

Seasonal altitudinal/latitudinal model results for F(O3), D(O3), and P(O3): (a) F(O3), spring; (b) D(O3), spring; (c) P(O3), spring; (d) F(O3), fall; (e) D(O3), fall; (f) P(O3), fall. (Blank grid boxes represent a lack of data needed for model runs.)

[18] Ozone destruction during springtime (Figure 2b) shows a very distinct altitudinal gradient. Values are seen falling rapidly above 4 km for all latitudes, with those between 4 and 12 km being a factor of 5–50 times lower than those observed for the BL. This strong altitudinal trend in D(O3) leads to 71–81% of the total column tropospheric O3 loss occurring in the bottom 4 km of the atmosphere. Not surprisingly, this low altitude D(O3) trend is seen tracking the trends in H2O and O3 where the major O3 loss process is R5 (O1D/H2O) [DiNunno et al., 2002]. The importance of the latter reaction also explains why D(O3) values for the midlatitude zone are only half those in the transition zone.

[19] By combining the spring F(O3) and D(O3) values from Figures 2a and 2b, one can evaluate the net effect of photochemistry on ozone (e.g., O3 tendency). These results are shown in Figure 2c. Here it can be seen that when below 6 km, values of P(O3) are generally negative, while at altitudes above this very modest net production occurs. Net destruction maximizes at 0.9–1.2 × 106 molecules/cm3/s between 5° and 30°N latitude. When converted to a per day value, this corresponds to 3–5 ppbv of photochemically destroyed BL ozone. This loss clearly necessitates a significant influx of new ozone if BL levels in this region are to be maintained. As discussed in the text that follows, this cannot be achieved by simple vertical mixing.

[20] Integrating the F(O3) and D(O3) values over the altitude range of 0–10 km represents yet another way of evaluating the net effect of photochemistry on ozone. For example, as shown in Figure 3 the column integrated O3 loss and formation rates are given along with the contribution made by each of the major chemical processes causing the loss or formation. Here it is seen that between 0° and 25°N the formation rate is relatively constant at ∼1.5 × 1011 molecules/cm2/s but shifts to higher levels (i.e., 1.5–2.5 × 1011 molecules/cm2/s) between 25° and 40°N. This shift to higher values demonstrates the influence of elevated NOx at these higher latitudes. As shown in Figure 3a, reaction R1 involving HO2 is the dominant process responsible for converting NO to NO2 and thereby providing a source of O3. The methyl peroxy reaction, R2, is of considerably lower importance and NMHC's are seen as playing only a very minor role in O3 formation.

Figure 3.

Spring latitudinal trends for column integrated (0–10 km) photochemical ozone formation and destruction. Major chemical species/processes are shown: (a) F(O3); (b) D(O3). Values are calculated for 5 degree latitude bins.

[21] Column integrated ozone destruction during springtime (see Figure 3b) reveals far more latitudinal structure than seen for formation. Values for destruction are shown increasing from 2.2 × 1011 molecules/cm2/s at the equator to 4.7 × 1011 molecules/cm2/s at 20°N. At this latitude it is more than 3 times larger than column formation. These high levels of destruction persist up to midlatitudes where they start dropping off, with a value of 1.7 × 1011 molecules/cm2/s between 40° and 45°N. As already noted, in the tropics the O1D/H2O reaction (R5) dominates destruction with reaction R6 (HO2/O3) playing an increasing role at higher latitudes. The reduction in O3 destruction from 15° to 45°N is primarily due to a reduction in the rate of R5, reflecting a reduction in both UV solar flux and H2O levels.

4.1.2. Fall: Formation, Destruction, and Net Tendency

[22] The O3 formation rate during fall (Figure 2d) is seen following a similar pattern to that for spring, showing only modest formation rates for most altitude/latitude grid boxes. As during spring, the highest values typically occur at low altitudes. The tropics, however, appear to be the exception in that the maximum is between 6 and 8 km (2.0–6.0 × 105 molecules/cm3/s), coinciding with observations of elevated NOx. By comparison, formation rates between 15° and 45°N, and 0 and 4 km, typically lies between 2.0 and 4.0 × 105 molecules/cm3/s. Note that the elevated levels of NOx during spring, attributed to Asian plumes, are not seen at lower altitudes during fall. This follows from the difference in meteorology for the two seasons where during spring there is much stronger westerly transport [Bachmeier et al., 1996; Hess et al., 1996; Merrill et al., 1997].

[23] Ozone destruction during fall (Figure 2e) maximizes in the lower 2 km of the troposphere just as seen in spring. The highest values are found between the latitudes 0°N and 30°N (i.e., 6–10 × 105 molecules/cm3/s). Overall, ozone destruction is lower than that during spring with BL values, on average, being reduced by 31%. This decrease in destruction is largely as a result of lower O3 concentrations. The difference between seasons decreases with altitude and is only 16% for 1–2 km, and even smaller at higher altitudes.

[24] The lower O3 destruction rates found in fall produce a slightly altered profile, relative to spring, for ozone tendency as seen in Figure 2f. Although still strongly negative at low altitudes during fall, P(O3) is smaller than for spring. In addition, due to the absence of NOx plumes in fall, no grid boxes show net production below 4 km. Significant net destruction (e.g., >5.0 × 105 molecules/cm3/s) occurs up to 3–5 km in the tropics, but extends only up to 2 km in the transition zone and at midlatitudes such high levels are nonexistent. Modest levels of net ozone production are evident at altitudes above 5 km in fall, slightly exceeding those found in the spring. The latter trend tracks with the higher average levels of NOx seen during the fall season.

[25] Column integrated formation rates for fall (Figure 4a) are seen peaking at 1.5–2.0 × 1011 molecules/cm2/s between 15° and 25°N, being only slightly higher than those for spring. The peak value coincides with peak levels of NOx observed at high altitudes. Outside of the 15°–25°N region, column rates for the two seasons are quite similar. Fall values indicate a larger contribution from reaction R2 than seen for spring. Production from R3 is even more insignificant than for spring. These findings partly reflect the lower concentrations of CO and NMHC's observed during fall, while CH4 levels were found to remain nearly constant. Column destruction is seen as quite similar to that shown for spring when examined at latitudes south of 20°N. At latitudes >20°N, column destruction decreases more rapidly than that for spring. While solar flux is reduced for both seasons in the transition zone, lower tropospheric O3 levels in fall result in a decrease in column destruction rates for this region (e.g., down a factor of ∼1.5 relative to spring). The column destruction stabilizes at midlatitudes, where increasing O3 compensates for the further reductions in water and solar flux.

Figure 4.

Fall latitudinal trends for column integrated (0–10 km) photochemical ozone formation and destruction. Major chemical species/processes are shown: (a) F(O3); (b) D(O3). Values are calculated for 5 degree latitude bins.

4.1.3. Column Integrated Ozone Tendency and Stratospheric Flux

[26] Figure 5 shows a latitudinal bar plot of column integrated P(O3) (0–10 km) for spring and fall seasons. Not surprisingly, the overall trend most closely resembles that of D(O3), given that there is little variation in F(O3) versus latitude. Negative values of P(O3) tend to peak between 5° and 25°N (−18 to −32 × 1010 molecules/cm2/sec), and are more negative in spring than fall. They are much reduced between 25° and 45°N with typical levels ranging from −5 to −10 × 1010 molecules/cm2/s. This is more true for spring than fall, but still applies to both seasons. The overall deficit seen in column P(O3) values suggest that ozone is most likely being added to the region either by advection or from stratospheric transport. However, it also must be acknowledged that the total tropospheric column in the tropics can extend to altitudes of 17 km, thus additional column increases in F(O3) are possible. We believe that these enhancements, in accordance with observed lowered concentrations of upper atmospheric HO2 [Tan et al., 2001], to be less than 20% of the total F(O3) column. This would lead to a change in the tropics of 2–3 × 1010 molecules/cm2/s and would therefore not alter significantly the column value of P(O3) at these latitudes.

Figure 5.

Latitudinal trend for column (0–10 km) P(O3) for spring and fall. Values were calculated for 5 degree latitude bins.

[27] The northern hemispheric stratospheric O3 flux maximizes during the time period of January to May and tends to go through its minimum in the fall [Appenzeller et al., 1996; Wang et al., 1998]. Since 80–90% of the stratospheric flux also occurs over the latitude range of 20° to 60°N, this is the region where a further assessment could prove to be most useful. In this context, the springtime average stratospheric flux is estimated at 8.7 × 1010 molecules/cm2/s [Appenzeller et al., 1996; Wang et al., 1998; Logan, 1999]. But this value decreases to 2.5 × 1010 molecules/cm2/s in the fall. For comparison purposes, the average 0–10 km photochemical destruction flux over the latitude range of 20°–45°N is 3.2 × 1011 for spring and 2.2 × 1011 molecules/cm2/s for fall. These values indicate that by itself, the stratospheric flux falls factors of 4 to 8 times below what is required to maintain ambient ozone levels. A comparison with both spring and fall F(O3) values also reveals that, on average, our estimated column-integrated ozone formation is also 2 to 6 times larger than that from the stratosphere. Overall, therefore, it is quite apparent that tropospheric photochemical ozone formation is essential if the ozone budget is to be balanced.

[28] The stratospheric flux, although falling short of that required for the entire latitude range of 20–45°N, does appear to make up some of the budget deficit for some of the high latitude bins shown in Figure 5. This is especially true for springtime where the stratospheric flux is capable of closing the photochemical column ozone deficit in three of the four latitude bins north of 25°N. Thus, at some of these northern latitudes it is quite possible that column equilibrium may be reached during springtime without further horizontal transport input. Such input, however, would still be necessary to maintain ozone levels for the transition zone and the tropics. During the fall time period, the deficit exceeds the average stratospheric flux by factors of 2–4 over the entire range of 20°–45°N. Again, this would necessitate an advective source if ozone levels were to be maintained at their specified levels.

[29] While the above assessment of the tropospheric column ozone budget for spring and fall seasons has been informative, yet a different approach involves a comparison of the column ozone trends evaluated in this work with those based on an analysis of satellite data. Such a comparison can be qualitatively carried out using the Fishman and Brackett [1997] tropospheric ozone residual maps. For example, these data reveal that for the CENP tropical and transitional zones, ozone levels peak in March–May; but for midlatitudes the peak in column O3 is delayed until June–August. Thus, the overall higher value for column ozone in spring reported by DiNunno et al. [2002] would appear to be consistent with the Fishman and Brackett tropospheric maps. However, when the detailed O3 column seasonal trends are examined in the context of our seasonally calculated P(O3) values, some problems surface. Clearly, the general overall increase in column O3 from the fall/winter time period to spring can not be explained by the fall photochemical ozone trends given here. Recall, the fall period had negative values for P(O3) at all latitudes, but especially so at latitudes <25°N. Furthermore, there is no way to know from the existing data whether they become positive before spring, but it seems unlikely. Thus, for these latitudes the most likely explanation for the Fishman and Brackett ozone trends would seem to be that much of the springtime/summer maximum in ozone in the CENP region is due to horizontal transport from upwind locations.

[30] The decrease in column ozone values in moving from the spring/summer time period to fall would appear to be less problematic. For instance, that column P(O3) values during fall are quite negative at all CENP latitudes is quite consistent with the Fishman and Brackett satellite ozone maps showing column ozone moving toward a minimum. While the estimated net destruction for fall is lower than that for spring, it is still large enough that the modest decrease in the ozone column predicted from the satellite maps might still require some influx of outside ozone rich air.

4.2. CENP Ozone Lifetime Considerations

[31] The photochemical lifetime of ozone primarily reflects the value of D(O3) and the measured ambient level of ozone. As can be seen in Figure 6a, there is a sizable latitudinal trend in the percentage of ambient ozone destroyed per day when the evaluation is based on the tropospheric column. For both spring and fall, the tropics (0°–15°N) typically show a column (0–10 km) ozone decrease of between 6% and 9% each day. This compares rather closely with tropical Pacific (10°S–10°N) values of 6.4–6.8% cited by Olson et al. [2001] for PEM-Tropics A and B data. The column loss for the transition zone (15°–30°N) is more consistent (4–8%), the fall season always being higher than spring. This primarily reflects higher levels of water in this region during fall versus spring. At midlatitudes column destruction is much reduced relative to the tropics, falling to 3–5% of the column ozone each day. As expected, the percent ozone destroyed per day in the BL (Figure 6b) is considerably higher than for the overall column. Tropical BL destruction is seen reaching values as high as 24% a day, and clearly suggests the need for a significant influx of new ozone if ambient levels are to be maintained. As shown in Figure 6b there is a very significant decrease in the per day destruction rate with latitude when at low altitudes with midlatitude values dropping to 4–8%. Although the two seasons differ slightly in their respective destruction rates in the tropics, at higher latitudes they are nearly the same. This near balance occurs as a result of higher water vapor in fall being compensated by higher HOx concentrations in spring.

Figure 6.

Percentage of ambient ozone destroyed [D(O3)] per day for spring and fall as a function of latitude: (a) 0–10 km column; (b) 0–1 km column.

5. Comparison of Western and Central/Eastern North Pacific

[32] Earlier studies of western North Pacific ozone photochemistry were reported by Davis et al. [1996] and Crawford et al. [1997b]. Davis et al's. analysis was based on airborne data collected during NASA's PEM-West A (September–October 1991) field study. This study revealed that near equilibrium conditions had been established for column ozone during the fall season when the latitude range was restricted to 18°–42°N. This conclusion was based on P(O3) values which showed that modest levels of destruction at low altitudes (≤4 km) were compensated by modest levels of net ozone production at high altitudes (6–12 km). Quite significant in the Davis et al. analysis was the fact that the database was restricted to a region relatively close to the Asian mainland (e.g., 115°–150°E). Crawford et al. [1997b] also analyzed NASA airborne data for the same general region but during spring, e.g., generated during NASA's PEM West-B field program. These investigators reported finding net ozone production occurring at all altitudes during spring, resulting in a large positive value for column integrated P(O3).

[33] To identify an appropriate latitudinal zone for comparison with the CENP region, the fall and spring climatological flow patterns and streamlines for the North Pacific at 925, 850, 500, and 300 mb have been used (e.g., see DiNunno et al. [2002]). Due to the location of the subtropical high centered near 25°N and 155°E, the streamlines during fall tend to have a northward drift. Thus, for fall, the western Pacific zone identified by Davis et al. (e.g., 18°–42°N) appears to be reasonably well encompassed by the 25°–45°N CENP data. For spring, the flow pattern shifts quite significantly and the 20°–30°N zone identified by Crawford et al. aligns itself with the CENP latitude range of 20°–35°N. This largely reflects the fact that the subtropical high shifts further to the south during spring, resulting in more direct flow across the Pacific at these latitudes. For purposes of future discussion, the label “midlatitude CENP” will be referring to the central/eastern Pacific latitude range of 20°–35°N for spring, and 25°–45°N for fall.

5.1. NOx Distributions

[34] As noted throughout the text, a critical species required for efficient ozone formation is NOx. An examination of the concentration levels of this species for the midlatitude CENP and WNP regions is shown in Figure 7. Quite apparent from both the spring and fall profiles is that WNP NOx levels are significantly greater than those for the midlatitude CENP region. This comparison is valid for all altitudes (i.e., 0–12 km). The difference is most striking in the spring where NOx concentrations in the western Pacific are 2–6 times higher than those for the central/east. The highest NOx levels are seen at low altitudes (0–4 km), and most likely reflect the transport of anthropogenic surface emissions into the Pacific. At still higher altitudes (8–12 km), the situation is somewhat different in that the dominant NOx sources are lightning and convected surface emissions with smaller contributions coming from aircraft emissions and stratospheric intrusions [Crawford et al., 1997b].

Figure 7.

Altitudinal profiles of median NOx concentrations (NO observed plus NO2 calculated) for WNP(20–30°N) and CENP (25–35°N): (a) comparison with PEM-West B (20°–30°N) to CENP spring profile (20°–35°N); (b) comparison with PEM-West A (18°–42°N) to CENP fall profile (25°–45°N).

[35] It can be seen from Figures 7 and 8 that the variations in NOx as a function of altitude, and geographical location are far greater than those observed for O3 and CO. The difference in O3 between the WNP region and CENP is quite modest with no general trend toward either region. Similarly, differences in CO are small, but in this case there is a trend toward somewhat higher CO in WNP. The highest values are seen occurring at low altitudes during springtime, reflecting the strong outflow of continental emissions during this season.

Figure 8.

Altitudinal profiles of median O3 and CO concentrations for WNP(20–30°N) and CENP (25–35°N): (a) comparison with PEM-West B (20°–30°N) to CENP spring profile (20°–35°N) for O3; (b) comparison with PEM-West A (18°–42°N) to CENP fall profile (25°–45°N) for O3; (c) comparison with PEM-West B (20°–30°N) to CENP spring profile (20°–35°N) for CO; (d) comparison with PEM-West A (18°–42°N) to CENP fall profile (25°–45°N) for CO.

[36] During fall, NOx levels are generally lower than for spring. This is most likely due to weaker outflow from Asia with the prevailing wind field creating a slower, less direct, path from the continent to the Pacific. This is particularly true at low altitudes where WNP NOx levels are seen as being only slightly higher than those observed for midlatitudes in the CENP region. The most important factor here is the low altitude wind field which frequently delivers clean tropical marine air into the midlatitude zone of the western North Pacific [Crawford et al., 1997a; Merrill et al., 1997].

[37] Yet another aspect of understanding the NOx patterns shown in Figures 7a and 7b is the NOx lifetime. These are shown in Table 2 in the case of the midlatitude CENP region. Most striking here are the much shorter NOx lifetimes shown for low altitude. This is primarily due to two factors: 1) most NOx at low altitudes is in the form of NO2 which is the form most readily removed (e.g., OH/NO2); and 2) the highest levels of OH are typically found at lower altitudes. The difference in lifetime for spring versus fall at 0–1 km can best be understood in terms of spring ozone levels. Higher ozone shifts more NOx into the form of NO2, and also leads to higher average levels of OH. Both changes promote a shortened lifetime for NOx.

Table 2. Model Estimated Photochemical Lifetimes for NOx at CENP Midlatitudes for Spring (20°–35°N) and Fall (25°–45°N)
Altitude, kmLifetime of NOx (days)

[38] Since the CENP region is many days removed from continental emissions (∼3–10 days), and thus horizontal transport is significantly longer than the predicted NOx lifetime at low altitudes, transport of longer-lived forms of reactive nitrogen can be the controlling factor defining NOx concentration levels. For example, Moxim et al. [1996] concluded that during the fall season, 70–90% of low altitude NOx in the northeastern Pacific may result from the decomposition of PAN descending from the upper troposphere. These investigators suggest that this percentage could be as high as 90%+ during springtime.

[39] As noted above, shifts in NOx/HOx chemistry from low altitudes to high altitudes result in major increases in the estimated lifetime for NOx. Thus, in combination with higher average wind speeds at the higher altitudes, one can estimate that NOx should traverse the large distance from the WNP to the CENP region within one chemical lifetime. Although further dilution would be expected during this transit due to atmospheric mixing, the levels of NOx observed at each location should be within factors of 3 or so. As seen from Figures 7a and 7b, this appears to be true for fall but not spring. With the higher wind speeds for spring, it would seem that this season should show the better agreement. That it does not would suggest that current models may be lacking in their description of NOx sinks; however, it may also be an indication of the weakness of the databases being used. For example there is only one year's worth of data in the western Pacific and for the CENP region there is limited data above 8 km north of 25°N.

5.2. Ozone Tendency

[40] A comparison of WNP ozone photochemistry with that for midlatitudes in the CENP region is shown in Figure 9. Here it can be seen that for spring, F(O3) values are strikingly different for the two regions. WNP (20°–30°N) values are more than double those for the midlatitude CENP (20°–35°N). As noted above, this primarily reflects elevated NOx (>50 pptv) levels in the WNP region. Low altitude NOx was reported by Crawford et al. [1997b] as being elevated at distances of up to 2000 km off the Asian coast. In comparison, for the midlatitude CENP region (4000–9000 km downwind from Asia), only sporadic NOx events (e.g., plumes) were seen and median levels of this species were typically below 30 pptv NOx. Integrated over the entire column, springtime F(O3) is 2.5 times higher in the WNP region (Table 3). The contrast between the two regions during fall season shows a similar pattern, but the difference is smaller (e.g., 1.9, Table 3). As shown in Figure 9b at low altitudes (0–4 km), formation rates during fall are a factor of 1.5 higher in the western Pacific; but this difference increases to 2.6 at higher altitudes (6–12 km). As shown earlier in Figure 7, WNP NOx values in fall varied from 65 to 75 pptv over the altitude range of 8–12 km; whereas, for CENP the range was 20–40 pptv.

Figure 9.

Altitudinal profiles of F(O3), D(O3), and P(O3) calculated for WNP and midlatitude CENP. Comparison of PEM-West B (20°–30°N) and A (18°–42°N) to CENP spring (20°–35°N) and fall (25°–45°N) profiles: (a) spring, F(O3); (b) fall, F(O3); (c) spring, D(O3); (d) fall, D(O3); (e) spring, P(O3); (f) fall, P(O3).

Table 3. Altitudinal Profiles for F(O3), D(O3), and P(O3) Calculated for WNP (20–30°N) and CENP (25–45°N) for Two Seasons
Column (0–12 km) Total Ratesa
SeasonFormationDestructionNet Tendency
  • a

    All values are in molecules/cm2 × 1011.

  • b

    20°–30°N in the West, 20°–35°N in the Central/East.

  • c

    18°–42°N in the West, 25°–45°N in the Central/East.


[41] Unlike F(O3), D(O3) rates for the two regions are seen as being rather similar (Figures 8c and 8d). During spring, the total column D(O3) value for WNP is only 4% higher than for the midlatitude CENP region. This difference increases during the fall season, with the WNP value being a factor of 1.4 times higher than for CENP. The absence of a large difference in spring would appear to be primarily the result of there being similar levels of O3 and water vapor in both regions. For fall, ozone levels are also similar, but water levels in the WNP region are significantly higher (∼44%) [Davis et al., 1996; DiNunno et al., 2002]. These higher levels of water are most likely due to the periodic influence of typhoons in the western Pacific. These systems are capable of transporting large quantities of water vapor from the surface to the upper troposphere [Newell et al., 1996].

[42] When evaluating the O3 tendency, it is found that the above cited variations in water (and the corresponding shifts in D(O3)) are not sufficiently large as to overcome the more dramatic differences seen in NOx levels at each location. For example, as shown in Figure 9e, springtime P(O3) values in the WNP region are positive at all altitudes [Crawford et al., 1997b]. For the midlatitude CENP region, they are positive only at altitudes above 4 km, and even here they are quite small in magnitude. For CENP, low altitude net ozone destruction leads to a net total column value for P(O3) of −12.4 × 1010 molecules/cm2/s. This value can be contrasted to 21.4 × 1010 molecules/cm2/s for the WNP region.

[43] During fall, at which time water level differences are largest, Figure 9f shows P(O3) values for both CENP and WNP regions as being negative when at altitudes of <6 km. Above 6 km, however, WNP shows net formation at a magnitude that compensates for the low altitude destruction, yielding a total column P(O3) value of 0.6 × 1010 molecules/cm2/s. For CENP this does not occur because of lower NOx levels, and the total column P(O3) value is −5.7 × 1010 molecules/cm2/s.

[44] One approach that has been previously used [Davis et al., 1996; Crawford et al., 1997b] to gain further insight about the trends in tropospheric P(O3) has involved examining the quantity known as “critical NOx”. Critical NOx represents the NOx level required for F(O3) and D(O3) to be in balance. As such, it can prove useful in predicting the potential change that might occur in ozone with further changes in ambient NOx levels. Figure 10 illustrates this approach for both the spring and fall seasons for the CENP region. In both cases it can be seen that critical NOx exceeds observed NOx between 0 and 4 km. Parity is achieved between 4 and 8 km, and between 8 and 12 km observed NOx is shown exceeding critical NOx. The largest difference in these profiles is seen in the large values of critical NOx required for springtime at low altitudes. This is a direct result of there being elevated levels of ozone during spring [DiNunno et al., 2002]. Elevated ozone has a two-fold effect: first, it leads to a larger value for D(O3), thus requiring larger levels of NO to compensate; and second, it results in a shift in the partitioning of NOx toward NO2, reducing the amount of NO available for reactions R1, R2, and R3 which lead to ozone formation. The current analysis suggests that for spring BL conditions, NOx would need to be increased by a factor of 4 to reach ozone equilibrium. This increase drops to a factor of 2 for fall conditions. Not surprisingly, a similar analysis of the WNP region indicates that during spring, at all altitudes, observed NOx exceeds the critical NOx level. For fall, this was found to be true only for altitudes above 6 km.

Figure 10.

Altitudinal comparison of observed NOx to “Critical NOx” for midlatitude CENP. (a) values from spring composite, 20°–35°N; (b) values from fall composite, 25°–45°N.

6. Summary and Conclusions

[45] An examination of O3 photochemistry in the central/eastern North Pacific using a multiyear airborne data set has revealed rather clear seasonal trends. For tropical and transitional zone latitudes, springtime F(O3) was, on average, quite low, typically ranging from 2.0 to 4.0 × 105 molecules/cm3/s. For the same season, the highest values for D(O3) occurred between 0 and 4 km, accounting for 71–81% of the total destruction. Springtime values peaked in the BL transition zone (e.g., 1.5 × 106 molecules/cm3/s) with midlatitude values being a factor of two lower. Because of the large D(O3) values during spring at altitudes below 4 km, P(O3) was also found to have a large negative value for this altitude range. Net ozone destruction therefore prevailed, peaking at 3–5 ppbv/day. When evaluated in terms of an integrated column, ozone tendency values were found to be negative at all latitudes during spring with the peak value again occurring at “transition” latitudes.

[46] Fall F(O3) values were also shown to be low, with typical values between 0 and 4 km again in the 2.0–4.0 × 105 molecules/cm3/s range. D(O3) in the fall, however, was not as high as in the spring but still resulted in P(O3) values that were negative in the BL ranging from −0.2 to −1.0 × 106 molecules/cm3/s. On average, these values were a factor of 1.7 lower than spring.

[47] The potential importance of stratospheric O3 as a tropospheric source was estimated to be significant between 20° and 45°N, but the average value was still far smaller than that required based on the calculated value of D(O3). It was also smaller than the estimated value for F(O3) for both spring and fall. By combining ozone sources, stratospheric plus F(O3), it was found that for both seasons there was still net column destruction for the CENP region. Independent satellite derived column ozone trends [Fishman and Brackett, 1997] reveal that summer and winter ozone concentrations are typically lower than for the season preceding them, the midlatitude spring to summer increase being one of the clearest exceptions. For the latter location, the seasonal trend would clearly require an outside source of new ozone. And even for the other CENP latitudes, the very modest decrease seen in the seasonal trend for column ozone based on satellite analysis, as compared to the relatively large negative P(O3) values determined for spring and fall, would suggest that these seasonal ozone column trends can not be satisfied without an influx of ozone from outside the CENP region. The source of this ozone would most likely be upwind of the CENP region, e.g., Pacific Rim countries or over the contiguous western Pacific Ocean.

[48] This analysis also led to estimates of the ozone lifetime. Not surprisingly, the shortest of these was calculated for the tropics. Both spring and fall were found to lose between 6% and 9% of their ambient column ozone per day (e.g., D(O3)). This destruction was reduced to 3–5% for midlatitudes. As noted above, by far the largest fraction of the photochemical loss was found in the BL, with tropical destruction approaching 24% per day.

[49] A comparison of ozone photochemistry for the midlatitude CENP region versus that in the western North Pacific revealed some very substantial differences in NOx levels but not for other critical species such as O3, CO, and water vapor. With the exception of a very limited number of small plumes, NOx showed little evidence of crossing the Pacific without a significant attenuation in its concentration. Thus, given its key role in O3 formation, its low level in CENP, relative to WNP, resulted in a major contrast in the values of both F(O3) and P(O3). This was particularly true during the spring high outflow time period observed during PEM-West B. The large net production of O3 estimated for WNP is in sharp contrast to the net destruction of O3 found during spring for the CENP region. Thus, the capacity of the Pacific basin to both chemically remove, as well as physically dilute NOx emissions from the Pacific Rim is strongly reflected in the data recorded in the western and central North Pacific. The resulting negative ozone tendency estimated for the CENP region, therefore, could serve as a buffer for any future increases in emissions from Eurasia. However, far more detailed assessments of Pacific Rim and of the still larger Eurasian emissions are needed. And chemical-transport models will be required to more accurately evaluate the buffering capacity of the Pacific basin. In this context it should again be noted that at midlatitudes the western Pacific data are focused at around 150°E longitude, while for the central/eastern Pacific the center is around 130°W longitude. Critical to any new assessment will be new aircraft data for the North Pacific region between the WNP and CENP regions. In this regard, new satellite data for critical photochemical species would go far to help link these regions together.


[50] The authors would like to acknowledge the partial support of this research from NASA grant NCC-1-306. We also would like to thank the NOAA-CIRES Climate Diagnostics Center, Boulder, Colorado for the use of the NCEP reanalysis data. The work performed by the various investigators who gathered the data over the previous two decades was essential to this study.