Implications of large scale shifts in tropospheric NOx levels in the remote tropical Pacific

. A major observation recorded during NASA's western Pacific Exploratory Mission (PEM-West B) was the large shift in tropical NO levels as a function of geographical location. High-altitude NO levels exceeding 100 pptv were observed during portions of tropical flights 5-8, while values almost never exceeded 20 pptv during tropical flights 9 and 10. The geographical regions encompassing these two flight groupings are here labeled high and low NO x regimes. A comparison of these two regimes, based on back trajectories and chemical tracers, suggests that air parcels in both were strongly influenced by deep convection. The low NO x regime appears to have been predominantly impacted by marine convection, whereas the high NO x regime shows evidence of having been more influenced by deep convection over a continental land mass. DMSP satellite observations point strongly toward lightning as the major source of NO x in the latter regime. Photochemical ozone formation in the high NO x regime exceeded that for low NO x by factors of 2 to 6, whereas 0 3 destruction in the low NO x regime exceeded that for high NO x by factors of up to 3. Taking the tropopause height to be 17 km, estimates of the net photochemical effect on the 0 3 column revealed that the high NO x regime led to a small net production. By contrast, the low NO x regime was shown to destroy 0 3 at the rate of 3.4 % per day. One proposed mechanism for off-setting this projected large deficit would involve the transport of 0 3 rich midlatitude air into the tropics. Alternatively, it is suggested that 0 3 within the tropics may be overall self-sustaining with was also found to be consistent


Introduction
To understand the photochemistry of tropospheric ozone, the distribution of NO and NO x (NO + NO2) must be understood. It is the cycling of NO x as facilitated by peroxy radicals that governs the photochemical formation of ozone [Chameides and •School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta. 2Solar Terrestrial Environmental Laboratory, Nagoya University, Toyokawa, Aichi, Japan.
On the basis of meteorological considerations the PEM-B sampling region has been separated into two large regional areas as illustrated in Figure 1. These regions are defined by the large-scale outflow from the Asian continent. Latitudes north of 20øN were heavily impacted by continental outflow [Merrill, et al., this issue]. By contrast, to the south of this latitude and east of 120øE, continental influences were found to be minimal, leading to our tentative conclusion that this region was representative of a remote tropical-marine environment. It is the latter region that exhibited the large shifts in NO and is the focus of this paper. As shown in Figure 2, the remote tropical component of the PEM-B flight track contained all or portions of flights 5-10.
During flights 5-8 the free tropospheric mixing ratios of NO were found to be highly elevated (e.g., often in excess of 100 pptv), whereas during flights 9 and 10 levels almost never exceeded 20 pptv. The former sequence of flights will be defined hereafter as the "high" NO x regime, while the latter flight sequence (i.e., 9 and 10) will be referred to as "low" NO x. Note also that although geographically in the tropics, the near-coast portions of flight 10 and all of flight 11 (i.e., longitudes west of 120øE) have not been included in this analysis because both showed evidence of having been significantly influenced by low and medium

NO and Other Photochemical Measurements
While atmospheric NO measurements over the last decade have shown a trend of increasing reliability [Hoell et al., 1985[Hoell et al., , 1987[Hoell et al., , 1996Gregory et al., 1990], observations from independent instruments still serve to increase one's confidence in unexpected findings. In the case of PEM-West B, measurements were reported by two groups: Georgia Institute of Technology and Nagoya University. The former group used the two-photon laser-induced fluorescence (TP-LIF) technique, while the latter group employed the more common NO detection method of 0 3 chemiluminescence. The results from these two approaches compared quite well for all PEM-West B flights. For example, based on 30-s sampling intervals, a standard regression analysis of almost 5000 paired NO measurements gave a regression line having a slope of 1.25 ñ 0.01, an intercept of 1.2 ñ 0.2 pptv, and an R 2 value of 0.95 with Nagoya (the dependent variable) having the higher NO values. Although this comparison only involved data with mixing ratios in the range of 0-200 pptv, these data constituted nearly 99% of the total data set. For further details concerning the experimental hardware and measurement methodology, the reader is referred to Hoell  While their designations are in many ways similar to ours, differences exist since their data separation is based predominantly on isentropic trajectories; also, they include those portions of flights 10 and 11 that we have omitted. In order to facilitate a statistical comparison of the two regimes, all flight data falling within a solar zenith angle window of 0-60 ø were grouped into altitude bins of 0-1, 1-2, 2-4, 4-6, 6-8, and 8-10 km. Comparisons between regimes therefore are based on median conditions for each altitude block for each regime. Table 1 presents the distribution of data available for this analysis. In this case, the number of data points (i.e., 30-s averages) as well as the number of flights contributing data are listed for each altitude block in each regime. Since the high NO x regime had more flights that contributed data, this regime is clearly more robust than the low NO x regime. For example, only for the altitude range of 8-10 km are both regimes nearly equal in terms of total available data. This can be largely attributed to the fact that low "NO x" flight 10 sampled almost exclusively at 9.5 km between 120 ø  Median values were derived from data binnod into 1 km increments from 0-2 km and 2 km increments from 2-10 km. and 145øE. Strictly speaking, therefore, the flight map shown in Figure 2 should be viewed as primarily providing the geographical range spanned by both regimes at high altitudes (i.e., 8-10 km). For lower altitudes the in situ data for the two regimes are more restricted as the bulk of these data come from a longitudinal band spanning 143-153øE. Figure 3 shows a comparison of median NO mixing ratios as a function of altitude for the "high" and "low" NO x regimes. These result show that NO levels are considerably larger for the high NOx regime at all free tropospheric altitudes. The difference between the two regimes is seen to range from a factor of 2.5 for altitudes of 2-4 km to nearly 1 order of magnitude (e.g., 6 versus 60 pptv) at altitudes of 8-10 km. Quite noteworthy also is the observation that for all altitude blocks above 2 km the median values for the two regimes differ by more than one quartile. Interestingly, NO levels for the low NOx regime remain essentially constant with altitude; therefore, the large differences observed between the two regimes at the higher altitudes are seen to be entirely due to increases in NO in the high NOx regime.
As shown in Figure 4, the trends in NO x deviate only slightly from those shown for NO in Figure 3. The steady increase of NO x with altitude in the high NO x regime is similar to the trend for NO but is somewhat less pronounced since at low altitudes there is a significant amount of NO x present as NO 2 . At the higher altitudes, NO x is predominantly in the form of NO due to a significant decrease in the reaction rate between NO and 0 3 and a modest increase in the rate of NO 2 photolysis. In sharp contrast to the NO x profiles shown for the high NO x re- Calculated median values were derived from data binned imo 1 km increments from 0-2 km and 2 km increments from 2-10 km.
gime, levels slowly decrease with altitude for low NO x. Thus at 8-10 km the difference in NO x also reaches 1 order of magnitude (e.g., 7 versus 70 pptv).

Back Trajectories and Chemical Tracers
To further explore the key factors responsible for the tropical shift in NO x, both back trajectories and chemical tracers were examined. This analysis included all flights listed earlier in the text as being identified with either the high or the low NO x regimes. Representative trajectories are presented in Figure 5, and chemical tracer data are presented in Figures 6a-6f. 2.3.1. Back trajectories. Back trajectories can provide some indication of both where a sampled air mass may have originated and how that parcel may have been influenced as it tracked to the point of sampling. Shown in Figure 5 are several 10 day isentropic back trajectories for selected air parcels. (Note that the symbols at the end of each trajectory indicate the location of the aircraft at the time of sampling.) In this case, because of the higher density of data in the 8-10 km range and the fact that the difference in the NO mixing ratio was largest at this altitude, the trajectories shown are based on aircraft locations within 8-10 kin. From Figure 5 it is quite apparent that the source regions for the two regimes are distinctly different. For example, high NO levels are primarily associated with trajectories that are confined to the northern hemisphere. This is true not only for those trajectories shown in Figure 5 but for all isentropic trajectories calculated for altitudes above 2 kin. Furthermore, approximately 80% of the trajectories for the high NO x regime are indicated to have recently passed over the Asian continent (i.e., within 2-3 days of sampling). By contrast, the trajectories associated with the low NO x regime are seen as being exclusively from the southern hemisphere. Again, this is not only true for the examples shown in Figure 5 but for all trajectories above 2 km.
Approximately 50% of the low NO x trajectories track back to remote South Pacific points of origin. The remaining 50%, although from the southern hemisphere, pass over New Guinea.
For purposes of comparison with the isentropic trajectories of Figure 5, a set of 6 day kinematic trajectories were provided by S. McKeen (private communication, 1996). For both regimes, these trajectories tended to travel more slowly covering roughly half the distance of the isentropic trajectories per day. For the high NO x regime the two methods compared quite well. Both indicated air parcels being transported eastward from southeastern Asia at high altitude. In the case of low NO x the kinematic trajectories differed somewhat in that they did not cross the equator or remain at high altitude. Instead, air parcels were shown rising steadily from altitudes as low as 2 km while at the same time being transported westward from the tropical, central Pacific. Thus the kinematic trajectories offer a somewhat different perspective on the origin of the air parcels sampled in the low NO x regime; however, they do not alter the earlier conclusion, based on the isentropic trajectories, that this regime was predominantly influenced by marine sources.
While back trajectories are much less reliable for BL conditions, especially after more than 2 days, it is still noteworthy that near-surface isentropic trajectories do not follow the pattern of those in the free troposphere. For both regimes, most of the trajectories have the central North Pacific as a source region.
2.3.2. Chemical tracers. As noted above, chemical tracer data can also be an indicator of the origin and/or history of a sampled air parcel. With the notable exceptions of NO x and 03 the source of many chemical tracers at high altitudes is the direct result of the coupling between vertical transport and the release of a species at or near the surface. In these cases, one of the more important vertical mixing processes can be deep convection. During PEM-B the tropics was convectively quite active as evidenced by numerous visual as well as satellite observations. Evidence supporting the notion that deep convection was indeed important in the redistribution of several tracers can be found in many of their altitude profiles. As illustrative of this, Figure 6a shows median values versus altitude for "fine" aerosol particles as expressed as CN (condensation nuclei) number density. These data clearly indicate that for both the high and the low NO x regimes, median CN values significantly increase with altitude. As suggested by other investigators [Clarke et al., 1993;Thornton et al., 1996] a likely explanation for this observation is that for tropical conditions, significant amounts of volatile surface sulfur (e.g., volcanic, anthropogenic, and/or biogenic) are pumped to high altitudes via deep convection. At high altitudes this reduced sulfur can be oxidized by OH radicals, and the resulting sulfuric acid, at very cold temperatures and in the absence of significant populations of larger particles, leads to the formation of new particles via nucleation processes. Thus, these data support the hypothesis that both the high and the low NO x regimes experienced the effects of deep convection. That deep convection occurring over both environments could lead to elevated high-altitude sulfur levels and hence CN levels is  Figure 6b, the altitude distribution of CH3I for the low NO x regime shows a significant enhancement over that for the high NO x case. At 8-10 km this results in median CH3I mixing ratios that are more than twice that for the high NO x regime. In fact, for all altitudes above 2 km, the lower quartile for the low NO x regime exceeds the upper quartile value for the high NO x. These results thus are consistent with the hypothesis that the low NO x regime was more heavily influenced by marine convection. Although dimethylsulfide (DMS) might also be considered as a useful tracer of marine convection, the measured median concentration levels for this species were found to be too near the limit of detection (LOD) to justify a detailed quantitative comparison between the two regimes. Even so, quite relevant to the above argument were those DMS observations recorded for the highest-altitude data bin at 8-10 km. In this case, the upper quartile value for the low NO x regime was nearly a factor of 3 higher than the LOD for DMS. By comparison, the high NO x upper quartile DMS value was still at the LOD level. Thus the DMS results are qualitatively consistent with those for CH3I.
Just as CH3I and DMS may be viewed as excellent tracers of marine BL air, CO and NMHCs can be considered equally good tracers of surface continental air. As shown in Figure 6c the median values for CO versus altitude indicate that above 2 km the difference between the two NO x regimes averages less than 7 ppbv and never exceeds 11 ppbv. Not surprisingly, the respective median values also fall within one quartile of each other. Thus these results suggest that neither regime was preferentially influenced by continental emissions. The median levels for the high and low NO x regimes (i.e., ranging from 83 to 97 ppbv) would also argue that the source strength of CO from combustion that had been advected into both regions was relatively low. However, in contrast to CO the results for several NMHC species did show a systematic difference. A likely reason for this is that the latter species have very low background levels in the upper atmosphere relative to CO. This characteristic can be attributed to their having typically shorter atmospheric lifetimes versus CO and their having sources which are limited to the surface only. As illustrated in Figure 6d, a significant difference is seen in the mixing ratios of C3H 8 for the two regimes. Differences between 4-8 km are larger than one quartile; but for the 8-10 km range, where C3H 8 is roughly 1.5 times higher for high NO x, median values actually fall within one quartile. This reflects the greater variability in C3H 8 levels at high altitude as one might expect due to its short lifetime and surface source. Differences reflecting a similar trend were also observed for the NMHC species C2H6, C2H2, and C6H 6. that for 2•øpb in that the high NO x regime showed significantly higher levels, on the average being nearly a factor of 2 larger than those recorded for low NO x. 0 3 is typically not considered a simple tracer of either continental or marine surface air because of its high photochemical reactivity and its highly variable mixing ratio as a function of altitude. In cases, however, where independent data suggest that a given air parcel was influenced by deep convection over a marine region versus having been more influenced by deep convection over a continental area, some Although the DMSP lightning flash data together with trajectory and tracer data suggest that a major source of NO x in the high NO x regime could have been lightning, the possible contribution from surface emissions still requires some investigation. In the text that follows we explore this question from several points of view. The first of these involves the observed trends in the trace gas CO. Recall that earlier in the text, when presenting the comparison of CO values, we found only a very modest difference between the two regimes. If surface combustion were a significant contributor to the elevated levels of NO x, one might expect that the difference in CO levels between regimes should have been much larger. On the other hand, the fact that there were significant differences recorded in the levels of C3H 8 at high altitudes does suggest some contribution from surface emissions. In

Recently, a more extensive examination of C3H8/NO x data for PEM-A as well as PEM-B and the GTE program TRACE
A has shown that the 3' 1 value for this ratio does not apply to regions where NMHC sources are dominated by biomass burning (e.g., the South Atlantic during the burning season). In this case, the partitioning of hydrocarbons is shifted away from that observed from high temperature combustion sources as evidenced by the ratios of C3H8/C2H 6, C3H8/C2H 2, and C3H8/C6H 6. On the basis of industrial emissions near Atlanta, Georgia [Jeffries, 1995], these ratios are expected to be around 0.75, 2.0, and 4.5, respectively. In fresh biomass burning plumes sampled during TRACE A, these ratios were 0.2, 0.  While an understanding of the large-scale tropical processes driving the NO x shift is of considerable importance, equally important is the closely coupled question concerning which of these regimes might be the dominant one in the tropical-western North Pacific. This issue is particularly important in the context of global modeling efforts that require a "representative" tropical NO x database. In this context, even though the high NO x regime was found to be much more robust in terms of the number of flights and amount of data (see Table  1), one cannot assume that this equates to this regime being temporally and/or spatially the most dominant. To further illustrate this point, Plate 1 shows the longitudinal distribution of 0 3 at low latitudes (20øN to 10øS) as derived from UV DIAL data and supplemented by more limited in situ 0 3 data, particularly that in the marine B L. This represents the most robust spatial coverage obtainable for any constituent measured during PEM-West B. While there is some longitudinal overlap between the high and the low NO x regimes as described earlier, it is still rather apparent from these data that there is a significant reduction in 0 3 levels at nearly all altitudes west of 140øE, a picture that is consistent with Figure 6f. This represents the best evidence available, which suggests that the two NO x regimes did span a longitudinal band at altitudes below 8 km which was comparable with that at 8-10 km where the bulk of the in situ data was recorded. In fact, it is quite possible that these two regimes persisted in their respective regional dimensions for the duration of the tropical sampling time period; but it is also possible that temporal changes did occur. For example, sampling of the high NO x regime took place over the time period of February 8-17, 1994, and that for the low NO x was February 19-21, 1994. Thus it is also possible that each NO x regime spatially spanned the entire tropical-western North Pacific sampling region during its respective time period. Quite obviously, it is also possible that the real answer involved some combination of spatial and temporal separation. Further exploration of this important issue should be a target for future field programs in the tropical Pacific.

Ozone Photochemistry
Radical formation in the troposphere is initiated by the photolysis of 0 3 in the wavelength range 290-320 nm. For the PEM-B high/low NO x data set, the effect was typically less than 1%.

Model description. As suggested by reactions (R4)-(R6), the photochemical consequences of the shift in NO levels is most directly seen in the 03 budget. This has been explored
in the text that follows using the output from a time dependent (TD) photochemical box model. The model employed has been previously described by Davis et al. [1993Davis et al. [ , 1996a]. The current model differs from that discussed by Davis et al. only in the manner in which NO x sinks and sources were handled. In earlier applications the only sink for NO x was the daytime reaction of NO 2 with OH. Accordingly, the NO  Table 2. For altitudes below 2 km the difference between the two regimes is seen to be quite small. Above 4 km, however, ozone formation is indicated to be 2 to 6 times larger for the high NO x regime with the difference increasing with altitude.

The closeness of values below 2 km reflects the similarity in chemical conditions for the two regimes and is thus not
unexpected. For altitudes above 4 km, the large difference in NO would seem to suggest that the difference in F(O3) values should have been greater. The reason it is not is due to the much higher water vapor levels associated with the low NO x regime (see Figure 8). Elevated levels of H20 result in higher predicted peroxy radical concentrations which tend to offset the effects of higher NO mixing ratios associated with the high NO x regime. Accordingly, the difference in F(O3) between the two regimes is seen to increase with altitude much slower than might be expected based on the difference in NO levels.

NO (symbols) versus altitude for the (a) high NO x regime and (b) low NO x regime. Error bars encompass the inner quartiles.
Although not explicitly shown in equation (2), the possible destruction of 03 from iodine reactions was also examined. Davis et al. [1996b] previously explored this possibility in the case of the tropical PEM-A data. These investigators estimated the iodine chemistry sink to be no larger than 1.8x10 •ø molecules/cm2/s or approximately 6% of the total 03 loss.

Scaling this estimate to the PEM-B low NO x regime, with adjustments for the differences in CH3I levels, results in an estimated increase in D(O3) of •-4 %. This estimate would be shifted still lower for the high NO x regime since the latter regime had lower CH3I concentrations. The net effect of all photochemical processes as related to in situ 0 3 levels, P(O3), has been estimated from equation (3).
These results are again provided in Table 2. Quite evident from these is the effect of combining a larger formation rate from the high NO x regime with a larger destruction rate from the low NO x regime. In summary, the difference between the two regimes, as expressed in terms of P ( Davis et al. [1996a], whose value was -12x10 xø molecules/cm2/s. Thus when looked at in terms of the total tropical western Pacific, the net effect on the 0 3 budget during each campaign was not significantly different. It is still noteworthy, though, that both column formation and destruction rates in PEM-B exceed those in PEM-A by 30% and 21%, respectively. This is primarily due to the higher NO x levels at high altitude and higher ozone levels at low altitudes in PEM-B. 3.2.4. Comparison with other high altitude tropical data. Of the other available major tropical data sets (e.g., NASA GTE CITE-l, ABLE-2A/B, and TRACE A), only TRACE A was found to encompass both a comprehensive suite of chemical measurements and a sampling that involved high altitudes. It must be noted, however, that the emphasis of the TRACE program differed from that of PEM-B in that it focused on a region that typically experienced large seasonal influxes of biomass burning emissions (i.e., the South Atlantic basin) [Thompson et al., 1996;Jacob et al., 1996]. Additionally, the TRACE A database encompassed sampling of both continental and marine regions. Despite these differences, both this study and TRACE A have concluded that deep convection in conjunction with transport of NO x via the Walker circulation were major factors in controlling the tropical ozone budget. A similar hypothesis was put forward earlier by Davis et al. [1996a] in their analysis of the PEM-A data.
For the upper free troposphere (i.e., 6-10 km), NO levels observed in TRACE A were somewhat higher than those found for the PEM-B high NO x regime. This is not unexpected in that additional contributions to the NO x burden were present due to biomass burning. Ozone levels at these altitudes were also higher by nearly a factor of 1.6, indicating higher net photochemical 0 3 production. At lower altitudes, differences in both NO and 0 3 were even more pronounced (i.e., factors of 2 or more). Collectively, these differences resulted in columnintegrated formation and destruction rates that exceeded PEM-B tropics (high/low NO x) by factors of 5 and 3, respectively. Not surprisingly, the TRACE A modeling results produced an overall photochemical 0 3 picture which involved a balanced budget. Recall that a similar situation was found for the high NO x PEM-B regime. One condition not observed in Plate 2 is high 03 in the presence of low NO x. Such air parcels would represent conditions where the NO x responsible for the buildup of 0 3 has been removed via reaction of NO 2 with OH, leaving elevated 03 behind. That such air parcels are not observed is most likely an indication of the strong role of vertical mixing in the tropical marine troposphere. In addition to bringing low 03 up from the marine BL, this mixing process also delivers 03 rich air from the upper troposphere to lower altitudes where higher H20 levels lead to rapid destruction. This appears to happen with such high frequency that NO x typically does not have sufficient time to reach its full potential for 03 production.
This interpretation of the relationship between NOx-O 3 in the tropics is undoubtedly an oversimplification of the complex interplay between dynamics and photochemistry. It does make the point, however, that any critical assessment of the 03 budget for the "greater" tropical Pacific will require an examination of the tropical troposphere over both a wide range of altitudes and an expansive range of longitudes. Based on PEM-B results, it is unlikely that such an assessment can be carried out in any simple way using currently available satellite observations of 03. The strong relationship between NO x and 03 also underlines the continuing need for further improvement in our understanding of NO x sources in the tropics. More than any other factor it is pivotal to our reaching a comprehensive understanding of the tropical 03 budget.

Summary and Conclusions
PEM-B flights in tropical regions of the western Pacific (i.e., 10øS -20øN) revealed large shifts in observed NO levels as a function of geographical location. This shift was most pronounced over the altitude range 8-10 lcm where the median NO level for flights 5-8 was found to exceed that from flights 9 and 10 by 1 order of magnitude (i.e., 60 versus 6 pptv). The geographical regions encompassing these two flight groupings define what have been labeled "high" and "low" NO x regimes. Isentropic back trajectories indicate a clear difference in the source regions for these two regimes. Air parcels from the low NO x regime appear to track back to the remote South Pacific, whereas the high NO x regime involves parcels that originated in the northern hemisphere and that tend to track back to the Asian continent.
Further differences were found when comparing chemical tracers in each regime. The low NO x regime saw enhancements in CH3I and DMS in conjunction with depressed but uniformly mixed 03 levels. These observations are seen as being most consistent with marine convection. Thus it is not surprising that this region also revealed little evidence of lightning-generated NO x. In sharp contrast the high NO x regime showed significant enhancements in C3H 8, 2•øpb, and organic acids such as CH3C(O)OH. These observations are more consistent with air parcels having experienced deep convection over a continental landmass. The observed high altitude levels of C3H 8, when ratioed against estimated values of NO x, tend to argue that the major source of NO x in this regime was lightning, not transport of surface emissions. This conclusion was also found to be consistent with DMSP satellite observations recorded during the PEM-B sampling period. Even so, we cannot at this time rule out the possibility that much of the NO x was being generated by the efficient recycling of unidentified NOy species through yet unknown/nondocumented mechanisms.
Time-dependent photochemical box-model simulations revealed large differences between the two regimes as related to their respective photochemical 03 budgets. These differences mainly could be attributed to the large differences between the two regimes in the levels of NO and H20. For free tropospheric altitudes, in situ 03 formation in the high NO x regime exceeded that for low NO x by factors of 2-6. By contrast, in situ 03 destruction for the low NO x regime exceeded that estimated for the high NO x regime by as much as a factor of 3. The latter difference was primarily driven by high H20 levels associated with the low NO x regime.
The The large rate of 03 destruction in the low NO x regime was found to be inconsistent with seasonally averaged tropospheric TOMS data, which suggest that tropical 03 levels should either have been at steady state or have been increasing with the onset of the spring season. Two basic mechanisms were proposed to reconcile this situation. The first involved the transport of 0 3 rich midlatitude air into the tropics from midlatitudes. Analysis of the PEM-B midlatitude data base has revealed that this region was both 0 3 and NO x rich. The second mechanism allowed for the existence of still other high NO x regimes in the "greater" tropical Pacific as well as other low NO x regions. It was proposed that this ensemble of regimes might lead to nearself-sustaining levels of column 0 3 in the tropics. Thus in the latter scenario 0 3 would be predominantly controlled by photochemistry. Obviously, combinations of mechanisms 1 and 2 would also be possible. Expanding on the earlier hypothesis by Davis et al. [1996a], the existence of Pacific high NO x regimes as well as possibly "super" NO x regimes (i.e., regimes showing significant positive P(O3) values) are viewed here as regions uniquely modified by transport processes. They are regions that most likely experienced the effects of long-range transport (e.g., Walker circulation) of air parcels that originated over a landmass that had experienced significant deep convection. The photochemical effects from these parcels can be most clearly seen at high altitudes, but they also become evident in those regions where sinking motion brings these NO x-energized parcels (i.e., due to lightning and/or surface NO x emissions) down to lower altitudes. This process in combination with the countereffects resulting from deep marine convection along the ITCZ now appear to be the major components impacting on the photochemical 0 3 budget of the tropical Pacific. However, the database from which this expanded hypothesis is based is still far too limited, and new observations at different geographical locations and times of year are needed. Evidence supporting the notion that photochemistry in the tropics routinely has a major impact on the levels of 0 3 was demonstrated in our finding a strong correlation between 0 3 and NO x levels at free tropospheric altitudes for the Pacific field programs PEM-A and PEM-B and also for the South Atlantic study TRACE A. In addition, these observations again point to the long-standing critical need of a comprehensive understanding of both primary and secondary sources of NO x for purposes of understanding the tropical 0 3 budget.