Climatology and trends of tropospheric ozone over the eastern Pacific Ocean: The influences of biomass burning and tropospheric dynamics

Jae H. Kim
Earth System Science Division, NASA/Marshall Space Flight Center, Huntsville, Alabama

M. J. Newchurch
Atmospheric Science Department, University of Alabama in Huntsville, Huntsville, Alabama

Abstract. To study the latitudinal variation of tropospheric ozone over the eastern Pacific Ocean, we derive ozone in the lower troposphere by subtracting TOMS column ozone over the Andes from the ozone column over the eastern Pacific Ocean. The analysis of tropospheric ozone climatology in conjunction with meteorological and biomass-burning data suggests three latitude bands with different characteristics for tropospheric ozone. (1) Air in the latitude range 3° N-7° N experiences persistent upward motion in the middle troposphere over the eastern equatorial Pacific Ocean. This condition results in relatively low ozone amounts in this region. (2) Tropospheric ozone between latitudes 2° N - 22° S shows strong seasonal variation that is well correlated with the biomass burning season over southern tropical South America. Persistent subsidence in the middle troposphere and prevailing easterlies over this region provide favorable conditions for enhancing ozone amounts in the lower troposphere. (3) Tropospheric ozone over 23° S-36° S does not appear to be influenced by biomass burning. Regression analysis of deseasonalized tropospheric ozone shows positive trends from 12° S to 23° S and from 2° N to 5° N with essentially zero trends elsewhere. Trends north of the equator are generally smaller and less consistently positive than trends from 12° S-23° S. With tropospheric ozone seasonally in phase in the presence of enabling horizontal and vertical winds, the average increasing trend of +0.9± 0.3(2s )[%/yr] or +0.14± 0.04(2s )[DU/yr] between 12° S and 23° S over the 14 years analyzed here is consistent with an increasing biomass-burning source.

Fishman and Larsen [1987] and Fishman et al. [1990, 1991], using indirect satellite measurements, identified a plume of elevated tropospheric ozone extending from the west coast of Africa into the Atlantic during the biomass-burning season. Since then, efforts have focused on understanding biomass-burning-induced chemistry and transport from both Africa and South America to the Atlantic [Kirchhoff et al. 1996; Thompson et al., 1996]. However, few studies have focused on transport of tropospheric ozone from South America to the eastern Pacific Ocean. Recently, Hsu et al. [1996] have shown enhanced aerosol concentrations over the eastern Pacific Ocean (5° N-15° S) using Total Ozone Mapping Spectrometer (TOMS) data. They concluded this aerosol was associated with biomass burning over South America. Jiang and Yung [1996] introduced a method for deriving tropospheric ozone by subtracting TOMS total ozone over the Andes from the ozone over the adjacent eastern Pacific Ocean. They concluded that the seasonal variation of derived tropospheric ozone in the layer from 0 to 6 km over the subtropical eastern Pacific Ocean (20° S-30° S) is consistent with the seasonal variation of biomass burning and that tropospheric ozone has increased by about 1.48± 0.4% per year for the period of 1979-1993.

Using this method, we study how the seasonal and annual variations of tropospheric ozone over the eastern Pacific Ocean are linked to biomass burning and vertical wind velocity in the upper troposphere. In addition, we discuss the influence of biomass burning and tropospheric dynamics on the tropospheric ozone trend.

To derive tropospheric ozone, we used monthly averaged Version-7 TOMS total ozone data on a 1° latitude by 1.25° longitude grid from January 1979 to April 1993 [jwocky.gsfc.nasa.gov]. The averaged height of the Andes corresponding to the TOMS grid derives from High Density TOMS data that contains height information over the TOMS field-of-view of about 50´ 50 km [McPeters et al., 1993]. To investigate the influence of tropical dynamics on the derived tropospheric ozone, we used meteorological data on a 2.5° latitude by 2.5° longitude grid from January 1982 to December 1994 from the National Centers for Environmental Prediction/National Center for Atmospheric Research re-analysis (NCEP/NCAR) data [Kalnay et al., 1996]. The vertical pressure velocity, used here in -Pascal/second, represents assimilated data.

The TOMS algorithm determines total column ozone from a reflecting surface to the top of the atmosphere. Therefore, by subtracting column ozone over the high elevation of the Andes from the ozone over the adjacent eastern Pacific Ocean, one can determine tropospheric column ozone from sea level to the averaged height of the Andes over the TOMS grid [Jiang and Yung, 1996]. At each particular latitude where the topographic difference between the ocean and the Andes (2.2 to 4.3 km) is great enough to retrieve statistically significant ozone amounts, we select only one pixel over the Andes to minimize mixed-elevation pixels and average two pixels over the adjacent Pacific Ocean, located 3.75° and 5° westward from the pixel over the Andes.

Figure 1 shows the annual variation of monthly mean tropospheric ozone in the latitude bands 5° -7° N (N57) and 0° -1° S (S01). The regions between the vertical bars in the figure correspond to the El Niño periods, which produced strong convective activity over the eastern Pacific Ocean. The two latitudes are distinctively different from each other both in the amount of tropospheric ozone (more in S01) and in their seasonal variations (greater in S01), although the height of the Andes is about the same over the two locations. Because the upper-level wind is persistently easterly in the equatorial region [Kalney et al., 1996], tropospheric ozone over South America can be transported to the eastern Pacific Ocean as long as the ozone is convectively lifted high enough to pass over the Andes, whose peak elevation (not averaged over a TOMS pixel) is about 4-5 km in these latitudes. Ozone transported from South America over the Andes to the eastern Pacific Ocean will subsequently enhance tropospheric ozone amounts only if the vertical wind is in the downward direction (subsidence).

Vertical wind velocities at 500 mb over N57 and S01, also shown in Figure 1, reveal persistent rising motion over N57, but show persistent subsidence over S01 except for the El Niño periods 82-83, 86-87, and 91-92 [NOAA, 1993; Kalney et al., 1996]. Analysis of the vertical wind velocity at other pressure levels in the middle troposphere indicates that the rising motion over N57 and subsidence over S01 are persistent features. The lowest ozone over S01 is recorded during the strongest El Niño period, 82-83, and is due to the strong upward motion during this period. Periodogram analysis shows a strong signal in the ozone with the period of approximately 58 months, which is consistent with the El Niño period. This result suggests that vertical wind velocity in the middle troposphere plays an important role in controlling the ozone amount in the lower troposphere over the eastern Pacific Ocean.

Figure 2 shows the seasonal variation of tropospheric ozone as a function of latitude for seven locations spanning 7° N to 36° S. Most of the difference in the mean ozone levels is due simply to the different heights of the Andes (~2.3 km at 1° S-7° N and at 35° S-36° S; ~4.3 km at 24° S-25° S; intermediate values elsewhere). Relatively low ozone is observed over N57 with a marginal maximum in June and July within the annual variation of only ~3 DU. This behavior in the ozone is not correlated with biomass burning over South America, which has an active burning season in July through November [Jiang and Yung, 1996; Hao and Liu, 1994; Fishman et al., 1991, Kirchhoff et al., 1991]. Figure 3, the seasonal variation of vertical pressure velocity at 500 mb, suggests that persistent rising motion throughout the entire year over N57 prevents biomass-burning-induced ozone from sinking to enhance lower-tropospheric ozone amounts.

Seasonal ozone variation is consistent from 2° N to 22° S, showing a maximum in August, September, and October and a minimum in January, February, and March. This seasonal behavior is very similar to the seasonal variation at Natal (6° S, 35° W) presented by Logan and Kirchhoff [1986] and Kirchhoff et al. [1991]. It is also similar to the seasonal behavior derived by Fishman et al. [1990] from the SAGE/TOMS tropospheric residual method. The surface measurements at the remote Southern Hemisphere sites Samoa (14° S), Cape Point (34° S), and Cape Grim (41° S) [Oltmans and Levy, 1994] all suggest peak surface ozone concentrations in June, July, and August - 2-4 months earlier than lower tropospheric ozone peaks seen in this analysis. Photochemical loss in the marine boundary layer would contribute to the summer ozone minimum while photochemical production from biomass-burning precursors would contribute to enhanced ozone during the burning season.

Figure 3 shows that the dominant vertical-velocity feature is the persistent subsidence throughout the year from 0° to 20° S as opposed to the strong seasonal variation and rising motion at 7.5° N throughout the burning season. Because intensive burning occurs in July through November and the dominant zonal wind in the middle and upper troposphere is easterly [Kalnay et al., 1996], the seasonal maximum in ozone provides strong evidence for the influence of biomass burning. As shown by Krishnamurti et al. [1993] during a particular case study of October circulation, 200-mb winds can form a cyclonic circulation advecting air from equatorial latitudes to as far south as 25° .

The seasonal variation in the ozone between 23° S and 36° S shows a maximum in October, November, and December (spring/summer) and a minimum in the May-June (autumn/winter) period. Because the zonal wind in the upper and middle troposphere south of 20° S is westerly and the seasonal variation and activity of biomass burning over this area is low compared with that in the subtropics, biomass-burning-induced ozone over South America is less likely to affect latitudes south of ~20° S. Our results in the Pacific Ocean show a different seasonal variation from that suggested by the shipborne ozonesondes in the Atlantic Ocean made by Weller et al. [1996], who observed a minimum in October-November-December and a maximum in May-June south of 20° S in the lowest 4 km of the troposphere. Figure 3 shows rising motion from September to January at 30° S. This rising motion provides unfavorable conditions for enhancing the ozone amounts in the lower troposphere, suggesting that the seasonal maximum in the ozone at these latitudes may be coupled with stratosphere-troposphere exchange.

Figure 4 shows tropospheric ozone linear trends as a function of latitude, resulting from a regression of the deseasonalized, monthly averaged ozone amounts versus time with the El Niño periods omitted. Trends computed including the El Niño periods are slightly higher between 10° S and 5° N, but are not significantly different from the Figure-4 results at the 95% confidence level. The results show increasing trends from 23° S to 12° S and from 2° N to 5° N with essentially zero trends elsewhere. Because tropospheric ozone is seasonally in phase with the biomass burning in the presence of enabling horizontal and vertical winds, the increasing trend between 23° S and 12° S is consistent with an increasing biomass-burning source. The large percentage increases north of the equator result from relatively small Dobson Unit increases in small mean ozone levels. They are, nonetheless, significantly positive at the 95% level at half of the sampled latitudes north of the equator. These mixed trends reflect the complex meteorology and steep latitudinal gradients in both vertical velocity and biomass burning.

The latitudes 22° to 26° S are most comparable to the Jiang and Yung [1996] study area. In these latitudes, we find a latitudinal average trend of approximately +0.52 ± 0.31(2s )[%/yr] compared to their +1.48 ± 0.40[%/yr], using a somewhat different sampling and trend analysis techniques. To gain further insights into mechanisms responsible for trends in tropospheric ozone, Logan [1994] suggested that we should know (1) the factors that control NO_x concentrations in the middle and upper troposphere and (2) the dynamical factors that influence the interannual variability of ozone.

Tropospheric ozone in the lower troposphere derived by subtracting TOMS V7 column ozone over high elevations in the Andes from the ozone over sea level provides an opportunity to study lower-tropospheric ozone climatology and trends. The latitudinal variation of tropospheric ozone over the eastern Pacific Ocean shows three domains that have different characteristics. (1) Persistent rising motion over the northeastern Pacific Ocean (5° -7° N) prevents any westward-transported ozone from sinking to enhance lower tropospheric ozone amounts. This rising motion results in the smallest amounts of tropospheric ozone in the lower troposphere relative to amounts southward. The small annual variation is out of phase with the biomass-burning variation. (2) Persistent sinking motion over latitudes 2° N-22° S coupled with strong seasonal variation that is well correlated with the biomass-burning season over South America suggests that a significant amount of tropospheric ozone is transported from South America over the Andes to the eastern Pacific Ocean with the prevailing easterly wind during the biomass-burning season to enhance ozone in the lower troposphere. (3) Vertical motions between neutral and rising over latitudes 23° -36° S coupled with no annual signal in biomass burning indicate tropospheric ozone in these latitudes experience little influence from biomass burning.

Regression analysis of deseasonalized lower-tropospheric ozone indicates increasing trends from 12° S to 23° S and from 2° N to 5° N with essentially zero trends elsewhere. Trends north of the equator are generally smaller and less consistently positive than trends from 12° S to 23° S. The average increasing trend of +0.9± 0.3(2s )[%/yr] or +0.14± 0.04(2s )[DU/yr] between 12° S and 23° S over the 14 years analyzed here is consistent with an increasing biomass-burning source.

Acknowledgments: We thank the Goddard Ozone Processing Team for use of TOMS version 7 data. We especially thank E. S. Yang for assistance in the trend analysis. The insightful reviews by J. Logan and an anonymous referee improved the paper significantly. JHK was supported by a National Research Council Post-doctoral Associateship. MJN was supported by the DOE Atmospheric Chemistry Program and the NASA Atmospheric Chemistry Modeling and Analysis Program.

_________

J. H. Kim (jaek@knuecc-sun.knue.ac.kr Earth System Science Division, NASA/Marshall Space Flight Center, Huntsville, Alabama, 35812, USA.

M. J. Newchurch (mike@atmos.uah.edu) Atmospheric Science Department, University of Alabama in Huntsville, Huntsville, Alabama, 35899, USA.

_________