Atmospheric sampling of Supertyphoon Mireille with NASA DC-8 aircraft on September 27, 1991, during PEM-West A

. The DC-8 mission of September 27, 1991, was designed to sample air flowing into Typhoon Mireille in the boundary layer, air in the upper tropospheric eye region, and air emerging from the typhoon and ahead of the system, also in the upper troposphere. The objective was to find how a typhoon redistributes trace constituents in the West Pacific region and whether any such redistribution is important on the global scale. The boundary layer air (300 m), in a region to the SE of the eye, contained low mixing ratios of the tracer species 03, CO, C2H6, C2H2, C3H8, C6H6 and CS2 but high values of dimethylsulfide (DMS). The eye region relative to the boundary layer, showed somewhat elevated levels of CO, substantially increased levels of 03, CS2 and all nonmethane hydrocarbons (NMHCs), and somewhat reduced levels of DMS. Ahead of the eye, CO and the NMHCs remained unchanged, 03 and CS2 showed a modest decrease, and DMS showed a substantial decrease. There was no evidence from lidar cross sections of ozone for the downward entrainment of stratospheric air into the eye region; these sections show that low ozone values were measured in the troposphere. The DMS data suggest substantial entrainment of boundary layer air into the system, particularly into the eye wall region. Estimates of the DMS sulphur flux between the boundary layer and the free tro-posphere, based on computations of velocity potential and divergent winds, gave values of about 69 gg S m -2 d-(cid:127)averaged over a 17.5 ø grid square encompassing the typhoon. A few hours a(cid:127)er sampling with the DC-8, Mireille passed over Oki Island, just to the north of Japan, producing surface values of ozone of 5.5 ppbv. These 03 levels are consistent with the low tropospheric values found by lidar and are more typical of equatorial regions. We suggest that the central eye region may act like a Taylor column which has moved poleward from low latitudes. The high-altitude photochemical environment within Typhoon Mireille was found to be quite active as evidenced by significant levels of measured gas phase H202 and CH300H and model-computed levels of OH.

In the typhoon system, there is a substantial lateral inflow of air in the lower troposphere to compensate for the rising convective motions and outflow aloft, and these quasi-horizontal flows are also important in redistributing trace constituents.An opportunity to carry out an experiment to check this redistribution was presented when Typhoon Mireille approached an area to the south of Japan which could be accessed from Tokyo.Air emering the typhoon in the surface boundary layer, in the eye itself, emerging from the eye and surrounding wall cloud region in the upper troposphere, was sampled.
Typhoon Mireille was first noted on the weather maps as a significant system on September 13, 1991, when positioned at 12øN, 172øE moving westward (see Figure 1) [from Rudolph and Guard, 1992] and qualified as a typhoon, with 1-min sustained winds exceeding 32 ms 'l, on September 16 while at 15øN, 157øE.By September 22 it had turned toward the northwest and winds had increased to 67 ms -1, qualifying Mireille as a "super typhoon."After September 23, Mireille began to slowly weaken, although its size, as measured by the diameter of its outermost closed isobar, continued to increase.On September 26, Mireille turned northward, then northeastward, and made land in western Kyushu at about 0600 UT on September 27.During the evening of September 27 it passed over the Sea of Japan and gradually lost force, passing over northern Honshu and southern Hokkaido early on September 28.The aircraft-reported winds are plotted in Figure 3a-3c for these three regions.Where available, radiosonde winds (see, for example, Figure 3c) and winds from commercial aircraft flights (AIREPs) are also included (see, for example, Figure 2).Most of the body of this paper deals with the atmospheric chemistry findings in these three regions and their possible interpretation in terms of atmospheric motions.In addition to the time series of constituents themselves, time series of ratios of two pairs of constituents which can be interpreted in terms of air mass age are presented in Figure 4g.As discussed elsewhere [Gregory et al.,   The spatial separation of the descending dry plumes is tens of kilometers, as may be seen from a map of specific humidity during the boundary layer sampling (not shown).The higher temperatures at 300 m are produced by subsidence, implying a downward heat transport by these motions.Because sinking motions are relatively dry and, as will be seen later, there is overall convergence into the typhoon region in the lower levels, rising motions in the same area will be relatively moist, and the net effect is an upward transport of moisture.•'here is a large vertical gradient of H20 mixing ratio in erie lowest 2 kin, as seen from vertical profiles made on descent to and ascent from the boundary layer (an example is shown in Figure 6, to be discussed below).This is maintained by the strong winds (>15 ms -l) and consequem high evaporation from the    4b).After the DC-8 left the boundary layer, DMS up to 40-50 ppbv in the central drier region, but them are values returned to their limit of detection until 0454 UT, no high values (>100 ppbv) as might be expected if just before entering the wall cloud convection region; at stratospheric air was present.Suggestions have been made that point they increased to 80 pptv.In the eye, there were that stratospheric air could be entrained into the eye, but three observations: 25, 42, and 48 pptv (see Figure 4b).On these ozone measurements suggest that Bergeron [1954] leaving the eye at 0537 UT, high values (78 pptv) were provided a more attractive explanation of the higher again associated with the wall cloud region; values are temperatures in the eye by arguing, as noted earlier, that considerably lower in the outflow region, as may be seen the subsidence responsible occurred within the troposphere.Nevertheless, there may be small-scale   '•'  ,' ,',, --•d'-'   The ECMWF cross section of vertical motion through the storm (Figure 18) is not inconsistent with these low values.In fact, the subsidence in the eye is rather limited, and the cylindrical volume shown could be regarded as a Taylor column, as outlined further below.These conclusions do not agree with the relatively large sinking motion at 5 km in the core suggested by Willoughby [19881.

Large-Scale Mass Flux
A 17 1/2 ø x 17 1/2 ø latitude-longitude box was drawn surrounding the typhoon (Figure 16), and flow into and out of the box was evaluated each day from the divergem wind component.An example appears in Figure 19 where it is evident that the majority of the i•fflow occurs below 500 hPa.We are interested here in the mass flux passing between the boundary layer and the free troposphere and have therefore calculated the vertical flux through 850 hPa.Values in Table 2 show that Mireille was sampled on the day of maximum mass flux, 29 x 109 kg s -1 These values are compared with Typhoon Orchid sampled on October 8, in Table 2; Mireille has about double the flux by Orchid but clearly a larger sample needs to be studied and compared with the flux by middle-latitude cyclones before the overall significance of these values from Mireille can be assessed.It would also be valuable to make an estimate of the actual area involved in the rising motion, perhaps by using weather radar.It is also worth noting from Figure 19 that as well as the boundary layer, the next two layers above also contribute substantially to the mass flux through the system.For examp•le, the 700-to 500-hPa layer contributes about 20 x 10 v kg s -• with at least one third of this material coming in from the adjacent Asian continent to the west, as may be seen from Figure 17.
Is the typhoon important in the general circulation in carrying trace constituents into the upper troposphere?This question is examined below for the case of DMS.
The PEM-West missions showed that DMS only reached the free troposphere from the boundary layer on three occasions: on September 16 over the North Pacific on a day of very strong convection as seen from the cloud videos; on September 27 in association with typhoon Mireille; and on October 8 in association with typhoon Orchid (Table 3).For typhoon Mireille the mass of sulphur carried as DMS to the free troposphere from the boundary layer can be estimated from the values for total mass flux derived above in conjunction with the mass mixing ratio of DMS.The molecular weight of DMS is 62, and the conversion to equivalent mass of sulphur from volume mixing ratio of DMS is therefore (ppv) x (62/28.9) x (32/62) = 1.11.For a boundary layer mixing ratio of 80 pptv and a mass flux from Table 1 of  x 10 -3 kg m -2 s -I evaluated over an area about 6 times as large.

Small-Scale Features
During the descent to the boundary layer at 0228 UT near 28øN, 136øE the aircraft measured the vertical profiles of ozone and water vapor shown in Figure 6.The layered structure between about 5 and 6 km shows that watervapor-rich air is associated with lower ozone than watervapor-poor air.These layers seem to be about 400 m thick; similar layers show in the lidar echoes near to the center of the eye (Plate 3b).They appear to provide evidence of air from the boundary layer folded into higher levels, probably by convective processes.It is noteworthy that for this to occur, there must be substantial instability; that is, 30e/3Z < 0. (0½ is equivalent potemial temperature and z is altitude.) Where 30½/3z > 0 and the system is stable, the air spreads out horizontally, forming the layers noted.A summary of the properties of these layers is given elsewhere in this issue [Newell et al., this issue], where it can be seen that the characteristics of Figure 6 are fairly typical.

Ozone From Surface Stations
The only surface station sampling trace constituems which was close to the path of Mireille was Oki Island (36.3øN, 133.2øE), which was in operation as part of the experimem "Perturbation by East Asian Continental Air Mass to Pacific Oceanic Troposphere" (PEACAMPOT).It is noteworthy that ozone mixing ratios as the eye passed over Oki at-4200 UT were down to 5.5 ppbv (Figure 20a).The station at Kenting, Taiwan (22øN, 120.9øE) reported a value of about 3 ppbv on September 27 (Figure 20b), when the surface pressure in the Kenting area was about 1007 hPa.At 0600 UT on September 26, Mireille was about 550 km to the northeast of Kerning, and there was a trough  20b) may have been associated with air which had been brought up from the lower latitudes by the typhoons.Unfortunately, the station power was turned off during the passage of Typhoon Nat on September 23.In the case of Oki the transfer is more easily understood, as the eye passed over the station (see Figure 1).The low (5 ppbv) ozone values here are in general agreement with those reported by lidar in the tropospheric region of the eye (see Plate 3).We speculate that the eye region is a cylindrical sample of tropical air "packaged" by the circulation which has a continuity rather like that of a classical Taylor column.Taylor [1923] performed some experiments to check an analysis by Proudman [1916] which showed that all slow steady motions of a rotating liquid must be two dimensional.Taylor showed that when a sphere or cylinder is pulled across a rotating tank of water, it is observations were not taken in the maximum velocity region gives a reasonable time of about 4 days, well within the C2H6 lifetime.It is clear that convection anywhere along the back trajectory could have introduced the material into the upper troposphere and that the vertical shear could have produced a vertical gradient of C2H6, such as was observed landing at Tokyo, where values at 7.5 km were about double those in the boundary layer.

Photochemical Environment
Although a detailed description of the photochemical environment of typhoon Mireille is not the main focus of this paper, for purposes of completeness the authors have summarized some of the general photochemical features of this system and have then comrasted these to air parcels sampled in the same geographical region under nontyphoon conditions.For example, in Figure 4d the observed mixing ratios for the photochemical products accompanied by a cylinder of water above it having the H202 and CH3OOH are reported as a time series, together same diameter.He did this by ejecting dye into the water with model-calculated profiles of the centrally important above the sphere or cylinder and noting that it diverged oxidizing species, the hydroxyl radical.If the midday time around an imaginary cylinder above the sphere or cylinder.period of 0330 UT to 0540 UT is examined, it can be If dye was ejected into the imaginary cylinder, it stayed shown that the median level of H20 2 (based on 3-min inside the cylinder and moved across the tank with it.It is averaged data) is 1032 pptv, whereas that for CH3OOH is this imaginary cylinder which is called a Taylor column.considerably lower, being 193 pptv.Model-estimated OH Yih [1977] reports that Taylor also experimemed with fish concentrations for this same time period ranged from 1.6 x in the tank, and they would try to avoid the cylinder as if 106 to 8.1 x 106 molecules cm -3, giving a median value of sensing the presence of a solid obstacle.Low mixing 3.6 x 106 molecules cm '3.Some of the highest estimated ratios of tropical ozone found in both the laser images OH values are seen for those times when the aircraft was (Plate 3) and in the Oki surface station data (Figure 20a) approaching or passing through the eye wall region; it was are thought to correspond to the dye in the experiment.at these times when highly elevated levels of NO were The equivalent potential temperature computed from the observed by the Georgia Tech instrument.By way of ECMWF data (Figure 21) shows neutral stabili .tynear the contrast to the typhoon environment, clear air observations typhoon center, but again the linfited horizontal resolution must be borne in mind.

Total Ozone Relationships
There have been several studies of the association between total ozone and tropical cyclones, including those in the western North Pacific.Some rather complex relationships were found which depended on whether the systems were intensifying or not intensifying [e.g., Stout and Rodgers, 1992].We were provided with Total Ozone Measurements Satellite (TOMS) data each day in the field, and after the mission we obtained digital data.The maps for September 26-27 showed no special relationship between Mireille and total ozone in accord with the lidar and cirrus cross sections discussed earlier.These comments apply to the region we could measure directly.The question concerning stratosphere to troposphere exchange outside the eye cannot be examined experimentally with our measurements.However, it was noted that high potential vorticity (PV) values were indicated on the ECMWF 315K analysis drawn up for this experiment.In sampling Typhoon Mireille, it is important to recognize that both continental and marine boundary layers were mixed together by the typhoon's circulation patterns.Thus in view of the fact that the continental boundary layer was not sampled directly, it is difficult to distinguish contributions from the continental boundary layer and the continental free troposphere.

Possible Remote
The low ozone in the eye region, shown by the lidar measurements and by Oki Island station, may have been transported directly from low latitudes in a process akin to a Taylor column.Low ozone values aloft would also fit with an ultimate low-latitude source for the air in the eye region.Ostlund (1968) also mentioned that some of his tritium results could be explained if the eye contained old air which had been trapped since the formation of the circulation.Finally, the NMHC data suggest the possibility that distant sources to the west may contribute to material found in the typhoon.
Figure 1.Track of Typhoon Mireille from September 13-28, 1991 [from Rudolph and Guard, 1992].Positions of ground stations Kenting and Oki Island are noted.
Time series of selected constituents are examined first to place the boundary layer inflow region in proper perspective.Figure 4 includes time series for the entire mission of CH4, CO, 03, H20, OCS, C82, CH3SCH3 (DMS), 802, C2H2, C2H6, C3H8, C6H6, NO, NOy, CH3C(O)OONO 2 (peroxyacetyl nitrate (PAN)), H202, CH3OOH, OH (model calculated), and aerosol.Figure 4a also contains the flight altitude, and Figure 4e includes ambient air temperature.Time resolution of the data used in Figure 4a is given with the Figure 4a caption.For other constituents the data are collected at variable intervals and are plotted at the midpoint of the collection interval.Each time series set has the boundary layer and eye regions marked on the time axis.Constituents having their lowest values in the boundary layer included CO, 03, SO2, C2H2, C2H6, C3H8, C6H6, NO, NOy, and PAN.PAN is actually below the detection limit; its concentration is very sensitive to temperature, decreasing at higher temperatures [Singh, 1987].Thus indications are that the air entering the typhoon from the south in the southeastern sector is clean marine air.The marine origin of the inflow was also indicated by the fact that the highest DMS values of the mission occurred in the boundary layer.The higher aerosol concentrations in the boundary layer (Figure 40 are thought to be associated with sea salt.

Figure 6 .
Figure 6.Vertical profiles of specific humidity and ozone from DC-8 measurements.This profile was made during the descent to the boundary layer between 0223 UT and 0242 UT on September 27, 1991.

Figure 7 .
Figure 7. Map of DC-8 air temperature in vicinity of eye at about 11.3 km on September 27, 1991.Points are 60 s apart.Units, degrees Celsius.

Figure
Figure 9. Map of ozone in vicinity of the eye at about 11.3 km September 27, 1991.Points are 60 s apart.Units, parts per billion by volme.

Figure 10 .
Figure 10.Map of carbon monoxide in eye region at about 11.3 km on September 27, 1991.The time interval between closely spaced points is about 60 s.There are a number of missing data points.Units, parts per billion by volume.

FigureFigure 13 .Figure 14 .
Figure 12.Map of DC-8 water vapor mixing ratio in the outflow region at about 11.2 km on September 27, 1991.Time interval between data points is 5 min.Units, parts per million by volume.

Figure 19 .
Figure 19.Example of mass flux components across the walls of the box in Figure 16 for September 27, 1991, based on divergent wind calculated from ECMWF data.Units, 10 9 kg s -1 .appear to be only a small fraction of that carded upward in the cloud regions.In an effort to estimate the vertical motion in central core, it is assumed that the heat balance within the core is between adiabatic compression and radiative cooling.The relationship is Figure 20a.Surface ozone variation at Oki Island (36.3øN, 133.2øE) in the period September 17 to October 23, 1991 (provided by H. Akimoto).Units, parts per billion by volume.
a DMS sulphur flux into the upper free troposphere comparable to that normally associated with the flux from the ocean into the boundary layer.The use of machine analysis may underestimate the mass flux circulating through the typhoon system, as the radiosonde wind fields may be smoothed excessively.There are relatively few estimates of the mass flux in the literature.McBride [ 1981] used about 800 rawinsondes to make a composite of Pacific typhoons and computed •,,.,,,.:•oencefrom the surface to 350 hPa as 28 x 10 -3 kg m -2 s -I evaluated over an 8 ø diameter circle, while our value from Figure l0 for the surface to 400 hPa is 21

Figure 21 .
Figure 21.Potential temperature (solid lines) and equivalent potemial temperature (dashed lines) for points near the eye center on September 27, 1995 at 0000 UT, based on ECMWF analysis.

Velocity potential and divergent wind component at 1000 hPa for 0000 UT, September 27, 1991. Wind vectors in meters per second. Velocity potential in 10 n m 2 s ']. Maximum vector (near eye) is 9 meters per second. Figure 18. Cross section of vertical motion through Mirefile (located at about 30øN, 127.5øE) from ECMWF analysis at 0000 UT, September 27, 1991 based on divergent wind calculated from ECMWF data. Units, Pascals
per second.Contour interval is 0.2.