4.2. SPECIAL PROJECTS
4.2.1. THE MAUNA LOA OZONE PROFILE INTER-COMPARISON
From August 14 through September 1, 1995, the CMDL Ozone Group measured daily ozone profiles during the Network for the Detection of Stratospheric Change (NDSC) Stratospheric Ozone Profile Intercomparison (MLO3) held at Mauna Loa Observatory and Hilo, Hawaii. Participating groups in Hawaii included the JPL lidar, NASA Goddard lidar, NASA Langley Microwave, and NOAA umkehr and ozonesondes. All the remote sounding instruments were operating at MLO (3397 m). Seventeen electrochemical concentration cell (ECC) ozonesondes were launched from Hilo (11 m) to coincide with the evening MLO observations. Six of the balloons launched were carrying a new "triple" ozonesonde platform. The "triple" flights showed very good precision that can be seen in the August 30, 1995, profiles (Figure 4.4a). These profiles were fairly typical of all the triple flights with an average relative standard deviation of 2-4% in the lower troposphere (0-5 km) and in the stratosphere from 15 km to burst altitude (35 km). The poorest precision was in the mid- to upper troposphere, where very low concentrations of ozone approach the background levels of the ozonesonde. In this region the relative standard deviation was as high as 20%.
Figure 4.4b shows the average ozonesonde profile from the August 30 flight,
the lidar and microwave profiles, and a SAGE overpass ozone profile measurement.
All of the profiles agreed very well above 15 km, except for the JPL lidar
between 15 and 18 km. At the ozone peak, the ozonesonde profile is about
8-10% greater than the ground based instruments. The ozonesonde profiles measured
during MLO3 were decreased by a maximum of 6% at burst altitude due to an observed
increase in response to ozone as the 1% potassium iodide solution evaporates
during the flight and becomes more concentrated. Additional testing has shown
that the sonde may be reading even higher than the 6% correction. This would
reduce the ozonesonde profile further resulting in less than 5% differences
between the ozonesondes and ground based instruments.
Fig. 4.4. (a) Three vertical profiles of ozone concentration (1012 cm-3) measured
by the August 30, 1995, triple ozonesonde flight. (b) The average of the triple
ozonesonde profiles from (a) and the profiles of ozone concentration from the
two MLO lidars and microwave ozone profilers and the coincident Stratospheric
Aerosol and Gas Experiment II (SAGE II) satellite profile.
4.2.2. OZONE VERTICAL PROFILES OVER THE NORTH ATLANTIC
As part of AEROCE and NARE, several intensive ozone vertical profiling campaigns were carried out. During these campaigns, ozonesondes are launched on a near daily schedule for periods of approximately 1 month. The focus of these intensive measurements is to describe the ozone variations throughout the troposphere over the North Atlantic and to relate the processes identified as being responsible for surface ozone variations with changes throughout the troposphere [Oltmans and Levy, 1992; Moody et al., 1995]. The results, primarily from the 1993 and 1994 campaigns, have recently been prepared for publication [Oltmans et al., 1996].
During spring 1995, an intensive series of measurements was made at Bermuda. Over the month the soundings were carried out and several profiles were also obtained at the University of Rhode Island and at the University of Maryland. These soundings on the east coast of the United States were made in order to look at the ozone content of air masses as they moved from the United States to Bermuda. The profiles from May 6, 1995, at Rhode Island and Maryland (Figures 4.5a and b) and for May 7 at Bermuda (Figure 4.5c) illustrate how such a system can produce high ozone amounts in the midtroposphere over the North Atlantic. At the more northerly location (Rhode Island), the upper troposphere from 8-11 km has ozone concentrations over 125 ppb (Figure 4.5a) and large ozone amounts (85 ppb) extend down to 6.5 km. At Maryland (Figure 4.5b), on the other hand, the layer of elevated ozone over 100 ppb is confined to a relatively thin layer at about 7 km. On May 6 at Bermuda (profile not shown) ozone throughout the troposphere is 50-70 ppb. Early on May 7 at Bermuda (Figure 4.5c) there is a peak of 120 ppb at about 6.5 km.
The Bermuda trajectories for May 7 (Figure 4.5d) show that for the 0000
UT arrival time, air had passed over the northeastern United States about 12
hours earlier at an altitude of 8 km. The larger ozone amounts seen at
Maryland and Rhode Island (1400 Z on May 6) are clearly part of the same system
producing the large ozone peak over Bermuda seen at 0100 Z at about 6.5 km.
Air travels rapidly from near the Arctic Circle and descends over 2.5 km
in just 2 days before reaching Bermuda. This meteorological pattern with flow
behind the upper air trough brings air from a region where significant transfer
of air from the stratosphere into the troposphere is likely to take place [Merrill
et al., 1996]. The presence of these layers of larger ozone concentration
aloft at Bermuda and other sites over the North Atlantic are invariably associated
with transport from higher altitudes and latitudes [Oltmans et al., 1996].
These layers are also very dry, which is another indication that the air was
mixed down from the stratosphere. It is likely that this process is an important
source of ozone in the troposphere [Oltmans et al., 1996].
Fig. 4.5. Vertical profiles of ozone mixing ratio (ppbv), temperature (C),
and frost-point temperature (C) at (a) University of Rhode Island for May 6,
1995, at 1355 UT, (b) University of Maryland for May 6, 1995, at 2020 UT, and
(c) Bermuda Naval Air Station for May 7, 1995, at 0100 UT. (d) The isentropic
back trajectories for May 7 at 0000 UT (solid line) and 1200 UT (dashed line)
arriving at 6.5 km at Bermuda.
During July 1995, intensive profiling campaigns were carried out at the
Canary Islands, Azores, Bermuda, and Newfoundland. For each of these sites,
ozone amounts near the surface are much smaller than are seen at the surface
during the spring. In the middle and high troposphere, however, intrusions of
large ozone amounts are common during the early summer and the peak values are
larger than those seen in the spring [Oltmans et al., 1996]. The presence
of these layers of large ozone concentration is strongly tied to flow characteristics
and water vapor amounts that demonstrate the stratospheric origin of these layers
in a manner very similar to what is seen in the spring. Transfer of ozone-rich
air from the stratosphere into the troposphere during the summer is significant.
4.2.3. WATER VAPOR AND OZONE PROFILES AT MCMURDO, ANTARCTICA
The very cold temperatures of the Antarctic winter stratosphere lead to the formation of polar stratospheric clouds (PSCs). The formation of these clouds, particularly when temperatures reach the frost point of water (type II PSCs), leads to significant dehydration of the Antarctic stratosphere [Vömel et al., 1995b; Peterson and Rosson, 1994]. In order to more completely describe and better understand the process of dehydration within the Antarctic stratospheric vortex, a program of 19 frost-point soundings was carried out at McMurdo between February and October 1, 1994. Each frost-point sounding was accompanied by an ozone vertical profile as well. In addition, ozonesondes were flown between the water vapor profile measurement times on about a one-per-week schedule. Measurements of PSC particles and nitric acid in the stratosphere were made by groups from the University of Wyoming and Denver University, respectively.
About the middle of June the coldest portions of the Antarctic stratospheric
vortex reach the water vapor saturation temperature leading to rapid formation
of ice crystals which fall with sufficient speed to rapidly dehydrate the stratosphere
[Vömel et al., 1995b]. By late July continuing into October (Figure
4.6), much of the stratosphere in the vortex between 12 and 20 km remains
highly dehydrated. During this period of sustained dehydration, there does not
appear to be significant continued removal of water vapor, and also little moisture
is added from outside the vortex. By the time the sun reappears over Antarctica,
the air in the entire stratospheric vortex is highly processed and ozone depletion
can take place over a broad geographical scale and through a significant depth.
Fig. 4.6. Vertical profiles of water vapor mixing ratio at McMurdo, Antarctica,
during the sustained dehydration period of the winter of 1994. The June 13,
1994, profile shows conditions before dehydration begins.
Early in the dehydration process (June 19), an ozone profile was observed
with significant depletion between 12 and 20 km (Figure 4.7). This
depleted ozone layer corresponds very closely to the region where significant
dehydration was first detected at McMurdo. Since it is dark over the Antarctic
continent at this time, sufficient sunlight to allow ozone depletion can only
be attained if the air parcel moves to a higher latitude after it is dehydrated
and the air is significantly processed. The trajectory analysis [Vömel
et al., 1995a] shows that indeed the dehydrated and ozone-depleted parcel
passes through the coldest temperatures over the continent and subsequently
moves to a sunlit region near 50S latitude. Furthermore, this analysis shows
that significant ozone depletion can take place on a relatively short time scale.
An analysis of the time when the observed air masses could have formed type
II PSCs (ice particles) for the first time, limits the time scale for the observed
ozone destruction to about 4 days [Vömel et al., 1995a].
Fig. 4.7. Vertical profiles of ozone partial pressure (mPa) and ozone mixing
ratio (ppmv) at McMurdo on June 13 and June 19, 1994. The June 19 case shows
significant ozone depletion and dehydration between 11 and 24 km compared
to the unperturbed June 13 profiles.
4.2.4. FLOW PATTERNS FOR SMO DESCRIBED WITH CLUSTERED TRAJECTORIES
Ten years of isentropic trajectories (1986-1995) were calculated for SMO (14.25°S, 170.56°W) and were summarized for various time frames using cluster analysis.
Ten-day back isentropic trajectories were produced twice per day at 0000 and 1200 GMT arrival times. Trajectories were calculated to arrive at approximately 500 m asl, an elevation close to that of the observatory (77 m) and representative of the boundary layer. Input to the transport model consists of gridded meteorological data from the European Centre for Medium Range Weather Forecasts.
An earlier study by Halter et al.  identified
three distinct source regions contributing their signals to atmospheric measurements
made at SMO, in particular those of carbon dioxide. Figures 4.8-4.10 show
trajectories depicting transport from these three regions. Figure 4.8 shows
trajectories arriving at SMO in the trade wind regime. At most, 1 km of
descent is seen in this type of trajectory. Ten days back from SMO air parcels
are in the remote southeastern Pacific, never reaching the South American continent.
The second transport type is illustrated in Figure 4.9, which shows flow
from southern midlatitudes to SMO. This trajectory type may descend 3 km
or more as it curves anticyclonically en route to SMO. Note that as air parcels
descend in elevation, wind speeds decrease significantly. For this transport
type, approach to SMO is often from the southeast. The third least frequent
transport type brings air from the northern hemisphere to the observatory. Figure 4.10
shows trajectories with origins well beyond the equator. Halter et al.
 found that carbon dioxide transported to SMO in this manner was in phase
with that measured at Kumakahi, Hawaii, although the amplitude of the signal
was reduced. To quantify the frequency of SMO transport types, the complete
set of trajectories was summarized using cluster analysis.
Fig. 4.8. Ten-day back isentropic trajectories arriving at SMO at 500 m
on August 7, 1989, showing an example of trade wind type transport. The
numbers along the trajectories give days back from SMO.
Fig. 4.9. Ten-day back isentropic trajectories arriving at SMO at 500 m
on June 30, 1990, showing an example of transport from midlatitudes. The
numbers along the trajectories give days back from SMO.
Fig. 4.10. Ten-day back isentropic trajectories arriving at SMO at 500 m
on February 1, 1987, showing an example of transport from the northern
hemisphere. The numbers along the trajectories give days back from SMO.
Cluster analysis is a multivariate statistical technique used here to group trajectories by shape and length. Each group or so-called cluster is represented by an average trajectory called the cluster mean. Figure 4.11 shows six cluster means derived for the entire 10-year period. Each cluster mean is marked with plus signs indicating the 1-day upwind intervals. At 10 days upwind, the cluster means are labeled with two numbers; the top number gives the percentage of trajectories that occurs in the cluster, and the bottom number identifies the cluster. The cluster identifiers (1-6) are used to distinguish one transport type from the other; the numbers themselves have no inherent significance and are not related from one clustering to the next.
Figure 4.11 gives a quantitative estimate of the frequency of each transport
type. For example, the rapid transport from southern midlatitudes is represented
by cluster 2 and cluster 3, with 2% and 7% frequency of occurrence,
respectively. These two clusters differ in the average length of trajectory
contained in the cluster, but both bring air parcels to SMO from elevations
of 3 km or more. As a result of subsidence on approach to the observatory,
wind speeds decrease significantly compared with their initial magnitudes several
days earlier. A third cluster with transport from the south, cluster 4,
exhibits much lower wind speeds, but shares the characteristics of 10-day average
origin south of 30°S and anticyclonic curvature
on approach to SMO from the southeast. This cluster, comprising 18% of trajectories,
has a mixed character because it contains some trajectories with tropical origins.
Fig. 4.11. Atmospheric flow patterns for SMO depicted by cluster-mean back
trajectories arriving at 500 m asl for 1986-1995. Plus signs indicate
1-day upwind intervals. At 10 days upwind the cluster means are labeled with
two numbers; the top number gives the percentage of trajectories that occurs
in the cluster and the bottom number identifies the cluster.
Clusters 1, 5, and 6 in Figure 4.11 contain trajectories that correspond to the easterly trade wind transport type. Cluster 5 trajectories have a northerly component on approach to SMO and occur 15% of the time. Cluster 6 trajectories approach from the due east 32% of the time. Cluster 1 trajectories have lower wind speeds, a southerly component, and occur 26% of the time.
Transport from the northern hemisphere occurs in cluster 5. This cluster
also contains trajectories that do not cross the meteorological equator. To
estimate the frequency of transport of true northern hemisphere air, the number
of trajectories that had 10-day origins north of 5°N
were counted. Trajectories of this transport type have frequency of 2-3% per
year. They occur most often in February and March at about 10% frequency, but
can also occur in December, January, and April though usually at lower frequency.
The occurrence of transport from the northern hemisphere is quite variable from
year to year with frequency in the two most likely months (February and March)
ranging from 0 to 25%. Trajectories were also clustered by month in order
to learn more about the seasonality of flow patterns at SMO. Figures 4.12a
(January) and 4.12b (July) represent yearly extremes in flow patterns. The marked
differences between these plots show a strong seasonality in SMO transport.
Summer months are characterized by minimal flow from southern midlatitudes,
a northerly component in some of the trade wind trajectories, and a cluster
with trajectories approaching SMO from the northwest (for example, cluster 2,
Figure 4.12a). Winter months are characterized by both easterly trades
and strong flow from the south. In fact this "winter" pattern is present
from May through November, with July having the highest frequency of strong
southerly flow. Trade winds are evident during every month.
Fig. 4.12. Atmospheric flow patterns at SMO for 1986-1995 depicted by cluster-mean
back trajectories arriving at 500 m asl for (a) January and (b) July. Symbols
are as in Figure 4.11.
In terms of 10-day origins, air transported to SMO is predominantly from remote
tropical-marine regions. The frequency of such origins is about 50% in winter
to 80% in summer. The remaining trajectories are from more southerly latitudes
usually from elevations of 3 km or higher. Northern hemispheric air is
transported to SMO at a rate of 2-3% per year, exclusively during the summer
and fall months.