OMI total-ozone anomaly and its impact on tropospheric ozone retrieval

1
OMI total-ozone anomaly and its impact on
tropospheric ozone retrieval
Jae Kim1, Somyoung Kim1, K. J. Ha1, and Mike
Newchurch2 1Department of Atmospheric Science,
Pusan National University, Korea
2Atmospheric Science Department, University of
Alabama in Huntsville, USA
Introduction There have been a series of
algorithm modifications to the TOMS total ozone
retrieval. The first major modification was the
ozone error due to inaccurate cloud
characterization in version-6 TOMS retrievals
Hudson et al., 1995 Thompson et al., 1993,
which were adressed in version7 (v7), concerned
mostly low-altitude marine stratocumulus clouds.
One significant discrepancy still remained in the
difference between the assumed cloud-top pressure
and the actual pressure. Because TOMS (OMI) can
retrieve ozone above a cloud, the climatological
tropospheric ozone amount below cloud is added to
it for determining total column ozone. Therefore,
the errors in the cloud top pressure and
tropospheric ozone climatology directly propagate
into the error in total ozone derivation. Because
for clear-sky conditions the contribution of
backscattered radiation from the atmosphere
dominates over the reflected radiation from
Earths surface, the effective scattering surface
for the backscattered UV radiation to the TOMS
instrument is at the middle to upper troposphere
(Hudson et al. 1995 Klenk et al. 1982). This,
the so-called the low retrieval efficiency of
lower-tropospheric ozone, causes an additional
error in the ozone retrieval. In order to reduce
this error, better a priori information is
particularly important for the lower troposphere.
Ozone Mapping Instrument (OMI), uses the v8
algorithm, which employed a cloud-pressure
climatology based on thermal infrared cloud-top
pressures and a new a-priori ozone profile
(McPeter et al., 2007). However, in OMI v8, there
are still cloud-pressure-related errors that
affect the ozone retrieval above clouds mostly
due to convective clouds. These include 1) error
in estimating multiple scattering between cloud
and overlaying atmosphere 2) errors in estimating
the effects of rotational Raman scattering (RRS)
and O2-O2 absorption and 3) errors in the
application of the aerosol correction (Joiner et
al., 2006). The new OMI v8.5 uses the optical
centroid cloud-top pressure in deep convective
clouds, which is several hundred hPa lower than
the thermal infrared cloud-top pressure. This
results in v8.5 OMI total ozone being less than
v8.0 OMI total ozone. Figure 1 shows the
comparison of version 8.0 and 8.5 OMI total ozone
over the topics. However, if the assumed
tropospheric ozone part in the a-priori is
different from the actual ozone profile, the
total ozone still contains errors. The objective
of this work is to estimate the cloud and
a-priori related errors.
Figure 5-a 15 January 2005 15
September 2005
1
Figure 5-a. Correlation between OMI V8.5 ozone
above cloud (OAC) and 360nm Reflectivity in 15
January (left column) and 15 September (right
column) 2005 over ITCZ regions with marks A, B, C
in Figure 4. The ozone and reflectivity show a
good negative correlation both for January and
September. Clouds with high reflectivity
generally for high altitude clouds.
January 2005 September 2005
Figure 1. Difference between OMI total ozone V8.0
and OMI total ozone V8.5. We selected the period
January 2005 corresponding to northern
biomass-burning season (left panel) and September
2005 corresponding to southern biomass-burning
season (right panel). Notice significant ozone
decreases over the ITCZ region, western Pacific,
and the South America continent. Increases
elsewhere.
Figure 5-b and c. When a cloud is present, OMI
can only measure column ozone above the cloud.
Then, total ozone is determined by adding
climatological tropospheric ozone (Table 1)
between surface and the cloud altitude to OAC.
Figure 5-b and c show that the correlation
between OAC and OBC over the Pacific and Atlantic
ocean, respectively. Tropospheric ozone colum
(TOR) can be determined by subtracting OAC from
total ozone over clear sky (Figure 5-b). If
tropospheric ozone climatology is true, OBC must
be the same as TOR. Then total ozone over clear
sky will be the same as total ozone over cloudy
regions. If not, we will retrieve different ozone
amount between over clear and cloudy condition.
TOR in Figure 5-b is slightly higher than OBC
over area B and C in Figure 4. TOR in Figure 5-c
is significantly higher than OBC over area A in
Figure 4. For this case, total ozone over cloudy
region becomes smaller than over clear region.
Therefore, the negative anomaly in Figure 3
illustrates the regions where OBC is less than
TOR. In order to correct the negative anomalous
regions, OBC must be increased. There are two
ways to fix this. One is to increase tropospheric
ozone amount in the climatology. However, from
Table 1, the ozone amounts from surface to the
various altitudes are higher than the ozone
amounts observed over the Pacific ocean (Kim and
Newchurch, 1996). Therefore, this can not be
accepted. The other way is to raise the reported
OMI cloud reflecting surface to higher altitude
and so we can add more ozone corresponding to the
increased altitude. Then we can make TOR equal to
OBC. This suggests that the current OMI algorithm
retrieves the altitude of reflecting cloud
surface lower than actual altitude.
Figure 5-b
Figure 2. Difference between ozone above cloud
(reflectivity gt 30) and ozone over clear region
(reflectivity lt 15) with OMI version 8.0. Strong
positive anomaly of about 7 DU is observed over
equatorial Pacific ocean
TOR
area B
area C
OBC
Figure 5-C
Figure 3. Difference between ozone above cloud
(reflectivity gt 30) and ozone over clear region
(reflectivity lt15) with OMI version 8.5. Strong
negative anomaly of -7 DU is observed over the
equator where strong positive anomaly is
observed. Especially, its interesting to see
that the reduction of 14 DU over equatorial
regions is observed over equatorial region in
comparison of V8.5 with v8.0. On the contrary, a
positive anomaly is still observed over south
eastern Pacific Ocean near South America and the
southern Atlantic Ocean.
Figure 6. The definition of TOR and OBC
Table 1. altitudes corresponding to ozone amounts
below clouds
Month Latitude Height for ozone amount Height for ozone amount Height for ozone amount
Month Latitude 10DU 15DU 20DU
JAN 5S 4.5 km 7.1 km 10.2 km
5N 4.0 km 6.3 km 8.8 km
SEP 5S 3.3 km 5.1 km 7.2 km
5N 3.9 km 6.5 km 8.6 km
Figure 4. The upper panels represent cirrus cloud
reflectance, which indicates the location of
convective clouds, in January and September 2005
from MODIS and the lower panels represent marine
stratocumulus clouds, which indicates low clouds,
from ISCCP. From the cloud information, the
negative anomaly and positive anomaly in Figure 3
correspond differently to convective cloudy
regions in ITCZ and low marine stratocumulus
cloudy regions, respectively.
A
B
C
A
B
C
Figure 7. 15 January 2005 15
September 2005
Figure 7. Correlation between ozone column above
cloud (OAC) and reflectivity over eastern Pacific
and southern Atlantic where the marine
stratocumulus clouds are persistently located (D
and E in Figure 4). The lower figures shows OAC
vs. OBC. Because OBC is just about 2-3 DU, these
clouds are low clouds, which are consistent with
ISCCP in Figure 4. Therefore, the correlation
between OAC and reflectivity is caused by OBC as
of Figure 6. The only OMI total ozone error
related to reflectivity is associated with the
tropospheric ozone retrieval efficiency. The
efficiency is high over high reflecting surface
(HRS). For this case, the algorithm retrieves
total ozone close to the truth. If climatological
tropospheric ozone is much smaller than the
truth, ozone over HRS retrieves much more than
over low reflecting surface (LRS) which results
in a steep slope between total ozone and
reflectivity in Figure 7. We can observe this
case over the southern Atlantic in
September-October period (area E in 15
September). However, the climatology is slightly
smaller than the truth, the ozone over HRS is
slightly more than the ozone over LRS. This
results in gentle slope in Figure 7. This case is
over marine stratocumulus cloudy regions the
eastern Pacific ocean and the southern Atlantic
in January when the biomass burning occurs over
the northern Africa. Therefore, the positive
anomaly in Figure 3 could be due to OMI retrieval
error related to tropospheric ozone retrieval
efficiency.
area E
D
D
E
E
area D
January 2005 September 2005
Figure 8. OMI-MLS tropospheric ozone for January
and September 2005
Conclusion (1). Compared to OMI V8.0, V8.5 shows
significantly less ozone over the ITCZ where
high-altitude convective clouds exists. No change
occurred over low, marine stratocumulus cloudy
regions. (2). V8.5 total ozone over convective
cloudy regions such as ITCZ was 3-7 DU lower than
over clear regions. This negative anomaly is due
mostly to OMI cloud-top altitudes that were lower
than the actual altitude. (3). On the contrary,
V8.5 total ozone over low marine stratocumulus
was higher than V8.0. This positive anomaly is
mostly like due to higher tropospheric ozone
retrievals over high reflecting surface than over
low reflecting surface. (4). When the
cloud-related error was removed, the OMI-MLS
tropospheric ozone discontinuity along the
coastline of west -Africa was reduced.
Figure 9. Corrected OMI-MLS tropospheric ozone
was derived by applying (OMI total ozone)
(MLS stratospheric ozone (OMI total ozone)
(OMI total ozone over clear sky condition) We
selected OMI V8.5 Level-2 total ozone over clear
sky condition was defined with OMI measurements
with reflectivity 15. The corrected
tropospheric ozone was marginally increased over
the northern equatorial Atlantic in January and
decreased about 2-4 DU over the southern Atlantic
in September. This decrease reduced the
discontinuity in tropospheric ozone between the
southern Atlantic and south African continent in
September.
Reference Hudson, R., J.-H. Kim, and A. Thompson
(1995), On the derivation of tropospheric column
ozone from radiances measured by the total ozone
mapping spectrometer, J. Geophys. Res., 100(D6),
11137-11145. Joiner J., A. Vasilkov, K. Yang, P.
K. Bhartia (2006), Observations over hurricanes
from the ozone monitoring instrument, Geophys.
Res. Lett., 33, L06807, doi10.1029/2005GL025592.
Kim, J., and M. Newchurch (1998),
Biomass-burning influence on tropospheric ozone
over New Guinea and South America, J. Geophys.
Res., 103(D1), 1455-1461 Klenk, K.F., P.K.
Bhartia, A.J. Fleig, V.G. Kaveeshwar, R.D.
McPeters, and P.M. Smith, "Total ozone
determination from the backscattered ultraviolet
(BUV) experiment", J. Appl. Meteorol., 21,
1672-1684, 1982. McPeters R. D., G. J. Labow,
J. A. Logan (2007), Ozone climatological profiles
for satellite retrieval algorithms, J. Geophys.
Res., 112, D05308, doi10.1029/2005JD006823.
Thompson, A., D. McNamara, K. Pickering, and R.
McPeters (1993), Effect of Marine Stratocumulus
on TOMS Ozone, J. Geophys. Res., 98(D12),
23051-23057.
Figure 10. Difference between (OMI-MLS) and
corrected (OMI-MLS) tropospheric ozone.