COMPARISON OF CO2 AND H2O ATMOSPHERIC MOTION VECTOR
HEIGHT ASSIGNMENT TECHNIQUES USING THE GOES-13 IMAGER
Anthony J. Schreiner1, W. Paul Menzel1, William Straka1, and Andy Heidinger2
1
- Cooperative Institute for Meteorological Satellite Studies
2
- NOAA/NESDIS
1225 West Dayton Street, Madison, Wisconsin 53706, USA
ABSTRACT
Atmospheric motion vectors (AMVs) estimated from time sequences of geostationary
satellite images are assigned heights by several techniques. In opaque clouds (usually not
chosen as a reliable AMV tracer), the level of best agreement between the infrared
window (IRW at 10.7 µm) brightness temperatures and the forecast temperature profile is
assumed to be the level of the tracer. In semi-transparent clouds or sub-pixel clouds
(preferred as an AMV tracer), corrections must be made for the radiation coming from
below the cloud. One approach is to use a simplified version of the carbon dioxide
slicing technique; radiances with similar ice cloud emissivities and sensitive to different
layers of the atmosphere (CO2 at 13.3 µm and IRW at 10.7 µm) are combined via a ratio
technique to infer the level of the cloud tracer. This is currently used on the GOES
satellites (after GOES-11). Another approach uses the IRW-water vapor (H2O) intercept
technique to perform a correction for cloud semi-transparency; the tracer cloud height is
extrapolated from the linear relationship between radiances sensitive to water vapor
(H2O at 6.5 µm) and radiances not sensitive to water vapor (IRW at 10.7 µm) as a
function of cloud amount. The H2O intercept approach has been used with data from
Meteosat and GOES (prior to GOES-12) and will be used on the INSAT 3D Imager.
Software for both algorithms has been revisited for the GOES-13 Imager and the cloud
heights inferred from CO2-IRW slicing and H2O-IRW intercept techniques have been
compared.
1.
TECHNIQUE DESCRIPTIONS
Cloud motion vectors are often inferred from tracking semi-transparent or sub-pixel
clouds. Height assignments for those clouds are difficult because cloud amount and
cloud emissivity are less than unity by unknown amounts; low thick clouds cannot easily
be separated from high thin clouds. The infrared window (IRW) technique works
reliably only in opaque clouds; IRW brightness temperatures can be compared to forecast
temperature profiles to infer the level of best agreement that is taken to be the level of the
cloud. In semi-transparent ice-clouds, the carbon dioxide (CO2) slicing technique
(Menzel et al., 1983) combines radiances that have similar ice cloud emissivities and that
are sensitive to different layers of the atmosphere (CO2 at 13.3 m and IRW at 10.7 m)
via a ratio technique to infer the level of the cloud. In semi-transparent water clouds or
sub-pixel clouds, the water vapor intercept technique (Szejwach, 1982) uses the linear
relationship between radiances influenced by upper tropospheric moisture ((H2O at 6.5
m) and radiances not sensitive to water vapor (IRW at 10.7 m) as a function of cloud
fraction to extrapolate the correct height.
The CO2 slicing technique cloud height uses a ratio of deviations in observed cloudy
radiances from corresponding clear air radiances for IRW and CO2 channels. The clear
and cloudy radiance differences are determined from observations with GOES-13 and
radiative transfer calculations. The cloud emissivities of the two channels are roughly the
same for ice clouds so the ratio of the clear and cloudy radiance differences yields a
solution for the cloud top pressure of the cloud within the field of view (FOV). The
observed differences are compared to a series of radiative transfer calculations with
different cloud pressures; the cloud top pressure belongs to the calculation that best
satisfies the observations (Menzel et al. 1983).
The H2O intercept height assignment is predicated on the fact that the radiances for two
spectral bands vary linearly with cloud amount. Thus a plot of H2O (6.5 m) radiances
versus IRW (10.7 m) radiances in a field of varying cloud amount will be nearly linear.
These data are used in conjunction with forward calculations of radiance for both spectral
channels for opaque clouds at different levels in a given atmosphere specified by a
numerical weather prediction of temperature and humidity. The intersection of measured
and calculated radiances will occur at clear sky radiances and opaque cloud radiances.
The cloud top temperature is extracted from the cloud radiance intersection (Schmetz et
al., 1993). This technique provides heights for mid- to upper tropospheric clouds.
The multi-spectral imager from GOES-13 measures IRW (centered at 10.7 m) and H2O
(centered at 6.5 m) and CO2 (centered at 13.3 m) radiances from 4 km FOVs. All
three cloud height assignment techniques can be compared with simultaneous
measurements from the GOES-13 imager.
2.
COMPARISONS AND VALIDATIONS
2a. Comparison of CO2 and H2O heights
Comparison of these height assignment techniques was accomplished with data from the
GOES-13 on 14 December 2010. Some details regarding the processing follow:
* The box size (line x element) used is 5 x 7 (35 observations), which is roughly 20 km
on a side at the GOES-13 Imager satellite subpoint.
* The IRW only algorithm uses the measured 11 µm brightness temperature (BT) and an
atmospheric profile from the Global Forecasting System (GFS) to determine Cloud Top
Pressure (CTP) at each cloudy field of view (FOV). Effective Cloud Amount (ECA) for
each cloudy FOV is assumed to be 100%.
* For each FOV, CO2 Slicing (CO2/IRW) determines a single CTP and ECA using 13.3
µm and IRW radiance measurements.
* The Water Vapor Intercept (H2O/IRW) technique generates a single CTP for the entire
box, where the assumption is that the cloudy FOVs represent a single cloud layer and
only the ECA is changing.
* The atmospheric first guess is based on hourly interpolated forecasts from the 3 hourly
GFS. Horizontal resolution of the first guess is 0.50 deg latitude / longitude with a
vertical resolution of 25 hPa from 1000 to 900 hPa and 50 hPa from 900 to 100 hPa.
* The surface analysis (temperature at sea level) is based on hourly surface observations
over land and buoy observations over water using the atmospheric guess as a background.
The daily Sea Surface Temperature (12 UTC) is based on NOAA polar orbiting satellite
observations.
* Hourly AMVs are generated using the XX:45 UTC as the processing time. This allows
for one hemispheric image every three hours (00, 03, 06, 09, 12, 15, 18, and 21 UTC).
* CO2/IRW and IRW CTPs at full resolution (single FOV) and H2O/IRW at box
resolution (5 line x 7 elements) are generated simultaneously.
* The statistics in this paper are based on CO2 and IRW point data and H2O box data.
* CTP is converted to Cloud Top Height (CTH) using the GFS.
When the three techniques are compared as a function of the ECA (see Figure 1), the
difference between the techniques decreases with increasing cloud amount. The
H2O/IRW intercept heights are consistently lower (higher CTP) than the CO2/IRW
slicing heights. IRW heights are very unreliable in thin clouds.
Figure 1: Comparison of IRW and IRW/H2O CTPs to CO2/IRW CTPs between 440 and
100 hPa. (Left) Number of occurrences of CO2/IRW CTP's in the indicated ECA
intervals. (Right) Average CTP for the three techniques at the varying ECA categories.
The x-axis (y-axis) plots ECA in % (CTP in hPa).
Figure 2 shows the scatter plot of H2O/IRW and CO2/IRW cloud top heights for clouds
above 4 km on 14 December 2010. H2O/IRW and CO2/IRW estimates show substantial
disagreement at all levels; correlation between the heights determined by the two
techniques is only 0.36. H2O/IRW heights tend to be 1 km lower overall and 4 km lower
in thin cirrus.
.
Figure 2: Scatter plot of the H2O/IRW (y-axis) and CO2/IRW (x-axis) CTH assignments
for clouds above 4 km. The resolution of the bins is 0.5 km. The bright vertical line
along the right hand side (dark red) is primarily a result of the situation where the
CO2/IRW technique finding the high thin cirrus, where the H2O/IRW version tends to
jump around a lot more. A logarithmic scale has been used to define the color bar
indicating the number of occurrences.
2b. Comparisons with CALIPSO
AMV heights were compared with collocated CALIPSO observations. Figure 3a shows
the GOES-13 Imager IRW image for 1845 UTC on 14 Dec 2010 (2010348). The white
line represents the estimated track of the CALIPSO satellite for this particular day.
CALIPSO co-locations were provided by the Atmospheric PEATE. The CALIPSO
attenuated 0.532-µm backscatter data indicates the location and thickness of the clouds.
Figure 3b shows superimposed on the CALIPSO attenuated cloud backscatter (shown in
gray), the cloud heights determined by the three GOES techniques and the CALIPSO
estimates of the cloud top and earth surface. In the region around the Equator (0 degrees
latitude), CALIPSO appears to be showing a very thin layer of cirrus that the CO 2/IRW
technique is picking up, but the H2O/IRW technique is missing.
Figure 3a: GOES-13 Imager IRW image for 1845UTC on 14 Dec 2010 (2010348). The
white line represents the CALIPSO nadir track; the yellow circle is the area (near the
Equator) containing high thin clouds.
Figure 3b: CALIPSO/CALIOP attenuated cloud backscatter (gray), the cloud heights
determined by the three GOES techniques (CO2/IRW in red, H2O/IRW in blue, and IRW
in green) and the CALIPSO/CALIOP estimates of the cloud top (black) and earth surface
(light blue).
Correlations of the three CTH techniques with CALIPSO are found to be as follows:
for all clouds along the transect
CO2/IRW vs CALIOP = 0.37
H2O/IRW vs CALIOP = 0.29
IRW vs CALIOP = 0.26
for high clouds above 4km (as determined by CALIPSO) along the transect
CO2/IRW vs CALIOP = 0.48
H2O/IRW vs CALIOP = 0.32
Overall, CO2/IRW is performing quite well in thin cirrus, while the H2O/IRW is
underestimating the CTH by 4 km.
3.
CONCLUSIONS
GOES-13 results presented in this paper suggest that
(1) H2O/IRW & CO2/IRW CTP determinations show modest correlation for AMV cloud
tracers above 4 km. H2O/IRW CTH estimates are about 1 km lower than CO2/IRW on
average, and for semi-transparent ice clouds this increases to 4 km. CO2/IRW CTH
estimates are in better agreement with CALIOP than H2O/IRW CTHs.
(2) INSAT winds derived using only H2O/IRW CTHs will necessarily have height
assignment biases with respect to GOES and Meteosat.
(3) ABI AMV CTH estimates are anticipated to be of better quality (better spatial
resolution, spectral characterization, and radiometric calibration) with improved AMV
tracer characterization (cloud phase, thickness, microphysics, …).
4.
REFERENCES
Menzel, W. P., W. L. Smith, and T. R. Stewart, 1983: Improved cloud motion wind
vector and altitude assignment using VAS. J. Clim. Appl. Meteor., 22, 377-384.
Nieman, S. J., J. Schmetz and W. P. Menzel, 1993: A comparison of several techniques
to assign heights to cloud tracers. J. Appl. Meteor., 32, 1559-1568.
Nieman, S. J., W. P. Menzel, C. M. Hayden, D. Gray, S. T. Wanzong, C. S. Velden, and
J. Daniels, 1997: Fully automated cloud drift winds in NESDIS operations. Bull.
Amer. Meteor. Soc., 78, 1121-1133.
Schmetz, J., K. Holmlund, J. Hoffman, B. Strauss, B. Mason, V. Gaertner, A. Koch, and
L. van de Berg, 1993: Operational cloud motion winds from METEOSAT
infrared images. J. Appl. Meteor., 32, 1206-1225.
Szejwach, G., 1982: Determination of semi-transparent cirrus cloud temperatures from
infrared radiances: application to Meteosat. J. Appl. Meteor., 21, 384.