JP1.25
Characterization of upper tropospheric temperature and water vapor using UW/SSEC S-HIS, RAMAN LIDAR and other insitu and remotely sensed data collected during the Joint Airborne IASI Validation Experiment (JAIVEx)
Sarah Bedka1, Paolo Antonelli1, Robert Knuteson1, Henry Revercomb1, William Smith2, David Tobin1 , David Turner1 and Hal Woolf1
1 Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin, Madison WI
2 Hampton University, Hampton VA
Abstract
The University of Wisconsin Space Science and Engineering Center Scanning
High-resolution Interferometer Sounder (UW/SSEC S-HIS) is an aircraft-based
high spectral resolution infrared spectro-radiometer that has flown on high
altitude research aircraft during numerous field experiments. This paper
presents a physical retrieval algorithm that may be applied to data from S-HIS.
Application of this retrieval to hyperspectral satellite platforms such as IASI
(Infrared Atmospheric Sounding Interferometer) is preliminary, and early
results are shown here. The algorithm uses a Maximum A Posteriori (MAP)
approach to the inverse problem, which has an advantage over statistical
regression in that it provides error characteristics and an explicit measure of
convergence. This methodology requires a priori knowledge of the solution
space, and as such, emphasis has been placed on understanding and
developing appropriate experiment-specific climatologies. We present case
studies showing T/Q retrievals from JAIVEx (Joint Airborne IASI Validation
Experiment). Comparisons will be made with other field experiment data in
order to provide preliminary validation for the S-HIS and satellite retrievals.
S-HIS Physical Retrieval Algorithm
Bayesian Maximum A Posteriori Retrieval Methodology
The Bayesian MAP is a method for simultaneously retrieving temperature, moisture
and ozone profiles from S-HIS radiances. This technique, described by Rogers
(2000), is a method of mapping the pdf of the measurement to the pdf of the state,
using a priori knowledge of the state in the form of a climatology. A line-by-line
forward model (AER LBLRTM v10.3) was used to minimize the observed minus
calculated radiances and compute with analytic jacobians (Clough). The NewtonGauss method was used to find the zero of the derivative of the joint probability
distribution (pdf) corresponding to the optimal solution. The actual form used is:
x = xa+(KTSe-1K+Sa-1) -1KTSe-1*[R-F(x)+K*(x-xa)]
Where xa is the first guess, R is the observation, F(x) is the forward model, K is
the jacobian (dF/dx), Sa is the covariance matrix of the a priori climatology, Se
is the covariance matrix of the measurement error.
Error Estimation
One advantage of using the Bayesian MAP retrieval method, is that it provides
a quantitative assessment of error. The error sources can be split up as:
^
x - x = (A - I)(x - xa)
^
+ GyKb(b-b)
+ Gy∆f(x,b,b')
+ Gyε
smoothing
model parameters
modeling error
measurement error
The measurement error results from the noise of the remote sensing
instrument. The two types of modeling error are not currently included in our
analysis, but will be in the future. They are systematic, and result from the
errors in representation of absorptive gasses within the forward model, as well
as the mathematical representation of complicated physical processes.
The smoothing error results from the fact that
passive infrared remote sensing instruments
that measure emitted radiance can not see
fine spatial structure in the vertical. In this
study, we have assumed that the retrieval is
an estimate of the true state, with an error
contribution due to smoothing, rather than
assuming that the retrieval is an estimate of
the smoothed state (not the true state). The
error contribution due to smoothing is
Sample averaging kernel (A). Each curve
determined by the averaging kernel.
represents the averaging inherent in
determining a solution for that level.
S-HIS Instrument and Field Experiment Details
The S-HIS is a scanning interferometer which measures emitted thermal
radiation at high spectral resolution between 3.3 and 18 µm
(specifications). The measured emitted radiance is used to obtain
temperature and water vapor profiles of the Earth's atmosphere. The SHIS produces sounding data with 2 kilometer resolution (at nadir) across
a 40 kilometer ground swath from a nominal altitude of 20 kilometers
(such as onboard a NASA ER-2 aircraft).
Retrieval Results
All of the following S-HIS retrievals use:
• Unfiltered, unapodized S-HIS radiances
• LBLRTM version 11.3
• 10 iterations (K updated on every iteration)
April 19, 2007
• Convergence threshold of chi_square < 9
• 61 level climatology (surface to 50 mb)
• √(NESR) as a measure of the total uncertainty
of the instrument noise and forward model error
ARM CART Site
Joint Airborne IASI Validation Experiment (JAIVEx): 14 Apr – 4 May 2007
JAIVEx was designed as a joint US and European collaboration focused on the validation of radiance
and geophysical products from MetOp-A. The central location was Houston, TX, however, several
flights extended north into the region around the ARM CART site. This site provided a wealth of groundbased data useful for validation.
Selected Instrumentation:
• S-HIS on the NASA WB-57
• In-situ cloud physics measurement suite on the UK FAAM BAe
146-301
• 22 Vaisala RS-92 Rawinsondes launched from the ARM CART
site
• Concurrent satellite overpasses of MetOp-A (including IASI) and
the NASA A-train satellites (including AIRS, MODIS, and TES)
Observed - calculated radiances
are within a reasonable range, as
is the S-HIS noise. Shown is from
335 UTC on 4/19.
Houston, TX
WB-57 flight track for April 19, with IASI
radiances and FOV. Flight was from
Houston to OK, and mostly clear.
Converged UWPHYSRET retrieval profiles
from the vicinity of the ARM CART site, with
reference sonde and LIDAR profiles
The NASA WB-57 carried the S-HIS during JAIVEx
A Priori Climatology Details
The Bayesian MAP method of retrieving temperature and moisture profiles from S-HIS radiances
requires the input of a realistic climatology of temperature, water vapor, and ozone. The input of
an accurate climatological covariance matrix is crucial for the physical retrieval algorithm.
Additionally, the climatological mean temperature, water vapor, and ozone profiles may be used
as a first guess. For this study, we have compiled a climatology of National Weather Service
standard rawinsondes (VIZ-B2, which measures water vapor using a carbon hygristor) from
within the experiment domain, over the season of interest. Since upper tropospheric rawinsonde
moisture profiles have documented inaccuracies (Ferrare et al., 2004), profiles are filled in above
300 mb using an inverse cubic power law.
This methodology shows potential for
retrieving both temperature and water
vapor accurately, and currently may be
applied to S-HIS and IASI. Preliminary
comparisons show IASI may be useful
for retrieving upper level water vapor.
April 16, 2007
The WB-57 flight track on 4/16 was very similar to 4/19
(over the ARM CART site). No IASI overpass was
available near the rawinsonde launch time, however, an
AIRS overpass occurred within a few minutes. The
UWPHYSRET algorithm is not yet being applied to AIRS,
though the L2 v.5 retrieval is shown for comparison.
Climatology Details
Date Range: 01 Apr – 16 May
Year Range: 2003 - 2006
Standard NWS sondes: VIZ-B2 sondes, launched at 0 Z or 12 Z)
Clear sondes only: 5813 profiles (2212 cloudy profiles excluded)
Note: Ozone profiles are not available from standard rawinsondes,
so we have estimated the ozone using regression
ARM CART Site
Houston, TX
Future Work
For comparison, NWS sonde T and
W profiles are shown with Vaisala
RS-92 profiles from a similar time
period, but launched only from the
ARM SGP site. The RS-92 sondes
measure water vapor using a twinsensor thin film capacitor. The mean
profiles of the NWS and
RS-92
sondes are shown in yellow or cyan,
respectively. The mean temperature
profiles are very similar, however,
the mean water vapor profile from
the VIZ-B2 is significantly drier at
high altitudes than the RS-92.
UWPHYSRET is a reference retrieval algorithm intended to be flexible and implemented
within MATLAB. Preliminary results show good agreement between SHIS T/Q retrievals
and rawinsondes, however, additional work is needed to ensure stable use with IASI
and AIRS data. Future work will also include retrievals of trace gasses (e.g. CO2 and
O3), additional characterization of the forward model error, inclusion of uncertainties
inherent in the a priori, and evaluation of uncertainties in surface emissivity over land.
Selected References
•Clough, S.A., M.J. Iacono. 1995. Line-by-line calculations of atmospheric fluxes and cooling rates: Application to water vapor. J. Geophys. Res., 97, 15761-15785.
•Rodgers, C. D. 2000. Inverse Methods for Atmospheric Sounding: Theory and Practice. World Scientific Publishing Co, 238 pp.
•Ferrare, R.A., E.V. Browell, S. Ismail, S. Kooi, L.H. Brasseur, V.G. Brackett, M.B. Clayton, J.D.W. Barrick, G.S. Diskin, J.E.M. Goldsmith, B.M. Lesht, J.R. Podolske, G.W. Sachse, F.J.
Schmidlin, D.D. Turner, D.N. Whiteman, D. Tobin, L. Miloshevich, H.E. Revercomb, B.B. Demoz, and P. Di Girolamo, 2004: Characterization of upper tropospheric water vapor measurements
during AFWEX using LASE. J. Atmos. Oceanic Technol., 21, 1790-1808.
This work was supported under NASA grant NNX08AH96G. Thanks to Rob Newsom for providing
RAMAN LIDAR data. AIRS data was provided by the Goddard DAAC.
Contact: Sarah T Bedka, sarah.bedka@ssec.wisc.edu