P2.36
Generation of Simulated Top of Atmosphere Radiance Datasets
for GIFTS/HES Algorithm Development
Jason A. Otkin*, Derek J. Posselt, Erik R. Olson, James E. Davies, Wayne F. Feltz, Robert O.
Knuteson, and John R. Mecikalski
Cooperative Institute for Meteorological Satellite Studies, Space Science and Engineering Center, University of Wisconsin–Madison
1. INTRODUCTION
4. MM5 SIMULATIONS: SENSITIVITY STUDIES
6. SIMULATED RADIANCE DATA
The anticipated launch of the Geosynchronous Imaging Fourier Transform
Spectrometer (GIFTS) represents a significant advance in our ability to image
and sound the atmosphere from a geosynchronous orbit. GIFTS is targeted at
a horizontal resolution of 4 km, a vertical resolution of 1-2 km, and a maximum
temporal resolution of 10 seconds. As such, it will produce retrievals of
temperature, moisture, and wind with much greater temporal and vertical
resolution than is currently available from any existing geostationary instrument.
SSEC/CIMSS is currently tasked with testing and developing the GIFTS fast
forward radiative transfer model and retrieval algorithms. In support of this
work, numerical model simulations with high spatial and temporal resolution are
used to produce a “truth” atmosphere that is then passed through the GIFTS
fast forward radiative transfer model to generate simulated top of the
atmosphere (TOA) radiances. Retrievals of temperature and water vapor
generated from these radiances are subsequently compared with the original
simulated atmosphere to assess retrieval accuracy.
Extensive sensitivity testing was undertaken in order to determine the optimal configuration of physical parameterizations for this simulation. Each simulation was initialized
at 1200 UTC 11 March 2003 and allowed to spin-up for 33 hours before the start of the ER-2 flight at 2100 UTC on 12 March 2003. Such a long spin-up period was chosen in
order to allow sufficient time for the MM5 simulation to properly generate realistic fine-scale cloud and moisture features from the original coarse-resolution GDAS analyses.
Simulated IR fields at 2100 UTC on 12 March 2003 are shown below for each sensitivity test.
Configuration details for each sensitivity run:
• All simulations were run on a Lambert Conformal conical projection grid with 50 vertical levels and fully explicit convection on the innermost 4-km domain.
• Control: Goddard microphysics, RRTM/Dudhia radiation, MRF boundary layer scheme, and Grell cumulus convection scheme on the outer two domains.
• ETA: as for the control simulation, but with the ETA boundary layer scheme in place of the MRF scheme.
• KAIN-FRITSCH: as for the ETA simulation, but with the Kain-Fritsch cumulus parameterization scheme used in place of the Grell scheme on the outer two domains.
• NUDGED: as for the ETA simulation but nudged toward the GDAS analyses using the MM5 four-dimensional data assimilation system. The outermost 36-km domain was
nudged during the first 24 hours of the simulation while the inner 12-km domain was nudged for the first 12 hours. Each domain then had a 3-hour period after their respective
"nudging periods" during which the nudging coefficients were ramped down to zero in order to allow the model simulation to transition gracefully between the forced and
freely-evolving states. The innermost 4-km domain was not nudged during any portion of the simulation.
UW-CIMSS has developed a fast forward radiative transfer model that
ingests atmospheric profiles of temperature, water vapor mixing ratio, liquid and
ice cloud, and ozone, and generates top of atmosphere radiances in the GIFTS
spectral range.
GIFTS Spectrum: Clear Sky
Observed and Simulated GOES-10 10.7 micron Brightness Temperatures
2. CASE DESCRIPTION
GIFTS Spectrum: Low-Level Cloud
Current work involves a high-resolution MM5 simulation of an intense uppertropospheric jet streak that occurred over the north-central Pacific Ocean during
12-13 March 2003. This particular jet streak case was selected because it
occurred during a day specifically targeted for the study of jet streak dynamics
during the Pacific THORPEX Observational Science Test 2003. Observational
datasets collected during a NASA ER-2 flight between 2100 UTC 12 March and
0400 UTC 13 March included remotely-sensed observations from the NPOESS
Aircraft Sounder Testbed-Interferometer (NAST-I), Scanning High-resolution
Interferometer Sounder (SHIS), MODIS Airborne Simulator (MAS), and Cloud
Physics Lidar (CPL), as well as dropsondes across the jet streak core.
Case specifics:
• Maximum analyzed zonal wind speed at 300 hPa was over 90 m s-1
• Upper-tropospheric jet streak was characterized by very strong horizontal
and vertical wind shear
Observation/Analysis
Control
ETA
KAIN-FRITSCH
NUDGED
The goal of this simulation was to achieve a high degree of realism between the observed and MM5-simulated cloud fields. Since observations of cloud microphysical
properties are not available for this case, a subjective comparison between the simulated and observed GOES IR fields was undertaken in order to determine which MM5
simulation performed best. Examination of the simulated IR fields indicates that both the ETA and Kain-Fritsch simulations most realistically depicted the cellular cumulus
cloud field observed across the southern three quarters of the domain while still maintaining the cirrus cloud shield associated with the cyclone along the northern periphery
of the domain. Since the Kain-Fritsch cumulus scheme has been shown in prior studies to have some difficulty simulating clouds over the ocean, it was determined that the
ETA simulation most faithfully represented the observed GOES IR imagery across the region of interest.
GIFTS Spectrum: Upper-Level Cloud
5. MM5 SIMULATION: RESULTS
300 hPa zonal wind isotachs (m/s) at
0000 UTC 13 March 2003
Meridional cross-section of zonal wind
isotachs (m/s) at 0000 UTC 13 March 2003
The observed GOES-10 IR brightness temperatures and the simulated brightness temperatures from the ETA simulation are shown below. Comparison between the
observed and simulated IR fields indicates that the ETA simulation accurately portrays the location of the cloud field associated with the northern cyclone as well as
realistically depicts the low-level cumulus cloud field to the south of this cyclone. It is also evident that the fairly smooth appearance of the observed IR field along the
southern portion of the domain is adequately portrayed in the simulated IR field by the relatively large spatial scale of the individual components of the stratus cloud deck
within this region. One notable difference, however, between the observed and simulated IR fields is the absence of upper-level cirrus within the southwestern portion of the
simulated IR field. Even with this error, the fact that the MM5 simulation was able to realistically simulate much of the observed cloud field renders it a useful example to
generate simulated GIFTS top of the atmosphere radiances over a maritime atmosphere characterized by strong horizontal and vertical wind shear.
Observed GOES-10 10.7 micron Brightness Temperatures
3. MM5 SIMULATIONS: DATA INGEST
Simulated atmospheric fields were generated using version 3.5 of the 5th
generation Penn State/NCAR Mesoscale Modeling system (MM5). Atmospheric
fields required to initialize the MM5 and provide lateral boundary conditions
during the subsequent simulation were obtained from six-hourly GDAS
analyses with 1° x 1° horizontal grid spacing. Each six-hourly GDAS file
contained three-dimensional fields of temperature, zonal and meridional
components of the horizontal wind, geopotential height, and relative humidity on
26 pressure levels between 1000 and 10 hPa. The GDAS analyses of skin
temperature, sea level pressure, soil temperature, soil moisture, surface height
and other near-surface atmospheric fields provided the lower boundary
conditions for the MM5 simulations.
Geographical region covered by the three
two-way interactive nested domains used for
each simulation. Horizontal grid spacings
for each domain were 36 km, 12 km, and 4
km, respectively. Only atmospheric fields
from the innermost 4-km domain were used
to produce the simulated TOA radiances.
Simulated radiances from the GIFTS forward model with spectra in the GIFTS
longwave band for clear (top), low cloud (middle) and high cloud (bottom) regions.
7. ATMOSPHERIC RETRIEVALS
Top of the atmosphere radiances generated by the fast forward radiative
transfer model are used to retrieve profiles of atmospheric temperature and
water vapor. Horizontal fields constructed from these profiles are then
compared to the original atmospheric fields to assess the robustness of the
retrieval method. A comparison of the simulated and retrieved (with double
noise added) water vapor mixing ratio fields at 750 hPa is shown below.
Simulated GOES-10 10.7 micron Brightness Temperatures
Simulated (left) and retrieved (right) water vapor at 750 hPa, 0000 UTC 13 March 2003
ACKNOWLEDGEMENTS
This work was funded by NASA NMP contract NAS1-00072
*Contact: Jason A. Otkin • Address: 1225 W. Dayton Street • Madison, WI 53706 • Phone: 608/265-2476 • Email: jason.otkin@ssec.wisc.edu