An Experiment of the ICI3 Scheme for Retrieving
Atmospheric Parameters over China
Xuebao Wu 1, 2 , Fengying Zhang 1 , Lydie Lavanant 3 ,
Pascal Brunel 3 , Wenjian Zhang 1 and Yuanjing Zhu 2
1
2
National Satellite Meteorological Center, Beijing, China
Department of Atmospheric Science of Peking University, Beijing, China
3
Centre de Météorologie Spatiale, Météo-France, Lannion, France
Abstract
This report reviews briefly the ICI retrieval method. An experiment was carried
out to retrieve the temperature and humidity profiles from NOAA-16 ATOVS data over
China. The ICI physical retrieval scheme with some adaptations is implemented in
this study. The analyses of the Chinese regional NWP model are utilized to create a
rolling library for initial guess field. Retrieval results validated against both NWP
analyses and radiosondes indicate good agreement between the retrieved
temperature and humidity profiles and conventional observations.
ATOVS Preprocessing
The AAPP (ATOVS and AVHRR Preprocessing Package, Klaes 1997) has been
installed to process the relay HRPT data from the 3 receiving stations (Beijing,
Guangzhou and Urumqi). The radiances from ATOVS instruments are interpolated to
the HIRS sounder’s field of view. With high-resolution AVHRR data as well as AMSU
data, the cloud detection technique for current ATOVS is improved greatly. Figure 1
shows such an example of the cloud cover. When available in the FOV of the HIRS,
such information enhances the cloud detection for selecting initial guess and
inversion process.
Retrieval Atmospheric Profile
The radiance leaving the earth-atmosphere system which can be sensed by a
satellite-borne radiometer is the sum of the radiation emissions from the earth’s
surface, each atmospheric level and those reflected by the surface. For a
non-scattering stratified atmosphere in local thermodynamic equilibrium, the radiative
transfer equation can be expressed as
E   s   ( p s ,0)  B(Ts )  
ps0
 (1   s )   ( p s ,0)  
0 ps
[ B(T p )( ( p,0) / ( ln p)) d (ln p)]
[ B(T p )( ( p s , p) /  ( ln p)) d (ln p)]
It is necessary to acquire realistic initial guess in ICI scheme. Generally the
analysis of NWP model is more representative of the real atmosphere than
climatological data. In our processing, the analyses of the Chinese regional NWP
model HLAFS (High-resolution Limited Area Forecasting System) are utilized to
create a rolling library of vertical profiles over the last ten days. Once an initial guess
profile has been obtained, inversion consists of perturbing this profile under certain
limits in order to minimize the error between the measurement and the synthetic
brightness temperature of the final profile. The most probable solution is the function
minimizing the cost function (Eyre 1989)
J (Ta )  (Ta  Tag ) t  G 1  (Ta  Tag )  (Tbmes  Fw(Ta )) t  (O  F ) 1  (Tbmes  Fw(Ta ))
Figure 2 is the diagram of ICI scheme implemented in this experiment.
Retrieval Results and Their Validation
The ICI package can be run in many different configurations. In this application,
ICI3 (Lavanant et al. 2000) has been run at a thinned resolution of 2 by 2 HIRS pixels
which is about 50km*50km. The retrievals are separated due to surface types (sea
and land) and cloud conditions (clear and cloudy). The cloud classification is based
on AVHRR cloud cover and AMSU precipitation flag. For monitoring the retrieval
results and their statistics, a few examples are given below. Figure 3 shows the
statistics between ICI retrievals and observations (NWP analyses plus RAOB
measurements) for a ten-day period for NOAA-16 ATOVS retrievals over China. The
average standard deviation for temperature profile is found to be about 2 Kelvin on 7
layers. Over land the standard deviation reaches 3K near the surface and is about 2K
on the 500hPa layer. Figure 4 shows the similar comparison but for humidity profile.
Summary
The ATOVS preprocessing and the physical retrieval method ICI have been
briefly described in this report. The retrieval results from ATOVS observation are
validated against Chinese NWP analyses and radiosondes. These results show good
agreement between ICI retrievals and conventional observations. Many Chinese
meteorologists are very interested in the retrieved atmospheric parameters for the
weather analysis and forecasting etc. The ATOVS radiances are also needed to
enable the assimilation of satellite data in a numerical weather prediction model. A
joint assimilation study of ATOVS data in the new generation Chinese NWP model is
underway between the Chinese Academy of Meteorological Science and NSMC.
References
Eyre, J.R. 1989. Inversion of cloudy satellite sounding radiance by non-linear optimal
estimation: theory and simulation for TOVS, J. Q .R. Meteor. Soc, 115,
1001-1037.
Klaes, K.D. 1997. ATOVS data processing in Europe, Technical proceedings of the
9th International TOVS Study Conference, 20-26 February, Igls, Austria,
267-274.
Lavanant, L. and P. Brunel et al. 2000. ICI Version 3 Retrieval Package, Technical
Proceedings of the 11th International TOVS Study Conference, September
20-26, Budapest, Hungary, 203-205.
Fig.1: AVHRR Cloud Cover in HIRS/3 Field of View
Forecast
Analysis
RAOB
Creating Rolling
Library
Level 1D
Guess
Inversion
Collocation
Cloud Cover
Tsurf
Validation
Retrieval
Files
Tuning
Visualization
Fig.2: The system schematic diagram of ICI inversion processing
Fig.3: The statistics between temperature retrievals and observations (NWP
analyses plus RAOB measurements) for a ten-day period for NOAA-16
Fig.4: The statistics between humidity retrievals and observations