|
|
GOES
R Hyperspectral Sounder Data Compression
This web page contains access to presentations on compression of atmospheric data presented at:
1. A first
Data Compression meeting held 25 January 2002 at the facilities of NOAA NESDIS, in Suitland MD and
2. A Second Data Compression meeting held May 22, 2003 at the facilities of both NOAA and NESDIS in Silver Spring MD.
The work explores methods of compressing the next generation Hyperspectral Sounder data of the GOES R satellite series.
GOES R the first of a new satellite series is to be launched in the 2012 time
frame. Planning for this new series is underway at NOAA and NASA. Though the
requirements for
GOES R are still evolving, historically the two prime sensors on GOES have been
the imager and sounder. With GOES R, both sensors will change to provide increased
performance. One characteristic of these new sensors is a significantly increased
data rate. This applies to the raw data rate that is first sent down to the ground
from GOES for processing as well as a processed data that is “globally” broadcast
out from GOES out to the users.
The present GOES series transmits its raw imager and sounder atmospheric and
ground
data measurements to NOAA’s ground site at Wallops, Virginia where it is calibrated
and given navigation grids. The data are then transmitted back up to the GOES
satellite from Wallops, VA by NOAA and then “globally” broadcast from GOES satellites
located at 75W and 135W. This is the present position of the satellites. This
flow of data was shown in the January 25, 2002 introductory talks. The data are
broadcast out to points on the ground to about a 0 to 5 degrees nominal elevation
angle to each satellite.
The future GOES R sounder was identified as the Advanced baseline Sounder (ABS)
and the imager as the Advanced Baseline Imager (ABI), at the time of the January
25, 2002 meeting. The ABS is referred to in the 25 January 2002 presentations.
Since that 2002 time period, NESDIS has moved to recognizing this “ABS” sounder
as part of a suite of instruments and as such has renamed the sensor the “Hyperspectral
Environmental Suite” or “HES”.
NASA through its “New Millenium Program” (NMP) is undertaking a program to develop
new earth observing technologies to support earth research and operational observations.
And as such is supporting their goals and NOAA in the form of a next generation
high
fourier transform spectral (FTS) resolution sounder technology development program.
The sounder being developed by the NMP program is identified as the Geostationary
Infrared Fourier Transform Spectrometer (GIFTS). It is a Michelson Interferomter.
GIFTS is referred to in the talks. NOAA will fly either an FTS or a grating spectrometer
sounder; and will draw heavily on the NASA GIFTS and EOS Aqua AIRS programs.
Today, in
the June 2003, GIFTS, HES, and ABS data rates are judged to be about 65 Mbps.
The current GOES 8-12 series imager and sounder sensors together have a raw unprocessed
data stream of about 2.6 Mbps and a processed data rate of about 2.1 Mbps. This is a
relatively small data rate compared to that expected for the GOES R series.
The GOES R series could have a combined 130 Mbps of raw data for both the imager and sounder,
and as much as 85 Mbps for the combined processed data of the imager and sounder. The GOES
R requirements have evolved between the January 25, 2002 and May 22, 2003 meetings; and as
such the data rates have increased.
A particular concern with the new higher ABI and ABS/HES data rates is the impact on future
GOES R spacecraft design in terms of additional power, mass, and cost to broadcast rates of
as much 85 Mbps. Of concern also is the limited available L band RF spectrum used for global
broadcasts. This cannot handle more than 32 Mbps. See the May 22, 2003 presentations for
the latest discussion of this.
The Cooperative Institute for Meteorological Services (CIMSS) is a unit of the Space Science
and Engineering Center (SSEC) of the University of the University of Wisconsin, located in
Madison Wisconsin. Further, CIMSS is a joint university of Wisconsin, NASA, and NOAA Cooperative
Research Institute. CIMSS conducts research on weather and climate through the use of satellite
observations and remote sensing. There has been a long history of collaborative research among
NOAA, NOAA-NESDIS, NASA, and CIMSS at the University of Wisconsin-Madison.
Two scientists at the University of Wisconsin-Madison, Dr. Allen Huang and Dr.
Paolo Antonelli published a paper in the March 2001 Journal of the American Meteorological
society titled: “Application
of principal Component Analysis to High-Resolution Infrared Measurement Compression and Retrieval.”
Their research indicated that data compression might be able to significantly reduce the volume of
GOES R sounder data. Further the quality of that data might be sufficient for National Weather
Service model prediction and its weather forecast offices’ use. Their research at that time focused
on the application of a mathematical technique known as “Principal Component Analysis” (PCA
or PC). CIMSS research has expanded since that paper to include other mathematical
approaches such as
wavelets and hybrid wavelets, etc.
Based on the work identified in their paper, Tim Schmit Senior Scientist of the Office of Research
and Applications (ORA) and Roger Heymann Senior Engineer of the Office of Systems Development (OSD)
undertook a joint research project to explore the development of acceptable data compression methods
for the GOES R era sounder. Richard Reynolds head of the OSD Ground Systems Division has also been
a major supporter of this work.
A 25 January 2002 meeting was a first meeting to begin a dialogue and possible
research to evaluating available data compression techniques to reduce the future
sounder’s data rate. A series of
presentations were presented at this January 25, 2002 meeting. Two particular
classes of compression techniques were identified for discussion at that meeting.
They are Principal Components (Karhunen
Loeve Transforms - KLT) and Sub-band Filters such as Wavelet Transforms and Modulated
Lapped
Transforms- MLT.
After the January 25, 2002 meeting a hyperspectral data compression team was formed. This compression
team has worked continuously to the present, as of June 2003, and will likely be working for the
foreseeable future 1-2 years or more on this. The team has met over the past 16 months through conference
calls with its members working at their own respective facilities. Their work is reflected in the
presentations of a May 22, 2003 second meeting. Also presented at this meeting were some papers from
other individuals/organizations doing related work in the field.
Provided are the agenda and talks of the first meeting January
25, 2002 and the second meeting of
May 22, 2003.
|
|
|
|
