The Earth Observer

The AIRS team held a very informative meeting at Santa Barbara, CA on February 27-29, 1996. Following is a brief summary of the meeting presentations.

The next AIRS team meeting will be held on June 25-27, 1996 in the NOAA Building #3 in Silver Spring, MD. Details will be provided to AIRS team members via e-mail about a month before the meeting.

General Information

Mous Chahine, AIRS Science Team Leader, reported on the EOS status.

1. Restructuring is anticipated for the data processing system (EOSDIS) starting with the EOS PM-1 mission. This restructuring may change the organizational responsibility for and/or physical location of the operational (routine) AIRS data processing facility. The restructuring will not change the responsibilities of the AIRS Science Team or the Team Leader Computing Facility with respect to delivering the algorithms and the operational retrieval software system, developing new data products, and making sure that the data processing generates valid data.

2. The New Millennium technology development of QWIP detectors, MMIC radiometers, and small, 45 K 0.5 W coolers will benefit a lightweight follow-on Integrated Multispectral (infrared/microwave) Atmospheric Sounder, IMAS.

3. IMAS specifications include the following functional requirements related to science:
   • specifications for the infrared wavelength coverage, spatial and spectral resolution, signal-to-noise ratio, and calibration accuracy of IMAS will be the same as the AIRS specification.

   • the microwave channels of IMAS will include the 54 GHz channels of the AMSU-A, but the 24 and 31 GHz channels will be deleted. Instead we will have about six 118 GHz channels (picked by Staelin) to replace the current single 150 GHz channel of AMSU-B. There will be about six channels in the 183 GHz region. The 118 GHz and 183 GHz channels will have the same footprint as the infrared channels (1.1 degree), the 54 GHz channels will have a 2.5 degree footprint (all dimensions refer to the full-width-at-half-peak). Since the infrared and microwave channel footprints are all scanned with the same scan mirror motor (the antennas in the current concept are on opposite output shafts of the motor), the spatial co-alignment of the infrared and microwave channels should be better than what we can do with the separate AMSU, AIRS, and MHS units.

4. Details of the Brazilian offer to build a copy of the AMSU-B for the EOS PM-1 spacecraft are being worked out. The proposed reconfiguration of an AMSU-B copy with the 150 and 183 Ghz channels, but without the 89 GHz channel, has been accepted by the EOS PM project.

5. LORAL has been bought by the Lockheed/Martin group, pending some antitrust investigations.

Fred O'Callaghan, AIRS Project Manager, reported on the status of the AIRS hardware and presented some details of the IMAS design concept. All major subcontracts for AIRS are in place. The Engineering Model (EM) CDR (Critical Design Review) was held in March 1996, the CDR for the PM-1 model will be in January 1997. Testing of the EM will be completed by the middle of 1997.

The IMAS will use new technology (QWIP, MMIC, small 45 K 0.5 W cooler, SiC lightweight structures and, possibly, information-preserving data compression) to combine the functionality of the AIRS, AMSU, and MHS into an integrated system. As much as possible the IMAS will build on the experience with and possibly component inheritance from AIRS. The goal is to fit the IMAS into an 0.8 x 0.9 x 0.4 meter volume, mass 110 kg, 100 W power, and 0.3 Mbps data rate.

Bjorn Lambrigtsen, JPL AMSU/MHS instrument representative, presented the AMSU-A and MHS status.

• The EOS PM-1 AMSU instrument CDR and Calibration Peer Review were scheduled for April 16-18, 1996. The plan for the characterization of the AMSU antenna pattern was not adequate.

• There was a choice of accepting an AMSU-B copy without either the 89 Ghz or the 150 Ghz channel. Since the AMSU 89 GHz channel is available, and we expect by the year 2001 to have many years of NOAA experience in comparing the AMSU 3.3 degree and AMSU-B 1.1 degree 89 GHz channels, retaining the 150 GHz channel was the right choice.

George Aumann, AIRS Project Scientist, announced the start of the AIRS home page at http://www-airs.jpl.nasa.gov. The AIRS home page is part of an educational outreach program. The home page includes papers giving details of the design of the AIRS instrument, the retrieval algorithms and the data processing system. Science Team members are encouraged to check it out and submit comments and contributions.

Core Retrieval Algorithm Developments

Joel Susskind discussed improvements in the GSFC retrieval code, since its delivery to the Team Leader Facility at JPL on December 15, 1995. Uncertainty of ozone amount is now included in the AIRS noise covariance matrix. The latest retrieval revision also solves for surface emissivity as a function of wavelength at day and night. Many retrieval options can be selected via a name list.

Mitch Goldberg gave an update of NOAA/NESDIS's AIRS core algorithm development:

• The software, including principal component regression, was delivered to JPL on December 15, 1995. Evaluation of pseudo channels for the first guess solution is continuing. Larry McMillin, who designed the cloud-clearing module of the NOAA algorithm, pointed out that the accuracy of the cloud-cleared data is limited by the microwave channels of the AMSU.

• Work on adding rocketsonde extensions to the NOAA matchup profiles has been completed, but the data have not been transferred to JPL.

Allen Huang discussed the Spatial and Spectral Simultaneous Retrieval Analysis developed in collaboration with Bill Smith at the University of Wisconsin. Unlike the NOAA and GSFC algorithms, which remove the effects of clouds with a cloud clearing algorithm using a 3 x 3 footprint pattern, this algorithm attempts to account for the effects of clouds. Retrievals under partly cloudy conditions can thus be accomplished with a single AIRS footprint. Retrieval accuracy using a single footprint under clear conditions is 1 degree K, but performance degrades significantly under cloudy conditions. The algorithm delivered in December 1995 works with night-time data only. The algorithm has to be modified to handle the reflected sunlight for day-time data. The higher spatial resolution obtained from single-footprint retrievals (versus 3 x 3 patterns) makes this algorithm an interesting candidate for mesoscale research.

Phil Rosenkranz has tried a version of the "microwave first guess" algorithm using DMSP SSM/T and SSM/T2 data (54 Ghz and 183 Ghz) to demonstrate the feasibility of single-footprint water vapor profile retrievals. The ability of the algorithm to reproduce the different moisture patterns at different pressure levels is very encouraging. The AMSU-A and MHS data from the EOS PM mission will have the same frequency coverage as the SSM/T and SSM/T2, but the spatial resolution will be a factor of three higher.

Larrabee Strow discussed the status of the fast transmittance algorithm and the water vapor spectroscopy. CO2 is treated as a fixed mixing ratio gas in the rapid algorithm and a mixing ratio has to be selected. Mous Chahine pointed out that a change in the CO2 mixing ratio of about 5 ppm can be detected on a yearly time scale from the evaluation of tuning residuals. This concept has been tested with TOVS data.

Larry McMillin has started to develop the radiance-tuning algorithm for AIRS and demonstrated NOAA's current algorithm using HIRS2 data. Since the HIRS2 has 19 channels versus the AIRS 2400 channels, the approach needs to be modified. He needs AIRS simulated data for algorithm testing. George Aumann will develop a simulation concept.

Data Processing and Instrument Operations (DPIO) Developments

Denis Elliott, DPIO Manager, presented the status of the DPIO and the team algorithm plan. The first delivery of core algorithms from Science Team members was received on December 15, 1995. Integration of the algorithms into a single retrieval system is progressing. The AIRS retrieval system preliminary design review (PDR) will be held at GSFC in October 1996, in conjunction with an AIRS team meeting.

Ed Olsen presented details of the acceptance status and a first cut at resource requirements for the algorithms delivered to the Team Leader facility on December 15, 1995. NOAA, GSFC, and MIT made a full delivery; the University of Wisconsin delivered night-time retrieval capability only. Although the code was developed on four different processors, installation on a fifth processor, a SUN 1000/4 SuperSparc went smoothly. Memory and CPU resource requirements were evaluated. Relative CPU times per retrieval for the code as delivered by NOAA, GSFC, and Wisconsin are 1 : 3 : 50. Typical tropospheric retrieval accuracy under 50% cloud cover for the NOAA and GSFC code is 1 degree K rms for temperature and 10% for water profiles.

Sung-Yung Lee, Ed Olsen, Luke Chen and John Gieselman comprise the team working on restructuring the team algorithm, selecting the best modules, and designing The AIRS temperature and moisture retrieval algorithm. The first version will be available for testing in July 96. The software will ultimately be in Fortran 90.

Toni Palmieri and his team (A. Revilla, R. Davidson, S. Gaiser, E. Manning, B. Morrison, R. Oliphant, H. Stone, and B. Weymann) have developed a complete AIRS prototype system, called prototype 4.0. They demonstrated prototype 4.0 using simulated data sets, starting with data ingest of packets, conversion to Level 0 data, calibration coefficient determination, application of the calibration to create Level 1b data, and following with retrievals to create the Level 2 output data. The input data come from a mesoscale GCM model developed by E. Kalnay and co-workers at NOAA/NMC. Prototype 4.0 allows a choice between the NOAA and the GSFC retrieval as delivered on December 15, 1995. The first version of The AIRS temperature and moisture retrieval algorithm developed by Sung-Yung, was to installed in the system starting in April 1996. Access to the code by team members and/or their designated programmers (under strict configuration control) is expected to start this summer. Hands-on-training will be made available for programmers in special workshops.

Dave Gregorich reminded the team of the various simulated data sets currently on line. The next simulation will include clouds with spectral emissivity/reflectivity dependence and simulations using the NOAA matchups (once they are received from Goldberg).

Mark Hofstadter showed model calculations of emissivity as function of wavelength for altostratus and cirrus clouds. The calculations were made with a modified Ackerman model (U. Wisconsin). For altostratus the deviations from a typical 0.98 emissivity are smaller than the emissivity variations created by the current data simulation. For cirrus clouds (optical depth 0.6) the calculated emissivity may be as low as 0.3, but is also spectrally relatively flat. Hofstadter has also started to simulate visible channel data using the prescription developed by Gautier/Shiren at UCSB.

George Aumann presented details of the AIRS infrared radiometric and spectral calibration. The spectral calibration is now exclusively based on the upwelling spectral radiance. The data simulation in support of the calibration algorithm is being implemented by Evan Manning. Steve Gaiser is writing the calibration software. The IR calibration is implemented in the prototype 4.0 system demonstrated by Tony Palmieri.

Catherine Gautier and Jang Shiren, UCSB, are working on the absolute calibration concept of the visible light AIRS channels. The calibration will depend on vicarious calibrations over White Sands, where the physical conditions are reasonably well understood, using theoretical radiance computations. Over any large sandy region, 50 km or bigger, they plan to intercalibrate with MODIS. An AIRS visible light calibration approach document is in preparation.

Research Algorithm Development

Joel Susskind showed total ozone retrievals from NOAA 9-11 between 1985-1991, using the HIRS 9.6 um channel. The HIRS requires a guess of the emissivity (0.98 over ocean, 0.95 over land). Comparison with collocated SBUV data is very encouraging. The extension of this algorithm to ozone profile retrievals using AIRS is straightforward. The AIRS core retrieval algorithm explicitly solves for emissivity at several wavelengths in the 10 um window area. In addition to total ozone burden on a 50 km spatial scale, Susskind expects AIRS to be able to distinguish between 2-3 layers of ozone in the upper troposphere and stratosphere.

Paul Van Delst and Hank Revercomb, U. Wisconsin, are developing ozone weighting functions using a Fast Jacobian calculation. This effort could directly support Joel Susskind's ozone retrieval algorithm.

Catherine Gautier is continuing to develop a cloud-altitude retrieval algorithm using visible data. The algorithm uses the AIRS3/AIRS1 wavelength ratio. This effort directly supports the diagnostic effort expected for dealing with low clouds in the infrared. Cloud height determination with +-200 m accuracy may be possible.

|Table of Contents| |Previous| |Next|