--Reinhard Beer,
beer@caesar.jpl.nasa.gov,
Jet Propulsion Labratory
After a general welcome from our Harvard colleagues (Daniel Jacob and Jennifer Logan), Reinhard Beer (PI) gave a brief outline of what we hoped to accomplish during the course of the meeting, the primary emphasis being (as always) on the algorithm and Algorithm Theoretical Basis Document (ATBD) developments.
The new CHEM Platform Scientist, P.K. Bhartia, then outlined the events of the past several months with particular attention to the "ESSP Mode" studies, and the CHEM platform science and overall program balance review by M. Prather's board. The general finding of the first set of studies showed that the baseline CHEM mission (4 instruments on 1 spacecraft) was, by a large margin, the cheapest approach. The Prather board endorsed this finding, in particular as a means of freeing up funds for the R & A program which has seen significant declines in recent years. Other issues raised by P.K. were the paucity of funding for cal/val and the fact that the activities that are ongoing need better coordination. He also announced that a CHEM platform experimenters group is being formed and will meet for the first time in Boulder in early August.
The JPL Project Manager, Tom Glavich, then provided an update on the TES instrument development, which is proceeding well. The next major milestone is the Preliminary Design Review (PDR) early next year. This discussion led naturally into the ATBD and algorithm schedule. While the Level 2 algorithm is undoubtedly the largest task, Level 1 has the highest priority because it is needed for Engineering Model testing beginning in June 1998.
Helen Worden provided a progress report on the Level 1B algorithm, where it was recently demonstrated that correction for self-apodization of off-axis pixels is feasible and successful. This is essential because the Science Team has requested that all spectra be put onto a common frequency and spectral resolution grid both for purposes of inter-comparison and to reduce the computational burden at Level 2. She was followed by Kevin Bowman, who has been investigating wavelet decomposition as an alternative approach to determining phase functions.
After lunch, a special briefing on TES was given to representatives of AER, Boston University, and local educators. Our first efforts at an outreach program are being made in the Boston area, and the team was very impressed by the enthusiasm shown for programs in atmospheric science in which students can participate. These programs currently target secondary school students but will soon apply to lower grades as well.
Tony Clough then reported on his January meeting with Clive Rodgers and Reinhard Beer (held under trying circumstances of high winds and consequent power outages) in which a detailed outline of the Level 2 algorithm was generated. This has proven to be a most valuable tool for setting up tasks and priorities. Following this, he discussed updates to a general strategy document that was generated about a year ago. A number of action items arose from this discussion.
The following morning discussion was begun by Ming Luo, describing the progress being made on the prototype Level 2 algorithm. The forward (radiative transfer) model is nearly complete, and work is beginning on the actual "retrieval" segment. We expect to have a functional version of the entire algorithm ready for testing by October of this year. Rob Toaz then discussed the actual C language implementation of the algorithm, with particular emphasis on the approaches used for temperature interpolation of the precomputed absorption coefficient (one set for each preselected pressure). It is clear that there is a strong trade-off between memory requirements and accuracy (e.g., 3-point Lagrangian interpolation is much better than 2-point linear but requires 50% more memory).
Pat Brown and Tony Clough then presented their on-going studies on limb pointing retrieval (using LBLRTM). They are investigating the errors associated with the spatial convolution of the radiance profile with a pixel model as a function of the number of rays (each requiring a forward model calculation) and the interpolation scheme.
The decision to use C and C++ for deliverable code was confirmed by the team, although FORTRAN inputs to the SCF at JPL will continue to be accepted.
Steve Larson discussed the logistics of storing and accessing the absorption coefficient tables (a non-trivial problem). He was followed by Jennifer Logan and Daniel Jacob, who provided the team with an updated list of trace species of atmospheric interest, whose feasibility for retrieval as Special Products should be investigated.
In a break from tradition, the meeting adjourned for the afternoon to permit R. Beer to give a special seminar to the Harvard faculty and students: "Airborne Emission Spectrometer (AES) results from the Southern Oxidants Study," a field campaign that took place in the summer of 1995. This was the first opportunity for many individuals (including some TES team members) to see the outcome of real retrievals on real remotely-sensed data, including some preliminary comparisons with in situ and sonde data.
The last morning began with a paper by Curt Rinsland, describing his retrieval of information from ATMOS tropospheric spectraa data set which has scarcely been examined. He has successfully extracted profiles of a number of new species (including HCN) but, particularly, has seen strong features attributable to cirrus clouds. Many of these characteristics should be readily seen in TES data.
Helen Worden and David Rider went into much greater depth on AES/Southern Oxidant Study (SOS) retrieval accuracy and error analyses than was possible in the previous day's seminar. While the temporal and spatial overlap between our results and those of other participants is quite small, where comparisons can be made they seem generally satisfactory, although some puzzling discrepancies remain among all the techniques employed (not just AES). The investigation of these differences continues.
The final two scientific presentations were made by X. Liu who is attempting to employ Singular Value Decomposition both as a means of noise reduction (by eliminating the higher order principal components) for smoothing black body calibration data and also as an interpolation scheme for absorption coefficient tables. The first is not obviously applicable to TES and the second may be quite risky and, in any case, offers no savings in computation time over direct interpolation (although it may offer mass storage savings).
The meeting closed with a brief overview of the highlights of the 7th Atmospheric Science from Space using Fourier Transform Spectrometry (ASSFTS) Workshop, held this year in Oberpfaffenhofen (near Munich) in May. As always, this is our best opportunity to learn of the progress of our international colleagues and to exchange views on many topics. This year's theme was "Calibration and Characterization," and one of the outcomes was a proposal to request the International Radiation Commission to sponsor a working group to arrive at a generally-agreeable algorithm for determining Noise Equivalent Spectral Radiance (NESR). Some existing methods give results that can differ by factors of 2-4, yet it is a critical performance parameter for any remote sensor (not just infrared FTS systems). The next workshop will probably be held in the south of France in October 1998, although some logistical issues remain to be worked out.
It was agreed that the next Science Team meeting would be held in Boulder (at NCAR if possible), probably October 7-9.