The MISR Science Team met in December 1998. The MISR Principal Investigator, Dave Diner of JPL, welcomed the meeting participants and outlined the meeting objectives, which were to review the status of the Terra spacecraft, MISR and AirMISR instruments, and MISR-2 Instrument Incubator efforts; review the status of the ground data-processing software; discuss product verification and quality assessment plans; discuss retrieval results obtained using AirMISR imagery; and demonstrate operational software and data analysis/visualization tools to the team.
The first session of the meeting was a discussion of AirMISR instrument and data-processing status. Carol Bruegge of JPL described the calibration and radiometric performance of the instrument, and Veljko Jovanovic of JPL described the georectification and geometric calibration of AirMISR data. Radiometrically, the instrument is meeting the established requirements. The primary issue with geometric calibration is the need to remove static offsets in the instrument orientation relative to the ER-2 navigation system. Bill Ledeboer of JPL talked about the software system that is used to process AirMISR imagery from raw flight data to calibrated, map-projected, and co-registered images. Jim Conel of JPL discussed flight plans for AirMISR for FY 1999, which include a vicarious calibration experiment that was conducted over Rogers Dry Lake in early December, for which an excellent field data set along with multiple AirMISR flight runs was obtained, plus planned flights over dark water (for aerosol studies), vegetation targets, and cloud fields later this year. Jeff Privette of GSFC gave an overview of plans for the EOS Validation Core Sites.
The next topic for discussion was MISR cloud mask techniques and a comparison with approaches used by other instruments, such as MODIS. MISR will use a variety of methods, including an automated, continuously updated histogram generation and analysis technique for establishing dynamic radiometric thresholds, plus stereoscopically-based cloud detection, and a multi-angle method for high-cloud detection. Using AVHRR as test data for the radiometric cloud-detection approach, Eugene Clothiaux of Pennsylvania State University and Larry Di Girolamo of the University of Illinois reported that discrepancies between MISR and MODIS algorithms were observed in 18-19% of the data. This needs to be followed up with actual data where longer time intervals will be available to improve the statistics used in developing the MISR thresholds, and more spectral bands will be available to MODIS. Over Greenland, where the radiometric thresholding technique is expected to have difficulties distinguishing snow from cloud, the MISR stereoscopic technique has been tested on Along Track Scanning Radiometer-2 (ATSR-2) data. Peter Muller of University College London reported a clear separability between the cloud layer and the underlying snow surface using this approach.
A discussion of aerosol retrievals followed. Ralph Kahn of JPL described sensitivity studies for MISR aerosol retrievals over ocean that indicate the ability to retrieve optical depth to an accuracy of 0.05 or 20% (whichever is larger), to distinguish among several size classes and compositional groups (characterized by refractive index), and to distinguish between spherical and non-spherical shapes. John Martonchik of JPL described his application of the heterogeneous land aerosol retrieval algorithm to AirMISR data acquired over Moffett Field, CA. Although optical depths that were similar to the ground truth values were retrieved, the data at high off-nadir angles in the blue and green bands appeared to require an unreasonable amount of absorption to be included in the aerosol model. The model results are sensitive to the pointing calibration at the oblique angles, and this is being investigated. Stu Pilorz of JPL then discussed the algorithmic approach he is using for building validation software that retrieves aerosol optical properties from field measurements. Wedad Abdou of JPL described lessons learned from a vicarious calibration experiment applied to AirMISR data. She also discussed the clear-sky anomalous absorption problem, in which certain observers have not been able to obtain consistency between direct field and diffuse field measurements of sunlight and skylight at the ground. She discussed the possibility of small particles as a possible explanation and follow-up studies are planned.
The next session focussed on surface retrievals. Bernard Pinty of the Joint Research Centre described the use of an algorithm based on the MISR surface retrieval approach that has been proposed for operational reprocessing of Meteosat data to retrieve surface albedo and aerosol optical depth simultaneously. Yuri Knyazikhin of Boston University discussed application of the MISR LAI/FPAR algorithm to POLDER data and the benefit of multi-angle viewing under LAI saturation. Peter Muller described a study examining the effect of surface albedo changes on climate forecasts with the UK Hadley Centre GCM, along with a comparison of POLDER-derived albedos with GCM 10-year mean forecasts.
The second day of the meeting began with a session on cloud stereo and albedos, in particular results from the AirMISR deployment over the SHEBA site in Alaska during FIRE III. Catherine Moroney of the University of Arizona first described several performance improvements that have recently been made to the MISR cloud-height stereo retrieval algorithms. Peter Muller then discussed application of the MISR stereo retrieval algorithms to ATSR-2 data. Using land surface topography for test purposes, differences between the stereo-derived elevations and an elevation map show the standard deviation of differences to be about 1.4 km for the ATSR-2 1-km nadir/1.6-km off-nadir data. (For MISR, stereo matching will be done at 275-m resolution.) Roger Marchand of Pennsylvania State University then presented results from the AirMISR Alaska deployment. Stereo retrievals over a cloud deck on June 3, 1998, show a mean cloud height of about 3 km, which agrees well with lidar results. Nadir radiances also agree well with the MODIS Airborne Simulator (MAS) radiances, but the radiances show a systematic offset from data acquired with the University of Washington Cloud Absorption Radiometer (CAR) on the CV-580 aircraft. Potential causes are under investigation. Tamas Varnai of the University of Arizona showed good agreement between cloud models and the AirMISR BRFs, however, the corresponding albedos are somewhat higher than expected. This is also being investigated.
Next, a discussion of the status of the MISR and the EOS Terra spacecraft was presented by Tom Livermore, the MISR Project Manager at JPL. Upgrades to the MISR flight software have been made to rectify several anomalies observed in earlier testing at Valley Forge. Susan Barry of JPL talked about MISR instrument operations and requested inputs from the Science Team on finalizing the Local Mode site list. Graham Bothwell of JPL discussed the status of the science data system along with the schedule for delivery, integration, and test of the at-launch version of the ground data-processing software.
Thomas Pagano of JPL presented the status of development of the MISR-2 prototype camera to be built under the Instrument Incubator Program. Plans for the MISR-2 cameras are to use a new optical design and chip-on-board electronics to reduce instrument size, while simultaneously including InGaAs detectors to extend the spectral coverage to 1.6 micrometers. Dave Diner informed the team of the status of NASA’s post-2002 mission planning.
Amy Braverman of UCLA talked about the approach that MISR is planning for its Level 3 global products. Presently, two types of products are planned, component and joint. Component products result from the processing of individual Level 2 Product Generation Executables (PGEs), whereas joint products will combine results from multiple PGEs. For the latter, a technique known as Entropy Constrained Vector Quantization will be used to summarize the results into a global map, while retaining covariance information across parameters. Mike Smyth of JPL described some of the design considerations being addressed in generating the Level 3 products.
Next, a session on quality assessment (QA) and verification of data products was held. Bob Vargo of JPL gave an overview of the process, and Ralph Kahn described the team’s approach to QA. Jia Zong described the verification approach for Level 1B2 products, and Larry Di Girolamo, Roger Davies of the University of Arizona, and John Martonchik discussed cloud mask, top-of-atmosphere/cloud product, and aerosol/surface product verification, respectively. Presentations by Scott Gluck and Carol Bruegge of JPL on Level 1A and 1B1 products were submitted for inclusion in the meeting notes but were not presented, in the interest of time. Earl Hansen of JPL led a discussion of the early mission data needs of the team.
The last day of the meeting focussed on demonstration of MISR operational software and visualization tools. Brian Chafin of JPL demonstrated the In-flight Radiometric Calibration and Characterization subsystem and Veljko Jovanovic, Mike Bull, Mike Smyth, Jia Zong, and Lisa Barge of JPL demonstrated the Geometric Calibration subsystem. Bill Ledeboer showed the AirMISR data processing system. Various visualization and analysis tools were demonstrated by Pranab Banerjee, Jeff Hall, Charles Thompson, Barbara Gaitley, and Robert Ando of JPL. Jim Galasso of the Langley DAAC demonstrated aspects of the EOSDIS Core System (ECS) processing, and Kyle Miller, Susan Paradise, and Kathleen Crean, assisted by Scott Gluck, Rick C. de Baca, David Nelson, Mark Apolinski, and Ruth Monarrez (all of JPL) demonstrated the Level 1 and Level 2 data-processing subsystems. A QA-based data query system being developed for MISR by the ECOlogic corporation was presented by Jennifer Carle, Renu Chaudhry, Guizhong Chen, and Scott Henderson of that company. Animations being developed in support of MISR outreach were presented by Eric de Jong and Shigeru Suzuki of JPL.
The meeting concluded with a presentation by Ralph Kahn on education and outreach, followed by a feedback and wrap-up discussion led by Dave Diner.