Moderate-Resolution Imaging Spectroradiometer (MODIS) Science Team Meeting
- David Herring (herring@ltpmail.gsfc.nasa.gov), Science Systems and Applications, Inc.
The May 1 - 3, 1996, MODIS Science Team Meeting was called to order and chaired by Vince Salomonson, team leader. Salomonson introduced and welcomed the new Science Team Members. (See the section below for details on the new members.)
Robert Murphy, MODIS project scientist, stated that the team needs to develop plans and increase efforts in the three following areas: 1) validation, 2) defining the characteristics of a technologically advanced MODIS, and 3) outreach. Murphy announced that a first draft of the MODIS Validation Plan is complete and will be revised over the next 3 months. He encouraged the team to put as much attention and energy into the MODIS validation efforts as possible.
Regarding plans for the advanced technology MODIS, Murphy reported that the EOS Program Office is discussing options. He said that we need to find ways of reducing the size and weight of MODIS. However, we need to also preserve and build upon the science of the current MODIS. Murphy asked the team to consider what lessons it has learned so far in the design and development of the current MODIS.
Murphy said there is a need for the MODIS Team to step up its outreach efforts. Specifically, it must work more closely with the general scientific community to help refine needs for its data products.
Richard Weber, MODIS project manager, reported that the integration and testing of the MODIS Protoflight Model (PFM) is underway. All optics and focal planes are assembled. By June 1996, Santa Barbara Remote Sensing (SBRS) will have the Spectroradiometric Calibration Assembly (SRCA), the Solar Diffuser Stability Monitor (SDSM), and instrument electronics assemblies in place. Weber stated that he expects the PFM to be delivered in December 1996; however, this assumes an "optimistic" schedule, with no major problems. He announced that the MODIS Flight Model-1 (FM-1) components are also being prepared now.
Tom Pagano, of SBRS, stated that some major events took place on the PFM since the last Science Team Meeting. SBRS realigned the optical system, completed vibration testing, and has almost finished integrating the onboard calibrators. Pagano proffered that MODIS has the most sophisticated focal planes for radiometry ever built in the remote sensing industry. (There may be detectors with higher resolution, but not with the combined dynamic range and sensitivity.) SBRS is currently working on characterizing the bidirectional reflectance distribution function (BRDF) of the solar diffuser.
Pagano reported that the PFM signal-to-noise ratio meets specifications for all bands. However, the dynamic range is a concern for the near-infrared detectors-their throughput is higher than specified and will affect the dynamic range. Pagano stated that the PFM meets specifications for radiometric accuracy, on board blackbody, and instrument internal temperature.
Pagano stated that the near-field response of MODIS compares very favorably with that of SeaWiFS, CZCS, and AVHRR. MODIS has an intermediate field stop that reduces far-field response over and beyond what its heritage sensors could do. Pagano stated that stray light analyses show that contamination dominates at level 400; however, MODIS must maintain a clean- room condition at level 300.
Ed Masuoka, Science Data Support Team (SDST) leader, reported that the MODIS beta software delivery was made by the January 1996 deadline. Focus has now shifted from integrating the software into the DAACs, to science integration-the code's ability to handle ancillary data.
Masuoka told the Team that the deadline for submission of Version 1 code is January 1997. The goals for that submission are to have the science algorithms implemented according to plans given in the ATBDs. Also, the code must use ancillary and Look-up Table data, and employ realistic resource usage, timing, and operations algorithms.
Masuoka stated that EOSDIS cost growth is a concern-it is currently $75 million over budget. Most of the extra cost is attributed to the cost of hardware for the "pull side"; i.e., robotics and storage media to support the user community. Masuoka hopes that better characterization of the "pull side" requirements will enable ESDIS to cut costs in areas where there will be lower demand. Also, SDST is working with ESDIS on ways to optimize MODIS code to improve its performance and lessen its processing requirements.
Al Fleig, of SDST, stated that it is the Science Team's responsibility to test the science content of their code, as well as perform its implementation. He stressed that not all testing will be done by SDST; most will be done by the Science Team members themselves. When the code is delivered to the DAAC, SDST will make sure that it runs properly in the DAAC environment. Fleig stated that SDST will write the software test plan, but it will need input from each Science Team member. The idea is to test all features of the science algorithms as an integrated process and as a total processing system.
Fleig said SDST plans to send the Science Team data sets that will stress their code in some ways. For instance, a data set will be provided in which each detector (there are 430 detectors on MODIS) is periodically labeled as a "bad detector." Algorithms will need to check for this label and decide what to do to compensate. Fleig also pointed out that the MODIS Characterization Support Team (MCST) plans to flag "noisy" detectors. So, in another of its test data sets, SDST will label some detectors as noisy, to which the Version 1 code is expected to be able to respond.
Fleig proffered that the quality assurance (QA) effort after launch will consume 20-to-50 percent of each algorithm's processing resource requirements. He said that the QA algorithms must be available at launch, so development is needed now. Fleig recognized that the QA effort will tremendously impact funding, and reminded the Team that the MODIS QA Plan is due this fall.
Bruce Guenther, MODIS Characterization Support Team leader, reported that Version 1 of the Level 1B software has been delivered, and Version 2 will be delivered in early 1997. The file specification was updated on April 8, 1996, and is now frozen. The new file specification for Version 2 is in progress and will be frozen in December 1996. Guenther feels it will remain frozen until launch.
Janet Campbell, U. of New Hampshire, presented an overview of her proposed research efforts. Her goal is to develop the scientific and statistical basis for monitoring algal pigments and primary productivity in coastal, estuarine, and inland ecosystems using satellite data and complementary surface measurements. Campbell's primary research objective for MODIS is to establish a protocol for developing and validating regional or site-specific algorithms for estimating surface chlorophyll-a concentration and primary productivity, while accounting for the optical variability of other water constituents.
Bo-Cai Gao, U.S. Naval Research Laboratory, stated that his proposed research includes thin cirrus detection and correction, radiative transfer modeling, and airplane contrail cirrus studies. Gao's research on atmospheric corrections is primarily of interest to the MODIS Ocean and Land Discipline Groups. He showed sample image data taken over Coffeyville, KS, and pointed out that in the 1.375-µm channel, cirrus clouds and not surface features are seen. MODIS will have the 1.375-µm channel for cirrus corrections. Gao proposes developing techniques for the operational removal of thin cirrus from MODIS data acquired over both ocean and land.
Ranga Myneni, Boston U., showed a 1995 data plot of average normalized differential vegetation index (NDVI) anomaly as compared to biospheric carbon and sea surface temperature. Myneni stated that, globally, there is some correlation between the three variables up until about 1990; afterwards, they do not appear as closely correlated. He hopes to perform more-intensive calculations of these variables using MODIS data, rather than produce more data plots. Specifically, he plans to derive leaf area index (LAI) and fraction of photosynthetic active radiation (FPAR) absorbed by green vegetation. Myneni will then develop a look-up table algorithm for estimating LAI and FPAR for a given MODIS scene. Ancillary data layers will include biome type, such as grasses and cereal crops, shrubs, broadleaf plants, needle forests, etc.
John Townshend, U. of Maryland, stated that he is interested in land cover characterization and monitoring land cover change. He noted that the currently planned MODIS land cover data set is based on multispectral and multitemporal data using a neural net approach. He proposes to enhance this product by making an at-launch product available using AVHRR data, and by creating additional planes of land cover characterizations depicting continuous fields, all based on the AVHRR sensor's data. Consequently, land cover change results will be available shortly after EOS AM-1 is launched, rather than during the second year after launch as was originally planned. Specifically, Townshend plans to monitor land cover change, showing where change occurs, and what sort(s) of change(s) occur, every 1 to 2 months.
Eric Vermote, U. of Maryland, presented an overview of his proposed land surface reflectance product. Vermote said that his algorithm will be used for atmospheric corrections; in turn, his products are heavily dependent upon the aerosol product. Vermote stated that his reflectance product will be an important input into many of the Land Group's algorithms-the quality of his product will influence the quality of products downstream in the processing flow. From a heritage perspective, Vermote points out that his product contains new features, such as aerosol correction, adjacency effect correction, and correction for BRDF-atmosphere coupling-all of which have never been included before in a surface reflectance product. Vermote also hopes to provide atmospheric correction for coastal water reflectance. His product will play an important role in validating the calibration of the MODIS reflectance bands.
Additionally, Vermote will work on development of the MODIS aerosol climatology product prior to launch. He feels that aerosol correction is the most important source of error in the surface reflectance algorithm.
Paul Menzel, U. of Wisconsin, told the Team that he would like to take advantage of MODIS' 36 spectral bands but there is currently no adequate data set. So, his team at the University of Wisconsin-Madison is working to simulate MODIS using MAS (MODIS Airborne Simulator) data. (He noted that MAS has no water vapor absorption channels, which is a problem.) The purpose is to produce a synthetic data set to facilitate and enhance MODIS algorithm development. Menzel explained that the synthetic data sets will be used for testing algorithms and strings of algorithms. He plans to use real data where possible and approximate the co-registration of MODIS. Output data will be Level 1B and geolocated, and will provide the best possible radiometric calibration.
Menzel explained that the synthetic data set is being put together as if the 50-m MAS footprint equals a 250-m MODIS nadir footprint. In the first quarter of 1996, a data set of clear sky over water scenes was produced. In the second quarter, a cloudy scenes data set over water was produced and in the third quarter, clear scenes with limb-corrected infrared data will be produced at true MODIS spatial resolution. Production in the fourth quarter is still to be determined.
Al Fleig said that the Team needs to generate a QA Plan by September 1996. Fleig has agreed to produce a draft document stating the purposes and objectives of the MODIS Quality Assurance Plan, as well as how QA will be used.
Robert Wolfe, of SDST, reported that progress has been made toward establishing the temporal compositing periods for MODIS data-consensus was for an 8-day period. There was some question as to whether the Team should simply resynchronize at the beginning of each month, or each year. Wolfe announced that the majority of Team members favored resynchronizing each year. That way, data users can compare the same periods for each year over a given region. Esaias stated that the Ocean Group favors the 8-day compositing grid, but it doesn't want the grid resynchronized ever.
Bill Barnes, MODIS instrument scientist, told the Team that there is interest in building a smaller, lighter version of MODIS for future missions. In scoping the new MODIS, he stated that emphasis will be on maintaining all of the requirements for the current MODIS. Barnes said he is putting together a specification for the follow-on MODIS sensor. This specification will include the same channels, radiometric requirements, and calibration requirements as the current MODIS; the difference will be in size and weight constraints.
Phil Slater, Calibration Discipline Group leader, summarized discussions at the MODIS Calibration Working Group meeting. He stated that he is concerned about the truncated test and calibration schedule at SBRS. He suggested that they should endeavor to use any opportunities to study the long-term stability of MODIS in the pre-flight phase. He also feels that they should re-expand the thermal vacuum testing that was cut back to 15 days.
He said there is a need to accurately determine the BRDF of the solar diffuser. He pointed out that this test was canceled at SBRS, but feels that there is still a need to characterize the solar diffuser. Slater requested that SBRS measurements of BRDF be compared to those of other institutions.
Slater recommended that SBRS study the feasibility of measuring far-field stray light effects to validate a basis for determining Level 1B radiometric uncertainties. For the ocean color bands, Slater noted that Esaias's study shows that about 56 percent of "clear" ocean pixels will have a scene-dependent error greater than 1 percent. Slater suggested that MCST should study, in consultation with the Science Team, the provision of an estimate of the radiometric error for the Level 1B product.
Slater feels that SBRS should employ the SRCA more frequently during testing and calibration activities to check the long-term stability of the SRCA and MODIS as a whole. He said SBRS should also perform system-level tests of stray light when the SRCA is in use.
Slater announced the upcoming first joint vicarious calibration field campaign, to be conducted at Lunar Lake and Railroad Playa, Nevada, from May 30 to June 7. The purpose is to compare TOA (top of atmosphere) radiances predicted by the various participating groups when measuring the same playa area at the same time. Up to three such estimates will be made each day to simulate the acquisition times of the AM-1 platform sensors at the solstices and equinoxes. MODIS, MISR, and ASTER calibration scientists will participate in the campaign.
Salomonson said that the report from SBRS on MODIS development was exciting and positive. He thanked SBRS and the MODIS Project for their efforts. He noted that algorithm development is going well and that the beta and Version 1 code delivery experience has been positive. He told the Team to expect challenges in data processing and storage requirements in the coming months.
Salomonson announced that the dates for the next MODIS Science Team Meeting are October 9 - 11, 1996, at a site near Goddard Space Flight Center.
[Table of Contents]
[Previous]
[Next]