Introduction
During the third week in February, some 80 oceanographers, engineers, computer scientists, and program and project managers within the international Earth science community held a 3-day meeting at the University of Miami to discuss the future strategy for handling global ocean color remote sensing data from multiple platforms. Co-chairs for the meeting were Robert Frouin, MODIS Co-Program Scientist, and Wayne Esaias, MODIS Ocean Discipline Group Leader.
At the meeting, participants created the framework for reaching their goal as presented by NASA Headquarters: to develop a plan and approach for conducting coordinated cross-calibration and validation of ocean color satellite sensors. NASA is specifically interested in establishing data system requirements necessary for the combined use of satellite ocean color products from SeaWiFS, MODIS, OCTS, GLI, MERIS, POLDER, and other ocean color sensors, to address the needs for decadal-scale observations within the NASA Mission to Planet Earth (MTPE) and international Global Change research community. The meeting participants' objective is to submit a report by early May 1995 to NASA MTPE that addresses the following:
On Wednesday, February 22, the international contingent met in an all-day plenary session to share status reports on their respective projects, and to set the stage for addressing the issues listed above. On Thursday, meeting participants were divided into five groups to discuss discipline-specific concerns. On Friday, the attendees reconvened for a Final Plenary Session to report on each groups' conclusions and/or recommendations.
Radiometric Calibration and Characterization of Sensors
Chuck McClain, SeaWiFS Project Scientist, reported on Group 1's recommendation to build on the framework developed under the joint SeaWiFS-MODIS calibration and validation program. McClain suggested that a U.S. Ocean Color Intercalibration Executive Committee be formed to oversee this program. McClain also urged the ocean color community to continue developing measurement protocols--laboratory and field.
Regarding laboratory calibration efforts, McClain stated that the ocean color community should strive to expand the scope of the SeaWiFS Calibration Round-Robin beyond the present radiometric source round-robins which have been hosted at San Diego State University Center for Hydro-Optics and Remote Sensing (although round-robins of this nature should be continued). For instance, the National Institute of Standards and Technology (NIST) could initiate training workshops to facilitate these additional round-robins. McClain recommended that the ocean color community develop a pre-launch sensor characterization standard that describes the key parameters and tests that should be performed and documented.
Regarding post-launch on-board calibration and stability, McClain suggested that the community could develop a cumulative description of the solar calibration, internal lamp calibration, calibration pulse, dark current, and sensor engineering data collection schemes for the present suite of ocean color sensors. He endorsed the support that the EOS Project is giving Hugh Kieffer for the lunar measurement program. McClain said the community should obtain, if possible, witness filter samples for all U.S. ocean color instruments and maintain samples in a vacuum environment. NASA HQ should also support efforts, such as field studies, to evaluate and correct sensor anomalies such as stray light and bright target recovery.
Regarding vicarious calibration, McClain stated that the community should support additional calibration mooring sites, and other vicarious calibration programs at both high latitudes and high altitudes. Atmospheric optical measurements near calibration mooring sites should be supported. Additionally, U.S. initialization cruises for every ocean color mission launch should be supported. A common atmospheric correction scheme for all applicable ocean color sensors should be implemented by the community. McClain recognized that international agreements for data exchange must be established whereas few currently exist. Ultimately, a plan must be developed and supported by the community for evaluating and comparing onboard and vicarious calibration information and associated uncertainty budgets.
McClain stated that NASA should support an ocean color calibration data archive for pre- and post-launch satellite calibration, characterization, and sensor engineering data. Additionally, match-up data and calibration round-robin data should be archived.
Global Geophysical Product Validation
Wayne Esaias and Frank Muller-Karger, University of South Florida, presented a summary of Group 2's deliberations. Esaias defined validation as "the process of defining the spatial and temporal error fields and regional limits for a given biological/geophysical product throughout the mission." Esaias stated that comparison of satellite-derived values with real in situ values is the basis for determining the accuracy of a data product in extended ranges. Error fields can also be interpreted in terms of the spatial and temporal statistics of the geophysical variables of interest. Every mission has a very basic, minimal validation program, but none are global in scope. Some validation programs are tuned to specific regions and sensors of interest. Esaias said the international sharing of validation data (in situ and ancillary) increases the spatial and temporal coverage by about a factor of 10 over individual projects, and enables cross-comparison of data products. Esaias added that the community should use the sun photometer network to help with vicarious calibration.
Esaias said there is a need to identify key regions that may have inadequate in-situ validation activities ongoing. Global survey cruises and focused field expeditions can be used to collect uninterrupted time series data on important variables. Esaias pointed out some key areas of concern: extreme optical environments, high latitude bio-optical moorings, increased sampling of the Southern Ocean, the tropical Atlantic (and northwest Africa to characterize its dust), and major river plumes. He recommended that the GSFC Wallops Flight Facility be used to obtain aerosol optical depth data.
Validation data collection efforts should emphasize normalized water-leaving radiance, chlorophyll-a, aerosol optical depth, diffuse attenuation coefficient (k), sea surface temperature, productivity, coccolith, suspended sediments, and fluorescence. Esaias stressed that there is a need to implement quality control measures in collecting validation data.
Muller-Karger stated that there is a need for data consistency, as well as standardized, simplified collection methods. He recommended establishing an ocean color working group with an international forum to define methods and protocols for compiling and distributing data products. Muller-Karger also recommended augmenting each international partner's validation database several fold so that the effort becomes global in scope. In short, access to data should be easy and completely open to international partners.
Muller-Karger suggested developing a global automated observation network for physical oceanography and ocean biogeochemistry. The intent is to establish a near real-time validation database that facilitates fine tuning of data products. He pointed out that the basic technology for such a network does exist, but needs more development (for example, biofouling is a concern). Muller-Karger encouraged the community to take advantage of ships of opportunity wherever possible--consideration should be given to using commercial vessels that frequent shipping lanes and fishery areas, as well as operational assets such as NOAA and U.S. Navy vessels. He recommended that ocean color community delegates conduct a feasibility study to identify vessels and shipping lanes. Additionally, technology and protocols must be refined to ensure data quality, while an international framework is established to ease sampling restrictions in foreign waters.
Multisensor Product Comparisons and Merging Data
Janet Campbell, University of New Hampshire, asserted that the only methods for monitoring global oceanic (or terrestrial) primary production require observations from space. Therefore, she said, our long-range goal is to produce a continuous time series of bio-optical and geophysical variables derived from ocean color satellite data. This database will enable oceanographers to monitor changes in coastal and open ocean biological production that might occur as a direct or indirect result of climate change and human population growth.
The time series will begin in 1996 with SeaWiFS and OCTS data and, subsequently, data from the MERIS, MODIS, and GLI sensors will also be incorporated. Campbell pointed out that all of these sensors draw from the common heritage of the Coastal Zone Color Scanner, and thus, there is a basis for merging data. Because of their high degree of compatibility, data from SeaWiFS and OCTS will be the easiest to merge. Changes in spatial and spectral resolution will make the task more challenging as the later sensors come on line.
It is unclear whether data from other sensors (POLDER, MOS Priroda, and others) will be merged because these sensors employ techniques or have other differences that may render them incompatible. Group 3 proposed creating a data set that would allow oceanographers to determine whether the data from these sensors can be merged directly, versus providing important comparative information.
The purpose of the time series is to monitor environmental change. Thus, the variables chosen include (to begin with): a CZCS-like pigment concentration (derived from CZCS bands) that will enable us to begin the time series in 1978 with CZCS data, chlorophyll-a, diffuse attenuation coefficient, and aerosol optical depths. Other variables (e.g., primary productivity, coccolithophore concentration, etc.) will be added as these become operational products at a later date.
Campbell does not recommend the production of time series for variables (e.g., water leaving radiances, epsilons, etc.) solely for the purpose of interpreting the higher-level derived variables. These data sets will exist within the project. However, she anticipates the need to make adjustments or corrections to make data sets compatible. No doubt, data from the earlier satellites (SeaWiFS and OCTS) will have to be "corrected" to make them compatible with later sensors. To this end, she recommends the creation of a Test Data Set (or Diagnostic Data Set) that will contain the information necessary to figure out how to accomplish the adjustments. This information (calibration constants, sensor gains, raw digital counts, algorithm parameters, etc.) is readily available and accessible during the initial processing of the data, but is highly inaccessible after the data are processed. Thus, Campbell recommends a Diagnostic Data Set should be created by each sensor project at the time the initial data are processed.
The Diagnostic Data Set will contain data and ancillary information for individual pixels located at a fixed spatial grid. The grid-point spacing will be relatively large in open ocean areas, but will get finer near shore. The total number of grid points will be on the order of 30,000 globally. Thus, the diagnostic data volume will be manageable, not overly burdensome, but extremely valuable in later years as oceanographers work out the details of how to merge the data from multiple sensors over a 15-year time period.
Data System Requirements
Gene Feldman, SeaWiFS Data Processing Manager, stated that recommendations on a data system could not be spelled out until processing requirements from Groups 1 - 3 are better defined. Feldman suggested that he should establish a home page on the World Wide Web for the ocean color community. The page could help clearly identify who the members of that community are, as well as the members' respective missions. Additionally, a Mosaic-like browser could be implemented for every ocean color project affiliated with distributed data servers running SEABASS, or a similar system in which in situ data are available for public use.
Feldman recommended putting together an actual data set package from a field program (e.g., the Southern Ocean JGOFS [Joint Global Ocean Flux Study] October 1996 - April 1997 campaign) that includes current ship data, as well as data from buoys, moorings and other sources. Metadata for each data set should also be preserved and made available. Additionally, while JGOFS is in operation, spacecraft ocean color missions also in operation should provide access to coincident satellite data in near real-time to the JGOFS researchers and data users.
Feldman also recommended designating a group/program/investigator to develop an ongoing effort to support the collection, formatting, cataloging, and distribution of ocean color, in situ, and/or field support data. Feldman said that this designate may provide links to a highly distributed system, or coalesce the data into a DAAC. Feldman added that a standard grid, such as the ISCCP 1-km nested grid, must be adopted.
National and International Coordination
Robert Frouin stated that oceanographers need long-term data sets to study interannual phenomena, such as global and coastal change, carbon cycling, and spatial scaling. Currently, there are six satellite sensors capable of global coverage that are scheduled to launch within this decade--a great opportunity to start building a long-term database. To take advantage of this opportunity, the ocean color community must develop composite calibration/validation data sets; pool metadata on sensor characteristics; compare and assess algorithms for product extraction; arrange for Level 1 data exchange after launch; and evaluate and distribute final products.
Frouin reiterated the need to establish a science working group on ocean color (such as an extended JUWOC, or Japanese-U.S. Working Group on Ocean Color) to make recommendations and take the lead on strategic planning. International space agencies, government bodies, and the Committee on Earth Observations Satellites (CEOS) should take the lead in setting policies, establishing multilateral agreements, writing memoranda of understanding, or whatever is needed for sharing data among the international community. Additionally, the United States should establish an interagency work group on ocean color to facilitate the coordination of programs, funding, etc.
Frouin recommended using pilot projects to lay the foundations for international collaboration. These pilot projects could help force the community to resolve problems relating to multi-sensor calibration and validation, data formatting and exchange issues, sharing agreements, etc.