ARTICLE IN PRESS

Deep-Sea Research II 51 (2004) 5–42

An overview of the SeaWiFS project andstrategies for producing a climate research quality global bio-optical time series

Charles R. McClain*, Gene C. Feldman, Stanford B. Hooker Code 970.2, Office for Global Carbon Studies, NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA

Received1 April 2003; accepted19 November 2003

Abstract

The Sea-viewing Wide Field-of-view Sensor (SeaWiFS) Project Office was formally initiated at the NASA Goddard Space Flight Center in 1990. Seven years later, the sensor was launched by Orbital Sciences Corporation under a data- buy contract to provide 5 years of science quality data for global ocean biogeochemistry research. To date, the SeaWiFS program has greatly exceeded the mission goals established over a decade ago in terms of data quality, data accessibility andusability, ocean community infrastructure development,cost efficiency, andcommunity service. The SeaWiFS Project Office andits collaborators in the scientific community have madesubstantial contributions in the areas of calibration, product validation, near-real time data access, field data collection, protocol development, in situ instrumentation technology, operational data system development, and desktop level-0 to level-3 processing software. One important aspect of the SeaWiFS program is the high level of science community cooperation andparticipation. This article summarizes the key activities andapproaches the SeaWiFS Project Office pursuedto define,achieve, and maintain the mission objectives. These achievements have enabledthe user community to publish a large andgrowing volume of research such as those contributedto this special volume of Deep-Sea Research. Finally, some examples of major geophysical events (oceanic, atmospheric, andterrestrial) capturedby SeaWiFS are presentedto demonstratethe versatility of the sensor. Publishedby Elsevier Ltd.

1. Introduction Planning activities (Ocean-color Science Working Group, 1982; Joint EOSAT/NASA SeaWiFS The Sea-viewing Wide Field-of-view Sensor Working Group, 1987) paralleledandsupported (SeaWiFS) was the result of a persistent effort by the argument to include global ocean-color ob- the ocean biogeochemical commu- servations in the (EOS; nity to have an operational ocean-color satellite Asrar andDozier, 1994 ; King et al., 1999) that following the great success of the experimental ultimately resultedin the ModerateResolution Nimbus-7 Coastal Zone Color Scanner (CZCS). Imaging Spectroradiometer (MODIS) on the andAqua platforms. The strategy was to *Corresponding author. Fax: +1-301-286-0268. have SeaWiFS launch precede the MODIS launch E-mail address: charles.r.mcclain@.gov by several years to initiate a global ocean-color (C.R. McClain). time series in support of the Joint Global Ocean

0967-0645/$ - see front matter Publishedby Elsevier Ltd. doi:10.1016/j.dsr2.2003.11.001 ARTICLE IN PRESS

6 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42

Flux Study (JGOFS) process studies in the Calibration, andValidationelements were merged Arabian Sea, equatorial Pacific, andSouthern to form the Calibration andValidation (CV) Ocean and to provide ample time to design and element. validate sensor calibration strategies and algo- The prelaunch phase was characterizedby an rithms in preparation for the MODIS time series. unprecedented level of cooperation between In addition, SeaWiFS was to be the first in a series NASA, OSC, andthe instrument subcontractor, of US andinternational ocean-color missions that Hughes Santa Barbara Research Center, given wouldeventually providea long-term recordof that the contract did not require a high degree of ocean biological andoptical properties for climate interaction. In fact, GSFC engineering groups research. In 1990, the NASA Goddard Space made many voluntary contributions towards Flight Center (GSFC) initiateda competitive resolving a number of technical problems with ocean-color data-buy procurement and established spacecraft components andsubsystems, e.g., ra- a SeaWiFS Project Office (SPO). diation hardness, power, navigation and attitude A data-buy contract with Orbital Sciences control, andtelemetry. Throughout the prelaunch Corporation (OSC) was finalizedin 1991 with an phase, OSC demonstrated a firm resolve to achieve expectedlaunch in mid-1993.NASA was to have success, even at considerable expense to the insight, but not oversight, into the spacecraft and company. In addition, NASA Headquarters pro- sensor design, construction, and testing. Under the vided steadfast support to the SPO. Despite the contract, OSC was responsible for building, best efforts of all parties, the launch schedule launching, andoperating the spacecraft. Origin- slipped4 years. In August 1997, the prelaunch ally, the spacecraft was calledSeaSTAR, but it was phase culminatedin a flawless Pegasus-XL launch, renamedafter launch to OrbView-2 when Orbital andSeaWiFS has delivereda continuous stream of Imaging Corporation (ORBIMAGE) purchased GAC and LAC data of unprecedented quality the spacecraft from OSC. The payment schedule since September 1997. was front-endloadedso that most of the fixed The original science goals andproject objectives, costs were paidat the completion of specific listedin Table 1 (Hooker et al., 1992), were defined milestones during the prelaunch system develop- to support quantitative research. The initial ment andpostlaunch dataacceptance phases. product suite was very similar to that of the CZCS After acceptance, fixedmonthly payments have and included total radiances (level 1 data), been made, subject to penalties if the data quality normalizedwater-leaving radiances, chlorophyll- is less than nominal (to date, no penalties have a, diffuse (490 nm), and certain atmo- been applied). NASA’s responsibilities included spheric correction-relatedparameters (level-2 sensor and onboard recorder scheduling (sensor products), and binned and standard mapped tilting, solar calibrations, lunar calibrations, inter- products at various temporal resolutions (level-3 nal sensor test sequences, andhigh-resolution data products), e.g., daily, 8-day, and monthly (Darzi, recording), postlaunch sensor calibration, derived 1998). The radiometric and chlorophyll-a accuracy product algorithm development, data acquisition goals are quite rigorous andhave provedchallen- [coarse resolution global area coverage (GAC) and ging to meet. Given the aggressive launch sche- fine-resolution local area coverage (LAC)], data dule, calibration and validation activities were to processing, research data archival and distribu- be coordinated with those of the MODIS tion, andselection of high-resolution picture Team that were already underway. The SPO also transmission (HRPT) stations. With these activ- sought the science community’s involvement in ities in mind, the SPO was organized at GSFC each aspect of the program to capitalize on their with a project manager, a project scientist, and expertise, to explore new applications of the data, project element leaders, the elements being Data andto ensure the greatest possible utilization of Capture, Mission Operations (MO), Instrument the data for Earth system science. The SPO Science, Calibration, Validation, and Data Proces- benefitedgreatly from its collaborations with the sing (DP). Over time, the Instrument Science, MODIS Oceans Team, the science community at ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 7

Table 1 Original SeaWiFS program goals andproject objectives ( Hooker et al., 1992)

SeaWiFS program goals SeaWiFS Project Office objectives

1. To determine the spatial and temporal distributions of 1. To serve as the NASA liaison to OSC for the procurement blooms, along with the magnitude and of SeaWiFS data for the oceanographic research community variability of by marine phytoplankton 2. To facilitate the operation andschedulingof the SeaWiFS on a global scale sensor system 2. To quantify the ocean’s role in the global and 3. To develop and validate algorithms for bio-optical other biogeochemical cycles properties andatmospheric correction 3. To identify and quantify the relationships between ocean 4. To characterize andcalibrate the SeaWiFS system andto physics andlarge-scale patterns of productivity assess on-orbit performance 4. To understand the fate of fluvial nutrients and their possible 5. To achieve radiometric accuracy to within 5% absolute and effects on carbon budgets 1% relative, water-leaving radiance to within 5% absolute, 5. To identify the large-scale distribution and timing of spring andchlorophyll- a concentration to within 35% over the blooms in the global oceans range of 0.05–50.0 mg m3 6. To acquire global data on marine optical properties, along 6. To collect, archive, and process recorded data, as well as with a better understanding of the processes associated with global maps of bio-optical properties andchlorophyll- a mixing along the edge of eddies and boundary currents concentration for the research community 7. To advance the scientific applications of ocean-color data 7. To provide quality assurance monitoring of, and coordinate andthe technical capabilities requiredfor dataprocessing, collection of, direct broadcast data by NASA’s selected management, andanalysis in preparation for future LAC receiving stations missions. 8. To support the SeaWiFS Science Working Group (SWG).

large, andthe Sensor Intercomparison andMerger tal Satellite System (NPOESS) Preparatory Project for Biological andInterdisciplinary Oceanic Stu- (NPP). dies (SIMBIOS) program (Fargion andMcClain, 2002; McClain et al., 2002) which provided processing algorithms, in situ data sets, and much 2. Project philosophy, functions, and approaches guidance. The purpose of this paper is to review the Because SeaWiFS was a data-buy rather than a original science goals andproject objectives, out- standard NASA mission, the SPO was established line the approaches pursuedto achieve them, and in the Earth Science Directorate (Code 900) at assess (as quantitatively as possible) the overall GSFC with almost all elements of the SPO being level of success in attaining them. During the 12- co-locatedandleadby ocean scientists. By having year history of the SPO, a number of innovative scientists with vestedinterests in the dataas approaches have been developed in addressing element leaders, a commitment to data quality certain problems, e.g., the lunar time-dependent anddata access, anda focus on research and calibration correction. In some cases, the SPO has development in every aspect of the mission was needed to adjust its approach when initial strate- ensured. While the SPO managers and element gies provedineffective or to take advantageof new leaders have been civil servants, the majority technology, e.g., the evolution of the data system (roughly 80–90% in the postlaunch phase) of the design. In the end, it is hoped that the SeaWiFS SPO personnel have been on-site contractors program can be usedas a modelfor future provided under support services contracts defined, missions andthat the infrastructure the SPO competed, and negotiated by the SPO manage- helpedestablish can be continuedandexpanded ment. This helpedensure that the SPO receivedthe in anticipation of future missions such as the best support possible. Nonetheless, training was National Polar-orbiting Operational Environmen- necessary, as the majority of the contract staff ARTICLE IN PRESS

8 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 were from other science disciplines. For instance, et al. (2002) to be consistent with the third the CV element conducted a comprehensive SeaWiFS reprocessing in 2000. literature review to familiarize the staff with all 5. The processing system must be tightly coupled aspects of ocean-color satellite calibration and to the science. None of the NET members were algorithm development, as well as the community locatedat GSFC where the processing was with whom they wouldbe working. being conducted. NET meetings (Acker, 1994) The SPO has been committedto a number of provided opportunities to review results and specific objectives (listedbelow) relatedto, but not advise the Nimbus Project, but these meetings explicitly identified in, the overall SeaWiFS were relatively infrequent. program objectives outlinedin Table 1. Many While none of the SPO staff were members of aspects of the SeaWiFS program were a result of the NET, many hadworkedextensively with the the lessons learnedfrom the proof-of-concept CZCS data and some had worked closely with the CZCS program andincludethe following. NET in the postlaunch period. Thus, there was an understanding of what needed to be done differ- 1. A comprehensive calibration andvalidation ently in an operational setting as opposedto a program spanning the entire mission. For the proof-of-concept scenario. Specifically, particular CZCS, the postlaunch calibration andvalida- attention was paidto the following strategies. tion program was supportedfor only the first year and, therefore, did not include time series 1. Direct community involvement in, and awareness observations to track sensor performance. In of, all SPO activities. This has been accom- addition, the field program for algorithm plishedthrough open project reviews, work- development was limited mainly to areas near shops, calibration anddataanalysis round North America. robins, periodic project status reports broad- 2. Rapid user access to data products. CZCS data cast to the community (especially important in did not become generally available until the the prelaunch phase), annual science team global reprocessing was completed( Feldman meetings, professional conference presenta- et al., 1989), several years after data collection tions, andsupport of a SeaWiFS booth at ended. major US conferences. The SPO has reliedon 3. Availability to the user community of a low the community to provide the atmospheric cost data processing capability. The Nimbus correction andbio-optical algorithms andmost Project did not provide data processing tools, of the in situ data (atmospheric and bio- not even to the Nimbus Experiment Team optical) for algorithm development and post- (NET). As a result, individual groups indepen- launch product validation. To finalize the dently developed their own software such as chlorophyll-a algorithm before launch, a spe- SEAPAK (McClain et al., 1991a, b, 1992a) and cial workshop involving in situ bio-optical data the University of Miami DSP software. processing andreal time algorithm compar- 4. The ability to reprocess the entire data set on a isons was conducted at the University of routine basis. The CZCS reprocessing required California/Santa Barbara ([91]O’Reilly et al., the entire data set be staged on optical media 1998). During each of the four reprocessings, from over 30,000 9-track tapes. This process the SPO actively sought the community’s input alone took over a year to complete. The by hosting workshops andposting descriptions vicarious calibration of the entire data set andanalyses of proposedprocessing modifica- requiredprocessing the datamany times to get tions on SPO Web sites. consistent results (Evans andGordon,1994 ) 2. Rapid turnaround and easy access to data andtook a similar amount of time. The final products. The goal was to process the data on processing andquality control took at least a same-day basis with release to the science another year. Recently, the CZCS data was community at the earliest possible time allowed reprocessed(at reducedresolution) by Gregg under the data-buy contract. One approach ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 9

was to design the data formats to be publication to a subscriber list of nearly 500 self-describing with the relevant ancillary individuals. andcalibration data files cross-referenced 7. A rigorous scrutiny of all aspects of ocean-color andautomatically distributedwith the asso- remote sensing technology, measurement ciatedimage datafile. Data selection was science, and algorithm development with an simplifiedwith on-line image browsing. emphasis on innovation and improving the 3. Focus on a few key geophysical products. science community’s technical infrastructure The number of archive products was kept to and ability to support ocean-color satellite a fairly small number of key geophysical missions. This has been accomplishedprimarily parameters with one product per parameter. through activities such as the calibration and Also, a single set of data quality masks data analysis round robins, measurement pro- andflags was used.The productsuite was tocol development, in situ instrument technol- revisitedat each reprocessing to remove ogy development, and the SeaWiFS Bio-optical products of little interest (e.g., CZCS pigment) Archive andStorage System (SeaBASS). Also, andaddothers(e.g., photosynthetically avail- while the project’s CV staff needed to be able radiation, PAR) based on community familiar with all processing algorithms and input. knowledgeable of the related scientific litera- 4. Affordable user-friendly end-to-end processing ture, algorithm development was left to the tools for the most common computer systems science community, so the SPO wouldbe used by the science community. This was impartial in its independent implementation considered essential if the data were to be fully andevaluation of algorithms. One exception is exploitedandresultedin the SeaWiFS Data the lunar calibration algorithm usedto track Analysis System (SeaDAS). SeaDAS was the sensor’s stability, which was developed by primarily supportedunderseparate funding SPO staff. This methodology continues to be from the NASA ocean biogeochemistry pro- refinedand,with recent US Geological Survey gram, but has required additional SPO invol- high-resolution reflectance models of the lunar vement andfinancial support (hardware, surface, may provide absolute calibrations of system administration, and SeaWiFS software the sensor. technical support). 8. SPO flexibility, evolution, and accountability in 5. Science quality data by the end of the 90-day its approach to scientific and operational devel- data acceptance period with occasional reproces- opment activities. Each SPO element manager sings. The CV element implementeda compre- was given complete control on defining the hensive set of test andevaluation criteria and technical approach. As a result, all elements analysis software prior to launch. During the adopted a strategy of continual evolution in the postlaunch data-acceptance period, all criteria technologies andapproaches to be pursued. were quantitatively evaluated, and the first For instance, the data processing element has reprocessing was initiatedimmediatelyafter- annually expanded and replaced systems to wards. Three subsequent reprocessings incor- continually upgrade with new technology. In poratedcalibration andalgorithm updatesto fact, the original SPO-wide computing design further improve the data quality. was three Unix servers connectedto X- 6. Full documentation and disclosure of SPO terminals. Over time, the X-terminals were supported activities, operational algorithms, replacedby Unix workstations, which have sensor characteristics, etc. The SPO has recently been replacedby PC Linux systems. supporteda full-time technical editorsince To ensure communication between elements early in the program andhas published andwith management, especially in the pre- nearly 70 technical reports, plus numerous launch phase, the SPO conducted periodic conference andjournal papers. These docu- open reviews (roughly every 6 months) of each ments have been distributed at the time of SPO element with members of the staff giving ARTICLE IN PRESS

10 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42

overviews of the components they were (AOT) andthe normalizeddifferencevegeta- responsible for developing. tion index (NDVI). Subsets over the MODIS 9. Community support and user services. The SPO landvalidation sites for the entire mission are saw the community as a vestedpartner and available via the SeaWiFS Web site. Examples recognizedthat overall success wouldbe of such interdisciplinary use of the data are largely determined by the community’s attitude Behrenfeldet al. (2001) , Wang et al. (2000), towards the SPO. With the element leads being and Chou et al. (2002). active scientists, they understood the commu- 12. Tightly coupled and highly interactive SPO nity’s needs. Examples of support services elements. It was recognizedthat all elements include SeaDAS, near-real-time image support, were interdependent and required a high degree andthe pre- andpostlaunch SeaWiFS Techni- of communication and mutual understanding. cal Report Series (STRS). For example, near-real time quality control 10. Fiscal accountability and openness. The budget procedures by the CV element were embedded guidelines for the SPO and each SPO element in the data processing sequence and affected were establishedwhen the SPO was organized the processing system design and throughput. andapprovedby NASA Headquarters. The Mission requirements. The SeaWiFS program allocations for the SPO elements have not goals andproject objectives ( Table 1) were changedsince then, although some elements considered to be requirements or criteria for have mergedover time. Each element manager mission success. was given complete control over the budget for 13. SPO focus on mission deliverables. The staff has that element, which gave the element manager always focusedon improving services to the the ability to adjust his budget priorities over community, the accuracy of the data products, time, e.g., manpower, hardware, university andoutreach. The SPO has purposely avoided grants, international agreements, andcon- committing project resources to investigations tracts. The SPO has always been committed of ocean or Earth science, although staff to operating within budget and not requesting occasionally assist groups in getting access to additional overguide funding. Because of the data products for specific studies. launch delay, however, a one-time request for overguide funding was made and approved in The following sections outline the activities and order to have flat funding during the opera- accomplishments of the SPO organizational ele- tional phase. ments. One of the important aspects of the SPO is 11. New products and applications beyond ocean that all elements are co-locatedallowing continual biogeochemistry. The SPO has striven to communication between the groups andthe promote the usefulness of SeaWiFS data by sharing of resources (personnel andequipment). exploring new ocean, terrestrial, andatmo- Co-location also helps the SPO to be very efficient spheric products, many of which are possible in terms of decision making and oversight of SPO because the sensor does not saturate, even over activities. The order of the sections reflect the bright landandclouds.After the third data flow, i.e. mission operations, data capture, reprocessing in 2000, the SPO began routine calibration and validation, data processing, and production of a number of ‘‘evaluation’’ data archival and distribution. Fig. 1 provides a products, e.g., PAR and color dissolved graphical description of the data handling process. organic matter (CDOM), for the community to consider as potential archive products. The 2.1. Mission operations and data capture evaluation products were derived using algo- rithms provided by the community. The SPO The goal of the MO element of the SPO was to has workedwith researchers in the atmospheric maximize the collection of scientifically useful data andterrestrial science communities to provide from the SeaWiFS instrument through a close products such as aerosol optical thickness working collaboration with the engineers and ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 11

Fig. 1. SeaWiFS data flow from satellite to data user.

mission planners at ORBIMAGE, the other Mission Operations group for final verification elements of the SPO, andthe global ocean-color before uplink to the spacecraft by ORBIMAGE. community. While ORBIMAGE is responsible for Special requirements—such as spacecraft emer- the actual operations of the OrbView-2 spacecraft gencies, downlink scheduling conflicts at WFF and andSeaWiFS instrument, NASA has the respon- requests for high data rate acquisition of telemetry sibility for (a) SeaWiFS instrument command information to help resolve potential spacecraft scheduling and onboard recorder management; anomalies—are handled on a case-by-case basis. (b) navigation, including position determination, Extensive prelaunch end-to-end system tests, and attitude determination and geolocation (Patt and regular postlaunch communications with ORB- Gregg, 1994; Patt et al., 1997; Patt, 2002); and(c) IMAGE, including joint annual reviews, have routine spacecraft andinstrument health and resultedin a smooth, efficient, andmutually safety monitoring. A representative daily SeaWiFS beneficial collaboration. data collection schedule is presented in Fig. 2. Over the 5 1/2 years of operation, there have To accomplish the data acquisition tasks in- been approximately 20 unscheduled safe-haven dicated in Fig. 2, NASA is responsible for events due to onboard system anomalies that have providing ORBIMAGE with the specific schedul- resultedin some loss of science data.During these ing requirements each week for data collection by events, the SeaWiFS instrument is stowed, and the the SeaWiFS instrument including defining the majority of the OrbView-2 systems are put into a operating parameters (gain, tilt, start/stop times, ‘protective’ posture waiting for a signal from targets), calibration activities (solar, lunar, time- groundcontrollers to resume routine operations delay integration) and coordinating the downlink after whatever event that triggeredthe safe-haven schedules with the primary S-band downlink site is identified, analyzed, and the corrective actions at NASA Wallops Flight Facility (WFF). ORB- needed for full recovery are executed. The majority IMAGE integrates NASA’s scheduling require- of these events have been associatedwith single- ments into their routine spacecraft operations event upsets in one of the spacecraft electronic schedules that are then provided back to the subsystems, andmost of these events have been ARTICLE IN PRESS

12 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42

Fig. 2. SeaWiFS data acquisition on 1 January 2002. Illustrated are the standard 14 orbits of GAC data collected each day (purple shading), the specific high-resolution LAC targets (blue shading with wider swaths), daily time-delay integration and solar calibration data (short LAC swaths at high southern latitudes colored red) and the two S-band downlink opportunities (red) over the US east coast. The GAC data collection is truncated at scan angles of 745. The limitedonboardLAC coverage ( B10 min per day) is scheduled in advance to cover in-water calibration targets (white elipses or circles), ships of opportunity collecting validation data, the solar calibration data, and regions of either specific scientific interest (white boxes; lowest priority).

shown to occur while the spacecraft is flying each downlink and posts this information on the through the South Atlantic Anomaly, an area east Web. Daily andlong-term plots of the behavior of of Argentina where solar particle fluxes are the major onboardsubsystems including battery elevated. As knowledge of these events developed, performance, horizon sensor parameters (phase andas confidencein the recovery procedures was andchord),telescope motor temperature and improved, the duration between the onset of safe- current, sun-sensor telemetry trends, and momen- haven events andfull recovery has greatly tum wheel performance are trackedandpermit decreased such that most events are recovered continual assessment of the health andsafety of within 12 h. In addition to these unplanned events, the spacecraft andinstrument. These analyses are the spacecraft has been intentionally placedinto a also crucial in maintaining functionality and protected, safe-haven posture during the peak improving the performance of the spacecraft over activity periods of the past four Leonid time. Many of the prelaunch concerns about the showers as a precautionary measure. In spite of longevity of a number of the spacecraft systems these unplannedinterruptions to normal opera- were completely resolvedthrough these analyses. tions, however, more than 98.5% of the potentially Plans for an extended mission beyond the original available data has been successfully acquired and 5 years have usedthe long-term trendanalyses to processed. assess the potential viability of the spacecraft The MO element works with ORBIMAGE in andinstrument. Presently, there is no indication of tracking the health of the spacecraft andthe any significant degradation in the spacecraft SeaWiFS sensor. Each conduct their own evalua- subsystems. tions. The MO element carries out regular analyses To meet the mission geolocation accuracy goal of the spacecraft andinstrument telemetry with (less than one pixel), members of the SPO ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 13 developed an intimate working knowledge of This fully automatedreceiving station serves as the the spacecraft design and operation, especially primary L-bandreceiving station for the US East the attitude control system (ACS). OrbView-2 Coast andas the S-bandbackup for the primary has a relatively simple ACS that consists of station at WFF. For a number of reasons, redundant sets of sun and horizon sensors that including conflicts at WFF caused by Space are usedto determine the pitch, roll, andyaw of Shuttle operations andantenna-relatedissues, the the spacecraft. More sophisticatedsatellites station at GSFC frequently serves as the primary have star trackers andgyroscopes. The ACS data telemetry site and has not missed a single sensors provide input that is used to control the downlink. momentum wheels andtorque rods(coils that In addition to the two NASA stations, the SPO work off the Earth’s magnetic field) that adjust collaborates with a network of approximately 124 the spacecraft attitude to maintain the sensor L-bandreceiving stations aroundthe worldthat at a nadir orientation. Maintaining the ACS in are affiliatedwith universities, research organiza- an optimal configuration includes periodic tions, andother space agencies to collect the LAC adjustments of some sensor parameters and fine data that are broadcast as the spacecraft passes tuning of sensor alignments. Bilanow andPatt over the stations several times per day. The vast (2004) provide a summary of some of the majority of these ground stations are independent major SPO contributions to the spacecraft ACS andare not supportedwith NASA funds.The SPO support. provides the coordination between the ground In addition, an automated technique of island stations andORBIMAGE for real-time decryp- target matching was developed (Patt et al., 1997) tion of the data in support of research activities or andthe results are continually updatedandposted demonstration projects and through NASA-pro- on the Mission Operations Web site with the vided software. This software is distributed to the resulting improvements to the navigation software groundstations, andfacilitates the collection, incorporatedinto the operational andSeaDAS processing, archive and distribution of data navigation procedures as appropriate. Over the life collectedby these stations. Underterms of the of the mission, an average of approximately 50 data-buy contract, NASA is authorized to collect islandtargets per GAC swath have been acquired and decrypt SeaWiFS data in near-real time from and analyzed. For the 1.1 km HRPT data that the any 12 of its authorizedreceiving stations (in SPO routinely receives, an average of more than addition to the station at GSFC) at any given time, 100 targets per scene are matched, yielding an while the remaining stations are allowedto collect average error of 1.03 km, meeting the mission’s SeaWiFS data, but must wait at least 14 days geolocation requirement. before ORBIMAGE provides them with the The SeaWiFS instrument has a nominal resolu- appropriate decryption keys, which allows them tion of 1.1 km at nadir, but because of limitations to process the data. At present, there are four in onboardstorage andtelemetry technology in stations covering the continental US that have the early 1990s, it was not considered cost effective been grantedpermanent real-time licenses, an to store and downlink the global 1-km data set. equal number of international stations have long- Consequently, the full-resolution data are sub- term floating licenses, andfour short-term floating sampledby every fourth pixel andevery fourth licenses are available at 2-week intervals to line, andthis global dataset is storedonboardwith support specific near-shore research activities not limitedamount of high-resolution data.Twice per coveredby the existing near-real-time stations or day, at approximately local noon and midnight, for ship-basedreceiving stations. The success of the spacecraft downlinks the recorded data as it this activity is quite remarkable andhas far passes over WFF. To minimize single-point fail- exceeded the initial SPO goal of coordinating a ures that wouldresult in an irrecoverable loss of network of 12 receiving stations as outlinedin the science data, the SPO installed its own dual S and contract. To date, the SPO has received approxi- L-bandreceiving station at GSFC prior to launch. mately 128,000 HRPT files from 88 receiving ARTICLE IN PRESS

14 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 stations. Like the GAC andonboardLAC data, Fargion andMcClain, 2002 ). The amount of data these are quality controlledandtransferredto the provided was greatly increased by the SIMBIOS GSFC DistributedActive Archive Center (DAAC) Project’s support of bio-optical andsun photo- for distribution. meter instrument pools andits augmentation of the Aerosol Robotic Network (AERONET; Hol- 2.2. Calibration and validation program ben et al., 1998) with 12 coastal andislandsites. A critical step in achieving the necessary As outlinedearlier, the scientific applications of radiance accuracies is to have a well-characterized satellite ocean-color data require an extremely satellite instrument. During the prelaunch phase, high radiometric accuracy of both the sensor considerable attention was focused on the calibra- calibration andthe derivedwater-leaving ra- tion andcharacterization of the SeaWiFS instru- diances. Because of these stringent accuracy ment (Barnes et al., 1994a, b; Johnson et al., requirements, a comprehensive calibration, valida- 1999a). After launch, a broad-based field program tion, andquality control program was conceived is requiredfor on-orbit calibration andproduct (McClain et al., 1992b, 1996) for the SeaWiFS, validation. These activities during the CZCS era which wouldcontinue throughout the mission. reliedexclusively on a small number of ship and The program included sensor calibration and aircraft campaigns, which provedto be inadequate characterization (pre- andpostlaunch), bio-optical to sufficiently quantify the degradation in the algorithm development, atmospheric correction CZCS instrument (Evans andGordon,1994 ). It algorithm development, postlaunch near-real-time was also recognizedthat the sensor characteriza- quality control, anda fieldmeasurement program tion andfielddatasets must be internally that included support for moored buoys (e.g., the consistent, readily available, and well organized Marine Optical Buoy, MOBY), oceanographic in order to expedite algorithm development and research cruises (e.g., the California Cooperative postlaunch validation. This led to the SeaWiFS Fisheries Institute, CalCOFI surveys), andtime Intercalibration Round-Robin Experiments (SIR- series stations (e.g., the Bermuda Atlantic Time- REX), the bio-optical measurement protocols, and series Station, BATS). The SeaWiFS CV element the development of SeaBASS (Hooker et al, 1994; hada budgetprofile that peakedat around$ 2 Werdell and Bailey, 2002), which was subsequently million per year early in the program in anticipa- expanded to include atmospheric validation data tion of a 1993 launch anddeclinedby 1997 to a under SIMBIOS support (Fargion et al., 2001). In steady level of about half of the peak year funding. the postlaunch phase, the calibration round The rationale was to take advantage of the robins, bio-optical protocol development, and MODIS Ocean Team activities, which were also SeaBASS have been largely supportedby the in the early phases of development, and accelerate SIMBIOS program. some of their activities, e.g., the atmospheric While the SPO purposely avoided algorithm correction algorithm, with additional funding to development, it did take the lead in organizing and meet the earlier SeaWiFS launch schedule. For- orchestrating a community-wide algorithm devel- tuitously, as the SeaWiFS calibration andvalida- opment andevaluation through a series of tion budget began to ramp downwards, the prelaunch workshops such as the SeaWiFS Bio- SIMBIOS program was initiatedin 1996 as a optical Algorithm Mini-workshop (SeaBAM; result of dropping the EOS Color mission, a O’Reilly et al., 1998). Summaries of each work- seconddata-buy to follow SeaWiFS, from the shop were publishedin the indexvolumes of the EOS program. SIMBIOS included a number of prelaunch STRS. Once the SIMBIOS program data collection investigations by a separate science started, the annual SIMBIOS science team meet- team andthe SIMBIOS Project Office, which ings provided a forum for such discussions and the provided a much more diverse global bio-optical needfor the SPO to host workshops was super- andatmospheric dataset than wouldotherwise ceded, except in special occasions, e.g., reproces- been available (McClain andFargion, 1999 ; sing workshops. With each reprocessing, both the ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 15

SPO andmembers of the user community have The strategies for using MOBY have differed suggestedprocessing modifications, which have somewhat between the MODIS andSeaWiFS. been openly discussed and evaluated with the SeaWiFS is able to view the moon monthly at the community at large. For instance, the chlorophyll- same phase angle, so lunar data are used to a algorithm was replacedwith the thirdreproces- characterize the time trends in the sensor respon- sing (O’Reilly et al., 2000). sivity (Barnes et al., 1999, 2001) andMOBY data One of the MODIS ocean team activities that are usedto adjustthe prelaunch gains ( Eplee et al., was acceleratedfor the original SeaWiFS launch 2001). MODIS is a more complex system andthe schedule was the development of the MOBY lunar measurements andsolar diffuser measure- (Clark et al., 1997). MOBY was to be the primary ments analyzedby the MODIS characterization vicarious calibration site for the MODIS and andsupport team (MCST) are not at the precision SeaWiFS missions andwas initiatedunder the requiredfor ocean-color applications, so MOBY MODIS program. The SPO provided substantial data are used by the MODIS Oceans Team to support for MOBY development prior to launch fine tune the time-dependent calibration gains through a contract with Moss Landing Marine provided by the MCST. Laboratory. Continuous MOBY deployments and By the thirdSeaWiFS reprocessing in 2000 data collection began in July 1997, just before the (McClain et al., 2000a, b), the strategy for cali- launch of SeaWiFS. Since then, financial support brating SeaWiFS on-orbit andfor validating from the SPO has been for MOBY maintenance products was fairly mature and incorporated a ship time, albeit at a much reduced level than variety of elements shown in Fig. 3. The strategy during the MOBY development phase. MOBY has included rigorous evaluations of sensor perfor- provided a continuous time series of high quality mance characteristics andcalibration prior to water-leaving radiances since the summer of 1997. launch, comparisons of the pre- andpost-launch

SeaWiFS Calibration Strategy

Preflight Instrument Complete sensor calibration Calibration and and characterization by Santa Characterization Barbara Research Center prior (1993-1994) to delivery of sensor to OSC

Preflight Instrument Calibration Recalibration by NIST at OSC just before launch Recertification (1997) Community-wide activities to Independent on-orbit calibration test ensure consistent data from after operations began in 1997 various sources

Solar Calibration for Calibration a Transfer-to-Orbit SeaWiFS Launch RR and Comparison Protocols Development

Onboard Vicarious Time-series Calibrations Calibration and Evaluations (time dependency Product Validation corrections)

In Situ Solar MOBY Bio-optical and Global 8-day Lunar Calibrations Calibrations Atmospheric Clear Water Calibrations (daily) (bands 1-6) (monthly) Data Comparisons Analysis

Fig. 3. SeaWiFS calibration andvalidationscheme showing pre- andpostlaunch activities andanalyses. ARTICLE IN PRESS

16 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 sensor calibrations (the so-called‘‘transfer-to- development andevaluation; andthe SeaWiFS orbit’’ experiment), lunar andsolar analyses of Project fieldprogram—summarize a number of the on-orbit sensor degradation, the MOBY-based subsequent activities andachievements of the vicarious calibration adjustments to the prelaunch SPO’s fieldmeasurement program. These activities laboratory calibration gain factors, time-series are interdependent and were conducted in parallel, analyses of global normalizedwater-leaving ra- but partitioning the material into three sections diances ðLWNÞ; atmospheric correction epsilon helps focus each discussion. values, and chlorophyll as additional sensor stability indicators, and match-up analysis of 2.2.1. SeaWiFS sea truth data accuracy satellite-derived versus in situ measured values considerations and advancements (bio-optical and atmospheric) for product valida- Ensuring the SeaWiFS radiometric retrievals of tion. The approach andresults of the CV element water-leaving radiance are within 5% over the life have been well documented in the STRS, con- of the mission requires a continuing commitment ference papers, andjournal articles as a result of to quantifying the uncertainties associatedwith the the SPO’s emphasis on information sharing, spaceborne andin situ instrumentation. This community infrastructure development, and ac- means the individual sources of uncertainty for countability. All these documents have been the acquisition of ground-truth data must be disseminated to a SPO distribution list of approxi- on the order of 1–2%, or what is referred to mately 500 individuals, universities, and libraries more generally as simply ‘‘1% radiometry’’. The andare available upon request from the SPO. sources of uncertainty for the groundtruth part Summaries of results have been published of the total uncertainty budget have a variety (McClain et al., 1998, 2000a, b; Hooker and of sources: McClain, 2000; Barnes et al., 2001, Eplee et al., 1. The measurement protocols usedin the field; 2001) andwill be updatedas new results become 2. The environmental conditions encountered available, e.g., the evaluations of the fourth global during data collection; reprocessing of the entire SeaWiFS data set. 3. The absolute calibration of the fieldradio- The original SPO strategy was to rely solely on meters; fieldbio-optical measurements for algorithm 4. The conversion of the optical measurements development and validation collected by the (in-water andabove water) to geophysical community, and funding was provided to augment parameters, e.g., diffuse attenuation and CalCOFI, BATS, andother datacollection water-leaving radiances, used in a data proces- programs. Indeed, the outside research community sing scheme; and has provided over 50,000 biological stations and 5. The stability of the radiometers in the harsh over 5000 discrete sun photometer observations environment they are subjectedto during (plus 94 fieldcampaigns of shadow-bandradio- transport anduse. meter data) to SeaBASS. It eventually became apparent, however, that a number of in situ The SeaWiFS CV element has sought to system- instrument design, calibration metrology, mea- atically identify and quantitatively address as surement protocol, anddataprocessing issues and many issues associatedwith all five sources of uncertainties couldnot be addressedbysimply uncertainty as was fiscally feasible. These efforts augmenting ongoing fieldprograms. Examples are outlinedbelow. include measurements in turbid water and proto- For the SPO, the first step in the process of cols for above-water measurements of water- controlling uncertainties in fielddatawas estab- leaving radiance. As a result, a separate SPO lishing, through consensus at a community work- capability focused on understanding and reducing shop, andthen publishing the SeaWiFS Ocean the uncertainties was instituted. The next three Optics Protocols (Mueller andAustin, 1992 ). The sections—SeaWiFS sea truth data accuracy con- protocols are a work in progress andandwere siderations and advancements; Field instrument initially revisedin Mueller andAustin (1995) , but, ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 17 under the SIMBIOS program, have undergone measurement wavelengths of approximately 1.5% substantial expansions (Mueller, 2000, 2002, 2003) (Johnson et al., 1998a). After repeateduse in by having the scientific community decide which SIRREX activities, the SeaWiFS calibration, and scientific areas would be updated. The SPO has an international integrating sphere comparison continually usedthe protocols as the requirements (Johnson et al., 1997), the SXR was commercia- for all ground-truth observations. lizedin limitednumbers anddifferent versions of The uncertainty in calibrations is the most the same design are being used in other round- fundamental, because all the others are only robin activities supportedby SIMBIOS (a con- quantifiable if the radiometers are properly cali- tinuation of SIRREX) andthe NASA EOS brated. To maintain internal consistency between calibration program. calibrations of the in situ sensors (andthe Instrumental drift due to filter deterioration and SeaWiFS instrument itself), the SPO required physical stresses, which can cause shifts in the calibration traceability to the National Institute optical alignment andelectrical characteristics of a of Standards and Technology (NIST) and imple- device, must be quantified even if a concerted mentedan ongoing series of SIRREX activities to effort is made to minimize these problems. To investigate andminimize calibration uncertainties. address this problem, NIST was contracted to In the progression from the first to the third jointly develop a highly stable portable field SIRREX (Mueller, 1993; Mueller et al., 1994, source, the SeaWiFS Quality Monitor (SQM; 1996; respectively), uncertainties in the traceability Johnson et al., 1998b; Hooker et al., 1998). to NIST for intercomparisons of spectral lamp Prior to the fieldcommissioning of the original irradiance and sphere radiance improved from SQM, many aspects of sensor performance 7–8% to 1–2%. The fourth through sixth SIRREX were maintainedby the instrument manufacturer activities further investigatedlaboratory andfield andwere not routinely measuredby the indivi- protocols (Johnson et al., 1996, 1999a, b;andRiley dual investigator. The first operational deploy- andBailey, 1998 , respectively), andshowedthat ment of the SQM demonstrated the importance of calibrations at an uncertainty level of approxi- independent evaluations of commercial equip- mately 2% were routinely achievable if the ocean ment. During that deployment, large changes in optics protocols were carefully implemented. This the responsivity of some of the radiometers (as culminatedin a detailedexperiment to quantify much as approximately 25% over a 1 month many sources of uncertainties not thoroughly period) were detected (Hooker andAiken, 1998 ). investigatedduringprevious activities at a single The SQM has a demonstrated capability of (commercial) calibration facility (Hooker et al., monitoring the stability of sensors to 2002a). Subsequent calibration experiments have within 1% in the field( Hooker andMaritorena, been supportedby the SPO ( Zibordi et al., 2002a) 2000) andwas sufficiently successful to be com- andthe SIMBIOS program ( Meister et al., 2002, mercializedby two different companies ( Hooker, 2003). 2002). One of the original concepts to be testedin the Regardless of the performance of an instrument SIRREX activity was to verify the sources and in the field, the absolute characterization of a calibration setup procedures at individual calibra- sensor in the laboratory is the starting point for tion facilities for both spacecraft andin situ almost all other subsequent evaluations. After the instruments. To do so required an accurate, stable, conclusion of investigating uncertainties in radio- andportable radiometer, a so-calledtransfer metric calibrations, the immersion factors of radiometer, designed specifically for SeaWiFS irradiance sensors were investigated during SIR- calibration applications. The SeaWiFS Transfer REX-8 (Zibordi et al., 2002a), andwere foundto Radiometer (SXR) was designed and built by be a significant source of uncertainty (more than NIST as part of an Interagency Agreement with 10% in the blue domain, and approximately 2–6% the SPO. The SXR has been proven to be a reliable in the green andredregions). The activity also transfer radiometer, with an uncertainty at all showedit was possible, however, to maintain a 1% ARTICLE IN PRESS

18 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 uncertainty budget for characterizing immersion 2.2.2. Field instrument development and evaluation factors (Zibordi et al., 2004a), especially with new Although the original calibration andvalidation protocols basedon a smaller laboratory apparatus planning did not consider instrument technology andmore time efficient procedures( Hooker and development, the limitations of the existing com- Zibordi, 2003a). In particular, the Compact mercial laboratory andfieldradiometers and Portable Advanced Characterization Tank (Com- instrument monitoring devices for ocean-color PACT) method, which uses a very small water remote sensing applications became apparent vessel (3 l versus as much as 3000 l for traditional fairly early in the program. As a result, the CV methods) provides the significant advantage of a element began investing in instrument evaluation quality-assuredandreproduciblevolume of pure in an effort to accelerate instrument design water (Zibordi et al., 2003). improvements andto identify subtle problems in The uncertainties associatedwith dataproces- instrument performance andcharacterization. sing are tiedto the protocols, but there are This activity is closely tiedto the calibration subjective aspects, like the choice of the in-water roundrobin andmeasurement protocol activities. extrapolation interval, which are not completely For example, the results of SIRREX-8 showed resolvedby a single protocol. The first SeaWiFS that the immersion factors suppliedby a commer- Data Analysis RoundRobin (DARR-94) investi- cial manufacturer were more than 10% in error at gateddataprocessing uncertainties andshowed some wavelengths (Zibordi et al., 2002a), andthere differences in commonly used data processing are other examples of the need for independent methods for determining primary optical para- confirmation of performance specifications in the meters from in situ light data were about 3–4% of literature (e.g., Mueller, 1995, Hooker andMar- the aggregate mean estimate (Siegel et al., 1995). itorena, 2000). These results appliedprimarily to Case-1 waters, This activity has ledto several improveddesigns but issues in turbidwaters remained.The focus of for both above- andin-water measurements. Fig. 4 the secondDARR (DARR-00) was to determine if is a mosaic of some of these instruments, and these results couldbe improved( Hooker et al., Table 2 provides information on the progression 2001). In terms of overall spectral averages, many of designs for both types. In each case, the basic of the DARR-00 intercomparisons were to within strategy was to move towardsmaller instrument 2.5%, andif the processing options were madeas packages andimprovedradiometric performance similar as possible, agreement to within less than andmeasurement configuration knowledge.Also, 1% was possible. for the in-water measurements, higher vertical The optical parameters do not account for all of resolution andsmaller size was a goal, so the the validation requirements. For instance, the instruments couldbe usedin more turbidwaters determination of chlorophyll-a concentration is where vertical gradients and instrument self- central to the SeaWiFS program pigment objective shading can be problematic. For example, in the of agreement to within 35%. An initial intercom- progression from the LoCNESS to the micro- parison of four laboratories using four different NESS, the in-air weight was reduced by almost a high performance liquidchromatography (HPLC) factor of 6, the cross-sectional area was reduced by methods for determining total chlorophyll-a con- almost a factor of 2, andthe full-scale depth centration showedthe overall accuracy of the four accuracy was improvedby a factor of 25. The methods in predominantly mesotrophic waters was microNESS andmicroSAS instruments now pro- within 8% (Hooker et al., 2000a, Claustre et al., vide highly accurate measurements of LWN and 2004). Subsequently, a more extensive pigment remote sensing reflectance (Rrs) in Case-1 and roundrobin of US laboratories was conductedby Case-2 waters from any standard deployment plat- the SIMBIOS Project (Van Heukelem et al., 2002) form. Furthermore, as a result of rigorous refine- andan international roundrobin jointly supported ments to above-water measurement protocols, by the SeaWiFS andSIMBIOS Projects is under- recent comparisons of above- andin-water mea- way (Hooker et al., 2003a). surements no longer differ by a methodological ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 19 bias—they agree to within the uncertainty in also provided a conductivity–temperature–depth calibrating the radiometers (Hooker et al., 2003b, (CTD) system with 30 l bottles to allow sufficient Zibordi et al., 2004b). water collection in highly oligotrophic waters, which was usedon four of the first nine cruises 2.2.3. The SeaWiFS project field program from September 1995 to June 1999. The instrument development and evaluation The AMT optical experiments were designed to activities andmeasurement protocol experiments compare a variety of deployment techniques used discussed above were linked to an extensive field to measure the in situ light fieldandto incremen- program. The fieldprogram revolvedaroundthree tally improve the methods and instrumentation primary sets of fieldstudies(collaborations). These employed. Both above- and in-water sensors and were the British Atlantic Meridional Transect methods were evaluated during AMT cruises, but (AMT; Aiken et al., 2000) program (Plymouth the latter constitutedthe majority of the effort. Marine Laboratory), studies from the Acqua Alta The culmination of the in-water experiments was a Oceanographic Tower (AAOT) in the Northern demonstration that the in situ part of the SeaWiFS Adriatic Sea (the Joint Research Center, JRC, in uncertainty budget (3.5%) could be satisfied with a Ispra, Italy), andvarious cruises in the Mediterra- dedicated effort of recurring calibrations, stability nean Sea, northwest African , andSouth monitoring, andstrict adherenceto the ocean Africa (Laboratoire d’Oceanographie! de Ville- optics protocols (Hooker andMaritorena, 2000 ). franche, LOV, inVillefranche-sur-Mer, France). Fieldexperiments with the JRC were conducted Each collaboration was formalizedwith a Letter at the AAOT, which is locatedapproximately of Agreement (LOA) between NASA andthe 15 km southeast of the city of Venice in the international institution. Table 3 provides a northern Adriatic Sea. The potential for using summary of the major fieldcampaigns in which oceanographic towers, as an alternative to buoy or the SPO participated. shipboardmeasurements, for ocean-color calibra- The AMT cruises took place on boardthe Royal tion andvalidationactivities is being realizedat Research Ship James Clark Ross when it steamed the AAOT (Zibordi et al., 1995). If the observa- between Englandandthe FalklandIslandsin tions are periodic (monthly), short deployments support of British Antarctic Survey (BAS) activ- are easily accomplishedwith the former ( Zibordi ities in the Southern Hemisphere. The odd- et al., 2002b), andif a modularin-water system numbered, southbound cruises sampled the boreal that allows the removal of the light sensors in autumn andaustral spring; while the even- between measurement campaigns is used, the most numbered, northbound cruises sampled the boreal difficult aspect of moored systems—bio-fouling of spring andaustral autumn. Because of the the in-water optical sensors—can be completely geographic extent of the transects (more than eliminatedanda high-quality time series can be 100 of latitude and 50 of longitude), the produced (Berthon et al., 2002). repetitive scheduling of the cruises (two per year Towers andresearch vessels are necessarily large lasting more than 30 days each), the diversity of structures, andthe primary advantage of the the environments encountered(oligotrophic gyres former over the latter is stability. Optical instru- to upwelling zones andeutrophic coastal regions), ments can be deployed on a tower with virtually no along with the use of the newest radiometer tilt andthe solar illumination geometry, which is designs (including calibration monitoring in the needed for an accurate removal of superstructure fieldwith the SQM), the AMT Program was a shading effects on in-water measurements (Zibordi particularly timely andsubstantial accomplish- et al., 1999; Hooker et al., 2002b; Doyle et al., ment. Deployments on the AMT cruises were 2003), can be accurately determined. The stability greatly simplifiedby having the SeaWiFS Portable and absence of in-water data degradation (no bio- Laboratory secured to the deck during the fouling plus the use of in-water correction schemes operations andstoredeither in the UK or the for perturbative effects or easily implemented FalklandIslandsbetween deployments. The SPO avoidance metrics for above-water measurements) ARTICLE IN PRESS

20 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 21 makes AAOT deployments excellent opportunities large offshore tower anddevelopedacorrec- for above- andin-water intercomparisons to tion scheme for the in-water sensors. examine many different aspects of measurement 2. Claustre et al. (2002) demonstrated that protocols. A 1 year intercomparison of water- Saharan dust deposition reduced blue reflec- leaving radiances derived from SeaPRISM and an tance andenhancedgreen reflectance causing in-water system, for example, showedthe overall anomalously high chlorophyll retrievals. spectral agreement was within 10% in the blue– 3. Hooker andMorel (2003) quantifiedship green channels (Zibordi et al., 2002c). Recent reflectance contamination in above-water mea- improvements in the above-water methodology surements andoutlinedmeasurement protocol that incorporate bidirectional corrections and a andquality control proceduresfor retrieving more accurate modeling of the surface reflectance data accurate to within 5%. (Hooker et al., 2003b), however, show a time series 4. Hooker et al. (2002b) conducted the first of data can be constructed using an autonomous rigorous comparison of above-water measure- above-water system that has an uncertainty in ment andin-water measurement methods. The keeping with calibration andvalidation uncertain- overall intercomparison of all methods across ties (Zibordi et al., 2004b). Case-1 andCase-2 conditions was at the 9% Additional field campaigns conducted with the level for the spectral averages. Laboratoire d’Oceanographie! de Villefranche have 5. Hooker et al. (2003c) mappedthe effect of an included the Productivite! des Systemes" Oceaniques! offshore tower on the above-water methodand Pelagiques! (PROSOPE) deep-ocean cruise to the establishedsampling metrics to prevent data Mediterranean Sea and the northwest African degradation (Hooker andZibordi,2003b ). If upwelling plus a satellite calibration cruise to the the sampling metrics are combinedwith the Benguela Current off South Africa (calledBEN- Hooker et al. (2003b) methodology, the above- CAL; Barlow et al., 2003). In addition, the andin-water determinations of water leaving SeaWiFS andSIMBIOS Projects have contributed radiances converge to within the total uncer- to the development of a new type of optical buoy tainties in the methods, about 3% (ignoring called Bouee! pour l0acquisition de Series! Optiques a" environmental variability). Long Terme (BOUSSOLE), which was operation- ally deployed in the Ligurian Sea between France In addition to improving the radiometric andCorsica, andis basedon commercial-off-the- reliability andaccuracy of the optical instrumenta- shelf sensors (Antoine andGuevel, 2000 ). tion, the development process greatly increased the Combining the careful metrology established amount of data collected during the field cam- during AMT cruises with the unique and well- paigns. For example, if the off-the-shelf approach establishedcapabilities of the LOV andJRC of AMT-1 is usedas a reference, there was groups resultedin several additionalnoteworthy approximately a 100% increase in the number of accomplishments: radiometric profiles collected during AMT-3, a 600% increase during AMT-5, and a 900% 1. Zibordi et al. (1999) measuredthe shading increase during AMT-7. Some of these increases induced on in-water optical measurements by a were the result of efficiencies that are always

Fig. 4. A collage of the primary above- andin-water optical instruments developedto support SeaWiFS calibration andvalidation fieldcampaigns: (A) SeaOPS (the inset magnification shows the orientation of the three optical sensors); (B) THOR (heldupright) with LoCNESS and miniNESS (on deck) in front of the SeaWiFS Portable Laboratory; (C) microNESS being operationally deployed for the first time from a very small (less than 5 m long) boat; (D) SeaSAS configuredto permit the intercomparison of three different above-water sensor designs (including the prototype microSAS sensor); (E) SUnSAS with the DIR-10 (directional) unit (the vertical, leftmost cylinder) and the viewing aperture open (a 25.4 cm plaque can be fitted into the aperture to allow for alternative measurement protocols); (F) microSAS fitted inside a cardanic gimbal and mounted at the end of an extensible deployment system built at the JRC; (G) SeaPRISM being operationally deployedatthe AAOT after 1 year of fieldtesting by the JRC; and(H) fieldtesting of the first gimbaledsolar irradiancereference on a very small boat. ARTICLE IN PRESS

22 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42

Table 2 Summary of field instruments developed with support of the SeaWiFS calibration and validation program

Instrument Characteristics References

In-water optical instruments * SeaWiFS Optical Profiling System Luðz; e.Þ; Ed ðz; e.Þ; and Robins et al. (1996) * þ (SeaOPS) Esð0 ; e.Þ; initially Aiken et al. (1998) * Euðz; e.Þ added later Hooker andMaritorena (2000) * Winch andcrane system (for large ships andplatforms) with 16-bit analog-to-digital (A/D) conversion SeaWiFS Free-Falling Advanced Light * Free-falling, 24-bit A/D conversion, Aiken et al. (1998) Level Sensors (SeaFALLS) integral unit (not modular) Hooker et al. (1999) Hooker andMaritorena (2000) SeaWiFS Square Underwater Reference * Two different surface reference Aiken et al. (1998) Frame (SeaSURF) andthe SeaWiFS systems that can be floatedaway from Hooker andLazin (2000) Buoyant Optical Surface Sensor a platform Hooker andMaritorena (2000) (SeaBOSS) Low-Cost NASA Environmental * Same as SeaFALLS, but with cheaper Aiken et al. (1998) Sampling System (LoCNESS) components (16-bit A/D) and Zibordi et al. (1999) modular Hooker et al. (1999) Hooker andMaritorena (2000) * THOR Luðz; e.Þ; Ed ðz; e.Þ; Euðz; e.Þ Hooker et al. (1999) * Longer than SeaFALLS Hooker andLazin (2000) * Built for Q-factor studies Barlow et al. (2003) miniNESS * Smaller than LoCNESS Hooker et al. (1999) * Suitable for small boats Hooker et al. (2000b) Doyle et al. (2003) microNESS * Smaller than miniNESS Hooker et al. (2003c) * Suitable for moderately turbid and Barlow et al. (2003) shallow water * Digital interfaces

Above-water optical instruments SeaWiFS Surface Acquisition System * Suitable for platforms andlarge ships Hooker et al., 1999 (SeaSAS) * Manual pointing Hooker andLazin (2000) Hooker et al. (2002b) SeaWiFS Shadow band (SeaSHADE) * Rotating shadow band sun Hooker andLazin (2000) photometer for attachment to Hooker et al. (2003c) irradiance reference sensors SeaWiFS Underway Surface Acquisition * Suitable for platforms andsmall ships Hooker andLazin (2000) System (SUnSAS) * Manual pointing Hooker et al. (2000b) Hooker andMorel (2003) Micro Surface Acquistion System * Suitable for very small ships Hooker et al. (2003b) (microSAS) * Gimbaledto reduceplatform motion Hooker et al. (2003c) Hooker andZibordi(2003b) SeaWiFS Photometer Revision for * Stable platforms only Hooker et al. (2000b) Incident Surface Measurement * AERONET compatiblewith Zibordi et al. (2002c) (SeaPRISM) automatedsatellite datatransmission Zibordi et al. (2004b)

The listings under the in-water and above-water categories are in chronological order and represent systematic improvements in radiometry, reductions in size, and, for the above-water instruments, more accurate pointing knowledge. The in-water systems measure þ upwelling radiance (Luðz; lÞ), downwelling irradiance (Edðz; lÞ), and downwelling surface irradiance (Esð0 ; lÞ; separately). The Three- Headed Optical Recorder (THOR) includes upwelling irradiance (Euðz; lÞ). The above-water instruments measure surface upwelling radiance and sky radiance. ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 23

Table 3 A summary of the major deep ocean and coastal field campaigns directly supported by the SPO

Campaign Investigator(s) Dates Stations

AMT-1 S. Hooker andG. Moore Sep.–Oct. 1995 25 AMT-2 G. Moore andS. Hooker Apr.–May 1996 24 AMT-3 S. Hooker, J. Aiken, andS. Maritorena Sep.–Oct. 1996 30 AMT-4 S. Hooker andS. Maritorena Apr.–May 1997 56 AAOT S. Hooker andG. Zibordi July 1997 6 AMT-5 S. Hooker, J. Aiken, andS. Maritorena Sep.–Oct. 1997 47 AMT-6B G. Moore, S. Hooker andS. Maritorena Apr.–May 1998 30 AMT-6 S. Hooker, J. Aiken, andS. Maritorena May–June 1998 61 SeaBOARR-98 S. Hooker andG. Zibordi July 1998 6 AMT-7 S. Hooker, J. Aiken, andS. Maritorena Sep.–Oct. 1998 57 AMT-8 S. Hooker andS. Maritorena May–June 1999 52 SeaBOARR-99 S. Hooker andG. Zibordi August 1999 4 PROSOPE S. Hooker, A. Morel, andS. Maritorena Sep.–Oct. 1999 21 Coastal-1 S. Hooker andJ. Brown Feb.–Mar. 2000 17 SeaBOARR-00 S. Hooker Apr.–May 2000 30 ADRIA-2000 S. Hooker, J-F. Berthon, andG. Zibordi July 2000 55 Coastal-4 S. Hooker andJ. Brown Feb.–Mar. 2001 29 SeaBOARR-01 S. Hooker andG. Zibordi June 2001 18 BOUSSOLE S. Hooker andD. Antoine July 2001 7 SeaBOARR-02 S. Hooker andG. Zibordi June 2002 15 BENCAL S. Hooker, J. Brown, J. Aiken, andA. Morel October 2002 42

The SeaWiFS Bio-Optical Algorithm RoundRobin (SeaBOARR) campaigns are associatedwith the general problem of investigating how methodological factors influence the data used in bio-optical algorithms.

gainedfrom repeating any exercise over andover of rapidprototype development andextensive again, but the majority were the direct conse- testing, andnew technology evaluation, the system quence of modifying the equipment and methods has been continually modified to handle additional used. These modifications included a custom-built requirements (i.e. new products, algorithm refine- data acquisition capability to permit the rapid ments, increaseddatavolumes, andnew satellite collection of radiometric data from the simulta- missions) while also improving the system’s cap- neous deployment of multiple instruments. ability to meet the operational mission requirement without requiring additional budget supplements 2.3. Data processing system over what hadoriginally been requested. The original objective was to have an ability to The design of the SeaWiFS data processing process 10 times the receiveddatavolume so as to system (SDPS) incorporatedmany of the lessons have the ability to reprocess the entire data set learned during the development and operation of within a reasonable amount of time without the CZCS global reprocessing (Feldman et al., affecting operational processing. As mentioned 1989). The initial design was a close collaboration earlier, the original SPO-wide computing design between the SPO andthe University of Miami was three Unix servers connectedto X-terminals. Rosenstiel School for Marine andAtmospheric Even during the prelaunch phase, this strategy Sciences (RSMAS), with the invaluable technical provedinadequate to meet the computational support from Silicon Graphics Incorporated(SGI). requirements andthe SPO replacedthe X-term- By designing the system to satisfy a number of inals with workstations. The approachedworked well-specifiedrequirements andthrough a process very well, but was expensive in terms of hardware ARTICLE IN PRESS

24 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 andmaintenance. Since launch, relatively inexpen- up approach designed by scientists for scientists) sive, yet powerful, PC-basedLinux systems have rather than by a more computer andinformation emergedandare systematically replacing the system approach, i.e. a top-down approach. The staff’s workstations . The Linux systems are also DP element has to interact with a large number of being clusteredto augment the Unix servers to groups to be able to carry out its functions. Some increase overall operational processing throughput of these are internal to the SPO (i.e. the other as HRPT station data volume and calibration and elements), some internal to NASA, but physically validation evaluation processing requirements separatedfrom the SPO (e.g., DAAC, WFF) and grow. In addition, process-control software has others often halfway aroundthe world,such as the evolvedfrom relying solely on the central servers remote HRPT receiving stations that collect and to being able to utilize all computer resources provide the SPO with copies of the SeaWiFS data within the SPO. they receive. The original products to be derived from the raw data stream (Darzi, 1998) included navigated 2.3.1. System requirements calibrated radiances (level-1), derived geophysical The five key functions that drove the design of products (level-2; normalized water leaving ra- the SDPS requiredthat the system are the diances, chlorophyll-a, CZCS pigment, diffuse following: attenuation at 490 nm, andcertain atmospheric Process large volumes of data in a timely fashion. parameters), binnedproductson a fixedgrid By using realistic estimates of processing/reproces- (level-3), and standard mapped products. Level-2 sing rates, data handling requirements, and products included a number of masks and flags through very extensive simulation of actual end- that are usedto identifypixels that are not to-endsystem operations, potential bottlenecks in processed (e.g., clouds) or indicate some special system throughput were identified and corrected circumstance, e.g., shallow water (McClain et al., prior to launch. The products and their volumes 1995). The mask andflags are incorporated were known andusedin the system design.The in the level-2 products and can be individually instrument- and spacecraft-driven delays in displayed. Some flags are used as exclusion launch were taken advantage of by the SPO criteria for level-3 binning. Changes in the pro- to simulate full mission functions from downlink ductsuite andthe masks andflags have been to DAAC distribution, day in and day out for at institutedat various reprocessings. For example, least one full year before launch. The end-to-end the CZCS pigment product has been discontinued, testing was possible because the MO element andPAR andNDVI have been added.Also, after generatedsimulatedlevel-0 datasets ( Gregg et al., the thirdreprocessing, a set of evaluation pro- 1994). It cannot be overemphasizedhow abso- ducts, e.g., CDOM, were included in the routine lutely critical the comprehensive prelaunch system processing stream in collaboration with the tests were to the generation of credible level-1, -2, scientists outside the SPO. Evaluation products and -3 products on the first day of routine are displayed on an evaluation product Web site operations. for consideration by the community. Once the Require as little human intervention as possible. community expresses sufficient interest in a pro- While the interactive quality control of each and duct, it is graduated to archive product status as every SeaWiFS data product was a key design was done with PAR. NDVI is supplied to a requirement, the interaction between the software terrestrial research group at GSFC who quality developed by the CV group to perform the quality control the product before sending it to the control andthe DP element to dealwith the DAAC. outcome of that process was optimizedthrough Because the SDPS was designed from the very the use of shared database tables and procedures. start to be a thoroughly integratedcomponent of Because the SPO had decided very early in the the SPO, driven by science, data quality and data planning process to maintain its own archive of the availability considerations (essentially a bottoms- complete mission data (in this case both level-0 ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 25 andlevel-1a), algorithm testing on large datasets, case, as an operational system to process Ocean- product validation subscene extractions, and color andTemperature Scanner (OCTS) dataand reprocessings couldproceedwithout many of the for MODIS, as a prototype data processing system coordination issues that have impeded other larger to demonstrate the ability to process MODIS data andmore complex missions. outside of the Earth Observing System (EOS) Allow changes to the processing methods to be DistributedInformation System (EOSDIS) Core easily implemented. By acknowledging that System (ECS). as understanding of the instrument and the science improves, there will be a needto modify 2.3.2. System design the algorithms andprocessing procedures, andthe The design philosophy that was used to develop system shouldbe able to accommodate these the SDPS was quite different from most other changes in a straightforwardmanner without major satellite missions. Once a top-level system affecting the operational processing. By designing requirement was identified, rapid prototyping of the system to have parallel development and basic core functions was usedto developthe key operational environments, changes to the system components of the SDPS. Once the prototype procedures could be easily implemented, thor- function hadenough capability to be tested,a oughly testedandevaluated,andmigratedinto comprehensive evaluation of it was carriedout, production with minimal impact. strengths and weaknesses were identified, and Permit multiple processing streams. It is com- refinements were made to eventually converge on pletely unrealistic to design a satellite data proces- a final implementation. As a result, the function- sing system without providing for the concurrent ality developed at the very beginning of the processing of multiple streams of data. For the SeaWiFS program evolvedover time to incorpo- SDPS, the system was scopedfor a minimum of at rate a wide range of additional requirements. The least three simultaneous andconcurrent processing four major components of the SDPS are the (1) streams for real-time operational processing, database, (2) the Services Layer, (3) the Scheduler, complete mission reprocessing, anddevelopment and(4) the Visual Database Cookbook (VDC). andtesting processing. This analysis resultedin a The heart of the SDPS is a relational database design goal of being able to have a 10 overall management system (RDMS), which provides all system capacity (process ten days of global data the controlling, scheduling, and cataloging func- per day). tions for the system. The SDPS uses a commercial Be easily understood and documented. Most off-the-shelf software (COTS) standard query interactions with the SDPS are carriedout through language server (Sybase), but any RDBMS can a series of simple graphical user interfaces (GUI) be usedbecause of the system’s database Services that interact directly with the various databases Layer, which contains all the vendor-specific that control all system functions so that even very database functions. If it were decided to use complex procedures such as redefining how a data another RDMS, only the Services Layer would file should be processed, what data products are to need to be updated because it shields all the other be produced, and at what step in the processing system functions from any RDMS-specific depen- each data granule resides at any moment is a dencies. There are two types of databases within relatively simple, andeasy to manage task. the SDPS: core databases and mission-specific Extensive Web-baseddocumentation of all key databases. The core databases support the proces- system components, public access to most system sing, cataloging, andadministration functions source code, and a complete, online log of system- within the SDPS, while the processing database’s relatedupdatesmake the SDPS a very open, easily primary function is to provide controlling and understood system. As a demonstration of this, the scheduling functions for the SDPS. entire SDPS has been portedto at least two other For scheduling, a table called todolist is used. groups (National Oceanic andAtmospheric Ad- This table contains records that describe jobs ministration andMODIS) to serve, in NOAA’s (tasks) that need to be done for the current day. ARTICLE IN PRESS

26 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42

The task’s attributes define the characteristics of virtual CPU is a work space defined on a machine the job such as its type, when it shouldbe run, that can be usedby the VDC. This work space is which machine it shouldrun on, andits status. one or two directories, defined as input and output Tasks can run at set intervals of time (monitor buffers, which will contain all of the files needed tasks), at a specific time (timedtasks), or be and produced by the individual steps of the triggeredby some event (triggeredtasks). The processing. The work spaces are defined in the Scheduler monitors the records in the todolist hosts andresources tables andare allocatedand table, andsubmits jobs to the operating system at freedautomatically by the VDC. Dependingupon the appropriate time. When a task completes, the capabilities of each computer defined as a successfully or unsuccessfully, it records its status resource, multiple virtual CPUs can be allocated in the todolist table. At any point, a new task can on any given machine. Currently, the SPO has be added, unwanted tasks suspended or deleted, or approximately 90 virtual CPUs to call on to meet tasks rescheduled through the Task Editor GUI. its processing needs, although the actual number Tasks not completedon any given daymay be of computers is significantly less. There are three rolledover to the next daywhen the system basic functions associatedwith the VDC: En- reconfigures itself each night at midnight. For trance, Master, andExit. The Entrance program is instance, data transfers between remote receiving the entry point into the VDC, andit monitors the stations andthe SPO can often take several hours entrance directory for new data to process, depending on the network links, so tasks not allocates the necessary resources, reads the job completedat midnight are carriedover andshow commandfile, stages the datafor processing, and up on the next day’s todolist. creates a stream recordthat signals the VDC to This scheme works well for high-level tasks and begin automatic processing. The Master program simple processing, but for low-level, complex is, in a sense, the master chef for the VDC. It processing (multi-step processing of SeaWiFS data monitors the Streams table for stream records that from level-1 through level -3 for example) , the are ready to be processed, marks them as busy, VDC is more suitable. When a file is queuedfor andinvokes the processing jobs according to the VDC processing, a new recordis insertedinto one steps of the assignedrecipe. Essentially, it moves a of the processing control database tables (the job from one step to the next until it is completed. ‘‘activeproc’’ table) where it waits in line until all The Exit program is the final stage of the VDC. of the ancillary data needed for processing have It monitors the Streams table for stream records been identified and staged. Once this occurs, the that have completedprocessing, de-allocates the file is admitted into the VDC and passed along to resource making it available for another job, any of the system resources available for proces- records the completion time, deletes the input data sing. One of the key design features of the SDPS is file, andmarks the stream recordreadyfor its distributed processing environment with system deletion. allocation of any/all available resources ranging A recipe is a processing scheme that consists of a from a multiprocessor SGI to desktop Linux set of steps, each step performing a separate task. machines. The system is scalable in size andis The VDC invokes the steps sequentially, just as a able to take advantage of resources as needed. For chef wouldfollow a cooking recipe. The recipes instance, many of the desktop machines used by are defined in the ‘‘recipe lookup’’ andrecipes SPO staff are available for processing when either tables. Because the recipe steps are designed to the loadon them is sufficiently low or when they perform as small a set of functions as possible, one have been idle for a predetermined amount of step in the recipe table can be usedby a large time. A simple entry in one of the database tables number of different processing scenarios. One can add or remove these resources from the step, CPYVDC, does nothing but copy all of the available pool. input data files, ancillary files, and parameter files The VDC uses two abstract concepts: a virtual into the processing directory and is used by many computer processing unit (CPU) anda recipe. A different recipes. ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 27

2.3.3. System evolution Hawaii. These sites could distribute real-time data The design goal of 10 (for global GAC) for validation cruise support with SPO permission. processing was met at launch and, through a To assist in validation efforts and cruise planning, combination of processor andnetwork speed the SeaWiFS andSIMBIOS Projects jointly advances, additional processors (mostly desktop developed a real-time user support Web site, systems), andprocess control software that dyna- which allowedthe user community to obtain mically utilizes all available computing resources overflight predictions and real-time data products within the SPO, the system has realizedthe for cruise support and demonstration studies. As a following throughput improvements: secondre- result, real-time data products have been provided processing (August 1998), 40 ; thirdreprocessing to nearly 400 fieldcampaigns. (May 2000), 130 ; fourth reprocessing (May Because the SeaWiFS mission was considered to 2002), 400 . Recent system benchmarks show a be an early component of EOS, the official data 3000 throughput which translates to a complete archive anddistribution functions were requiredto regeneration of all the standard archive products be handled by the EOSDIS Version 0 DAAC for the entire mission in less than 1 day. locatedat GSFC. The DAAC serves as the long- term archive anddistribution facility for all 2.4. Data distribution SeaWiFS data once the data has passed the 14- day embargo period, because there is a great deal Under the contract with ORBIMAGE, only of valuable science that needs to be conducted authorizedSeaWiFS researchers may have access using real-time or near-real-time data. In response, to the digital SeaWiFS data for research and the SPO developed an extensive Web-based set of educational purposes, and the SPO is responsible user-friendlybrowse, order,andimage anddata for providing the DAAC with a list of these distribution tools to service the science commu- authorizedresearchers. During the early part of nity’s needfor access to this dataprior to it being the SeaWiFS mission, researchers were requiredto archivedat the DAAC. In a typical month, submit a written proposal to the SPO for review, SeaWiFS sends 60–80 gigabytes of compressed and if the intended use of the data fell within the data to the DAAC and the DAAC distributes agreedupon activities outlinedin the contract, approximately 500–600 gigabytes (compressed then that researcher was added to the list and the volume), or about nine times the archivedvolume. DAAC was notified. Recently, this process has Acker et al. (2002) provide a comprehensive review been streamlinedthrough the provision of an of the DAAC’s SeaWiFS product distribution online request form where potential SeaWiFS data statistics. users can submit all the requiredinformation electronically, andwhere after review andap- 2.5. Outreach proval, acceptance messages are passedto the DAAC andto the researcher. This process is Outreach has many forms, several of which are generally completedon the same day that the mentionedabove, e.g., near-real-time support submission is made. While most requests are services. Other activities have been focusedon generally approved, submissions from commercial simply advertising the SeaWiFS program and the interests are referreddirectly to ORBIMAGE. applications of the data. The SPO has worked Since launch, there has been an almost constant closely with the GSFC Office of Public Affairs to rate of increase in the number of authorizedusers, provide coverage of newsworthy events to the which now exceeds 2300. news media. Also, for the first 3 years after launch, The original contract with Orbital Sciences the SPO put much emphasis on having a presence provided NASA with 13 real-time licenses that at all major Earth science-relatedconferences (US allowed decryption as the data were received. andinternational) by providing presentations or Aside from the SPO, licenses were distributed to an information booth to familiarize different stations in Mississippi, California, Alaska, and sectors of the potential user community with the ARTICLE IN PRESS

28 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 mission. Two outreach activities have been parti- andsource code.One important benefit from the cularly important, SeaDAS andthe STRS, which early prelaunch release of the SeaDAS source code are described in detail below. was that it enabledthe user community to examine SeaDAS: One of the lessons learnedfrom the the code and provide feedback to the SPO, e.g, CZCS experience was the needfor the user coding errors, and revised or new features. Having community to have an affordable and easy-to-use the processing flow documented before launch data processing capability so that the remote (Darzi, 1998) made the process of reviewing the sensing data products could be tailored to their code much easier. SeaDAS also supports the research andapplications. Also, having a hands- processing of CZCS, OCTS, Ocean Scanning on capability allows the user to become much Multispectral Imager (OSMI), andModularOpto- more familiar with the processing, helps connect electric Scanner (MOS) data, the display of all the research community to remote sensing tech- MODIS oceans products, and the generation of nology andmethodologies, andallows the user to SeaWiFS NDVI fields. SeaDAS has been down- integrate alternative algorithms andanalysis rou- loaded to more than 500 unique user sites in over tines into the software. SeaDAS (Fu et al.,1998; 45 countries since the fourth reprocessing and Baith et al., 2001) was an outgrowth of the receivedthe 2003 NASA Software of the Year SEAPAK system (McClain et al., 1991a, b, Award. 1992a) and was designed to meet this need, al- SeaWiFS Technical Report Series: Although the though it was not considered a project requirement CZCS mission andthe CZCS global reprocessing and did not have an explicit budget line. This being effort (Feldman et al., 1989) provided much of the the case, it was funded through the NASA Ocean blueprint for setting up the organization and Biogeochemistry Program with assistance (system responsibilities of the SPO, most of this important administration and equipment) from the SeaWiFS anduseful information was outsidethe subject andSIMBIOS Projects. This fundingsupported matter of peer-reviewedpublications andwas not the SeaDAS staff, which has never exceeded three available in a centralizedlocation, i.e. it was full-time individuals (usually only two). scatteredamongst the original participating in- SeaDAS development was initiated early on in dividuals, institutes, and companies. In some the prelaunch phase with the first user training cases, solution processes to important problems classes being heldin 1994. SeaDAS is not meant to were needlessly repeated, because the relevant be a comprehensive analysis system, but is material, although known to exist, was no longer designed primarily to replicate all the operational accessible. processing procedures and products (level-0 The STRS was createdearly in the development through level-3; GAC andLAC) while allowing phase of the SPO to ensure all the scientific, key processing parameters to be variedas the user technical, andmission-relatedaccomplishments finds appropriate. To do this effectively, SeaDAS andapproaches were documentedandavailable has emphasized product display functionality and from at least one source. It also provided a diagnostic analyses for evaluating the products. publishing forum for members of the global The system initially supportedSilicon Graphics ocean-color community collaborating with the andSun Microsystem Unix workstations andwas SPO. The STRS placeda high priority on quality later adapted to run on PC-based Linux systems. and consistency, and, therefore, standardized The Interactive Data Language (IDL) is usedfor document formatting, nomenclature, and symbols. the user interface, display functions, map projec- Index volumes were included at regular intervals tions, andbasic statistical analyses, which allows to include listings of citations, symbols, updates, the software package to be easily expanded by andcorrections. The STRS consists of 43 pre- users who write their own IDL routines. A run- launch and29 postlaunch volumes, andhas been time license is provided with SeaDAS for those routinely distributed to nearly 500 scientists, who do not plan to write IDL routines. SeaDAS is institutes, libraries, andmarine information cen- distributed in two forms, compiled executable code ters throughout the world. In addition, the ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 29 postlaunch series is available to the general OC4V4 algorithm, was adopted (O’Reilly et al., public worldwide as downloadable files from the 2000). The other major modification in the third SeaWiFS Web site (the prelaunch reports are in reprocessing was the switch to the SPO’s code the process of being convertedto downloadable (calledMSL12) from the original RSMAS level-2 files). Hardcopies of any or all of the pre- and processing code. Summaries of all the improve- postlaunch series are available upon request from ments associatedwith the secondandthird the SPO or the technical editor (Elaine Firestone). reprocessings are in McClain et al. (2000b). Finally, the fourth reprocessing was completedin 2.6. SeaWiFS reprocessings and data quality June 2002 (Patt et al., 2003) andincorporateda improvements revisedvicarious calibration basedon recalibrated data from one of the MOBY systems, which During the first 5 years of SeaWiFS operations, accountedfor stray light in the MOBY spectro- the SPO conducted four reprocessings. These were meter. A subsequent calibration of the second initiatedat the request of the CV element with operational MOBY system showedonly a slight input from the user community when improve- difference from the values derived for the first ments in the algorithms, fielddata,or sensor system. Also, a revisedNIR correction that characterization were demonstrated to have a incorporates the backscattering model of Gould significant effect on data quality. The reproces- et al. (1999) was adopted. These improvements sings were preceded by extensive analysis and further reduced the occurrence of negative water- evaluation which was sharedwith the user com- leaving radiances in turbid and coastal waters. munity for comment through workshops andWeb The quality of the data products is evaluated sites before final approval was given to reprocess. primarily on the comparisons with in situ data In addition, each reprocessing required consider- (Bailey et al., 2000; Werdell et al., 2003) and LWN able coordination between the SPO elements, stability analyses (Eplee andMcClain, 2000 ). The especially the CV andDP elements, as well as LWN andchlorophyll- a match-up comparisons with the DAAC. between the satellite retrievals andin situ data The first reprocessing was executedin January from the fourth reprocessing indicate excellent 1998, shortly after data acceptance, and incorpo- agreement over the entire range of values for all rateda number of codingerror corrections and parameters (Fig. 5). For example, the slope and r2 mask/flag adjustments based on the initial evalua- for the LWN(4 1 2) is 0.992 and0.82, respectively, tions of the data products. It also included the for over 200 match ups spanning a range from initial vicarious calibration basedon MOBY about 0.1 to nearly 3.0 mW cm2 mm1 sr1. This is observations. The secondreprocessing was exe- particularly important because the atmospheric cutedin August 1998 after it was determinedthat correction at this wavelength requires the greatest significant degradation in the near-infrared (NIR) extrapolation from the infraredwavelengths used bands was occurring. The most important im- to select the aerosol model. It is also the most provement associatedwith the secondreprocessing difficult wavelength in the SeaWiFS band set to was the use of the monthly lunar images to determine accurate in situ instrument calibrations estimate the time dependence of the sensor because of the relatively small standard calibration sensitivity (Barnes et al., 1999). The thirdrepro- lamp output. With retrievals of this quality, the cessing was completedin June 2000 andwas 412 nm bandcan now be usedfor new applications initiated to address the continuing problem of andalgorithms, (e.g., Siegel et al., 2003), especially negative water-leaving radiances in coastal areas. in coastal waters, e.g., discriminating viable This problem was addressed by including a NIR chlorophyll from degradation products (the origi- reflectance adjustment in the atmospheric correc- nal purpose for the 412 nm band). For chloro- tion model (Siegel et al., 2000). Also, a new phyll-a, the slope and r2 are 1.03 and0.85, chlorophyll-a algorithm that sequencedthrough a respectively, for 262 match ups over a range of set of three band-ratio relationships, the so-called 0.03 to around20 mg m 3. As a demonstration of ARTICLE IN PRESS

30 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42

Fig. 5. Comparisons of in situ andsatellite LWN andchlorophyll- a values, i.e., match-up comparisons, after the fourth reprocessing.

stability, Fig. 6 shows the annual cycles of average reprocessing in early 2004. As before, the SPO will deep-water (>1000 m) LWN values for 1998–2003. solicit the user community’s input on algorithm The plots show that the 510 and555 nm LWN and product improvements, including additional values are constant andthat the 412, 443, and products to be added to the archive product suite. 490 nm LWN values have regular seasonal patterns with little variation from year to year. This stability in the derived products is based solely 3. Major geophysical events captured by SeaWiFS on the lunar calibrations andis independentof the Earth-viewing data. These results and others are During the first 5 years of global observations, a available from the reprocessing Web site and great many geophysical events have been captured detailed descriptions of the reprocessing and the in the SeaWiFS data set. These range from local relatedanalyses are being documentedin the events such as anomalous phytoplankton blooms STRS. (Florida Bay; the southwest Florida Dark Water Finally, assuming the SeaWiFS data buy ends in Observation Group, 2002), volcanic eruptions December 2003, the SPO will conduct a final (Mt. Etna), wildfires (western UnitedStates and ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 31

Fig. 6. Comparison of annual cycles of global 8-day mean LWN values for depths greater than 1000 m during the years 1998–2003.

Fig. 7. Global seasonal cycle. Because SeaWiFS GAC data collection includes all sun lit portions of the orbit, including land masses, it is possible to derive NDVI and other terrestrial products from SeaWiFS. For example, Behrenfeldet al. (2001) generated global time series of marine and terrestrial primary productivity using SeaWiFS data showing that over the first 3 years of data, ocean productivity increased in the post-El Nin˜ o ocean, while terrestrial productivity remained constant. This four-panel figure shows the 5-year seasonal climatologies of ocean chlorophyll-a andterrestrial NDVI. ARTICLE IN PRESS

32 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42

Central America), and floods (US mid-Atlantic coast after Hurricane Floyd), to global phenom- ena such as the 1997–1998 El Nin˜ o-La Nin˜ a. SeaWiFS data also are being used for freshwater studies, e.g., the Great Lakes (Lesht et al., 2002). Data from the mission have been usedfrequently by the news media, not only because of its ease of access andrapidturnaround,but also because the SPO andthe GSFC Public Affairs staff madeit a high priority to provide information on news- worthy events. In this section, an image gallery with brief descriptions (captions) of some of the events of particular interest to the scientific community are provided. Fig. 7 shows the four seasonal climatologies of ocean chlorophyll-a and NDVI. Figs. 8–12 are sequencedfrom local to basin scale events (see figure captions for descrip- tions). False color images are composedof Rayleigh correctedcomposites of the 412, 555, and670 nm SeaWiFS bands. Fig. 8. Bering Sea coccolithophore blooms. During the summer andfall of 1997, a large coccolithophore bloom persistedin the 4. Summary eastern Bering Sea (Vance et al., 1998). Such blooms are very unusual in the Bering Sea andcausedwidespreadstarvation of SeaWiFS was to be the first in a continuous marine birdsandmammals andalso severely affectedfisheries which use the adjacent Alaskan rivers for spawning. The bloom series of international global ocean-color missions reappearedthe following spring andwas vividlycapturedin this which wouldprovidea high quality climate 25 April 1998 SeaWiFS image. research quality time series of marine biological andoptical properties. Because of the launch slip, OCTS was the first andprovided9months of global coverage. Fortunately, the SeaWiFS data collection began shortly after the endof the OCTS validation. SeaWiFS also has served a similar data record leaving only a 3-month gap. While purpose for a number of other ocean-color SeaWiFS was not launchedearly enough to missions including the European Medium Resolu- support much of the JGOFS program, which tion Imaging Spectrometer (MERIS) andthe was a primary rationale for the mission, the 4-year Japanese Global Imager (GLI). Table 4 sum- delay did allow the SPO to build much more marizes some of the most significant accomplish- comprehensive capabilities within each SPO ele- ments of the SeaWiFS program andunderscores ment, which ultimately allowedthe SPO to meet the point that the program has achieved, even its primary goals andobjectives early in the surpassedin most cases, its original objectives and program. Additionally, because of its bilinear gain goals. design and accurate calibration, SeaWiFS has With support from the SIMBIOS andMODIS proven to be an outstanding tool for terrestrial programs, the SeaWiFS CV element has under- ecology andaerosol research. SeaWiFS didpre- taken a very comprehensive andcoordinated cede the first MODIS instrument on the Terra effort, the components of which must be conti- platform by 2.5 years andthe secondon the nuedin some to-be-determinedmanner if projects platform by nearly 5 years, andwas invaluable for like the NPP/Visible andInfraredImaging Radio- MODIS prelaunch preparations andpostlaunch meter Suite (VIIRS) are to provide climate ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 33

Fig. 9. Asian and Saharan dust events. Because on the bilinear gain design of SeaWiFS, the sensor does not saturate in any of the bands even over the brightest targets. The figure includes examples of dust events off NW Africa (Saharan dust) and over Asia (Gobi dust). Saharan dust has been tracked using SeaWiFS data as far to the northwest as New England. Similarly, Asian dust events have been trackedacross the Pacific to the US west coast ( Husar et al., 2001). The detection of low levels of dust in the imagery continues to be problematic as it escapes the dust and cloud masks resulting in artificially elevated chlorophyll concentrations.

research-quality data consistent with preceding require a determination by both the government data sets from SeaWiFS and MODIS. Also, in the andthe commercial partners to deliveron their future, more emphasis will be placedon providing respective obligations no matter how difficult and broader and more accurate suites of products from costly the tasks may be. Success also requires a ocean-color missions for carbon cycle research and commitment to work openly andmaintain com- coastal zone management, e.g., primary produc- munications no matter how stressful the circum- tivity, particulate organic carbon (POC), CDOM, stances may be. Clearly, the numerous technical calcite, andtotal suspendedmatter (TSM). These problems prior to launch andthe financial burden will require new or improvedalgorithms and,in on both parties were challenging to bear andboth some cases, more accurate atmospheric correc- parties hadopportunities to terminate the mission. tions, especially over turbidwater andin areas In the end, the resolve was there, solutions were with absorbing aerosols. More accurate measure- found, SeaWiFS was launched, and the complete ments, especially in turbidwaters, will be neces- data set was delivered. This achievement was sary. In addition, some products also may require recently acknowledged when the SPO and OSC future missions to make observations further into jointly receivedthe prestigious Pecora Award,a the ultraviolet where instrument calibration is joint awardfrom NASA andthe Department of more challenging. Thus, it is critical that activities Interior, for the mission’s contributions to Earth such as SeaBASS, the calibration roundrobin, the science. protocol development, and the instrument evalua- tions anddevelopment activities be continuedafter the SeaWiFS andSIMBIOS Projects end. Acknowledgements Finally, the SeaWiFS mission has demonstrated that a data-buy approach to obtaining a science The authors acknowledge a number of indi- quality data set can succeed, although success does viduals, groups, and organizations for their ARTICLE IN PRESS

34 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42

Fig. 10. 2002 North Atlantic spring bloom. Early results from the CZCS global reprocessing highlightedthe extent andmagnitudeof the spring bloom (Esaias et al., 1986; McClain et al., 1990) andthe bloom was the subject of the first JGOFS fieldstudy.Of the spring blooms observedby SeaWiFS, the 2002 bloom was particularly intense. This sequence of monthly mean chlorophyll composites illustrates the northwardmigration of the ‘‘green wave,’’ a zonal bandof high biological production that results as solar illumination increases over the course of spring andsummer at higher latitudes.

contributions to the SeaWiFS program. OSC, Bill Townsend, Dixon Butler, and Stan Snyder, Santa Barbara Research Center, andORB- were steadfast in their support. Throughout the IMAGE deserve special recognition for building, process of getting the mission approvedand launching, andmaintaining a spacecraft and launched, several Ocean Biogeochemistry Program instrument that has workedexceptionally well. Managers have assistedandrepresentedthe SPO Several different GSFC engineering groups pro- at NASA HQ (Marlon Lewis, Greg Mitchell, vided critical assistance to OSC in diagnosing Robert Frouin, Jim Yoder, Janet Campbell, John problems with the spacecraft, launch vehicle, and Marra, Chuck Trees, andPaula Bontempi). instrument prior to launch. During the prelaunch Science Applications International Corporation phase, NASA HQ senior managers, particularly (SAIC) andits subcontractors (Science Systems ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 35

Fig. 11. 1997–1998 El Nin˜ o-La Nin˜ a. The most noteworthy event to date during the SeaWiFS mission was the 1997–1998 El Nin˜ o-La Nin˜ a, which was the most intense on record( McPhaden, 1999). The equatorial Pacific ecosystems transitionedfrom extremely low chlorophyll concentrations over the winter of 1997–1998 to the highest concentrations ever recorded during the summer of 1998 (Chavez et al., 1999; Murtugudde et al., 1999). These extremes are illustratedin the January andJuly 1998 monthly composites.

andApplications, Inc., andFuturetech Corpora- DavidAntoine, andAndr e! Morel, co-investiga- tion) have provided a talented and dedicated tors). The SPO is indebted to several members of support staff from the beginning of the SPO. The the MODIS oceans team who have made a variety GSFC Public Affairs Office, particularly Wade of major contributions to the SPO including Sisler, has workedvigilantly with the SPO to Wayne Esaias, Dennis Clark, Bob Evans, Howard quickly get data products to the news media Gordon, and Ken Carder. The SPO’s collabora- during special geophysical events. Similarly, the tions with NIST, mainly Carol Johnson, has been GSFC Visualization Laboratory has provided invaluable to the calibration of not only the many excellent video sequences for press confer- SeaWiFS sensor, but also in situ data. Watson ences andspecial presentations. The CV element Gregg deserves special recognition, as the Mission has workedclosely with a number of international Operations element leader from the beginning of programs, such as the AMT, under LOAs with the SPO through data acceptance. Four indivi- Plymouth Marine Laboratory in the UnitedKing- duals have served as SPO Managers: Bob Kirk dom (Jim Aiken, P.I.), the Joint Research Center oversaw the early phases of the SPO through the in Ispra, Italy (Giuseppe Zibordi, P.I.), and the acceptance of the SeaWiFS instrument; Mary Laboratoire d’Oceanographie! de Villefranche in Cleave guided the SPO through much of the Villefranche-sur-Mer, France (Herve! Claustre, spacecraft acceptance andlaunch; Chuck McClain ARTICLE IN PRESS

36 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42

Fig. 12. The 1998 La Nin˜ a equatorial Pacific bloom time series. While the July 1998 monthly chlorophyll composite indicates the areal extent of the bloom over a month’s time, the bloom was actually moving eastwardrapidlyandwas much smaller in size ( Strutton et al., 2001). This figure shows four 8-day composites of SeaWiFS chlorophyll and the corresponding composites of sea surface temperature (SST) illustrating the high coherence between the two fields and the wave structure associated with the bloom. directed the SPO through data acceptance and the the negotiations for the extended SeaWiFS mis- first three reprocessings; andGene Feldmanhas sion. Finally, the ocean-color research community guided the SPO during the fourth reprocessing and has provided unwavering support and assistance ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 37

Table 4 Antoine, D., Guevel, P., 2000. Calibration andvalidationof satellite ocean-color observations: The BOUSSOLE Project. Summary of major SeaWiFS program accomplishments Proc. Ocean Optics XV, Monaco, October 16–20, 2000. 1. First global biosphere data set from a single instrument. [Available on CD-ROM: Office of Naval Research, 2. Uninterrupted global data reception, product generation, Washington, DC]. andproductarchival andrelease since first dayof on-orbit Asrar, G., Dozier, J., 1994. EOS, Science Strategy for the Earth data collection (18 September 1997). Observing System. American Institute of Physics Press, 3. First mission to operationally use the moon to track sensor Woodbury, NY (119pp). stability on-orbit. Bailey, S.W., McClain, C.R., Werdell, P.J., Schieber, B.D., 4. Four major reprocessings completed. 2000. Normalizedwater-leaving radianceandchlorophyll a 5. Data product accuracy goals surpassed. match-up analyses. In: McClain, C.R., Hooker, S.B., 6. Over 2300 authorizedresearch users. Firestone, E.R., et al. (Eds.), SeaWiFS Postlaunch Calibra- 7. Routine distribution of 9 times the archive product data tion andValidationAnalyses, Part 2. NASA Tech. Memo. volume. 2000-206892, Vol. 10. NASA Goddard Space Flight Center, 8. SeaDAS user sites in nearly 50 countries. Greenbelt, MD, pp. 45–52. 9. Over 70 NASA technical memoranda published and Baith, K., Lindsay, R., Fu, G., McClain, C.R., 2001. SeaDAS, distributed to the user community. a data analysis system for ocean-color satellite sensors. 10. Over 400 fieldexperiments supportedwith coverage EOS, Transactions of the American Geophysical Union 82, predictions and/or real-time data. 202. 11. Over 120 HRPT stations established with data decryption Barlow, R., et al., 2003. In: Hooker, S.B., Firestone, E.R. capabilities. (Eds.), BENCAL Cruise Report. NASA Tech Memo. 2003- 12. Staff participation in over 20 fielddeployments. 206892, Vol. 27. NASA Goddard Space Flight Center, 13. Initiation andsupport of numerous calibration andpigment Greenbelt, MD (64pp). round-robins, field instrument design studies, and Barnes, R.A., Barnes, W.L., Esaias, W.E., McClain, C.R., measurement protocol development activities. 1994a. In: Hooker, S.B., Firestone, E.R., Acker, J.G. (Eds.), 14. Over 1250 cruise data sets ingested into SeaBASS. Prelaunch Acceptance Report for the SeaWiFS Radiometer. 15. Successful data buy contract with Orbital Sciences Corp. NASA Tech. Memo. 104566, Vol. 22. NASA Goddard andORBIMAGE. Space Flight Center, Greenbelt, MD (32pp). Barnes, R.A., Holmes, A.W., Barnes, W.L., Esaias, W.E., McClain, C.R., 1994b. In: Hooker, S.B., Firestone, E.R., Acker, J.G. (Eds.), SeaWiFS Prelaunch Radiometric Calibration andSpectral Characterization. NASA Tech. to the SPO throughout the SeaWiFS program Memo. 104566, Vol. 23. NASA Goddard Space Flight which has been a great encouragement. Center, Greenbelt, MD (55pp). Barnes, R.A., Eplee Jr., R.E., Patt, F.S., McClain, C.R., 1999. Changes in the radiometric sensitivity of SeaWiFS. Applied Optics 38 (21), 4649–4664. References Barnes, R.A., Eplee, R.E., Schmidt, G.M., Patt, F.S., McClain, C.R., 2001. The calibration of SeaWiFS, Part 1: direct Acker, J.G., 1994. In: Hooker, S.B., Firestone, E.R. (Eds.), The techniques. AppliedOptics 40 (36), 6682–6700. Heritage of SeaWiFS: A Retrospective on the CZCS Behrenfeld, M., Randerson, J., McClain, C., Feldman, G., Nimbus Experiment Team (NET) Program. NASA Tech. Los, S., Tucker, C., Falkowski, P., Field, C., Frouin, R., Memo. 104566, Vol. 20. NASA Goddard Space Flight Esaias, W., Kolber, D., Pollack, N., 2001. Temporal Center, Greenbelt, MD (43pp). changes in the photosynthetic biosphere. Science 291, Acker, J.G., Shen, S., Leptoukh, G., Serafino, G., Feldman, G., 2594–2597. McClain, C., 2002. SeaWiFS ocean-color data archive and Berthon, J.-F., Zibordi, G., Doyle, J.P., Grossi, S., van der distribution system: assessment of system performance. Linde, D., Targa, C., 2002. In: Hooker, S.B., Firestone, IEEE Transactions on Geoscience andRemote Sensing E.R. (Eds.), Coastal Atmosphere and Sea Time Series 40, 90–103. (CoASTS), Part 2: Data Analysis. NASA Tech. Memo. Aiken, J., et al., 1998. In: Hooker, S.B., Firestone, E.R. (Eds.), 2002-206892, Vol. 20. NASA Goddard Space Flight Center, AMT-5 Cruise Report. NASA Tech. Memo. 1998–206892, Greenbelt, MD (25pp). Vol. 2. NASA Goddard Space Flight Center, Greenbelt, Bilanow, S., Patt, F.S., 2004. In: Hooker, S.B., Firestone, E.R. MD (113pp). (Eds.), Pointing Performance for the SeaWiFS Mission. Aiken, J., Rees, N., Hooker, S., Holligan, P., Bale, A., Robins, NASA Tech. Memo. 2004-206892, Vol. 28. NASA Goddard D., Moore, G., Harris, R., Pilgrim, D., 2000. The Atlantic Space Flight Center, Greenbelt, MD (63pp). Meridional Transect: overview and synthesis of data. Chavez, F.P., Strutton, P.G., Friedrich, G.E., Feely, R.A., Progress in Oceanography 45, 257–312. Feldman, G.C., Foley, D.G., McPhaden, M.J., 1999. ARTICLE IN PRESS

38 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42

Biological andchemical response of the equatorial Fu, G., Settle, K., McClain, C.R., 1998. SeaDAS: The SeaWiFS Pacific Ocean to the 1997–98 El Nin˜ o. Science 286, Data Analysis System, Proceedings of the Fourth Ocean 2126–2131. Remote Sensing Conference, Qingdao, China, July 28–31, Chou, M.-D., Chan, P.-K., Wang, M., 2002. Aerosol radiative pp. 73–77. forcing derived from SeaWiFS-retrieved aerosol optical GouldJr., R.W., Arnone, R.A., Martinolich, P.M., 1999. properties. Journal of Atmospheric Science 59, 748–757. Spectral dependence of the coefficient in case 1 Clark, D., Gordon, H.R., Voss, K.J., Ge, Y., Broenkow, W., andcase 2 waters. AppliedOptics 38, 2377–2383. Trees, C., 1997. Validation of atmospheric correction Gregg, W.W., Patt, F.S., Mezaache, A.L., Chen, J.D., Whiting, over the oceans. Journal of Geophysical Research 102, J.A., 1994. In: Hooker, S.B., Firestone, E.R. (Eds.), The 17209–17217. SimulatedSeaWiFS Data Set, Version 2 NASA Tech. Claustre, H., Morel, A., Hooker, S.B., Babin, M., Antoine, D., Memo. 104566, Vol. 15. NASA Goddard Space Flight Oubelkheir, K., Bricaud, A., Leblanc, K., Queguiner,! B., Center, Greenbelt, MD (42pp). Maritorena, S., 2002. Is desert dust making oligotrophic Gregg, W.W., Conkright, M.E., O’Reilly, J.E., Patt, F.S., waters greener? Geophysical Research Letters 29, 10.1029/ Wang, M.H., Yoder, J.A., Casey, N.W., 2002. NOAA- GL014056. NASA coastal zone color scanner reanalysis effort. Applied Claustre, H., Hooker, S.B., Van Heukelem, L., Berthon, J.-F., Optics 41 (9), 1615–1628. Barlow, R., Ras, J., Sessions, H., Targa, C., Thomas, C., Holben, B.N., Eck, T.F., Slutsker, I., Tanre, D., Buis, J.P., van der Linde, D., Marty, J.-C., 2004. An intercomparison Setzer, A., Vermote, E., Reagan, J.A., Kaufman, Y.J., of HPLC phytoplankton pigment methods using in situ Nakajima, T., Lavenu, F., Jankowiak, I., Smirnov, A., samples: Application to remote sensing anddatabase 1998. AERONET-A federated instrument network and data activities. Marine Chemistry 85, 41–61. archive for aerosol characterization. Remote Sensing of Darzi, M., 1998. SeaWiFS Algorithm Flow Chart. NASA Environment 66, 1–16. Contractor Report 1998-206848. NASA Goddard Space Hooker, S.B., 2002. Stability Monitoring of FieldRadiometers Flight Center, Greenbelt, MD (36pp). Using Portable Sources. In: Mueller, J.L., et al., Ocean Doyle, P.J., Hooker, S.B., Zibordi, G., van der Linde, D., 2003. Optics Protocols for Satellite Ocean-Color Sensor Valida- In: Hooker, S.B., Firestone, E.R. (Eds.), Validation of an tion, Revision 3, Vol. 1. NASA Tech. Memo. 2002–210004/ In-Water, Tower-Shading Correction Scheme. NASA Tech. Rev3–Vol.1. NASA Goddard Space Flight Center, Green- Memo. 2003-206892, Vol. 25. NASA Goddard Space Flight belt, MD, pp. 91–99. Center, Greenbelt, MD (32pp). Hooker, S.B., Aiken, J., 1998. Calibration evaluation and Eplee Jr., R.E., McClain, C.R., 2000. SeaWiFS Global Clear- radiometric testing of field radiometers with the SeaWiFS water Analysis. In: Hooker, S.B., Firestone, E.R., McClain, quality monitor (SQM). Journal of Atmospheric and C.R., et al. (Eds.), SeaWiFS Postlaunch Calibration and Oceanic Technology 15, 995–1007. Validation Analyses, Part 2. NASA Tech. Memo. 2000- Hooker, S.B., Lazin, G., 2000. In: Hooker, S.B., Firestone, 206892, Vol. 10. NASA Goddard Space Flight Center, E.R. (Eds.), The SeaBOARR-99 Field Campaign. NASA Greenbelt, MD, pp. 29–33. Tech. Memo. 2000-206892, Vol. 8. NASA Goddard Space Eplee Jr., R.E., Robinson, W.D., Bailey, S.W., Clark, D.K., Flight Center, Greenbelt, MD (46pp). Werdell, P.J., Wang, M., Barnes, R.A., McClain, C.R., Hooker, S.B., Maritorena, S., 2000. An evaluation of 2001. The calibration of SeaWiFS, Part 2: vicarious oceanographic radiometers and deployment methodologies. techniques. AppliedOptics 40 (36), 6701–6718. Journal of Atmospheric andOceanic Technology 17, Esaias, W., Feldman, G., McClain, C.R., Elrod, J., 1986. 811–830. Satellite observations of oceanic primary productivity. EOS, Hooker, S.B., McClain, C.R., 2000. The calibration and Transactions of the American Geophysical Union 67 (44), validation of SeaWiFS data. Progress in Oceanography 835–837. 45, 427–465. Evans, R.H., Gordon, H.R., 1994. Coastal zone color scanner Hooker, S.B., Morel, A., 2003. Platform andenvironmental ‘‘system calibration’’: a retrospective examination. Journal effects on above- andin-water determinations of water- of Geophysical Research 99, 7293–7307. leaving radiances. Journal of Atmospheric and Oceanic Fargion, G., McClain, C.R., 2002. SIMBIOS Project 2001 Technology 20, 187–205. Annual Report, 2002. NASA Tech Memo. 2002–210005. Hooker, S.B., Zibordi, G., 2003a. Advanced methods for NASA Goddard Space Flight Center, Greenbelt, MD. characterizing the immersion factor of irradiance sensors. Fargion, G.S., Barnes, R., McClain, C., 2001. In Situ Aerosol Journal of Atmospheric andOceanic Technology, sub- Optical Thickness Collectedby the SIMBIOS Program mittedfor publication. (1997–2000): protocols data QC and Analysis NASA Tech Hooker, S.B., Zibordi, G., 2003b. Platform perturbations in Memo. 2001-209982. NASA Goddard Space Flight Center, above-water radiometry. Applied Optics, submitted for Greenbelt, MD, 103pp. publication. Feldman, G., et al., 1989. Ocean-color: availability of the global Hooker, S.B., Esaias, W.E., Feldman, G.C., Gregg, W.W., data set. EOS, Transactions of the American Geophysical McClain, C.R., 1992. In: Hooker, S.B., Firestone, E.R. Union 70 (23), 634. (Eds.), An Overview of SeaWiFS and Ocean-Color. NASA ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 39

Tech. Memo. 104566, Vol. 1. NASA Goddard Space Flight in Above-Water Radiometry. NASA Tech. Memo. 2003- Center, Greenbelt, MD (24pp). 206892, Vol. 23. NASA Goddard Space Flight Center, Hooker, S.B., McClain, C.R., Firestone, J.K., Westphal, T.L., Greenbelt, MD (35pp). Yeh, E.-N., Ge, Y., 1994. In: Hooker, S.B., Firestone, E.R. Husar, R., et al., 2001. The Asian dust events of April 1998. (Eds.), The SeaWiFS Bio-Optical Archive and Storage Journal of Geophysical Research 106 (D16), 18317–18330. System (SeaBASS), Part 1. NASA Tech. Memo. 104566, Johnson, B.C., Bruce, S.S., Early, E.A., Houston, J.M., Vol. 20. NASA Goddard Space Flight Center, Greenbelt, O’Brian, T.R., Thompson, A., Hooker, S.B., Mueller, MD (40pp). J.L., 1996. In: Hooker, S.B., Firestone, E.R. (Eds.), The Hooker, S.B., Shaw, P-S., Hooker, S.B., Lynch, D., 1998. Fourth SeaWiFS Intercalibration Round-Robin Experi- Radiometric and engineering performance of the SeaWiFS ment (SIRREX-4), May 1995. NASA Tech. Memo. Quality Monitor (SQM): a portable light source for field 104566, Vol. 37. NASA Goddard Space Flight Center, radiometers. Journal of Atmospheric and Oceanic Technol- Greenbelt, MD (65pp). ogy 15, 1008–1022. Johnson, B.C., Sakuma, F., Butler, J.J., Biggar, S.F., Cooper, Hooker, S.B., Zibordi, G., Lazin, G., McLean, S., 1999. In: J.W., Ishida, J., Suzuki, K., 1997. Radiometric measure- Hooker, S.B., Firestone, E.R. (Eds.), The SeaBOARR-98 ment comparison using the Ocean-Color Temperature FieldCampaign. NASA Tech. Memo. 1999–206892, Vol. 3. Scanner (OCTS) visible andnear infraredintegrating NASA Goddard Space Flight Center, Greenbelt, MD sphere. Journal of Research of NIST 102, 627–646. (40pp). Johnson, B.C., Shaw, P.-S., Hooker, S.B., Lynch, D., 1998a. Hooker, S.B., Claustre, H., Ras, J., Van Heukelem, L., Radiometric and engineering performance of the SeaWiFS Berthon, J.-F., Targa, C., van der Linde, D., Barlow, R., Quality Monitor (SQM): a portable light source for field Sessions, H., 2000a. In: Hooker, S.B., Firestone, E.R. radiometers. Journal of Atmospheric and Oceanic Technol- (Eds.), The First SeaWiFS HPLC Analysis Round-Robin ogy 15, 1008–1022. Experiment (SeaHARRE-1). NASA Tech. Memo. 2000- Johnson, B.C., Fowler, J.B., Cromer, C.L., 1998b. In: Hooker, 206892, Vol. 14. NASA Goddard Space Flight Center, S.B., Firestone, E.R. (Eds.), The SeaWiFS Transfer Radio- Greenbelt, MD (42pp). meter (SXR). NASA Tech. Memo. 1998-206892, Vol. 1. Hooker, S.B., Zibordi, G., Berthon, J.-F., Bailey, S.W., Pietras, NASA Goddard Space Flight Center, Greenbelt, MD C.M., 2000b. In: Hooker, S.B., Firestone, E.R. (Eds.), The (58pp). SeaWiFS Photometer Revision for Incident Surface Mea- Johnson, B.C., Early, E.A., Eplee Jr, R.E., Barnes, R.A., surement (SeaPRISM) FieldCommissioning. NASA Tech. Caffrey, R.T., 1999a. In: Hooker, S.B., Firestone, E.R. Memo. 2000-206892, Vol. 13. NASA Goddard Space Flight (Eds.), The 1997 Prelaunch Radiometric Calibration of Center, Greenbelt, MD (24pp). SeaWiFS. NASA Tech. Memo. 1999-206892, Vol. 4. NASA Hooker, S.B., Zibordi, G., Berthon, J.-F., D’Alimonte, D., Goddard Space Flight Center, Greenbelt, MD (51pp). Maritorena, S., McLean, S., Sildam, J., 2001. In: Hooker, Johnson, B.C., Yoon, H.W., Bruce, S.S., Shaw, P-S., Thomp- S.B., Firestone, E.R. (Eds.), Results of the Second SeaWiFS son, A., Hooker, S.B., Eplee Jr., R.E., Barnes, R.A., Data Analysis RoundRobin, March 2000 (DARR-00). Maritorena, S., Mueller, J.L., 1999b. In: Hooker, S.B., NASA Tech. Memo. 2001-206892, Vol. 15. NASA Goddard Firestone, E.R. (Eds.), The Fifth SeaWiFS Intercalibration Space Flight Center, Greenbelt, MD (71pp). Round-Robin Experiment (SIRREX-5), July 1996. NASA Hooker, S.B., McLean, S., Sherman, J., Small, M., Lazin, G., Tech. Memo. 1999-206892, Vol. 7. NASA Goddard Space Zibordi, G., Brown, J.W., 2002a. In: Hooker, S.B., Fire- Flight Center, Greenbelt, MD (75pp). stone, E.R. (Eds.), The Seventh SeaWiFS Intercalibration Joint EOSAT/NASA SeaWiFS Working Group, 1987. System Round-Robin Experiment (SIRREX-7), March 1999. concept for wide-field-of-view observations of ocean phe- NASA Tech. Memo. 2002-206892, Vol. 17. NASA Goddard nomena from space. Report of the Joint EOSAT/NASA Space Flight Center, Greenbelt, MD (69pp). SeaWiFS Working Group. Earth Observation Satellite Co., Hooker, S.B., Lazin, G., Zibordi, G., McLean, S., 2002b. An Lanham, MD, 92pp. evaluation of above- and in-water methods for determining King, M.D., Greenstone, R., Spangler, S., 1999. EOS Science water-leaving radiances. Journal of Atmospheric and Plan, NP-1998-12-069-GSFC. NASA Goddard Space Flight Oceanic Technology 19, 486–515. Center, Greenbelt, MD (397pp). Hooker, S.B., et al., 2003a. The SecondSeaWiFS HPLC Lesht, B.M., Stroud, J.R., McCormick, M.J., Fahnenstiel, Analysis Round-Robin Experiment (SeaHARRE-2). NASA G.L., Stein, M.L., Welty, L.J., Leshkevich, G.A., 2002. An Tech. Memo., NASA Goddard Space Flight Center, event-driven phytoplankton bloom in southern Lake Greenbelt, MD, in preparation. Michigan observedby satellite. Geophysical Research Hooker, S.B., Zibordi, G., Berthon, J.-F., Brown, J.W., 2003b. Letters 29(8), 10.1029/2001GL013533. Above-water radiometry in shallow, coastal waters. Applied McClain, C.R., Fargion, G., 1999. SIMBIOS Project 1998 Optics, in press. Annual Report. NASA Tech. Memo. 1999-208645. NASA Hooker, S.B., Zibordi, G., Berthon, J.-F., D’Alimonte, D., van Goddard Space Flight Center, Greenbelt, MD, 105pp. der Linde, D., Brown, J.W., 2003c. In: Hooker, S.B., McClain, C.R., Esaias, W.E., Feldman, G.C., Elrod, J., Endres, Firestone, E.R. (Eds.), Tower-Perturbation Measurements D., Firestone, J., Darzi, M., Evans, R., Brown, J., 1990. ARTICLE IN PRESS

40 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42

Physical andbiological processes in the North Atlantic NASA Tech. Memo. 2002-210006. NASA Goddard Space during the First Global GARP Experiment. Journal of Flight Center, Greenbelt, MD (65pp). Geophysical Research 95 (C10), 18027–18048. Meister, G., et al., 2003. The SecondSIMBIOS Radiometric McClain, C.R., Darzi, M., Firestone, J., Yeh, E.-N., Fu, G., Intercomparison (SIMRIC-2), March–November 2002. Endres, D., 1991a. SEAPAK Users Guide, Version 2.0, Vol. NASA Tech Memo. 2002-21006 Vol. 2. NASA Goddard I–System Description. NASA Tech. Memo. 100728. NASA Space Flight Center, Greenbelt, MD, 65pp. Goddard Space Flight Center, Greenbelt, MD (158pp). Mueller, J.L., 1993. In: Hooker, S.B., Firestone, E.R. (Eds.), McClain, C.R., Darzi, M., Firestone, J., Yeh, E.-N., Fu. G., The First SeaWiFS Intercalibration Round-Robin Experi- Endres, D., 1991b. SEAPAK Users Guide, Version 2.0, Vol ment, SIRREX-1, July 1992. NASA Tech. Memo. 104566, II–Descriptions of Programs, NASA Tech. Memo. 100728. Vol. 14. NASA Goddard Space Flight Center, Greenbelt, NASA Goddard Space Flight Center, Greenbelt, MD, MD (60pp). 586pp. Mueller, J.L., 1995. Comparison of irradiance immersion McClain, C.R., Fu, G., Darzi, M., Firestone, J.K., 1992a. coefficients for several marine environmental radiometers PC-SEAPAK User’s Guide, Version 4.0. NASA Tech. (MERs). In: Mueller, J.L., Hooker, S.B., Firestone, E.R., Memo. 104557. NASA Goddard Space Flight Center, Acker, J.G., et al. (Eds.), Case Studies for SeaWiFS Greenbelt, MD (408pp). Calibration andValidation,Part 3. NASA Tech. Memo. McClain, C.R., Esaias, W.E., Barnes, W., Guenther, B., 104566, Vol. 27. NASA Goddard Space Flight Center, Endres, D., Hooker, S.B., Mitchell, G., Barnes, R., 1992b. Greenbelt, MD (46pp). In: Hooker, S.B., Firestone, E.R. (Eds.), Calibration and Mueller, J.L., 2000. Overview of measurement anddata Validation Plan for SeaWiFS. NASA Tech. Memo. 104566, analysis protocols. In: Fargion, G.S., Mueller, J.L. (Eds.), Vol. 3. NASA Goddard Space Flight Center, Greenbelt, Ocean Optics Protocols for Satellite Ocean-Color Sensor MD (41pp). Validation, Revision 2. NASA Tech. Memo. 2000-209966. McClain, C.R., et al., 1995. In: Hooker, S.B., Firestone, E.R., NASA Goddard Space Flight Center, Greenbelt, MD, Acker, J.G. (Eds.), SeaWiFS Algorithms, Part 1. NASA pp. 87–97. Tech. Memo. 104566, Vol. 28. NASA Goddard Space Flight Mueller, J.L., 2002. Overview of measurement anddata Center, Greenbelt, MD (45pp). analysis protocols. In: Mueller, J.L., et al. (Ed.), Ocean McClain, C.R., Darzi, M., Barnes, R.A., Eplee Jr., R.E., Optics Protocols for Satellite Ocean-Color Sensor Valida- Firestone, J.K., Patt, F.S., Robinson, W.D., Schieber, B.D., tion, Revision 3, Vol. 1. NASA Tech. Memo. 2002–210004/ Woodward, R.H., Yeh, E.-N., 1996. In: Hooker, S.B., Rev3-Vol. 1. NASA Goddard Space Flight Center, Firestone, E.R. (Eds.), SeaWiFS Calibration and Validation Greenbelt, MD, pp. 100–122. Quality Control Procedures. NASA Tech. Memo. 104566, Mueller, J.L., 2003. Overview of measurement anddata Vol. 38. NASA Goddard Space Flight Center, Greenbelt, analysis protocols. In: Mueller, J.L., et al. (Ed.), Ocean MD (64pp). Optics Protocols for Satellite Ocean-Color Sensor Valida- McClain, C.R., Cleave, M.L., Feldman, G.C., Gregg, W.W., tion, Revision 4, Vol. III: Radiometric Measurements and Hooker, S.B., Kuring, N., 1998. Science quality SeaWiFS Data Analysis Protocols. NASA Tech. Memo. 2003– data for global biosphere research. Sea Technology 39 (9), 211621/Rev4-Vol. III. NASA Goddard Space Flight Center, 10–16. Greenbelt, MD, pp. 1–6. McClain, C.R., et al., 2000a. In: Hooker, S.B., Firestone, E.R. Mueller, J.L., Austin, R.W., 1992. In: Hooker, S.B., Firestone, (Eds.), SeaWiFS Postlaunch Calibration and Validation E.R. (Eds.), Ocean Optics Protocols for SeaWiFS Valida- Analyses, Part 1, 2000a. NASA Tech. Memo. 2000-206892, tion. NASA Tech. Memo. 104566, Vol. 5. NASA Goddard Vol. 9. NASA Goddard Space Flight Center, Greenbelt, Space Flight Center, Greenbelt, MD (43pp). MD (82pp). Mueller, J.L., Austin, R.W., 1995. In: Hooker, S.B., Firestone, McClain, C.R., et al., 2000b. In: Hooker, S.B., Firestone, E.R. E.R., Acker, J.G. (Eds.), Ocean Optics Protocols for (Eds.), SeaWiFS Postlaunch Calibration and Validation SeaWiFS Validation, Revision 1. NASA Tech. Memo. Analyses, Part 2, 2000b. NASA Tech. Memo. 2000-206892, 104566, Vol. 25. NASA Goddard Space Flight Center, Vol. 10. NASA Goddard Space Flight Center, Greenbelt, Greenbelt, MD (66pp). MD (57pp). Mueller, J.L., Johnson, B.C., Cromer, C.L., Cooper, J.W., McClain, C.R., Esaias, W., Feldman, G., Frouin, R., Gregg, McLean, J.T., Hooker, S.B., Westphal, T.L., 1994. In: W., Hooker, S., 2002. The Proposal for the NASA Sensor Hooker, S.B., Firestone, E.R. (Eds.), The Second SeaWiFS Intercalibration andMerger for Biological andInterdisci- Intercalibration Round-Robin Experiment, SIRREX-2, plinary Oceanic Studies (SIMBIOS) Program, 1995. NASA June 1993. NASA Tech. Memo. 104566, Vol. 16. NASA Tech. Memo. 2002-210008. NASA Goddard Space Flight Goddard Space Flight Center, Greenbelt, MD (121pp). Center, Greenbelt, MD, 63pp. Mueller, J.L., Johnson, B.C., Cromer, C.L., Hooker, S.B., McPhaden, M.J., 1999. Genesis and evolution of the 1997–98 El McLean, J.T., Biggar, S.F., 1996. In: Hooker, S.B., Nin˜ o. Science 283, 950–954. Firestone, E.R., Acker, J.G. (Eds.), The Third SeaWiFS Meister, G., et al., 2002. The First SIMBIOS Radiometric Intercalibration Round-Robin Experiment (SIRREX-3), Intercomparison (SIMRIC-1), April–September 2001. 19-30 September 1994. NASA Tech. Memo. 104566, ARTICLE IN PRESS

C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42 41

Vol. 34. NASA Goddard Space Flight Center, Greenbelt, Siegel, D.A., Maritorena, S., Nelson, N.B., Hansell, D.A., MD (78pp). Lorenzi-Kayser, M., 2003. Global distribution and dy- Murtugudde, R.G., Signorini, S.R., Christian, J.R., Busalacchi, namics of colored dissolved and detrital organic materials. A.J., McClain, C.R., 1999. Ocean-color variability of Journal Geophysical Research 107(C12), 21/1–14. the tropical Indo-Pacific basin observed by SeaWiFS South-West Florida Dark Water Observation Group, 2002. during 1997-98. Journal of Geophysical Research 104 Satellite images track ‘‘black water’’ event off Florida coast. (C8), 18351–18366. EOS, Transactions of the American Geophysical Union Ocean-Color Working Group, 1982. The Marine Resources 83(26), 281. Experiment Program (MAREX), Report of the Ocean-color Strutton, P.G., Ryan, J.P., Chavez, F.P., 2001. Enhanced Science Working Group. NASA Goddard Space Flight chlorophyll associatedwith tropical instability waves in Center, Greenbelt, MD, 107pp. the equatorial pacific. Geophysical Research Letters 28, O’Reilly, J.E., Maritorena, S., Mitchell, B.G., Siegel, D.A., 2005–2008. Carder, K.L., Garver, S.A., Kahru, M., McClain, C., 1998. Vance, T., et al., 1998. Aquamarine waters recorded for first Ocean-color chlorophyll algorithms for SeaWiFS. Journal time in eastern Bering Sea. EOS, Transactions of the of Geophysical Research 103 (C11), 24937–24953. American Geophysical Union 79 (10), 121. O’Reilly, J.E., et al., 2000. In: Hooker, S.B., Firestone, E.R. Van Heukelem, L., Thomas, C.S., Glibert, P.M., 2002. In: (Eds.), SeaWiFS Postlaunch Calibration and Validation Fargion, G.S., McClain, C.R. (Eds.), Sources of Variability Analyses, Part 3. NASA Tech. Memo. 2000-206892, in Chlorophyll Analysis by Fluorometry andHigh-Perfor- Vol. 11. NASA Goddard Space Flight Center, Greenbelt, mance LiquidChromatography in a SIMBIOS Inter- MD (49pp). Calibration Exercise. NASA Tech. Memo. 2002-211606. Patt, F.S., 2002. In: Hooker, S.B., Firestone, E.R. (Eds.), NASA Goddard Space Flight Center, Greenbelt, MD Navigation Algorithms for the SeaWiFS Mission. NASA (50pp). Tech. Memo. 2002-206982, Vol. 16. NASA Goddard Space Wang, M., Bailey, S., McClain, C.R., 2000. SeaWiFS provides Flight Center, Greenbelt, MD (17pp). unique global aerosol optical property data. EOS, Transac- Patt, F.S., Gregg, W.W., 1994. Exact closed-form geolocation tions of the American Geophysical Union 81 (18), 197. algorithm for Earth survey sensors. International Journal of Werdell, P.J., Bailey, S.W., 2002. In: Fargion, G.S., McClain, Remote Sensing 15, 3719–3734. C.R. (Eds.), The SeaWiFS Bio-Optical Archive and Storage Patt, F.S., Woodward, R.H., Gregg, W.W., 1997. An auto- System (SeaBASS): Current Architecture andImplementa- matedmethodfor navigation assessment for Earth survey tion. NASA Tech. Memo. 2002-211617. NASA Goddard sensors using islandtargets. International Journal of Space Flight Center, Greenbelt, MD (45pp). Remote Sensing 18, 3311–3336. Werdell, P.J., Bailey, S., Fargion, G., Pietras, C., Knobel- Patt, F.S., et al., 2003. In: Hooker, S.B., Firestone, E.R. (Eds.), spiesse, K., Feldman, G., McClain, C., 2003. Unique data Algorithm Updates for the Fourth SeaWiFS Data Repro- repository facilitates ocean-color satellite validation. EOS, cessing. NASA Tech. Memo. 2002-206982, Vol. 22. NASA Transactions of the American Geophysical Union 84 (38), Goddard Space Flight Center, Greenbelt, MD (74pp). 377. Riley, T., Bailey, S., 1998. The Sixth SeaWiFS/SIMBIOS Inter- Zibordi, G., Barale, V., Ferrari, G.M., Hoepffner, N., calibration Round-Robin Experiment (SIRREX-6) August– Alberotanza, L., Cova, P., Ramasco, C., 1995. Coastal December 1997. NASA Tech. Memo. 1998-206878. NASA Atmosphere andSea Time-Series project (CoASTS): an Goddard Space Flight Center, Greenbelt, MD. ocean colour remote sensing calibration–validation project. Robins, D.B., Bale, A.J., Moore, G.F., Rees, N.W., Hooker, Proceedings of the Third Thematic Conference on Remote S.B., Gallienne, C.P., Westbrook, A.G., Maran˜ on,! E., Sensing Marine Coastal Environments Seattle, Vol. 2, Spooner, W.H., Laney, S.R., 1996. In: Hooker, S.B., September 18–20, pp. 96–100. Firestone, E.R. (Eds.), AMT-1 Cruise Report and Pre- Zibordi, G., Doyle, G.P., Hooker, S., 1999. Offshore liminary Results. NASA Tech. Memo. 104566, Vol. 35. tower-shading effects on in-water optical measure- NASA Goddard Space Flight Center, Greenbelt, MD ments. Journal of Atmospheric andOceanic Technology (87pp). 16, 1767–1779. Siegel, D.A., O’Brien, M.C., Sorensen, J.C., Konnoff, D.A., Zibordi, G., D’Alimonte, D., van der Linde, D., Berthon, J.-F., Brody, E.A., Mueller, J.L., Davis, C.O., Rhea, W.J., Hooker, S.B., Mueller, J.L., Lazin, G., McLean, S., 2002a. Hooker, S.B., 1995. In: Hooker, S.B., Firestone, E.R. In: Hooker, S.B., Firestone, E.R. (Eds.), The Eighth (Eds.), Results of the SeaWiFS Data Analysis Round-Robin SeaWiFS Intercalibration Round-Robin Experiment (SIR- (DARR-94), July 1994. NASA Tech. Memo. 104566, Vol. REX-8), September–December 2001. NASA Tech. Memo. 26. NASA Goddard Space Flight Center, Greenbelt, MD 2002-206892, Vol. 20. NASA Goddard Space Flight Center, (58pp). Greenbelt, MD (39pp). Siegel, D.A., Wang, M., Maritorena, S., Robinson, W., Zibordi, G., Berthon, J-F., Doyle, J.P., Grossi, S., van der 2000. Atmospheric correction of satellite ocean-color Linde, D., Targa, C., Alberotanza, L., 2002b. In: Hooker, imagery: the black pixel assumption. AppliedOptics 39, S.B., Firestone, E.R. (Eds.), Coastal Atmosphere and Sea 3582–3591. Time Series (CoASTS), Part 1: A Tower-BasedLong-Term ARTICLE IN PRESS

42 C.R. McClain et al. / Deep-Sea Research II 51 (2004) 5–42

Measurement Program. NASA Tech. Memo. 2002-206892, sion Factors of Irradiance Sensors. NASA Tech. Memo. Vol. 19. NASA Goddard Space Flight Center, Greenbelt, 2002-206892, Vol. 26. NASA Goddard Space Flight Center, MD (29pp). Greenbelt, MD (34pp). Zibordi, G., Hooker, S.B., Berthon, J.-F., D’Alimonte, D., Zibordi, G., Hooker, S.B., Mueller, J., Lazin, G., 2004a. 2002c. Autonomous above-water radiance measurements Characterization of the immersion factor for a series of from an offshore platform: a fieldassessment experi- underwater optical radiometers. Journal of Atmospheric ment. Journal of Atmospheric andOceanic Technology andOceanic Technology 21, 501–514. 19, 808–819. Zibordi, G., Melin,! F., D’Alimonte, D., Hooker, S.B., Holben, Zibordi, G., D’Alimonte, D., van der Linde, D., Hooker, S.B., B., 2004b. Ocean-color calibration andvalidation using an Brown, J.W., 2003. In: Hooker, S.B., Firestone, E.R. (Eds.), autonomous above-water system. IEEE Transactions on New Laboratory Methods for Characterizing the Immer- Geoscience andRemote Sensing, in press.