Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 11
The Role of Small Satellites in NASA and NOAA Earth Observation Programs 2 Core Observational Needs This chapter summarizes National Aeronautics and Space Administration (NASA) and National Oceanic and Atmospheric Administration (NOAA) observational requirements for both research and operational satellite programs. Much of this discussion is not specific to small satellites but is provided to frame information presented in later chapters. Research programs such as the Earth Observing System (EOS) take advantage of satellites' ability to provide a consistent, global vantage point from which to observe land, ocean, and atmosphere processes. Operational programs such as the Polar-orbiting Operational Environmental Satellite (POES) program also rely on the global perspective of satellites, but they generally emphasize the rapid delivery of global observations to support weather forecasting. The distinction between research and operational programs has a profound impact on how NASA and NOAA manage their programs. The distinction between research data sets and operational data sets can sometimes be artificial, however, and both can be essential elements of a climate research and global change program. Operational systems are usually associated with the acquisition of long time series of data that may not meet the measurement requirements for climate research. An example of such a data set might be ocean topography, which appears in both the EOS and the National Polar-orbiting Operational Environmental Satellite System (NPOESS) observation sets. In this case, the accuracy requirements for climate research needs are much more stringent than the operational needs. On the other hand, some operational data sets do meet climate research needs, such as the NOAA POES and Geostationary Operational Environmental Satellites (GOES). In the next section, the committee examines measurements identified by the EOS and NPOESS program offices as critical to the success of their respective missions. These measurements are not examined individually in detail; instead, the focus is on identifying common attributes present in each class of variable (research and operational). REQUIRED MEASUREMENTS Measurements in Support of Climate and Global Change Research The specific variables that are currently measured from space are based on an understanding of Earth processes as well as on the technical ability to make measurements with the requisite accuracy and temporal and spatial resolution. The development of measurement requirements—and the generation of any list of critical
OCR for page 12
The Role of Small Satellites in NASA and NOAA Earth Observation Programs measurements—is thus in continuous flux as scientific understanding evolves. For any particular mission, science requirements must be translated into a set of satellite sensors with specific measurement and sampling capabilities. The actual sensor requirements are therefore a melding of these science requirements and existing capabilities. Many critical processes do not have an electromagnetic signal that can be measured by satellite. For example, the partial pressure of CO2 in the surface ocean cannot be measured remotely, although it plays a critical role in determining the flux of CO2 between the ocean and atmosphere. Also, many processes simply cannot be measured with adequate temporal and spatial resolution from space. For example, ocean salinity can be measured by satellite, but not with the required accuracy or spatial resolution of current microwave radiometer technology. Another example is the study of the ozone hole. In this case, ground observations first revealed the existence of the hole, which then stimulated a reanalysis of the satellite data sets. However, ground-based and in situ observations continued to be required to study the dynamics of the Antarctic ozone vortex in conjunction with satellite measurements. As these examples all show, Earth science measurement requirements are tempered by the reality of the technical capabilities of present and planned remote sensing systems. The objective of this section is to identify the processes that are used to develop Earth science requirements and how these are in turn used to define a satellite mission. In this regard, the suite of 24 EOS measurements1 shown in Box 2.1 represents the current understanding of the important processes related to Earth's climate and global changes as well as the ability of EOS sensors to make these measurements.2 However, even when a measurement is listed, it should not be assumed that it will meet the science requirement. This is a result of the gaps in our understanding of Earth system processes, not poor sensor design. With this in mind, the EOS requirements were designed to be broad in scope, with the expectation that new insights into climate and global change processes will arise from having long-term, consistent observations. The EOS measurement set was also based on the realization that multiple observations of the same variable would lead to a better understanding of the relevant processes. That is, each observation has its own sampling characteristics and measurement approach that, when combined with other measurements of the same variable, may lead to a higher quality measurement. In Earth system research, it is necessary to balance long-term observations with the need to study smaller scale events. Of particular interest are variations in the Earth system that occur on interannual and longer time scales. It will take many years to decades to observe such processes in a statistically robust manner. Processes that occur on much shorter time scales, such as severe storms or mesoscale ocean eddies, may drive the overall system, however. The Earth does not operate as a smoothly varying system but rather as a set of nonlinear processes that can change rapidly. Earth system research goes far beyond the realm of atmospheric dynamics. The ocean clearly provides strong feedback through the transport of heat and the exchange of water with the atmosphere. Moreover, both the marine and terrestrial components of the biosphere affect climate through their impacts on heat and moisture exchanges as well as through their modulation of biogeochemistry, especially greenhouse gases. In other words, there is no single measurement that will provide a comprehensive understanding of climate processes and their interaction with the biosphere. Any Earth observing system must consist of an integrated, comprehensive set of measurements. However, it must also have the capacity to include new measurements as our understanding of the Earth system evolves and our technical abilities improve. Measurements in Support of Operational Applications The measurement requirements for operational observing systems, such as NPOESS, are designed for a set of objectives that differ from those for research observing systems. In large part, this is a result of operational systems usually being focused on short-term, event-scale processes and the rapid delivery of near-real-time data. Such applications place less importance on long-term stability of data sets and more importance on data availability, for example, to protect life and property. Operational data also play an important role in numerical weather prediction 1 It is expected that the 24 EOS measurements discussed here—a set maintained during several previous program rescopings—will now change as NASA rethinks its plans beyond the first series of EOS spacecraft in light of this study's findings. 2 See the sections on calibration and validation later in this chapter.
OCR for page 13
The Role of Small Satellites in NASA and NOAA Earth Observation Programs BOX 2.1 Classification of the EOS Observation Set into 24 Measurement Categories Atmosphere Cloud properties, including amount, optical properties, and height Radiative fluxes at the top of the atmosphere and the surface Precipitation Tropospheric chemistry, including ozone and precursor gases Stratospheric chemistry, including ozone, BrO, OH, and trace gases Aerosol properties in both the troposphere and stratosphere Atmospheric temperature Atmospheric humidity Lightning, including events, area, and flash structure Solar Radiation Total solar irradiance Ultraviolet spectral irradiance Land Land cover and land use change Vegetation dynamics Surface temperature Fire occurrence, including extent and thermal anomalies Volcanic effects, including frequency of events, thermal anomalies, and impact Surface wetness Ocean Surface temperature Phytoplankton and dissolved organic matter Surface wind fields Ocean surface topography, including height, waves, and sea level Cryosphere Land ice, including ice sheet topography, ice sheet volume change, and glacier extent Sea ice, including extent, concentration, motion, and temperature Snow cover, including extent and water equivalence models that rely on data assimilation. Here, data and sampling needs are well understood from analyses of model performance. In fact, the models are often designed for specific data types with specific characteristics, so it is difficult to adapt them to assimilate new data sets. The primary user community is also well defined, and sensors and data products are often developed to meet particular application needs. There is, however, a large group of users outside the primary user community of weather services: The success of the POES data in furthering understanding of such long-term phenomena as El Niño has considerably broadened the user base. Secondary data products are available from the National Weather Service (NWS), the National Environmental Satellite Data and Information Service, and the National Centers for Environmental Prediction/Climate Prediction Center, which continue to expand the interest in these systems; research use of the data from these systems is increasing. It is not clear, however, that this use has driven the requirements for NPOESS. The net result of these requirements is a suite of satellite sensors that evolve slowly over time and that are designed to meet well-defined needs. Given the operational focus of these missions, the user community has a very low tolerance for gaps in the data. In contrast to the Earth science missions, the primary users of operational data tend to rely on recently collected data rather than analyses of historical data. Moreover, the emphasis on specific application needs generally results in less interest in complex sensor suites to study a wide variety of Earth system processes. This is not to say that the sensor suites are not complex, but rather that they are chosen for specific application to the meteorological problem. The weather services are not usually interested in more general sensors that have a wide application to other problems. Indeed, the aversion to risk that characterizes operational programs such as those of the NWS is reflected in the conservative choice of sensors. The emerging NPOESS program has developed a list of environmental data records (EDRs) that are intended to meet the needs of its broadly based user community. These EDRs are described in detail in the Integrated
OCR for page 14
The Role of Small Satellites in NASA and NOAA Earth Observation Programs Operational Requirements Document.3 For example, some of the service branches of the Department of Defense (DOD, which is NOAA's partner in the NPOESS program) require constant ground resolution imagery. Providing images in this format eases their interpretation and eliminates the need for image analysts to undergo the specialized training they would otherwise need to interpret images in the format provided by the Advanced Very High Resolution Radiometer instrument on POES. However, it is much less useful for quantitative analysis for scientific purposes because of the difficulty in obtaining quantitative radiometric measurements. The current list of EDRs (Box 2.2) represents a careful balance between NOAA and DOD needs. Each EDR has an associated threshold and objective for performance. The EDR objectives constitute a set of desired performance levels but are not strict requirements. If a sensor cannot meet the EDR threshold, however, it is deemed to have failed. The NPOESS 3 This and other NPOESS requirements documents are available online at <CS:WebLink>http://npoesslib.ipo.noaa.gov/Req_Docs.htm>. BOX 2.2 Environmental Data Records Identified for the National Polar-orbiting Operational Environmental Satellite System Key Parameters1 Atmospheric vertical moisture profiles Atmospheric vertical temperature profiles Imagery Sea surface temperature Sea surface winds Soil moisture Atmospheric Parameters Aerosol optical thickness Aerosol particle size Ozone total column/profile Precipitable water Precipitation (type and rate) Pressure (surface and profile) Suspended matter Total water content Cloud Parameters Cloud base height Cloud cover and layers Cloud effective particle size Cloud ice water path Cloud liquid water Cloud optical depth and transmittance Cloud top height Cloud top pressure Cloud top temperature Earth Radiation Budget Parameters Surface albedo Downward long-wave radiation at the surface Insolation Net shortwave radiation at the top of the atmosphere Solar irradiance Total long-wave radiation at the top of the atmosphere Land Parameters Land surface temperature Normalized difference vegetation index Snow cover and depth Vegetation and surface type Ocean and Water Parameters Currents Freshwater ice motion Ice surface temperature Littoral sediment transport Net heat flux Ocean color and chlorophyll Ocean wave characteristics Sea ice age and motion Sea surface height and topography Surface wind stress Turbidity Space Environmental Parameters Auroral boundary Auroral energy deposition Auroral imagery Electric field Electron density profiles and ionospheric specification Geomagnetic field In situ ion drift velocity In situ plasma density In situ plasma fluctuations In situ plasma temperature Ionospheric scintillation Neutral density profile/neutral atmospheric specification Radiation belt and low-energy solar particles Solar and galactic cosmic ray particles Solar extreme ultraviolet flux Suprathermal through auroral energy particles Upper atmospheric airglow Note: These were current as of this writing; the EDRs are, however, subject to change, and the latest requirements may be found a the NPOESS Web site at <CS:WebLink>http://npoesslib.ipo.noaa.gov>. 1 These are baseline measurements that must be part of NPOESS.
OCR for page 15
The Role of Small Satellites in NASA and NOAA Earth Observation Programs requirements methodology may appear to be more rigorous than the open-ended approach taken with climate research, but note that each approach has been chosen to meet the needs of its particular user community. Although there are apparent overlaps between the EOS 24 measurement set and the NPOESS EDRs, the lists are not interchangeable. Each measurement has its own set of performance and sampling requirements that are appropriate for a specific application. In general, sensor performance requirements for climate and global change research are more stringent than operational requirements, while the operational requirements for sampling and continuity are more stringent than the science requirements. However, as Earth system research matures, the operational observing systems probably will begin to assume responsibility for these more stringent measurement requirements. For example, the long-term variation in climate processes requires a monitoring approach that is appropriately done in an operational context. This transfer of measurement responsibility from research to operational systems must take place in an orderly manner.
OCR for page 16
The Role of Small Satellites in NASA and NOAA Earth Observation Programs CHARACTERIZATION, CALIBRATION, AND VALIDATION Prelaunch Sensor Characterization Prelaunch characterization is necessary for all sensors that are expected to return accurate data, such as those required for climate studies. Sensors require a suite of prelaunch tests to establish performance parameters to be used in data processing. These performance parameters are instrument characteristics such as spectral bandpass, polarization sensitivity, out-of-field-of-view responsivity (scattered light effects), deviations from linearity, and temperature sensitivity. The type of performance parameter, as well as the accuracy to which it needs to be measured, is determined by the requirements of the data processing algorithm(s); this is true for both climate and operational sensors. However, the two types of sensors will differ in the extent and degree of the characterization tests. Sensor characterization is different from sensor calibration. Calibration is the process of measuring the relationship between sensor output (digital counts) and absolute radiant (radiance) input. Sensor characterizations must be completed before launch in almost all instances. However, it is possible to calibrate a sensor's response ''vicariously" on orbit by observations of calibrated ground targets. It is difficult, if not impossible, to measure most sensor characteristics after launch. Sensor characterization measurements in the laboratory can be accomplished under reasonably controlled conditions compared to on-orbit measurements. It may be sufficient to perform some characterizations at the component level and others at the system level; for particularly critical performance parameters, both system and component level measurements may be required. The requirement for a complete set of accurately known sensor characteristics is independent of satellite size. Thus, it is not a unique problem of small satellites but may be more difficult to accomplish for them because of the funding and schedule limitations of their missions. To complete the tasks associated with full characterization of a sensor takes time and money, which may not be commensurate with the shorter schedules and lower costs desired for small satellite missions. Calibration Careful sensor calibration is essential if the typically small signatures of phenomena like climate change are to be recognized. For example, small increases in global sea surface temperature represent an enormous change in the heat content of the world ocean, and they are a valuable diagnostic of climate models. However, drift in sensor calibration or sensor performance could easily mask these changes. Monitoring solar output is another example where careful calibration is necessary; it is complicated in this case by the need to assemble a consistent time series across multiple copies of a sensor, variations in which are a significant component of the total apparent system variance.4 Calibration programs must be an ongoing part of any satellite system that is focused on Earth system studies. Much of the effort of calibration occurs before the satellite is launched. As with sensor characterization, this too requires time and money and may not be commensurate with the shorter schedules and lower budgets desired for small satellite missions.5 These programs will rely on a combination of in situ studies—vicarious calibration—as well as on-board calibration systems. The primary role of an on-board calibration system is to measure the sensor's precision—that is, its short-term stability on orbit and its calibration stability through the rigors of launch. On-board calibration systems take space and add cost. It has not yet been demonstrated that they can be implemented within the constraints of the "smallsat" paradigm.6 Calibration programs must pay particular attention to assessing the performance of new sensors, even if they are copies of old sensors. This assessment may mean overlapping the use of old and new sensors so it will be possible to ensure compatibility between data products. 4 See Willson (1997) and the references therein. 5 The care, planning, and effort that may be required are illustrated by the program to calibrate the EOS AM-1 Multi-Angle Imaging Spectroradiometer instrument. See Bruegge et al. (1996, 1998). 6 However, the excellent results obtained from SeaWiFS (Sea-Viewing Wide Field-of-View Sensor) are an example of what can be done and illustrate the value of on-board calibration for a small sensor.
OCR for page 17
The Role of Small Satellites in NASA and NOAA Earth Observation Programs Alternatively, carefully planned field campaigns may provide sufficient cross-comparisons to produce a consistent time series across multiple sensors. The present series of polar and geosynchronous meteorological satellites (POES, Defense Meteorological Satellite Program, GOES) were not designed to meet the rigorous calibration requirements of users such as climate researchers. While not yet defined, the calibration plan for the next generation of meteorological satellites, NPOESS, is likely to be less rigorous than that required for studies of climate and other aspects of global change. The committee notes, however, that satisfying many of the NPOESS operational requirements will necessitate comparisons between present measurements and climatologies derived from long-term observations. As undetected shifts in sensor performance may substantially reduce the quality of the operational product, regular assessments of instrument performance should be part of the calibration program. The importance of an adequate calibration program is illustrated by the following example. The blended sea surface temperature (SST) was a NOAA-generated data product of the early 1980s. The SST was a combination of satellite and in situ measurements. However, if the satellite observations exceeded the climatological SSTs by too large a value, it was assumed that they were in error. In that case, the satellite-derived SSTs were replaced by the climatologically derived SST values, as it was assumed that the on-board calibration information was insufficient to evaluate sensor performance. The result of this approach, however, was that the blended SST product did not reveal the 1982–1983 El Niño Southern Oscillation (ENSO) event until direct measurements were made by ship in the eastern tropical Pacific—6 months after the ENSO event began. The satellite values, which were much warmer than the climatology measurements, had been correct, but they had exceeded the quality threshold and were discarded. Validation Validation is the process of evaluating the algorithms that are used to convert instrument measurements into geophysical quantities and assessing the uncertainties in derived geophysical quantities. As with calibration, validation is required for both research and operational systems, although the requirements may differ. The need for validation applies to all remote sensing systems, regardless of mission or satellite size. Algorithms may change over time as new methods are developed. However, operational agencies are typically not interested in reprocessing old data in an attempt to extract new information. This essential difference between climatology and meteorology affects mission goals; it must be understood if the operational and research communities are to exchange data and rely on each other's satellites and databases. Validation should be done over the full range of possible environmental conditions. The EOS program is developing a plan for validation that should provide critical information for future researchers as they assemble long time series of observations from a variety of sensors. Validation can be seen as the counterpart of calibration, but it is applied to the software or algorithm component of the data set. DATA CONTINUITY A stable set of well-calibrated, reliable measurements is central to the U.S. Global Change Research Program. Indeed, much effort has been expended to design instruments that will be stable enough to ensure that they indicate actual changes in the Earth and its environs, rather than the effects of instrument artifacts and instabilities. The introduction of new instruments can undermine continuity and confidence in long-term measurements. Yet there are serious problems in attempting to field a particular set of instruments in perpetuity. For example, changes in technology will eventually make it impossible to reproduce a given set of instruments as the availability of the specific components and intellectual skills associated with a particular design vanish. Twenty years is probably the outside limit for any one design. In fact, some space instruments have been in production for nearly this interval, but closer examination would reveal that they have been continually upgraded to surmount some of the parts availability problems. Continuity of measurements is sometimes confused with continuity of sensors. Given budgetary as well as user-imposed constraints, continuity may be ensured by a strategy of launching an identical sensor upon antici-
OCR for page 18
The Role of Small Satellites in NASA and NOAA Earth Observation Programs pated failure of an orbiting sensor, although other approaches do exist. NPOESS and the operational user community obviously have much stricter requirements for continuity, especially for the six key parameters listed in Box 2.2. Operational Data Continuity Operational data requirements, such as those of the NWS, are often based on the need to protect lives and property; therefore, a long gap in a critical data set is unacceptable. A strategy to replace a failed sensor is thus a critical component of any operational program. This replenishment strategy is also dependent on the definition of sensor failure, which in turn may be different for the research and operational user communities. For example, an imager might lose a few channels or the signal-to-noise ratio might increase, and yet the sensor might still return useful data from an operational perspective. However, such data could be useless for the more demanding needs of Earth system and climate researchers. Data Continuity in Research In Earth system research, the broadly ranging requirements for data continuity are based on an assessment of the critical scales of variability of the processes under study. For example, if ENSO forcing is important and ENSO events are assumed to occur generally every 3 to 5 years, then a gap of 0.5 to 1 year will likely not compromise the quality of the data set for climate research. On the other hand, a gap of 2 to 4 years will seriously degrade the quality of the record. Another example is the Antarctic ice sheet. It has been suggested that the ice sheet need only be mapped every few years and that a continuous record is not required. The risk in this thinking is that catastrophic events may be a critical component of ice sheet dynamics, and these events might be missed with such a sampling strategy. Continuity requirements, therefore, will depend on the science objectives and our understanding of the scales of variability. Although the current EOS instruments were carefully designed, continuing improvements in technology can and will influence design trade-offs. There must be an effective mechanism for periodically revisiting these trade-offs and incorporating new technologies that can enhance capability or reduce cost. This introduces a classic cost trade-off: there are significant nonrecurring engineering costs associated with developing new instruments, so that the cost of developing the first of a series of new instruments is almost invariably higher than simply building another copy of an existing design. However, newer sensor designs could offer savings in total system cost if size, mass, and power reductions would permit corresponding reductions in satellite bus and launch vehicle size. An additional element in the trade is the potential nonrecurring investment to design the smaller satellite needed to realize these system-level savings. In essence, there are competing strategies for effecting economies. On the one hand, the most powerful economic strategy is to produce many copies of the same design, thereby deriving economies of scale from quantity production and amortization of nonrecurring development costs over many units. On the other hand, advances in technology will eventually change the cost-versus-capability analysis enough to overwhelm economies of scale. An effective strategy must balance these countervailing trends by embracing a methodical approach that captures some economies of scale by producing several copies of a particular design, but then periodically introduces significant design alterations—block changes—that exploit advances in technology (perhaps on every second series of satellites). Indeed, this approach has been successfully used by other satellite programs that faced challenges similar to those of NPOESS. The complex challenge of how to achieve technical renewal while maintaining data continuity and quality can best be addressed by embracing a concept the committee calls dynamic continuity. Specifically, the quality and continuity of measurements must be transparent to three levels of changes: Between successive flights of the same instrument design, Between successive generations of instruments, and Between similar instruments fielded by different countries.
OCR for page 19
The Role of Small Satellites in NASA and NOAA Earth Observation Programs Dynamic continuity can be achieved via a strategy that encompasses a rigorous calibration program, utilizing both in-orbit and in situ measurements, and a disciplined approach of overlapping measurements between successive generations of instruments over a sufficient interval to ensure accurate cross-calibration. In many respects, NASA did this very well in the Landsat (Land Remote Sensing Satellite) 4-5 era, where the then experimental thematic mapper instrument was carried alongside the existing operational multispectral scanner. This same basic approach can and should be carried forward as a technology insertion strategy, although it is not strictly necessary to carry the old and the new simultaneously on the same spacecraft. Indeed, small satellites become an ideal vehicle for the development and introduction of new technology in a manner that does not introduce risk into the mainstream scientific measurement program. Specifically, new payload instruments, developed in parallel with EOS and flown on small satellites (perhaps in formation with the EOS spacecraft), would be an excellent approach for establishing the validity and comparability of new and old measurements. This approach also provides an opportunity to develop and prove the algorithms for reducing and analyzing the data from these new payloads and to provide an orderly transition to these experimental instruments, which are first flown on small satellites and then become the next-generation mainline instruments on subsequent spacecraft. In addition, it encourages the pursuit of high-risk, high-payoff technologies. By conducting the development off-line without risk to the operational mission, fear of failure does not become an overriding principle that stifles innovation. Moreover, the development protocol for such technology demonstrations can be less formal and expensive. SIMULTANEITY Some EOS and NPOESS requirements necessitate multiple, nearly simultaneous observations of the same location on Earth. In some cases, this requirement can be met with a single highly capable instrument. For example, the Moderate-Resolution Imaging Spectrometer (MODIS) will provide nearly simultaneous observations in the infrared and visible portions of the spectrum for studies of cloud properties and illustrates the simplest level of integration necessary to achieve simultaneity. More difficult challenges in achieving simultaneity arise when data streams from multiple sensors must be combined either to derive a geophysical quantity or to study specific Earth processes. These challenges can often be met by co-boresighting multiple sensors on a common satellite and—in many cases—by deploying the sensors on multiple platforms flying in formation. Clouds are perhaps the most rapidly changing element of the Earth system: Cumulus cloud lifetimes can be as short as 15 minutes, and winds can move clouds at speeds greater than 1 km/minute. Therefore, measurements of the Earth's radiation budget (in which clouds play a dominant role) have the most stringent simultaneity requirements. The processes of cloud motion and cloud development will cause two different sensors to measure different portions of the cloud field. This discrepancy introduces random errors in the validation of cloud properties that might be derived from other sensors. More importantly, it introduces errors in the estimation of critical properties such as radiative fluxes within the cloud field. In the case of EOS platforms, this strict simultaneity requirement applies to measurements made by MODIS and CERES (Clouds and the Earth's Radiation Energy System) instruments. Researchers at NASA's Langley Research Center have performed analyses that account for two sources of co-location error: navigation errors between sensors and time differences between sensors. It was concluded that 6 minutes was the largest acceptable difference in time between the two sensors and that 3 minutes should be the goal. Other sensors require contemporaneous observations but do not need to be exactly simultaneous; the interval between these observations can vary from minutes to days. For example, studies of linkages between wind forcing over the ocean and primary productivity require that measurements be made on the same day. The critical fact, especially for climate research, is that the measurements are made during the same time period and for a sufficient length of time to resolve the important scales of variability. SAMPLING ERRORS When observing any process, the quality of the measurement is a function of both the intrinsic characteristics of the sensor—such as its sensitivity and dynamic range—and its sampling characteristics. In remote sensing, the
OCR for page 20
The Role of Small Satellites in NASA and NOAA Earth Observation Programs research community often emphasizes sensor performance, and technical innovation is thus driven to improve the sensor characteristics. Overall measurement quality is usually dominated by sampling errors, because the temporal and spatial scales of natural variability are not adequately resolved by the observing system. Although sampling theory will provide an initial quantitative estimate of the errors associated with a particular sampling strategy, evaluation of these errors depends in part on our understanding of the critical scales that must be resolved. For example, if we assume that ENSO events dominate primary production in the eastern tropical Pacific, we will design a sampling strategy to observe this system with acceptable error levels. But if smaller scale processes such as mesoscale eddies unexpectedly dominate the system, our observing strategy may not resolve them with sufficient accuracy. Sampling requirements are fairly well understood in operational satellite systems. Long time records and an extensive history of numerical weather prediction have provided a foundation for the development of a robust observing system. Earth system processes, on the other hand, are not as well understood, especially in the area of feedbacks and linkages between the components of the Earth system. Therefore, an Earth observing strategy for research must balance quality of measurement with quality of sampling. Multiple copies of lower performance sensors on a constellation of small satellites may provide a better data product than a single high-quality sensor that cannot adequately resolve key processes. The appropriate strategy will depend on the processes and scientific questions under study; there is no one correct answer. Increased sampling with multiple satellites does not necessarily improve the quality of the data product. The overall effect will depend on the details of the satellite sampling, the time and space variability of the geophysical field, and the scientific requirements. For example, Greenslade et al. (1997) analyzed satellite altimeter orbits in terms of their effective temporal and spatial resolution. Greenslade showed that the effective resolution is a function of the natural variability of ocean topography, the orbit characteristics of the platform (which drives sampling parameters such as repeat time), and the scientific criteria necessary to address a specific question (mean topography of the ocean). Although it would seem that multiple altimeters would automatically have better time/space resolution than a single altimeter, this was not the case. In fact, the orbits of the multiple platforms need to be studied carefully and coordinated in such a way that the sampling pattern does indeed improve the quality of the blended data product. SUMMARY Measurement requirements for operational programs and Earth system research reflect both scientific needs and the technical capability to acquire data of requisite accuracy and resolution. As scientific knowledge improves or as measurement capabilities increase, requirements can be expected to evolve. Although research and operational systems differ in their detailed requirements and approach, there are several common issues that must be addressed. These include instrument calibration, prelaunch sensor characterization, data record continuity, simultaneity, and sampling strategies. All of these issues influence mission architectures for small satellites, although the research and operational user communities may assign differing importance to them. Requirements for spatial resolution, calibration, and other sensor performance criteria greatly influence sensor size. In turn, sensor size affects the overall size of the platform and the launch vehicle. More stringent requirements, which are often associated with research missions, lead to larger satellite solutions. In contrast, operational missions need to ensure continuity of critical measurements in a cost-constrained environment. These requirements often lead to a design based on multisensor platforms using a block purchase approach. Very different success criteria can thus lead to similar approaches to system architecture. Small satellites can potentially alter significantly the approaches for both research and operational missions. Typically, the primary argument for small satellites in research is the ability to deploy low-cost missions, provide more flexibility in scientific and technical approaches, and obtain results more quickly because of a shorter development cycle. If small satellites and small launchers eventually follow a commodity pricing model (where the profit margin is small on individual units and revenue is generated from high-volume sales of these low-margin units), the science community will shift its perspective on remote sensing mission design.
OCR for page 21
The Role of Small Satellites in NASA and NOAA Earth Observation Programs Until now, most missions were designed to push the envelope in terms of technical ability and sensor performance. There was an implicit preference for a high-quality measurement once per day versus lower quality measurements several times per day. As the research community moves toward systematic observations of the Earth system, spatial and temporal sampling become more important factors. This is prompting a rethinking of the performance requirements of an individual measurement and the coverage requirements in time and space. Indeed, new satellite and sensor technologies are fostering a fundamental shift in Earth remote sensing measurement and satellite options. In particular, satellite constellations and clusters could provide significantly better coverage and open up new approaches for calibration and data continuity. The research community needs to evaluate the time and space scales of critical processes and match them with the appropriate sampling strategy. Such new satellite architectures no longer constrain the community to a single sampling approach, such as a Sun-synchronous orbit with a fixed equatorial crossing time. Although this chapter's discussion takes essentially a research perspective, the time sampling strategy of operational missions such as NPOESS could also be analyzed rigorously. If the EDRs were prioritized, a strategy different from the present small constellation of three medium-sized platforms might result. The committee recommends that both the research and operational communities perform a complete analysis of sampling strategies in the context of potential new mission architectures. The result of this analysis might be a different mix of sampling strategies, including small one-of-a-kind missions, constellations of small satellites, and a few mid-size multisensor platforms. As discussed in subsequent chapters of this report, the maturation of remote sensing science and the development of new sensor, platform, and launcher technologies allow for a more systematic approach to both research and operational Earth remote sensing. Just as personal computers and networks have revolutionized the way computational systems are organized, new technologies in remote sensing can shift the way we design observing systems. REFERENCES Bruegge, C.J., D.J. Diner, and V.G. Duval. 1996. The MISR calibration program. J. Atmos. Oceanic Technol. 13(2):286–299. Bruegge, C.J., V.G. Duval, N.L. Chrien, R.P. Korechoff, B.J. Gaitley, and E.B. Hochberg. 1998. MISR prelaunch instrument calibration and characterization results. IEEE Trans. Geosci. Remote Sensing 36:1186–1198. Greenslade, D.J.M., D.B. Chelton, and M.G. Schlax. 1997. The midlatitude resolution capability of sea level fields constructed from single and multiple satellite altimeter datasets . J. Atmos. Oceanic Technol. 14:849–870. Willson, R.C. 1997. Total solar irradiance trend during solar cycles 21 and 22. Science 277:1963–1965.
Representative terms from entire chapter: