Findings and Recommendations
Advances in technology and the response of the marketplace have led to smaller sensors, satellites, and launch vehicles capable of performing useful space missions at relatively low cost. The Committee on Earth Studies has examined the capabilities and potential roles of such small satellites in key National Aeronautics and Space Administration (NASA) and National Oceanic and Atmospheric Administration (NOAA) Earth observation programs. The committee's study focused in particular on the use of small satellites in the NASA Earth Observing System (EOS) program and the planned NOAA-Department of Defense National Polar-orbiting Operational Environmental Satellite System (NPOESS) program. This chapter reviews, in the broad context of an integrated study, the topics highlighted in the summary sections of the preceding chapters, and it presents key findings and recommendations.
Earth observation mission life cycle costs are the sum of all those incurred for developing, fielding, operating, and maintaining all elements of the system; these include the sensor payload, satellite, launch vehicle, and ground segments (command, communications, control, and data processing). It is often difficult to determine the true costs of some of these mission elements because prior expenditures on technology development, system infrastructure, and even some mission components (e.g., available sensors) may not be accounted for. When assessing and comparing costs for alternative mission architectures, comparisons must be made on an equivalent basis and not be biased by unequal treatment of these ''hidden" costs.
The lowest cost missions are achieved when the costs of all mission elements are minimized. For their space segments, the promise of low-cost small satellite missions is based on this approach—low-cost sensors accommodated on low-cost small satellite buses and launched on low-cost small launch vehicles. The committee found that small launch vehicles are available at substantially lower cost than the Delta or Atlas class used for mid-size or larger satellites, but at higher specific cost (cost per pound of satellite to orbit). Similarly, small satellite buses are available at substantially lower cost than the larger Delta or Atlas class satellites (depending on capability), but again at higher specific cost (cost per pound of satellite). Thus, low-cost sensors that can be accommodated on these smaller satellite buses and launch vehicles can be flown as low-cost small missions. Technology advances are helping reduce sensor size and cost (as long as complex deployables are not required). Simple missions with limited objectives will yield the lowest cost scenarios.
A different situation is extant in a trade between multiple small satellites versus a larger multisensor satellite to accommodate a given sensor payload. Here the higher specific costs for small satellites and small launch vehicles will generally result in a higher cost to field the system initially (but not necessarily to maintain it) than using a larger multisensor satellite and a matching launch vehicle. This is true irrespective of sensor size or cost.
MEETING MISSION GOALS: OPPORTUNITIES WITH SMALL SATELLITES
Much of the interest in small satellites stems from a desire to "do more with less" and an assumption that small satellite missions result in lower costs. A more pragmatic objective reflecting recent budget realities might be to "spend less and do as much as possible." Small satellites clearly provide a vehicle for accomplishing the latter. Here, the committee distinguishes between "small satellites'' (100 to 500 kg), "small missions" (low cost), and larger (higher cost) missions that are performed with multiple small satellites. Low-cost small satellites help enable low-cost small missions. This benefit is derived as much from the relative simplicity of many small missions (or the preexistence of mission elements) as from the size of the satellite. Small missions generally consist of only one or a few sensors and may have less stringent requirements as measured by performance, calibration, or longevity. Less complex missions require shorter development time, which goes to the heart of lower costs. Because simpler missions may also be less capable, the science and operational needs must be carefully evaluated to ensure that they are adequately addressed.
When considering small satellites to perform a larger mission involving a number of sensors, a mission architecture trade-off study is required. Alternative architectures include accommodating all sensors on a single larger platform, on multiple small satellites, or on a mixed fleet. Trade-off criteria may include programmatic flexibility, preferred measurement sampling strategies, risk tolerance, system robustness, schedule, and—of course—life cycle cost. The lowest cost architecture for such missions is not evident a priori, but depends on mission-specific parameters.
One of the emerging benefits of small satellite missions is a reduction in the "time to science." Large, complicated missions often take many years to develop, during which time both scientific understanding (and hence requirements) and technology may evolve substantially. In addition, an increasingly cost-constrained fiscal environment makes large missions especially vulnerable to budget instabilities. When a large mission can be accomplished with multiple small satellites, this approach may lead to faster science return—but this is not guaranteed. The overall schedule and cost must be examined to determine if the need for multiple satellites and launches increases or reduces the time interval to establish full capability. The potential for obtaining some (perhaps the most important) data sooner can be a compelling driver.
Small satellites offer the potential for new mission architectures, such as clusters or constellations. Such architectures may permit development of new observing strategies that alter the relative balance between observation error (as quantified by parameters such as signal-to-noise ratio) and sampling error. Employing constellations of small satellites to acquire large amounts of observational data, albeit of perhaps lower quality,1 may provide a more robust estimate of the overall statistics of the data field. This aspect of the scientific mission has not been examined in detail in this report and would have to be considered on a case-by-case basis.
OPERATIONAL AND RESEARCH EARTH OBSERVATIONS
Although there are differences between the operational measurement requirements of missions such as NPOESS and the science requirements of research-oriented missions such as NASA's EOS, there is clearly overlap as well. Moreover, many operational measurements are useful for research, especially for long-term climate studies. The separation of instrument variability from the often subtle, long-term variations in climate-related processes requires careful calibration and validation of the sensor and its derived data products. As sensors are replaced over time, it is essential to maintain "dynamic continuity" of the data product
despite changes in sensor performance. As a result of the need to protect life and property, operational systems generally have little tolerance for temporal gaps. Research systems can generally tolerate longer gaps, especially in the area of climate research, as long as dynamic continuity of the data can be achieved through calibration. Cross-calibration of sequential sensors by ensuring temporal overlap of their satellite platforms is a preferred method of guaranteeing such dynamic continuity.
Except for studies of clouds, most of the research systems do not need strict simultaneity with co-boresighting of multiple sensors on a single platform. Rather, it is more important to ensure that a full suite of sensors is available to measure processes related to coupling of various components of the Earth system, such as air/sea fluxes, and that this suite is continued for a sufficient period of time. Thus, the emphasis in research is on contemporaneous data sets rather than strict simultaneity. For operational systems, strict simultaneity is also not generally required. Because the sensors are not all co-boresighted and because some have inherently different sampling strategies, even operational satellite platforms that carry multiple sensors mostly provide contemporaneous rather than simultaneous observations. Distributing appropriate groups of sensors (e.g., an atmospheric sounding suite) on a cluster of satellites flying in close formation would likely meet requirements for spatially coordinated measurements in operational weather systems.
The requirements for research missions evolve rapidly with advances in science and technology. Long development times often run counter to this emphasis on flying the latest in sensor design. Moreover, research missions emphasize the quality of the individual observation and thus constantly push the technology envelope in an attempt to obtain better quality data. Operational systems, on the other hand, tend to evolve more slowly, in part in response to budgets, which grow more slowly, and in part in response to the well-defined operational nature of the missions. For example, the data processing infrastructure of the operational user community often involves numerical models that may be expressly designed to assimilate satellite measurements collected at specific times with specific observing characteristics.
Small satellites offer new opportunities to partition a program's space architecture between small, focused missions and larger, more comprehensive missions. This flexibility is of particular importance to meet the differing needs of operational and research missions. For example, operational missions may use small satellites in a replacement strategy to ensure minimum gaps of critical data records, whereas research missions may use small satellites for maximum programmatic flexibility and to ensure minimum "time to science."
The potential to design smaller satellites for Earth observation missions is driven by advances in microelectronics and other technologies that facilitate the design of smaller and lighter sensors and spacecraft subsystems. However, there are fundamental laws of physics that in some cases restrict the degree of miniaturization that can be achieved while retaining sufficient performance to meet the observation requirements.
Sensor design and size are determined by a complex trade-off among spatial, spectral, and radiometric performance. All three of these performance measurements are interdependent, and, barring compensatory design changes, each measurement is improved at the expense of the others. Developing a design that balances all of these performance parameters while minimizing size, cost, and technical risk is the essence of sensor system engineering.
Within these constraints, technology improvements may alter the various design trade-offs and permit improved performance, lowered costs, and/or more compact sensors. Reductions in sensor size, mass, and power can have substantial leverage on the entire space segment architecture and costs in that smaller sensors can be accommodated on smaller spacecraft and smaller spacecraft placed into orbit with smaller launch vehicles.
The fundamental philosophy of sensor design should also reflect the architectural trades available with small satellites and focus more on specific observing tasks than on general applications. The more observation requirements that a particular sensor attempts to fulfill, the more complex the design. Such general-purpose sensors often must balance conflicting requirements, sometimes resulting in poorer performance than would be achieved by a design focused on a subset of the requirements. Reduction in size and system complexity, as well as simpler
payload integration, can often be achieved by developing "task-specific" sensors. When measurement objectives are limited, task-specific sensors on small satellites are a promising approach to low-cost, focused missions.
Technically capable small satellite buses suitable for Earth observation missions are now available. Further development efforts should continue to reduce power, weight, volume, and other aspects of these platforms as well as enhance their payload accommodation capabilities. Present science and operational payloads are sometimes too big, require too much power, or create too much vibration to be accommodated on these platforms, but the trend toward smaller sensors should expand their utility in the future. Depending on capability, small satellite buses to support 100 to 500 kg spacecraft are available at recurring costs in the range of $10 million to $30 million. Low-cost "production" buses, developed primarily for telecommunications applications, generally must be tailored to meet the needs of Earth observation missions (for example, adding star trackers to provide more precise pointing); such customization can add substantially to costs. To take best advantage of such buses, instrument development must focus more on building to a standard platform interface rather than customizing the interface (and the platform) to accommodate the sensor. Such an approach is the reverse of the way sensors are traditionally developed, where sensor requirements drive platform design, and care must be exercised not to compromise the science objectives.
The launch vehicle must be matched to the mission if costs are to be minimized. Although the cost per kilogram of performance (maximum satellite payload mass to orbit) increases as the size of the launcher decreases, the total cost of the system will generally be lowest when launch vehicle performance is matched to satellite mass. Excess capacity represents wasted costs. Present launch vehicle capabilities do not effectively span the range of potential payloads, and there are notable gaps in performance, especially at the low end of the range. Fairing volume (which determines the stowed payload size) is also limited with small launch vehicles and sometimes determines the type and complexity of deployable systems such as antennas. More flexible launch systems are needed where volume constraints are less stringent. Launching multiple satellites with a single larger launch vehicle is a partial remedy to these problems but comes at the expense of programmatic flexibility—a potential feature of small satellite missions.
The present scarcity of reliable small launch vehicles is of great concern. In part, this lack of reliability is a result of the relative newness of these systems, and the situation should improve with time. Robust mission architectures can mitigate the problem in the interim. For example, where possible, parsing the payload among clusters or constellations of small satellites allows a mission to be fielded and/or maintained by small launch vehicles with only modest sensitivity to failure.
The relative merits of small, mid-size, and large platforms are a complicated function of the overall mission objectives, available budgets, and success criteria. These criteria are significantly different for research and operational missions. For example, operational systems are judged by performance, life cycle cost, and availability (the percentage of time the system can deliver timely data, often on demand). Long gaps between observations or loss of a single critical sensor can result in mission failure. With research missions, dynamic continuity and data quality as well as the flexibility to pursue new sensors and new science requirements are more important than just data availability. These differences may lead to different mission architectures.
Trade-offs between multisensor platforms versus multisatellite approaches arise whenever a mission requires multiple sensors on orbit simultaneously. Such trade-offs are analogous to the multimeasurement versus special-purpose sensor trades discussed under payloads. The lowest initial cost to place a given suite of sensors on orbit
is likely with multisensor platforms. However, the architecture with the lowest life cycle cost to field and maintain an operational suite of sensors in orbit is mission specific, dependent on the design lives and reliabilities of the system elements (for example, sensors, spacecraft bus, launch vehicles, and ground segment). With fewer launches and satellites involved, multisensor platforms offer a higher probability of successfully placing all sensors into orbit. Although the impact of a launch vehicle or spacecraft bus failure on a multisensor platform is obviously severe, these events have relatively low probability as multisensor spacecraft are typically designed with redundant systems,2 and candidate launch vehicles such as the Delta II have historically demonstrated high reliability. Managers of operational missions consequently typically choose the multisensor, common bus approach. Small satellites flying in formation with the larger platforms can play a critical role in timely replenishment or augmentation strategies for these systems.
For Earth science research missions, small satellites provide important flexibility to the overall program. They provide schedule flexibility when responding to advances in scientific understanding, changes in scientific priorities, or development of new technologies. New missions can be developed and launched without waiting for accommodation on more slowly evolving multisensor platforms. Small satellite clusters and constellations provide new sampling strategies that may more accurately resolve temporal and spatial variability of Earth system processes. New approaches to calibration may be possible as well. These and other characteristics of small satellites allow the mission developer to design a program that is more balanced between long-term monitoring observations and short-term measurements for research than would be possible if using only larger systems.
The design of an overall mission architecture, whether for operational or research needs, is a complex process and requires a complete risk-benefit assessment for each particular mission. A mixed fleet of small satellite and larger multisensor platforms may provide the best combination of flexibility and robustness, but the exact nature of this mix will depend on mission requirements.
Small satellites present several opportunities for both research and operational Earth observing missions. First, as noted previously, small satellites can enable low-cost missions and support rapid deployment. Their relative simplicity allows them to take quick advantage of innovations in both science and technology, thus leading to a shorter time to science. Second, small missions may enable new sampling strategies based on clusters or constellations. Such strategies may significantly enhance the scientific quality of the resulting data set and represent a new approach in Earth remote sensing. Third, missions built on a mixed fleet of multisensor and single-sensor spacecraft enable new maintenance and replenishment strategies, which may be particularly important for operational missions such as NPOESS. To realize these benefits fully, the management structure for the missions must be properly aligned as well. Small missions, which are generally less complex and can accept more risk, are best implemented with less oversight.
Management overhead can easily slow system development or discourage innovation, thus inhibiting many of the advantages of the small satellite approach. This argues for a management structure that is streamlined and less hierarchical than that typically employed. Small, tightly integrated teams have an advantage in such an environment; care must be exercised in how the teams are managed to avoid "burnout." In this context, for example, reducing team size too much in an effort to lower overhead costs becomes counterproductive. Excessive stress on the team can also be reduced if management has an understanding of and plans for the greater risk of failure that is typical in the "faster, better, cheaper" approach. It is also advantageous if government involvement is participatory and is characterized more by insight than oversight.
Implementing new mission architectures based on a mixed fleet requires changes in management as well. First, there must be appropriate mechanisms to ensure the design and maintenance of a coherent observing strategy. For example, in the EOS program, solicitations for new missions should be consistent with the overall science directions of the Earth Science Enterprise. Second, management must address the issues associated with maintaining dynamic continuity of long-term data sets where the specific sensors (and even measurement techniques) will change over time. For operational missions, this includes replenishment strategies to replace failed sensors. It also includes developing comprehensive plans for cross-sensor calibration, data validation, and prelaunch sensor characterization. Third, with the assistance of the science community, management should undertake quantitative evaluations of how data quality varies under different assumptions regarding sampling frequency and individual sensor quality. This evaluation should include an assessment of the impacts of data gaps as well as temporal and spatial resolution.
There is no single best strategy that replaces larger satellites with smaller satellites. However, an overall mission architecture that effectively combines the elements of large, mid-size, and small satellites can now be developed for Earth observing programs. Mission architecture choices must be driven by the requirements of the eventual users of the data. In planning for future missions, the committee recommends that NASA and NOAA consider the merits of small, medium, and larger satellites without prejudice, seeking the most appropriate system architecture based on mission requirements and success criteria.