Following the launch of Landsat 8 on February 11, 2013, there are several options for a sustainable land imaging capability. All approaches to sustainable land imaging require addressing programmatic as well as technical design. They require stable funding to escape from the chaotic on-again, off-again funding cycle that Landsat has experienced over the past 40 years. In addition to the requirements described in Chapter 2, all approaches need to address the biggest impediment to sustainability: cost. As Table 5.1 shows, life-cycle costs for each mission since Landsat 4 were about $1 billion, when adjusted to current-year dollars. Building an exact copy of Landsat 8 might seem to be the simplest approach for Landsat 9, but such an approach is not likely to substantially lower the cost. Exact parts are not likely to all be available. Moreover, it may not be possible to procure the spacecraft or instruments from the same providers, and even if the same providers were involved, the same teams of people may not be available.
The following options represent four different ways of creating an affordable, sustainable land imaging capability. Each option focuses on one aspect of affordability, but they can be combined intelligently. The committee does not assert that these are the only options, but they are representative examples.1
All options could benefit substantially from the utilization of a collaborative team approach between the U.S. government and its implementation partners, whether they are domestic contractors, international partners, or other teammates. By “collaborative approach,” the committee means that the government and its partners should operate as a single unit from an operations standpoint, not as employees representing separate entities each with its own unique goals and priorities. In this way, the parties are free of contractual and other impediments, such that they can truly work together to achieve fully successful solutions to problems as they present themselves with no fear of being “blamed” for any problem. This approach was recently successfully employed on the Air Force’s TacSat-3 program,2 as well as routinely in the high-resolution imaging industry.3
1 The committee’s recommended options are intended to apply in the timeframe after Landsat 9. However, they could also apply to Landsat 9, particularly if a decision on a successor to Landsat 8 is delayed past the fiscal year 2014 budget cycle.
2 T. Cooley, Air Force Research Laboratory, Space Vehicles Directorate, Kirtland Air Force Base, New Mexico. “Tactical Satellite 3: Mission Overview and Lessons Learned,” presentation to the meeting of experts titled “Towards the Use of Lower-Cost Platforms for the Acquisition of Environmental Data from Space,” March 30, 2012.
3 W. Scott, “Mission Assurance at DigitalGlobe: Success, Cost, and Schedule are Compatible,” presentation to the 2013 Mission Assurance Improvement Workshop,” April 30, 2013.
TABLE 5.1 Cost of Landsats 1 Through 8, Adjusted to 2012 Dollars
|Design Life (years)
|Original Cost ($ million)
|2012 Cost ($ million)
|$197 together with Landsat 2
|$197 together with Landsat 1
NOTE: 2012 costs calculated from http://www.bls.gov/data/inflation_calculator.htm, using year-by-year consumer price indices.
a See NASA ERTS-B Press Kit (NASA News Release 74-329), January 14, 1975, see http://www.scribd.com/doc/42461911/Erts-b-Press-Kit. This value includes research and development and the launch vehicles for both Landsat 1 and Landsat 2.
b See Landsat Policy Issues Still Unresolved: Report by the Comptroller General to the Congress of the United States, 1978, http://gao.gov/ products/PSAD-78-58.
SOURCE: Originally compiled by Tony Morse, Spatial Analysis Group, LLC, from the identified sources.
Several of the Landsat satellites have been acquired in a very expensive way. Particularly in the case of Landsat 7 and Landsat 8, each satellite included substantial new technology, was designed afresh, was acquired one at a time using cost-plus contracts, and was managed with a philosophy of over-engineering to minimize perceived risk, with the well-intended objective of improving the chances of mission success.
An acquisition model for a cost-constrained world is quite different. Rather than acquiring satellites one-off, this model makes block buys. Purchasing multiple spacecraft at once would reduce nonrecurring engineering costs and permit the advance purchase of parts, thus reducing their cost and improving availability later in the program’s life cycle. Additionally, a block-buy model would potentially enable the provision of spare spacecraft, either stored on the ground or in orbit (where the risky launch phase has been passed), which would make the program much more immune to unexpected failures. A long-term commitment would also result in the development and continuity of institutional memory in both the government agencies and aerospace contractors. This approach would be very similar to the model used by the National Oceanic and Atmospheric Administration for the provision of satellite observations to the National Weather Service for weather and severe storm forecasting.
Coupling the block-buy approach with a fixed-price contracting approach could reduce costs further. However, for a fixed-price contracting approach to be fully successful, the requirements must be well known and unlikely to be changed—for example, where the system being acquired is a copy of one that has already flown. And, after contract award, the government would need to minimize the number of contract change orders—ideally, to zero.
In the block-buy model, large-scale technological changes come with each new block, not within the block. In this regard, it is essential to only incorporate new technologies that do not compromise core operational capabilities. This could readily be done by leveraging industry, international, and/or other agency technology development activities. Additionally, each satellite in a block could accommodate a secondary instrument with a well-defined interface, on a noninterference basis, which would preserve the commonality between elements of the block while still allowing for modest, incremental technological insertion.
The acquiring entity must engage in a more collaborative relationship with the builder and be prepared to accept more perceived risk through less intrusive “light touch” oversight rather than the traditional very intrusive insight. While this seems unorthodox in light of several well-documented and high-profile acquisition failures over the past decade, it has been shown to work (for example, the Applied Physics Laboratory’s New Horizons mission, the University Corporation for Atmospheric Research’s COSMIC mission, the National Geospatial-Intelligence Agency’s (NGA’s) NextView and EnhancedView programs, the Air Force’s TACSAT-3 mission, NASA’s QuikScat mission, and so on), and it is particularly applicable to the block-buys-of-clones model that eschews new technology development for predictability.
The Landsat satellites are not the only source of Earth imaging data available today. By including other sources under the umbrella of the Sustained and Enhanced Land Imaging Project (SELIP), not only is it possible to mitigate risk (by having other sources to fall back on in the event of a premature satellite failure), but also it enables an even more cost-effective approach where the core program is not constrained to acquire all needed data on its own. The integration can create a more robust data set by using other existing or planned data sources.
Many of the possible options were exhaustively studied by the Landsat Data Gap Study Team from 2005 to 2007 after the scan corrector failure on the Landsat 7 ETM+ instrument.4 This excellent examination of the subject offers a framework for developing a robust and sustainable land imaging program that integrates sources of Landsat-type data from the international land imaging community. Although the United States started the Landsat series and has continued to exercise leadership over the past 40 years, leadership is not synonymous with going it alone. There is a long history of international partnering in other space endeavors. Burden sharing could take many forms: a foreign launch vehicle provided under a science-driven memorandum of understanding with no exchange of funds, instruments (such as thermal infrared, visible and near-infrared, or shortwave infrared) from an international partner, or a foreign satellite bus.
One example is the European Space Agency’s (ESA’s) Sentinel-2, which is planned to collect all but the thermal infrared bands of Landsat and does so in a wider swath for improved revisit.5 NASA is collaborating with ESA to calibrate the Landsat 8 and Sentinel-2 instruments to generate comparable data products. Such an arrangement could be complemented with data (also shared) from a U.S.-funded thermal-infrared-only small satellite. Other nations, such as India and Japan, operate their own remote sensing programs, which could potentially fill some Landsat user needs, and China is emerging as an Earth observing satellite operator in the coming decade. On the Suomi NPP satellite, the VIIRS instrument collects data at greater frequency though lower spatial resolution and may be suitable for some applications, particularly when sharpened by less frequently collected but finer resolution data to enable a degree of spectral unmixing.6 Finally, the EnhancedView contract, managed by NGA, collects commercial imagery that can be widely shared within federal government agencies, potentially satisfying some of their need for Landsat-type data, although the data from EnhancedView cannot be freely distributed to the public and, thus, does not offer the full value of a national land imaging program. None of these suggestions can replace a dedicated U.S. program for obtaining critical measurements; however, judicious use of other data sources may reduce risk, reduce cost in some cases, and enhance the SELIP.
A potential design modification, which applies to all other options, is to increase the swath width of the sensors, with the objective of shortening revisit time, a commonly sought characteristic of any new Landsat system. Historically, Landsat has acquired data over a 185-km swath, which, for a single satellite system, yields a 16-day
4 U.S. Geological Survey, Landsat Data Gap Studies, available at http://calval.cr.usgs.gov/satellite/landsat-data-gap-studies/.
5 European Space Agency, GMES Sentinel-2 Mission Requirements Document , available at http://esamultimedia.esa.int/docs/GMES/Sentinel-2_MRD.pdf.
6 B. Huang, Spatiotemporal reflectance fusion via sparse representation, IEEE Transactions on Geoscience and Remote Sensing 50:3707-3716, 2012.
revisit from a 705-km orbit altitude. Fortuitously, for many years we have enjoyed simultaneous coverage by both Landsat 5 and Landsat 7 (and now by Landsats 7 and 8), yielding an 8-day coverage pattern. However, flying two Landsat satellites in the future would likely be prohibitively expensive, except in cases where an earlier satellite exceeds its design life. Thus consideration should be given to increasing the swath width to reduce revisit time at far less cost than increasing the number of satellites. Landsat 8 can point its sensors off nadir ±15 degrees by a spacecraft yaw maneuver. This capability is implemented to enable data collection only for major disaster relief and recovery or other high-priority imaging.
ESA plans to fly a moderate-resolution multispectral system, Sentinel-2, with a 290-km swath width, which could improve revisit time to about 10 days with a single satellite and 5 days7 with the planned two satellites flying concurrently. With the current 185-km swath, the nadir view angle at the swath edge is 7.5 degrees, the sensor view angle (different because of Earth’s curvature) is 8.3 degrees, and the relative atmospheric path length is 1.010. With a 290-km swath, the corresponding angles are 11.6 and 12.9 degrees, and the path length is 1.026—a minor impact to angular viewing geometry at the edges of the field of view and, of course, no impact at all within the central 185-km swath for those applications that are particularly sensitive to angular viewing geometry. Generally the bidirectional reflectance distribution of most surfaces shows significant angular features at angles beyond 1 degrees from the nadir.8 Thus, the possibility of increasing the swath width for future U.S. systems needs to be explored in more depth, as it could help considerably with the goal of a shorter revisit time at lower cost.
Historically, every Landsat has included the full Landsat sensor suite of the time. Improved revisit times required more Landsats. Fortunately the extended life of Landsat 5 provided an 8-day revisit time, even though the original Landsat requirement was a 16-day revisit time. However, nothing compels future missions to involve only a single satellite, or for each satellite to contain the full sensor suite.
Smaller satellites can offer many benefits, either as an augmentation to a “mother ship,” such as Landsat 8 (with a full sensor suite), or as an ultimate replacement. RapidEye and the Disaster Monitoring Constellation (DMC) are already examples of less costly (though less complete) land imaging satellites that could augment SELIP by providing more frequent revisit times. A small satellite carrying only a thermal infrared sensor, placed in a phased orbit with the primary Landsat, could cut revisit time in half for much less than the cost of a duplicate Landsat, with the benefit of estimating evapotranspiration for practical water resource management. Alternatively, a small satellite carrying only a simple land imaging instrument, such as a slightly enhanced Multispectral Imager (MSI), routinely flown on the DMC of imaging small satellites, would cut revisit time in half for the nonthermal imaging channels. Two such small satellites, one with thermal and the other with VNIR and SWIR, flying in conjunction with the primary Landsat, might be able to provide near full capability at half the revisit time for dramatically less cost than two full Landsats.
Small satellites also offer several other benefits. They are intrinsically resilient, enabling intelligent trade-offs of redundancy at the constellation level, as opposed to requiring full redundancy in each spacecraft, allowing for lower cost. By being simpler (often single-payload), they have lower systems engineering, integration, and test costs. Their smaller size can enable them to fly as secondary payloads, reducing launch costs. They offer improved revisit because one can afford to acquire more satellites,9 so engineering teams can be continuously tasked instead of being organized and then dismantled for every mission. And by having more satellites, there are more opportunities for gradual introduction of new technology, enabling continuous improvement at lower cost and risk than wholesale replacement.
7 The planned revisit time is 5 days over the equator and 2 to 3 days over mid-latitudes. See European Space Agency, ESA-NASA Collaboration Fosters Comparable Land Imagery, February 13, 2013, available at http://www.esa.int/Our_Activities/Observing_the_Earth/GMES/ESA_NASA_collaboration_fosters_comparable_land_imagery.
8 M. von Schönermark, B. Geiger, and H.P. Röser, eds., Reflection Properties of Vegetation and Soil—With a BRDF Database , Wissenschaft und Technik Verlag, Berlin, 2004.
9 The RapidEye constellation of five small satellites cost $160 million, including launch (Space News, May 22, 2006, available at http://www.spacenews.com/archive/archive06/Briefs_052206.html).
To minimize risk, one or more low-cost small satellites could be launched before the end of the design life of Landsat 8. Not only would this demonstrate capability, but it would also allow for cross-calibration, as is common in many other scientific endeavors (Jason-1 was calibrated by underflying the gold standard TOPEX/Poseidon, not to mention the Landsat 7 underflight of Landsat 5 and the Landsat 8 underflight of Landsat 7).
To sustain U.S. land imaging, one would weigh the identified alternative approaches to implementing Landsat 9 and beyond and select a combination that best suits the circumstances of the moment. Fiscal resources are likely to be the leading constraint. One such approach might be to build Landsat 9 as a clone of Landsat 8. However, so much time has passed since Landsat 8 was procured and constructed that a true clone probably cannot be built. Some parts are likely to be unavailable; government procurement rules would make sole-sourcing the same contractors difficult; and the specific teams of people involved have gone on to other projects. Nonetheless, it might make sense to use Landsat 8 as a template for the next suite of missions, even allowing for some modest technological improvements (given the impossibility of building a true clone anyway), such as increasing the swath width. In this case, the desired approach would be a block buy of several identical units, perhaps Landsats 9 through 12. The design is fixed, the parts are all bought up front, and the same team builds all four units. With a fixed-price contract, the government making no changes along the way, and a collaborative team approach following “light touch” principles, significant savings would be realizable for Landsats 10 through 12. However, Landsat 9, a near clone of Landsat 8, would cost as much as its predecessor.
Therefore, if the overarching constraint is the cost of the next Landsat, then this approach is not viable. In such a case, one is forced to look at more creative, innovative, possibly riskier approaches such as constellations of small satellites. Considerable cost savings could result, especially for the first unit(s), but this approach would require the government to step outside its comfort zone and do something totally different, driven by the unavailability of funds that would allow doing otherwise.
Regardless of the approach selected, integration of the data from Landsat 9 and beyond with data from both commercial and international sources is necessary. Given these other factors, the committee does not recommend a specific course of action. The agencies and Congress must decide which combination of options to implement.
The Sustained and Enhanced Land Imaging Program will not be viable under the current mission development and management practices.
At least partly because of the unplanned, chaotic programmatic history of Landsat, the cost of each of five Landsat missions after the addition of the Thematic Mapper instrument has also been about $1 billion, when adjusted for inflation. Over the last 30 years, while there has been some technological improvement in the collection, processing, and use of Landsat data, there has been no reduction in the cost of a Landsat mission.
Building an exact copy of Landsat 8 might seem to be the simplest approach for Landsat 9, but that approach is not likely to substantially lower the cost for the next mission.
Nonetheless, options do exist to create a less costly, more robust SELIP, including the block buy of a sequence of missions, less cumbersome contracting processes, and technological innovations.
The Sustained and Enhanced Land Imaging Program should create an ambitious plan to incorporate opportunities to improve land imaging capabilities while at the same time increasing operational efficiency and reducing overall program cost.
The program should consider a combination of the following to increase capabilities while reducing the costs for land imaging beyond Landsat 8:
• Shift the acquisition paradigm by means of block buys and fixed-price contracting and by collaborating with commercial and international partners;
• Streamline the process by which satellites and sensors are designed, built, and launched, using a single organizational unit approach (a collaborative team approach) consisting of both government employees and contractors working together as a fully integrated team;
• Identify foreign sources of land imaging data that complement the U.S. core land imaging requirements and seek formal data-sharing agreements with them;
• Consider technological innovations, such as increasing the swath width and employing constellations of small satellites;
• Incrementally incorporate new technologies that leverage industry, international, and other technology development activities but do not compromise core operational capabilities;
• Accommodate candidates for improved or new instruments on a small satellite for the purpose of demonstrating new technologies; and
• Take advantage of opportunities to fly as a secondary payload or as a shared ride.