Menu
/ / / / GIF 2014 Volume 12 Issue 8 (November/December)

Hyperspectral Heads into Orbit

  • Written by  Henry Canaday, GIF Correspondent
  • Read 12075 times
  • Print
  • Email

Having already proved its value in identifying materials precisely from airborne platforms, hyperspectral imaging (HSI) is heading for orbit on satellites, where coverage can be persistent, much wider and unrestricted in reach. Several projects for putting HSI on satellites are under way or deep in planning stages.

Defense and intelligence customers have much to gain from satellite HSI, but reaping benefits will not be simple. The costs of orbital HSI may mean commercial revenues are also necessary to justify launches. Launch vehicles are limited and expensive. That may require multiple payloads to make HSI launches affordable, or it may mean schedule slippage. Choices and trade-offs must be made between the precision and resolution of HSI data and the size, weight and expense of platforms for obtaining it.

Even so, industry is pressing forward. In September, Boeing received an order for its 502 Phoenix small satellite from HySpecIQ, which seeks to launch high-resolution HSI on two satellites in 2018.

HySpecIQ processes and interprets HSI images for government and commercial customers. “We work for several verticals, mining and agriculture on the civilian side and defense and intelligence for government,” said Joseph Fargnoli, executive vice president for product development.

Briefly put, hyperspectral imaging sensors gather data from across the electromagnetic spectrum, seeking the spectrum for each pixel in an image. That enables users to find objects or identify substances based on their unique spectral characteristics.

HSI data is very rich and requires substantial experience to work with, Fargnoli emphasized. “You are collecting thousands of bands. That is a lot of knowledge, like focusing on a fingerprint.”

Airborne HSI is slow and yields limited data, so it is hard to justify the upfront investment in software necessary to process it quickly and profitably. The equation may work out differently, however, for satellite deployment.

“If we hit a point of scale economy so we can both collect and process enough to customize algorithms for mission applications, then we can automate it and exploit the power of the cloud,” Fargnoli said. “Then we can do it quickly enough to make intelligence available right after tasking.”

Satellite-borne HSI can generate the volume of imagery needed to reach those economies of scale. Moreover, it can view remote or restricted areas that airborne HSI cannot. “It can take years to get a license to stage an aircraft in some areas,” Fargnoli said. “A satellite can be there in days.”

So HSI in orbit could mean lower unit costs, faster processing and better coverage, which may eventually lead to much wider use. “We want to start with our verticals,” Fargnoli said. “Our ultimate goal is to democratize HSI intelligence for the masses.”

Airborne HSI has developed very good technologies, but is fundamentally limited in speed and volume, he suggested. HSI promoters gained confidence from an Air Force Research Lab proof of concept with ARTEMIS HSI on TacSat-3, which helped show that the degree of risk is reasonable.

With cloud capabilities reducing both computing and data storage costs, HySpecIQ hopes to be the first to offer satellite HSI commercially. Airborne HSI has been done commercially, but at prices that have severely limited markets.

“We can do high-value intelligence and combine HSI with other data,” Fargnoli argued. “Miners can see lithology. Oil men can see whether a substance is crude oil or another type of oil. Agricultural firms can track genetically modified seeds. Defense can see what is under that tarp and if we should bomb it.”

As prime contractor, Boeing provides the satellite, selects the HSI provider and does command, control, telemetry and preliminary data processing and calibration onboard and on the ground. HySpecIQ will do the high-value processing. Boeing will be the sales channel to the intelligence community and most defense customers, although HySpecIQ will deal directly with select defense users.

HSI’s advantage for defense and intelligence lies in its ability to discriminate among different materials more precisely and with more confidence. “Is a truck ISIS or Syrian Army?” Fargnoli asked. “HSI can pick up the difference in paint. Are emissions from a plant harmful or harmless? What emissions are coming from a North Korean nuclear plant?” Answering such questions requires getting the best spectral resolution, he continued. “For tactical defense, we can tell whether ground is marshy and how traversable it is. We can see 30 meters down in coastal littorals, which can be very useful to intelligence and defense.”

HSI does not have the spatial resolution of electro-optical (EO) imaging, but EO has just a few color bands. “We have thousands of color bands,” Fargnoli said, adding that HySpecIQ will have better spatial resolution than other HSI efforts.

The first block of two satellites will carry 600 kg each, with more to follow. Smaller satellites would yield much lower resolution.

Global Coverage

Boeing’s 502 Phoenix is its newest satellite, weighing 250 kg to 1000 kg and designed for single, dual and multiple-manifest launches. The HySpecIQ satellites will operate in sun-synchronous low earth orbit (LEO) after reaching escape velocity of 17,500 mph, noted Erik Daehler, deputy director of Boeing remote sensing programs and the 502 Phoenix product line, who said, “the sun-synchronous orbit provides global coverage opportunities every day with consistent lighting.”

The HySpecIQ system will provide the kind of high-resolution hyperspectral data needed by defense and intelligence users. A rival HSI satellite proposal would apparently offer broad area coverage and very low resolution, monitoring land resources with a ground-sampling distance of about 90 m per pixel. But Daehler suggested that the lower resolution probably means that the other proposal is addressing different markets and applications.

Boeing and HySpecIQ will collaborate on processing algorithms, data-fusion products and advanced, big data analytic solutions. HySpecIQ will deliver hyperspectral data products, decision-support tools and integrated analytics to commercial customers and limited defense segments, while Boeing will offer separate services to the U.S. intelligence community, other defense customers and select international partners.

“Providing the defense and intelligence community with first-of-a-kind, commercial high-resolution hyperspectral data will be game-changing,” Daehler predicted.

HSI provides significantly more information than the collectors now available today. EO and synthetic aperture radar images identify objects based on shape and texture, but HSI provides insights into materials and chemical composition.

“This product provides insights to the world that you cannot get from any other remote-sensing source,” he said.

The two HySpecIQ satellites offer an order of magnitude more spectral bands than any other commercial, space-based solution, Daehler contended. Higher spectral resolution enables finer discrimination among objects for identification and reduction of false alarm rates.

In addition, HySpecIQ will deliver answers to questions customers are asking, rather than just raw data. The project’s distinctions are thus delivering “unprecedented source data and solutions,” he said.

Under a contract announced as this issue was going to press, Raytheon will provide hyperspectral imaging sensors for the satellites.

Other Projects

There are other satellite HSI projects. More than two years ago, Headwall Photonics delivered HSI sensors to a Japanese consortium for an expected launch on a Japan Aerospace Exploration Agency rocket. The project, intended entirely for commercial purposes, has not yet been launched due to government launch schedules and mission priorities.

“There is a lot going on,” observed Headwall CEO David Bannon. “There are hundreds of satellite launches for communications and traditional radar and an emerging interest in hyperspectral.”

Headwall builds complete hyperspectral imagers and is the only manufacturer to build only diffractive optics and then integrate other elements. The company does not make focal plane arrays, which are supplied by major manufacturers that can grow the necessary wafers in sufficient volumes with sufficient quality controls.

Headwall does not process HSI data, but Bannon believes this has gotten easier. “Twenty years ago this was a complex technology, and you needed an optical Ph.D. to interpret it.”

Headwall provides the HSI data cubes—3-D digital portraits of the ground surveyed—that can now be interpreted by 
standard software.

Headwall’s sensors for the Japanese project capture visible and near-infrared (VNIR) waves from 400 nm to 1000 nm and short-wave infrared (SWIR) waves of 900 nm to 2500 nm. Each band is of interest to different potential users. For example, agricultural customers generally seek VNIR images, while mining and mineral industries usually seek images in the 2400 nm to 2500 nm range.

HSI combines all the wavelengths, but this puts burdens on bandwidth for download. Keeping images separate by wavelength yields multispectral images that satisfy some needs. “We can do either,” Bannon said.

The trend now is to launch many smaller satellites in order to provide frequent repeats, especially for disaster response purposes. HSI can fit on these smaller, more numerous platforms. Headwall works with customers to fit within size and weight constraints, and its typical imagers are 5 kg to 20 kg. Bannon said he expects that HSI satellites will fly in LEO of 600 km to 650 km.

“We build affordable imagers,” Bannon said, explaining that with a staff of only 50, Headwall usually charges a tenth of rival companies’ HSI costs. The company commonly works under firm-fixed-price contracts to meet agreed specifications.

Another benefit of Headwall HSI is that it gets a high signal-to-noise (S/N) ratio by maximizing both optical efficiency and quantum efficiency. “This is a technology differentiator,” Bannon said.

S/N can be 600 to 1 for VNIR and 300 to 1 at SWIR. Headwall HSI offers aberration-corrected imaging so that imagers can make wide swathes without distortions of the field of view. HSI technology is now proven, but one obstacle to satellite deployment is launch costs. Government funding is not plentiful now, so Bannon said he expects that future missions will have to serve both government and commercial uses.

Configurable Platforms

Some companies are potential end-to-end suppliers. Ball Aerospace & Technologies, for example, makes visible-light, multispectral and HSI sensors and the satellites that can carry them. They also process HSI data, noted John Sekula, staff director for Ball’s national defense strategic business unit.

The satellites, called Ball Configurable Platforms, range from BCP 100 to BCP 5000, with the numerals representing payload capacity in kilograms. The smallest BCP is a 3-foot cube.

Satellite HSI could be extremely useful for defense and intelligence purposes, Sekula said, due to its ability to detect weapons of mass destruction, IEDs, camouflage and targets, suspicious littoral conditions, illicit crops, factory effluents, disturbed earth and night illumination sources.

Ball is in discussions with Teledyne-Brown and NASA on putting HSI on the International Space Station in LEO. Teledyne Imaging Sensors would make the focal-plane array, the “digital film” that captures the HSI images. Neither timing nor a definite decision to launch has been settled, however, in light of budget pressures.

The best altitude for HSI orbit depends on the aperture of the telescope and the spatial resolution and S/N ratio desired. Higher altitudes gain resiliency and are harder for adversaries to defeat, but image quality can degrade quickly as altitude increases.

Ball satellites can host multiple payloads, so HSI might be an auxiliary payload; otherwise, Ball HSI might ride on another company’s satellite. Ball’s vibration-isolation technology allows a mechanically cooled Ball HSI to ride harmlessly along with other vibration-sensitive payloads.

Customers are increasingly looking at disaggregating sensors on multiple satellites to improve resiliency, Sekula explained, so Ball’s BCP 100 could be attractive for such an HSI constellation.

Ball’s Detector Technology Center can precisely design and integrate all elements of the HSI imaging system, including focal-plane arrays made by a variety of companies, such as Teledyne, Raytheon, DRS or Sarnoff. Moreover, Ball’s processing center in Dayton, Ohio, can do all the data extraction and analysis necessary to turn HSI data from any spectral range into useful intelligence. “We’ve been processing data for the National Air and Space Intelligence Center at Wright-Patterson AFB, Ohio, for 25 years,” Sekula pointed out.

Further, Ball’s HSI has a spectral range from visible to very-long wavelength IR, and so can co-register multiple focal-plane arrays to within a pixel width. “For an exquisite sensor, these capacities combine to limit our competitors to perhaps only one or two other companies,” Sekula argued.

Other HSI sensors might be useful for science, Sekula acknowledged, but he predicted that “national security will rely on the tried and true.”

Ball is further refining its capabilities, for example, by developing IR sensors that can operate at higher temperatures with large-format staring arrays, better spectral exploitation algorithms and countermeasures to spoofing. Sekula expects these advances to dramatically improve HSI performance relative to cost. For example, HSI that functions at 150 degrees Kelvin rather than 75 degrees enables designers to shed 200 pounds of cooling equipment.

One major challenge has been the cost and availability of launch vehicles. But HSI does not need large rockets, Sekula noted, and firms like SpaceX, Lockheed Martin and Orbital are bringing down the costs of putting smaller payloads in orbit.

Data Processing

A different perspective comes from Sean Anklam, president of Exogenesis, which designs software to process data from satellite-borne HSI. “The value was proved by airborne HSI, but airborne can’t do many things in space restricted for military, political and climatic reasons. It’s difficult to fly an aircraft on top of a mountain range or polar ice cap,” he said.

Exogenesis can help process and productize HSI data for specific industries. “We build cloud-deployable stream processing analysis nodes for this data,” Anklam explained. Exogenesis expects to process data for at least three of the four satellite HSI projects that Anklam is aware of.

Anklam believes defense and intelligence have “huge gaps” that satellite imagery could fill. Unlike the short-term ARTEMIS, “this would be persistent HSI,” he said, adding that he expects the first HSI satellite to launch in 2015, and a significant number to go up in 2018.

Sierra Nevada Corp. (SNC), which makes satellites, is seeing substantial interest in HSI payloads, said John Roth, vice president of business development for space systems. “The military did TacSat-3 and is interested, and commercial customers are interested. We are getting contacted by startups. At least three of these are interested in putting HSI on satellites.”

While Roth expects HSI to be launched into orbit, he also sees substantial challenges. “The imagery must be processed, and there is no generic solution; you must tailor it to each sensor,” he said.

There is no archived HSI data on terrain to compare new images with and detect changes, as there is for EO imagery. In addition, images can vary even on the same HSI satellite depending on time of day and look angle, which can pose difficulties in interpretation.

HSI can be put on small satellites, but not too small, he cautioned. TacSat-3’s 150 kg to 200 kg payload was sufficient for demonstration purposes, but possibly not for operational HSI. Roth reckons that would require satellites for 250 kg to 500 kg payloads.

“Launch costs and availability are perennial issues,” Roth said, noting that industry hopes that SpaceX Falcons could be held to $10 million or less per launch have been disappointed. “There is no good low-cost vehicle for 500 kg payloads.”

But when HSI developers are ready, SNC has the platforms. The company makes high-quality small satellites, larger than CubeSats but smaller than 1000 kg. “If a customer needs high point accuracy, rapid slews for multiple areas and high-power systems, we excel in that,” Roth said. ♦

Last modified on Tuesday, 16 December 2014 08:47

Additional Info

  • Issue: 8
  • Volume: 12
back to top