- Has anyone driven over certain desert dunes in Afghanistan during the past few hours?
- Is a ship on the high seas fishing illegally?
- Is subsidence threatening buildings in an urban area that previously had been mined for coal?
- How much biomass is there in a square kilometer of forest?
- How much ice is in a section of the Arctic Ocean?
- Satellite-based radar can answer all of these questions.
Currently, only the Canadian company MDA, the German company Infoterra GmbH, and the Italian company e-GEOS operate commercial radar satellites. All three missions — RADARSAT-2, TerraSAR-X, and COSMO-SkyMed, respectively — have similar resolution modes. RADARSAT-2 captures data at “high-to-low” resolutions, while TerraSAR-X and COSMO-SkyMed capture data at high- to mid-resolution. U.S. operators of Earth observation satellites have not deployed radar instruments, opting instead to focus on higher resolution optical imaging. However, two U.S. companies, Fugro EarthData and Intermap, acquire radar imaging from aircraft.
In October, South Korea is scheduled to launch its first radar satellite, the KOMPSAT-5 (Korean Multi-purpose Satellite 5) system, which will provide X band SAR images of up to 1-meter resolution. Also this year, India is planning to launch into a sun-synchronous orbit its RISAT-1 (Radar Imaging Satellite 1), which will carry an active C-band SAR imager with a maximum resolution of 2 meters.
How Radar Works
Similar to a flash camera, an imaging radar (acronym for radio detection and ranging) provides its own light to illuminate areas on the ground, except at radio wavelengths rather than using visible light. A radar emits microwave signals, measures the strength and round-trip time of the reflections (backscatter returns) off a distant surface or object, converts these reflections to digital data, and passes them to data recorders for later processing and display as images. The round-trip time for each of the pulses, which travel at the speed of light, provides the distance (range) to the reflecting objects, while the chosen pulse bandwidth determines the resolution in the range (cross-track) direction. As an imaging radar moves along its flight path or orbit, its footprint moves along the ground in a swath, building the images.
The image resolution in the azimuth (along-track) direction depends on the length of a radar’s antenna — the longer the antenna, the finer the resolution. Very long antennas are impractical, however, so synthetic aperture radar (SAR) was developed as a way to synthesize a very long antenna by combining signals the radar receives as it moves along its flight track. Typically, image analysts are not familiar with interpreting high-resolution SAR data and therefore tend to resist this type of imagery because of its complexity and the understanding required to benefit from it, explains Kern. “However, simply looking at SAR images is an absolutely viable way of using such data,” he adds. “We provide courses and other tools that help analysts get the full value out of SAR-detected images.”
Satellite-based radars are well suited and often used for monitoring large, remote regions. Therefore, radar imaging is commonly used in the energy sector, especially by oil and gas companies, as well as for ice mapping, maritime surveillance, border control, detecting pollution and illegal fishing, and monitoring pipelines and ship traffic. “We can provide routine views of remote areas and then aircraft or ships can follow up,” says John Hornsby, general manager of MDA’s Geospatial Services Division.
Interferometric SAR (InSAR or IfSAR) is used to develop accurate terrain elevation data. Differential IfSAR (comparing multiple IfSAR datasets) is used to monitor land deformation, with a precision in the order of centimeters. One application is to allow reservoir engineers to adjust operations so as to avoid well sheer. Similar techniques are used to detect subsidence in urban areas, for example, in ones built over former coal mines.
A new and important application is coherence map analysis, which involves combining multiple IfSAR datasets to analyze, for example, whether a vehicle has moved along a road or what kind of agricultural activity is taking place in an area. “Traditional change detection allows us to detect a new building, but now we can analyze changes in surface texture,” says Giorgio Apponi, director of COSMO-SkyMed International Development.
Satellites vs. Aircraft
Satellite-based radar imaging differs from aerial radar imaging in two key respects, explains Kern. First, it differs in terms of accessibility, as satellite-based radar coverage area is global and has no denied areas and second, in terms of scalability, because with satellite-based radar it is cost-effective to cover very small areas, for example, for a single map sheet. “Satellite radar mapping is faster and cheaper than airborne radar mapping,” he says. Satellites’ temporal revisits also make them better than aircraft for continuous, time-lapse monitoring and change detection.
Conversely, aircraft-based radar is optimally suited to cover large areas with multiple products in a very short period of time. It is also a much more flexible tool than satellites, which are limited by orbital dynamics as to where they can look, says Louis Dean, Fugro EarthData’s senior advisor for science and technology. However, any airborne system involves the challenge of operating aircraft, which are maintenance-intensive.
The U.S. government now is paying more attention to radar imaging for defense and security applications than in the past. To support its requirements, in December the U.S. National Geospatial-Intelligence Agency (NGA) awarded three Indefinite Delivery Indefinite Quantity (IDIQ) Multiple Award Contracts (MACs) for commercial satellite synthetic aperture radar (COMSAR) imagery, data products, and direct downlink services to MDA Geospatial Services, Inc., to EADS North America with the EADS Astrium subsidiary Infoterra GmbH in Germany, and to Lockheed Martin Space Systems Company in partnership with e-GEOS of Italy. Each contract includes a five-year ordering period with a ceiling of $85,000,000.
As the Department of Defense’s sole procurer of commercial remote sensing (CRS) data, the NGA routinely procures commercial imagery from both domestic and foreign sources. The agency and its mission partners utilize commercial SAR data for intelligence, military, and homeland security applications. According to the NGA, these contracts will improve its ability to provide intelligence in low light and bad weather conditions.
There are many levels of geospatial data, Kern points out. “The more sources analysts are able to consult —including optical and radar images, old maps, etc. — the more reliable and higher quality will be the resulting information products.”
Digital elevation models (DEMs) are the initial foundation of any accurate geospatial product, he argues, as they form the baseline on which other image fusion — a process that combines multi-source imagery and data — is achieved. Classical data fusion involves simply overlaying high-resolution optical images onto radar images and conducting image analysis, using different sources at different resolutions and maybe at different times. “Because fused data provides for robust operational performance — i.e. increased confidence, reduced ambiguity, improved reliability, and improved classification — accuracy of the underlying DEM is critical to mission success.”
Fugro EarthData also routinely performs data fusion. We have had good success in integrating our products with other geospatial products — using our GeoSAR DEMs to orthorectify other source imagery,” says Dean.
“While being a valuable data source in its own right,” Kern adds, “space-based radar data provides additional benefit when used in combination with data from other sources. For maritime surveillance applications, for instance, we combine our radar data with AIS ship transponder signals, in-situ information, and optical image data. When a ship does not send an AIS signal, it may be trying to hide something — for example, that it is engaged in illegal fishing or oil discharge. In an experiment that we conducted last year, we combined satellite and coastal radar as well as AIS data in support of identifying ships that were illegally dumping oil on open seas. If we detected the signature of oil and a ship but no AIS signal, it was an indication of potentially illegal oil discharging activities.”
He continued, “This is not a brand new application; it was already supported by mid-range radar with resolution of 30-50 meters. However, with 5-meter SAR images it is possible to detect much smaller vessels. Based on the results of the experiment, we developed a fully integrated service concept that combines radar data (location, heading, and speed) with AIS on/off data. This integrated product is helping to speed up the process of ship detection and identification for maritime authorities (coast guard, police, and navy).”
Another application where space-based radar provides unique capabilities is surface motion monitoring based on coherent multi-temporal acquisitions, integrating very precisely co-registered radar data. Any kind of additional data — such as land cover/land use, geological maps, or activity information — supports the final information production, but is not mandatory, says Kern.
For many years, MDA (MacDonald, Dettwiller and Associates, Ltd.) operated the only commercial radar satellite with RADARSAT-1, launched in 1995. The company built RADARSAT-2, an advanced C-band commercial Synthetic Aperture Radar satellite, which launched in 2007. MDA operates RADARSAT-2, sells the data it produces, and provides data analysis for specific industry applications like DEMs, maritime surveillance reports and oil spill detection. “We are vertically integrated, providing end-to-end solutions to customers who require real-time intelligence, surveillance, and reconnaissance information from space-based and airborne sensors,” says Hornsby. “We continue to make improvements and changes because the nature of the RADARSAT-2 sensor is such that we can modify the imaging to tailor it to our users’ changing needs.” The evolution, he points out, is toward higher resolution – from 3 meters when the satellite was launched, it is now down to 1 meter, but they are investigating increasing the swaths of some of the high and medium resolution beam modes in order to better meet the needs of some customers.
“One of the fundamental reasons for building our satellite, and a requirement of the Canadian government, was to monitor ice,” Hornsby recalls. “We’ve had NGA contracts for many years, especially supporting the National Ice Center (NIC), a multi-agency operational center operated by the U.S. Navy, NOAA, and the U.S. Coast Guard. RADARSAT is a work horse for that requirement, but it also supports other government agencies.”
Over the years, MDA has greatly improved the timeline for tasking satellites and for collecting data, Hornsby says. “We are often held to 30 minute delivery from time of collection. Our polar-metric product — now available but not yet fully exploited — represents a further opportunity. We are preparing a RADARSAT constellation, which will provide much greater revisit times. In the future, we will probably see more partnerships and collaborations to provide better solutions to our customers.”
Infoterra has seven product/application lines and all of them, according to Kern, benefit from data acquired by its high-resolution radar satellite TerraSAR-X, for which the company holds the exclusive commercial exploitation rights. “The majority of our revenue comes from TerraSAR-X image products, both with and without training,” he says. “TerraSAR-X’s work horse is the StripMap mode (up to 3-meter resolution with a scene width of 30 kilometers and length of 50 kilometers, extendable to up to 1,650 kilometers), which we consider the best trade-off in regard to resolution, scene size, and polarization options. Whereas COSMO-SkyMed has advantages when it comes to temporal resolution, with TerraSAR-X we clearly benefit from the unique horizontal location accuracy of our system (sub-resolution), which contributes to our capability to generate high quality DEMs.”
Infoterra’s clients and projects include emergency support for the European Global Monitoring for Environment and Security (GMES) program and equivalent projects through its private partners, who are, for example, clients in Australia, Japan and China; topographic and land use mapping in Southeast Asia; monitoring of surface movement phenomena in urban areas in Southern Europe and China; and global monitoring of sites and infrastructures.
e-GEOS is a public-private company: Agenzia Spaziale Italiana, the Italian space agency, holds 20 percent of the shares and Telespazio the remainder. The company, which took over all of Telespazio’s Earth-observation activities, owns and operates the COSMO-SkyMed constellation of three radar Earth-observation satellites. It will launch a fourth satellite in Q4 2010 and the second generation of COSMO satellites in 2013. In partnership with Argentina, it is developing two L band satellites and expects to launch the first one in 2011 or 2012.
e-GEOS’ mission is to collect and distribute commercial imagery, as well as to develop new products and support customers in deriving the data they need for their specific purposes. COSMO-SkyMed was supposed to have other European partners and focus on the Mediterranean area, but it now provides world-wide coverage in a sun-synchronous, near-polar orbit.
The company’s constellation allows it to provide more frequent revisit and more imagery than a single satellite would, points out Apponi. “We can offer up to four images per day of one location on the equator and even more at higher latitudes — for example, five per day at 40 degrees of latitude and up to 24 per day in the Arctic. All the satellites have the same revisit time. Furthermore, we have the only constellation based on identical satellites.”
“Our satellites have a resolution of up to 1 meter with Spotlight mode,” Apponi continues. “All other satellites can observe an area from 20 to 60 degrees from nadir, but only COSMO-SkyMed can guarantee a 1-meter resolution at any of these angles, over 100 square kilometers and a potential swath of 600 kilometers on the ground.”
For the NGA contract, e-GEOS teamed up with Lockheed Martin. “We have already begun receiving orders under the contract,” says Apponi. “We FTP the data from Italy through a high-speed connection, or downlink telemetry and images directly from the satellites to U.S.-based ground stations. There is one now in Miami and we expect the U.S. to soon set up more.
Fugro EarthData’s GeoSAR
Fugro EarthData owns and operates the GeoSAR radar mapping system. It is the world’s only dual-banded, dual-sided, single-pass airborne IfSAR commercial mapping platform, says Dean. GeoSAR is designed to cover large areas with multiple products in a very short period of time. It captures imagery in both the X band (surface features) and the P band (foliage penetration). There are several other commercial airborne IfSAR providers, but none are single-pass, dual-band, and dual-sided and they are less efficient in terms of acquisition, Dean adds. “We always fly with a triple-return, triple-intensity LiDAR profiler that looks near the nadir of the aircraft, so we can use those as virtual ground control points, instead of having to send someone in the field in a dangerous spot.”
The X band is the most common commercial SAR product, and Fugro’s X band product is similar in many ways to those of other companies. What differentiates Fugro EarthData’s products from those of satellite-based systems, according to Dean, is its P band data, which the company collects simultaneously. This dual-band system produces both magnitude imagery and elevation from a single flight. Both are ortho-rectified and meet the standards to interface with other geospatial products. “X band interferometry yields the canopy height for trees and other vegetation,” Dean explains. “P band is long-wavelength, UHF (ultra-high frequency - 300 MHz to 3 GHz). It penetrates foliage; therefore it can see the ground, and ground-trunk interactions have interferometric phase centers on the ground, so we get close to bare earth, plus a magnitude image to characterize the coverage. Because X band and P band are collected simultaneously and are co-registered, the difference between them tends to suggest tree height (and biomass) in forested areas.
GeoSAR is typically used for topographic mapping at scales of 1:25,000 or 1:50,000 over entire countries. “From GeoSAR data, we can develop the framework data layers needed to develop a national spatial data infrastructure (NSDI) in a fraction of the time that it takes using conventional satellite or airborne methods,” Dean argues. Fugro EarthData caters mostly to government agencies — including the NGA, NOAA, and the Australian government — and to oil & gas companies. “When the area to be mapped is much smaller than that of a single U.S. state, it becomes less cost-effective to use our system.”
The company operates primarily in equatorial regions (Southeast Asia, South America, Africa, etc.) where clouds and tropical forests are prevalent, but its system, it claims, is also highly effective in high latitudes where there is little sunlight, rugged terrain, and dense boreal forests.
The large volume of data produced by IfSAR systems is challenging to process and store. To process GeoSAR data quickly and efficiently, Fugro EarthData has developed a parallel-processing capability, as well as a data storage capacity of more than 1,500 terabytes of available space.
The X band resolution depends on the power (bandwith) of the radar, not on its altitude. However, longer band-widths suffer from atmospheric attenuation, so no one has been able to build a satellite-based P band SAR system.
“Standard remote sensing software took quite some time to get to the point where it can fully handle high-resolution SAR data, in particular the novel 32-bit complex data products,” says Kern. “This was longer than we expected. It was a ‘chicken and egg’ problem: software companies were understandably reluctant to invest in developing radar software until they were sure that there was sufficient demand to justify their efforts. We started working with several software providers very early, even before launching our satellite. We gave them our specifications and, as soon as the satellite was in space, we provided them with free sample data. Nevertheless, even by the second year of the satellite’s operation (which was last year) some software tools still had bugs, which slowed down our early market entrance. To speed up this process, we now work very closely with some of the vendors to develop radar data exploitation tools.”
In has become apparent in recent years that radar applications are increasingly useful and effective, to the point where “Neither snow nor rain nor heat nor gloom of night…” will stay these monitors from “their appointed rounds.” Like the postal couriers of old, they dependably bring needed information at regular, sometimes critical, intervals, impeded by very few circumstances.