SAR

Synthetic Aperture Radar (SAR) is a powerful remote sensing system, enabling observations of the Earth’s surface day or night, in all weather conditions from airborne platforms and from space. Unlike optical remote sensing systems, which rely on the Sun for illumination, SAR provides its own, illumination via microwave (Radar frequency) transmissions from the satellite. These transmissions are coherent, meaning that both amplitude and phase information are retained. The Sun’s illumination reaching the Earth is incoherent (scrambled), meaning that only amplitude information can be used. This leads to another important difference between optical and SAR systems. In many cases, the coherent phase information transmitted by a SAR satellite or aircraft is reflected from the surface back to the sensor with the phase more or less intact. A subsequent satellite pass, several days to several years after the initial pass, may also retain the phase information. A phase comparison of the two images via interferometric techniques may reveal subtle shifts in the position of the Earth’s surface (closer to or farther from the satellite). This technique is known as Interferometrtic SAR (InSAR). It can be done from either aircraft or spacecraft, but for most scientific applications, space-based platforms are preffered

The potential precision of InSAR depends on many things (e.g., the accuracy of the orbit determination for the two satellites passes, atmospheric conditions that affect the phase delay) but in principle the surface displacement measurement can have a precision of 2%-5% of the SAR wavelength. Typical SAR wavelengths are in the range 3-30 cm, implying millimeter to centimeter precision in surface displacement between two satellite passes. Refinements of this basic technique include “stacking” of multiple SAR passes to reduce noise [e.g., Gourmelen and Amelung, 2005]. InSAR has been used to study co-seismic offsets due to earthquakes, volcano deformation, and subsidence due to withdrawal of ground water, oil, or natural gas [e.g., Amelung et al., 1999].

Permanent Scatterer InSAR (PSInSAR)

While InSAR is a powerful technique for measuring changes in the Earth’s surface, it does have limitations. These include temporal and geometrical decorrelation (low signal to noise ratio in the phase change estimate), and variable tropospheric water vapor, which can generate variable phase delay due to the impact of water vapor on the propagation speed of microwave signals. The corresponding phase changes can be misinterpreted as surface change. The effects can be quite large, especially in tropical and sub-tropical regions. In tropical regions, up to 10 cm of variable path delay over several weeks has been observed [Dixon and Kornreich-Wolfe, 1990; Dixon et al., 1991]. An example of a tropospheric water vapor “signal” in a SAR interferogram in sub-tropical New Orleans is shown in Figure 4. Although the water vapor signal is only about 6 mm, the passes are separated by just 3 weeks, implying an annual average subsidence rate in excess of 100 mm/yr. The average subsidence rate here is known to be about 6 mm/yr, so clearly the tropospheric error can be significant.

Permanent Scatterer InSAR (PSInSAR) [Ferretti et al., 2000, 2001;2004; Colesanti et al., 2003a,b,c] exploits several characteristics of radar scattering and atmospheric decorrelation to measure surface displacement in otherwise non-optimum conditions. Atmospheric phase contributions are spatially correlated within a single SAR scene, but tend to be uncorrelated on time scales of days to weeks. Conversely, surface motion is usually strongly correlated in time. An example is surface subsidence, which is usually steady over periods of months and sometimes years. Thus, atmospheric effects can be estimated and removed by combining data from long time series of SAR images, averaging out the temporal fluctuations. Radar scatterers that are only slightly affected by temporal and geometrical decorrelation are used, allowing exploitation of all available images regardless of imaging geometry. In this sense the scatterers are “permanent”, i.e., persistent over many satellite revolutions (the technique is also known as Persistent Scatterer InSAR).

What constitutes a permanent or persistent scatterer? Man made structures in urban settings are most common. Inspection of the individual scatterers in a PSInSAR image of New Orleans indicates that many of the scatterers are located at the intersection of a street or sidewalk and vertical structure such as the side of a building, or a roof (Figure 3). In contrast, parks and other vegetated areas have no permanent scatterers, presumably because vegetation is not a strong radar reflector, and also undergoes significant wind-driven motion as well as growth over the imaging period (in this case, three years). This implies that the PSInSAR technique is ideally suited to studying surface changes in urban areas, e.g., due to subsidence or co-seismic offsets.

We used the PSInSAR technique to study subsidence in New Orleans [Dixon et al., 2006]. We used 33 RADARSAT (6 cm wavelength) scenes acquired between April 2002 and July 2005 in the ascending orbit, standard beam mode (S-2). A total of more than 1.8x105 radar targets were identified that retained some phase coherence over the three year study period, providing excellent spatial resolution for our space-derived surface velocity map (Figure 6). Both InSAR and PSInSAR are by definition relative (ambiguous), hence determination of the actual subsidence rate requires calibration with one or more ground control points of known elevation and motion. This is usually accomplished by referencing to stable areas ~50 km away from the locus of deformation, where motions are assumed to be minimal. This is problematic in the Gulf coast, where a large region is thought to be subsiding. We use a ten year time series from a high precision GPS station in the greater New Orleans area (ENG1) to provide an independent reference.

The mean and standard deviation range change rate for all the point targets is -5.6±2.5 mm/yr. The range change measurement is in the radar line of site direction, but can be considered nearly equivalent to vertical subsidence. The inset to Figure 6 shows an expanded view of one of the levees in the eastern part of New Orleans that failed during Hurricane Katrina (Figure 7). These levees have high subsidence rates (in excess of 20 mm/yr) and there may be a relation between high subsidence rates and levee failure, e.g, the underlying substrate may be weak.

	Figure 1 Sketch of the ERS-2 satellite, a primary source of SAR data for many scientific investigations. Note the large array of solar panels on the left, necessary to supply power to this active imaging system. The large panel on the right is the SAR antenna, which has the dual role of illuminating the Earth's surface with microwave energy, and receiving the resulting backscattered signal.
How SAR Works:		Figure 2 Cartoon illustrating some of the basic principles of SAR interferometry. The right hand inset show two satellites illuminating the same region of the Earth. In practice, a single satellite illuminates the area, followed several weeks to years later with a second image from the same satellite in the same nominal orbit. Displacement of the Earth's surface between the two successive satellite passes is estimated, following principles illustrated in the left hand side of the figure. The two basic requirements are that the orbits of both satellite passes are known precisely, and that the phase information inherent in the SAR signal is exploited (electro-optical satellites, using the visible and infra-red portion of the electomagnetic spectrum, utilize only amplitude information; SAR uses both amplitude and phase). Then phase information from the two satellite passes (each of which in effect is a type of distance measurement) is used to estimate the change in distance (range change) between the two passes, in the direction of the satellite line of site ("look angle"). This means that the measurement is inherently scalar. The use of both ascending and descending passes can be used to determine a two dimensional vector measurement.
	Figure 3 Galapagos Interferogram