Issue link: https://read.dmtmag.com/i/55888
The satellite data used in this pilot project were sup- plied by Infoterra from the German TerraSAR-X sensor. As the name suggests, this synthetic-aperture radar uses X-band wavelength radiation. This satellite can provide imagery at various resolutions (pixel sizes), depending on the intended application and desired area of coverage. For this project, 20 strip-map mode scenes with approximately two-meter pixels were analyzed. The scenes covered an area with dimensions of 32 by 56 kilometers (20 by 35 miles). One of the radar images is shown in Figure 2. On-board tracking and post-processing provide satellite positional information with an accuracy of less than 0.5 meters in space, which supports precise surface deformation measurements and geolocation. The software required for these precise calculations is complex, sophisticated and previously existed largely in the domain of universities and specialized research facilities. However, the software used for this project, the Radar Interferometry module in the IMAGINE Radar Mapping Suite, was developed specifically to bring this technology to organizations that need the results—dis- placement maps—but don't have or want specialized radar-processing expertise. In many instances, it's fully automatic; just input the two scenes and wait for the derived displacement image to be calculated. Figure 3 shows one of the user interfaces of the wizard workflow software. The images shown in Figures 4 and 5 were calculated by the software using a fully automated process, achiev- ing these results with no need for a specially trained radar-processing analyst. In addition to the DInSAR image pair, an accurate digital elevation model (DEM) is required for the displacement calculation. Conveniently, this can be extracted using the related InSAR DEM extraction process to "unwrap" the interferogram to reveal the elevation of each pixel. For this study, several of the TerraSAR-X images were used to create overlapping DEMs that then were averaged. Image Analysis The 20 TerraSAR-X scenes can be combined into inter- ferometric image pairs spanning various time periods. Each image pair can be processed to yield a surface displacement through that time period. This allows the monitoring facility to collect a progressive sequence of images illustrating surface movement over time. This time series of images enables measurement of the effect of increased extraction or the response to a remedial action. The displacement calculation process generates several intermediate images that analysts can view if necessary. These could be beneficial for evaluating whether a particular step was yielding the best pos- sible results or determining if processing parameters need minor adjustments. One of the most useful intermediate images is the interscene coherence image, a numerical (0-1) indication of measurement stability for each pixel that's calculated from the two images. High-coherence (bright) areas are more reliable than darker areas. As the coherence drops, the accuracy of the resultant dis- placement is lowered, and the signal becomes noisy. A sample coherence image is shown in Figure 4, along with a corresponding displacement image. In the coherence image (Figure 4, left), the dark pix- els indicate a significant loss of coherence along the river flowing east to west through the scene's center. There also are areas of very low coherence associated with vegetated areas along the river. These are very dark and rectangular. The center scene in Figure 4 is the corresponding interferogram (in color) overlain on the radar image. These phase fringes quantify the surface displacement between the two images of the InSAR pair. Unwrapping this series of fringes creates the final subsidence image, as seen in Figure 4 (right). Of particular interest are the two dark sinuous features in the center; these are subsidence areas. Each pixel in this image has a value indicating subsidence in millimeters. Figure 5 illustrates a closer analysis, enabling ana- Figure 4. An interscene coherence image (left), interferogram (center) and corresponding displacement image (right) can be created from satellite data. Red arrows indicate subsidence features of high confidence. These product images are derived from a TerraSAR-X InSAR pair with a six-month separation. 24 GEO W ORLD / FEB R UA R Y 2O12 lysts to interpret the displacement map in the context of interscene coherence. For example, the river shows a near total loss of coherence, as do several agricul- ture plots. But note that some agricultural fields show very high coherence. These may be bare earth awaiting sowing and correspond to mottled areas in the subsid- ence image, indicating erratic subsidence results. Resource Monitoring