Basic SAR Analysis. New York City. CEE 6100/ CSS 6600 Remote Sensing Fundamentals Lab #8: Radar

Transcription

1 1 Basic SAR Analysis Images for this tutorial were taken from the SIR C/X archive at This web site has a good collection of examples of multi frequency, multi polarization images for a variety of applications. The specific images discussed in this exercise may be downloaded from the class web site: New York City: NYC_ _p45621.tif Pishan: Pishan_China_ _p46177.tif Stockholm_ _p49516.tif New York City This is radar image of the New York city metropolitan area. The island of Manhattan appears in the center of the image. The green-colored rectangle on Manhattan is Central Park. This image was acquired by the Spaceborne Imaging Radar-C/X-band Synthetic Aperture Radar (SIR-C/ X-SAR) aboard the space shuttle Endeavour on October 10, North is toward the upper right. The area shown is 75.0 kilometers by 48.8 kilometers (46.5 miles by 30.2 miles). The image is centered at 40.7 degrees north latitude and 73.8 degrees west longitude. Red: L-band HH; Green: L-band HV; Blue: C-band HV In general: light blue areas correspond to dense urban development, green areas to moderately vegetated zones and black areas to bodies of water. The radar illumination is from the left of the image. The Hudson River is the black strip that runs from the left edge to the upper right corner of the image. It separates New Jersey, in the upper left of the image, from New York. The Atlantic Ocean is at the bottom of the image where two barrier islands along the southern shore of Long Island are also visible. John F. Kennedy International Airport is visible above these islands. Long Island Sound, separating Long Island from Connecticut, is the dark area right of the center of the image. Many bridges are visible in the image, including the Verrazano Narrows, George Washington and Brooklyn bridges. Examine the image in detail: 1. Download the file NYC_ _p45621.tif from the course web site. 2. Load the image into ENVI and display as a color image (the default). Figure 1: NY City radar image

2 2 3. Display the individual bands as separate images (Select the gray scale option and remember to select a new display for each image) 4. Link the set of images. In any of the image windows, select Tools > Link > Link Displays. When you change the size of a window or move the view in any one image window the other windows will track the move. 5. You may find it helpful to locate the area in Google Earth or Google Maps for reference and identification of individual features. 6. Corner reflection: Locate the area with the bright return in the L-band HH image in the lower left corner of the image. The bright sections correspond to areas where the streets are oriented perpendicular to the radar look direction, thus providing efficient corner reflectors The area is only bright in the like-polarization image, In the cross-polarization images, the return is relatively bright and uniform regardless of the orientation of the streets. (Why?) 7. Texture: JFK airport is located along the causeway below the bright areas. You can identify the airport in the cross polarization images by the dark runways bounded by the slightly brighter grass areas. This is most obvious in the C-band HV image. The runways are relatively smooth at the radar wavelengths and reflect most of the radiation away from the radar. Grass is slightly rougher and returns some radiation even though the vegetation is generally darker (water content). Figure 2: Area near JFK airport 8. Water is also smooth (in this image) at the radar wavelengths and is uniformly dark, generally contrasting well with the land. Note, however that the land-water difference is dependent on

3 3 look-angle and the relative roughness of the land surface. There are a number of small, marshy islands in the bay to the right of the airport in the images above. The islands are only apparent in the C-band image, in part because the marsh grasses are rough at the C-band wavelength and relatively smooth in L-band. Also, the grasses are probably partially submerged (water reflection). The land water distinction can also be confounded by smooth shores (e.g., mud flats). 9. Vegetation: Parkland is generally darker than the surrounding city areas. The vegetated areas do not contain coherent corner reflectors and the water content tends to subdue the return. Return from vegetation also tends to favor multiple scattering which favors depolarization. There is an interesting contrast between the parkland (toward the top of the image above) and the marshland around the bay. The marshland tends to be brighter than the parkland in the C-band HV image while both are about the same tone in the L-band HV. The marsh practically disappears in the L- band HH image. The parkland vegetation is predominantly trees and shrubs more responsive to longer wavelengths than the low and relatively uniform marsh grasses. Figure 3: New York Harbor 10. Shift to New York Harbor on the left side of the image. The bright spots in the harbor are ships (tankers, container ships, etc.) anchored in the harbor. Together with the water they form small corner reflectors and provide strong return in all three images. Even boats too small to be resolved in the image will act as point corner reflectors. Transmission line: In the upper right of L-band HH (red) L-band HV (green) C-band HV (blue)

4 4 11. Figure 3 is a series of dots arcing upward toward the top of the image. These are the metal towers of a transmission line passing through a marshy pond area. The towers provide a strong radar return in all the images. 12. Arched road: Just below the transmission line is a curving road passing through the flat, marshy area (highlighted by the zoom box). The portions of the road at an angle to the radar look direction provide a strong radar return. As the road becomes increasingly parallel to the radar look direction and the signal diminishes noticeably. When at an angle to the radar the sides of the elevated structure provide a surface to reflect radiation. When the road is parallel to the look direction the sides are not providing a scattering (reflective) surface and most of the radar radiation is reflected away from the radar by the relatively smooth road surface. Pishan, China This radar image is centered near the small town of Pishan in northwest China, about 280 km (174 miles) southeast of the city of Kashgar along the ancient Silk Route in the Taklamakan desert of the Xinjiang Province. Geologists are using this radar image as a map to study past climate changes and tectonics of the area. The irregular lavender branching patterns in the center of the image are the remains of ancient alluvial fans, gravel deposits that have accumulated at the base of the mountains during times of wetter climate. The subtle striped pattern cutting across the ancient fans are caused by thrusting of the Kun Lun Mountains north. This motion is caused by the continuing plate-tectonic collision of India with Asia. Modern fans show up as large lavender triangles above the ancient fan deposits. Yellow areas on the modern fans are vegetated oases. The gridded pattern results from the alignment of poplar trees that have been planted as wind breaks. The reservoir at the top of the image is part of a sophisticated irrigation system that supplies water to the oases. This image was acquired by the Spaceborne Imaging Radar-C/X-band Synthetic Aperture Radar (SIR-C/X-SAR) aboard the space shuttle Endeavour in April This image is centered at 37.4 degrees north latitude, 78.3 degrees east longitude and shows an area approximately 50 km by 100 km (31 miles by 62 miles). Figure 4: Pishan, China. From top to bottom: Color; L-band HH (red); L-band HV (green); C-band HV (blue).

5 5 Examine the image in detail: 1. Download the image file Pishan_China_ _p46177.tif from the course web site. 2. Load the image into ENVI, display and link the images as with the NY City image. 3. Locate the area in Google Earth by typing in the coordinates (as above). 4. Roughness scale: The area highlighted by the zoom box in Figure 4 demonstrates the dependence the radar return on the roughness scale relative to the wavelength. In C-Band, there is little differentiation between the poplar groves and the alluvial Figure 5: Google Earth image of the Pishan scene. fan both are roughly equally rough. As seen in Figure 6, there are clear differences in the structural detail, but the overall tone of the two areas is the same in C-band. Figure 6: Poplar groves and alluvial fan. 5. Polarization: There is clearly contrast between the L-band images as well. The like polarization images (Figure 6) shows significant structural detail in the grids of the poplar grove. The pattern is difficult to see in the cross-polarization L-band image, for which the return is dominated by depolarization due to multiple scattering, and effect which appears to blur the geometric details.

6 6 Illumination direction and layover: Move to the section of the image outlined in the false color image comprised of the three radar bands (Figure 7). A color image of Figure 7: Color image to locate the mountain scene. the highlighted area from Goo Google Earth (enhanced) is shown in Figure 8. The natural illumination from sunlight is from the south (bottom of the image in Figure 8. Thus, the outlined area is a valley with the brighter northern hillside tilted toward the south. Contrast this with the radar images in Figure 9 where the northern hillside is dark and elongated (tilted away from the radar look direction) and the southern hillside is bright and foreshortened (tilted toward the radar look direction). The south-tilted slope shows evidence of pseudo-shadowing and, in a few areas, radar shadow. Figure 8: Color image from Google Earth of the mountain scene. Figure 9: Pishan mountains

7 7 6. Range, wavelength and polarization: The L-band HH image in Figure 9 is the sharpest image of the set. This may be in part due to the wavelength, but is probably more a factor of the likepolarization. The greater part of the energy is returned after one scattering event leading to a stronger signal in the like-polarization image. Both of the cross-polarization images show fading toward the bottom. This is the portion of the image that is farthest from the radar and the signal is already relatively low. That coupled with the lossy nature of multiple scattering leads to an overall loss of contrast. 7. Overlay: I could not find an unambiguous example of overlay in this image. Perhaps you can find one. Stockholm, Sweden The flat, glaciated region that surrounds the city of Stockholm, Sweden is shown on this radar image. The southern portion of this capital city is the bright area in the upper right of the image. Dark areas on the image are the fjords, lakes and rivers that crisscross the region, forming thousands of islands that comprise the Stockholm archipelago. Green areas on the image are vegetation, blue areas have been partially cleared and white areas are urban and other settlements. The area shown is 27 by 40 kilometers (16 by 25 miles) centered at 59.2 degrees north latitude, 17.9 degrees east longitude. The image was acquired by the Spaceborne Imaging Radar-C/X-band Synthetic Aperture (SIR-C/X-SAR) imaging radar when it flew aboard the space shuttle Endeavour on October 2, Examine the image in detail: 1. Download the image file Stockholm_ _p49516.tif from the course web site. Figure 10. Stockholm Sweden 2. Load the image into ENVI, display and link the images as with the previous examples. 3. Locate the area in Google Earth by typing in the coordinates 59.2N, 17.9E. You will find a somewhat better correspondence of land use patterns if you display the 9/5/2005 image from the historical image set (click on the clock in the icon string at the top of the Google Earth window and use the slider to examine the different images). 4. Display the area outlined in the red box in Figure Corner reflection: The look direction in this case is from the bottom of the image. Compare the northern shores of the fjords with the southern shores. The northern (east-west trending) shores are generally brighter than the southern shores because the combination of the water, steep shoreline and the orientation form a corner reflector for the radar. The same phenomenon often occurs at the border of the cleared fields and forest.

8 8 6. Transmission line: Examine the area highlighted by the zoom box in Figure 11. The linear feature is very bright in the L-band HH image and the line extends east and west from that point, though not as consistently. The feature is not apparent in either of the cross-polarization images. This is a transmission line: a set of wires on towers. It is likely that the spacing and orientation of the wires at the outlined location are nearly ideal to serve as a reflector or antenna for the L-band radiation when looking in the like-polarization mode. With cross-polarization, the amplification effect is lost. The right of way for the transmission line is apparent in the Google Earth image and if you look carefully you will be able to locate at some of the towers. The transmission lines themselves are not discernible in the Google Earth image. Figure 11: West of Stockholm 7. Volume scattering: An agricultural field (white arrow in Figure 11) is quite dark in the L-band HH image, and is surrounded by forest which provides a bright return. The field probably has a relatively uniform canopy capable of reflecting the majority of radiation away from the radar in a single reflection whereas the forest canopy provides a much brighter return. In the crosspolarization mode, the field and forest are nearly indistinguishable. It is curious that this occurs in both L-band and C-band, and only at this location. It may be an unusual crop.