Scientific Investigations Report 2007–5251
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2007–5251
Version 2.0, June 2013
Using Interferometric Synthetic Aperture Radar (InSAR) is an effective way to measure vertical changes of land surface. InSAR is a satellite-based remote sensing technique that can detect centimeter-level ground-surface deformation over a 100 km2 area with a spatial resolution of 90 m or less. Synthetic Aperture Radar (SAR) imagery is produced by reflecting radar signals off a target area and measuring the two-way travel time between the target area and the satellite. InSAR uses two SAR scenes of the same area taken at different times and “interferes” (differences) them, resulting in maps called interferograms that show line-of-sight ground-surface displacement (range change) between the two time periods. The generation of an interferogram produces two components, amplitude and phase. The amplitude is the measure of the radar signal intensity returned to the satellite and shows roads, mountains, and other reflective features (similar to the image shown in fig. 5); the phase component is proportional to range change and shows the coherent displacements imaged by the radar (figs. 6A, 7A, 8A). If the ground has moved away from (subsidence) or towards (uplift) the satellite between the times of the two acquisitions (the “timeline”), a slightly different portion of the wavelength is reflected back to the satellite resulting in a measurable phase shift that is proportional to range change. The map of phase shifts, or interferogram, is depicted with a repeating color scale that shows relative range change between the first and the second acquisitions. In this report, one complete color cycle (fringe) represents 28.3 mm (0.09 ft) of range change. Assuming all the motion is vertical, the indicated range change is about 90- to 95-percent of true vertical ground motion, depending on the satellite look angle and the location of the target area. The direction of change—subsidence or uplift—is indicated by the color progression of the fringe(s) from the outer edge of a deforming feature toward its center. For interferograms in this report, the color-fringe progression of yellow-green-blue-pink indicates subsidence; the opposite progression indicates uplift.
InSAR signal quality depends partly on satellite position, atmospheric effects, ground cover, land-use practices, and timeline of the interferogram. Strict orbital control is required to precisely control the look angle and position of the satellite. Successful application of the InSAR technique is contingent on looking at the same point on the ground from the same position in space so that the horizontal distance between each satellite pass, or perpendicular baseline, is minimized. Perpendicular baselines generally greater than about 200 m (656 ft) usually produce excessive topographic effects (parallax) that can mask the deformation signal. Phase shifts can be caused by varying atmospheric mass that is associated with different elevations. A digital elevation model (DEM) is used in the interferogram generation process to reduce the atmospheric effects caused by elevation differences (and also to georeference the image). Phase shifts also can be caused by laterally variable atmospheric conditions such as clouds or fog, as the non-uniform distribution of water vapor differentially slows the radar signal over an image. Atmospheric artifacts can be identified by using several independent interferograms, which are defined as interferograms that do not share a common SAR image. When apparent ground motion is detected in only one interferogram, or a set of interferograms sharing a common SAR image, then the apparent motion likely is due to atmospheric phase delay or other error source and can be discounted.
The type and density of ground cover also can significantly affect interferogram quality. Densely forested areas are prone to reflect poor signals because the C-band wavelength (56.6 mm or 0.18 ft) cannot effectively penetrate thick vegetation and is either absorbed or reflected back to the satellite from varying depths within the canopy resulting in incoherent signal (shown as randomized colors on an interferogram). Sparsely vegetated areas and urban centers, however, generally reflect coherent signals because bare ground, roads, and buildings have high reflectivities and are relatively uniform during at least some range of InSAR timelines. Certain land-use practices, such as farming, which is prevalent in the Coachella Valley, also cause incoherent signal return. The tilling and plowing of farm fields causes large and nonuniform ground-surface change that cannot be resolved with InSAR. In both urban and non-urban areas, signal quality also is adversely affected by longer timelines, as there is more opportunity for nonuniform change and temporal decorrelation of the radar signal. Because Coachella Valley is fairly flat and contains several urban centers, the area lends itself to confident InSAR interpretations because most of the error sources are minimized. The agricultural fields in the area, however, cause significant InSAR signal incoherence that cannot be reliably interpreted using the coherent InSAR techniques described here.
Data from the European Space Agency’s (ESA) ENVISAT satellite were obtained for analysis. The multi-mission ENVISAT platform was launched in 2002 and is currently (2007) the only ESA-owned fully functional SAR satellite. (The previous studies [Sneed and others 2001, 2002] used ESA’s ERS-1 and -2 satellites; ERS-1 was turned off in 1999 and ERS-2 is no longer routinely suitable for interferometry). The satellite is side-looking and orbits the Earth at an altitude of approximately 800 km (500 mi), and has a 35-day repeat orbit. Twenty SAR images, acquired between February 26, 2003 and September 25, 2005, were used to produce 22 interferograms having timelines that range from 35 to 595 days. The three interferograms shown in this report were selected on the basis of overall interferogram quality and timeline length. Figures 6, 7, and 8 show interferograms representing timelines ranging from 35 days to 595 days; the variable timelines demonstrate temporal decorrelation of an interferogram (coherency) and on subsidence feature characteristics, such as geographic extent.
The interferograms of the Coachella Valley show significant land-surface-elevation changes in at least four areas. Land subsidence occurred in three of the areas—Palm Desert (Area 1), Indian Wells (Area 2), and La Quinta (Area 3) (figs. 6, 7, 8; table 2)—and both subsidence and uplift apparently occurred in one of the areas—Indio-Coachella—between February 26, 2003, and September 25, 2005. The interferograms show that other local areas in the Coachella Valley also may have deformed (figs. 6, 7, 8), but the extent of these areas and the amount of deformation generally are small compared with the three areas of land subsidence discussed in this section.
A subsidence signal was detected in Palm Desert (Area 1 in figs. 5, 6, 7, 8) this signal was previously detected by Sneed and others (2001, 2002) using InSAR methods. All 22 interferograms analyzed for this study show subsidence in this area (for example, figs 6, 7, 8; table 2). The part of the signal that has the largest magnitude is nearly circular (slightly elongated north to south); in longer-term interferograms, it is as large as about 2 km (1.2 mi) in diameter and has an area of about 4 km2 (1.5 mi2) (fig. 8). This part of the signal is approximately bounded by Clancy Lane on the north, Fred Waring Drive on the south, Highway 111 and Bob Hope on the west, and Monterey Avenue on the east (fig. 5). The part of the signal that is smaller in magnitude extends to the north and east, has a pronounced northwest–southeast elongation, and has a much larger extent. The extent of this part of the signal is larger in interferograms representing longer timelines (fig. 8) than interferograms representing shorter timelines (fig. 7), but generally extends to the northwest where Frank Sinatra Drive intersects the Whitewater River channel, to the north beyond Country Club Drive, and to the east as far as Cook Street (fig. 5). The San Jacinto and Santa Rosa Mountains, which are outcropping consolidated rock, may act as barriers to subsidence farther to the south and southwest; and the Indio Hills, which are outcropping partly consolidated deposits, may act as barriers to subsidence farther to the northeast (figs. 2, 5). A lack of barriers to the northwest and southeast may explain the pronounced elongation of the subsidence signal in these directions. Several deformation timeseries can be constructed from InSAR data. The longest timeseries timeline (840 days)—May 7, 2003 to September 25, 2005—was constructed using 3 interferograms (5/7/2003-9/8/2004; 9/5/2004-4/3/2005; 5/8/2005-9/25/2005) and indicates that about 180 mm (0.59 ft) of subsidence occurred in the Palm Desert area during this time—a rate of more than 6 mm/month (0.02 ft/month) (Area 1 in figs. 5,6, 7, 8,9A). Sneed (2001; 2002) reported that about 150 mm (0.49 ft) of subsidence occurred in this area during 1,541 days between 1996 and 2000, which is equivalent to a rate of about 3 mm/month (0.1 ft/month). These data indicate that subsidence rates have doubled since 2000.
In the Indian Wells area, two distinct subsidence signals were detected, termed the western bowl and the eastern bowl (Area 2 in figs. 5, 6, 7, 8). Sneed and others (2002) discussed a possible additional subsidence signal as suspect about 1 km (0.6 mi) southeast of the eastern subsidence bowl because of the proximity of this area to steep topographic terrain. The shorter timelines and multiple interferograms analyzed during this current study show that it is an extension of the eastern subsidence bowl (for example, fig. 6). Twenty of the 22 interferograms analyzed for this study show the western bowl and 19 show the eastern bowl (for example, figs. 6, 8; table 2). The two interferograms with the shortest timelines (35 days) do not show either the western or eastern subsidence bowl (for example, fig. 7) and one interferogram of relatively low quality (not shown) with a 105-day timeline does not show the eastern bowl (table 2). The western subsidence bowl is approximately bounded by Fred Waring Drive on the north, the Santa Rosa Mountains on the south, Rancho Palmeras Drive on the west, and Indian Wells Lane on the east (fig. 5). The eastern subsidence bowl is approximately bounded by Highway 111 on the north, the Santa Rosa Mountains on the south, Club Drive on the west, and Mountain Cove Drive on the east (fig. 5). Distinct northwest-southeast linearities shown near the northern margins of the pair of subsiding bowls suggest the presence of barriers to ground-water flow or abrupt changes in lithology that control the northern extent of subsidence (fig. 6). The western bowl, which is elongated northwest–southeast, is as large as about 2.3 km (1.4 mi) long and about 1.8 km (1.1 mi) wide (northeast–southwest) and has an area of about 4.1 km2 (1.6 mi2) in longer-term interferograms. The eastern bowl, which also is elongated northwest–southeast, is as large as about 2.3 km (1.4 mi) long and ranges from about 0.3 to 0.8 km (0.2 to 0.5 mi) wide (northeast–southwest) and has an area of 1.2 km2 (0.5 mi2) (Area 2 in figures 6 and 8). The maximum subsidence for the western bowl was about 105 mm (0.34 ft) and for the eastern bowl was about 75 mm (0.25 ft) from May 7, 2003 to September 25, 2005 (fig. 9A); the equivalent rates are nearly 4 mm/month (0.013 ft/month) and nearly 3 mm/month (0.01 ft/month) for the western and eastern bowls, respectively. Sneed (2001; 2002) reported that about 80 mm (0.26 ft) of subsidence occurred in each bowl between 1996 and 2000; the equivalent rate is about 1.6 mm/month (0.005 ft/month). These data indicate that subsidence rates have doubled since 2000.
A third area of subsidence was detected in the La Quinta area (Area 3 in figs. 5, 6, 7, 8). All 22 interferograms analyzed for this study show subsidence in this area (for example, figs. 6, 7, 8; table 2). Sneed and others (2002) analyzed 2 interferograms with timelines of 315 and 350 days and suggested Area 3 consisted of two separate areas of subsidence—termed the La Quinta and the Lake Cahuilla areas of subsidence—because the interferograms were not coherent throughout the area. Similarly for the interferograms analyzed for this current study, the longer-term interferograms were incoherent near the central part of the subsiding area (for example, figs. 6 and 8); however, the shorter-term interferograms show that this subsidence area likely is continuous, and is termed the La Quinta subsidence area for this report (for example, fig. 7A). The La Quinta subsidence area is about 10 km (6 mi) in length (northwest-southeast) and is as much as 3 km (1.9 mi) in width. The area is approximately bounded by 48th Street (if extended westward) on the north, 60th Street (if extended westward) on the south, the Santa Rosa Mountains on the west, and streets varying from Jefferson Street to Monroe Street on the east (because the subsidence area trends northwest-southeast) (fig. 5). Different portions of this elongate subsidence bowl are subsiding at different rates; for May 7, 2003 to September 25, 2005, subsidence was 160 mm (0.52 ft) in the northern part, 175 mm (0.57 ft) in the central part, and 120 mm (0.39 ft) in the southern part (fig. 9A); equivalent subsidence rates are about 6 mm/month (0.02 ft/month) for the northern and central parts and more than 4 mm/month (0.013 ft/month) for the southern part. Sneed (2002) reported that about 80 mm (0.26 ft) of subsidence occurred in the northern part between 1998 and 2000 and is equivalent to a subsidence rate of more than 3 mm/month (0.01 ft/month). Sneed (2001; 2002) reported that about 35 mm (0.11 ft) of subsidence occurred in the southern part between 1996 and 2000 and is equivalent to a subsidence rate of nearly 1 mm/month (0.003 ft/month). These data indicate that subsidence rates have doubled in the northern part and increased by about a factor of 4 in the southern part since 2000.
Indio-Coachella has not been previously discussed as an area of deformation. The shorter timelines of InSAR data presented in this report as opposed to those presented in Sneed and others, 2001 and Sneed and others 2002, show land-surface-elevation changes in the Indio-Coachella area which appear to mostly be uplift or stable; however, the apparent deformation is suspect. As was mentioned in the InSAR Methodology section, interferograms show relative deformation, not absolute deformation. Because much of the basin is subsiding as is shown by the GPS results, it is likely that the Indio-Coachella area only appears to be uplifting as proximal areas in the basin are subsiding around it. The Indio-Coachella area could actually be subsiding at slower rates than proximal areas of the basin, which would appear as relative uplift in an interferogram. Because the deformation interpretations are suspect, this area is not discussed further in this report.
All the areas where significant subsidence was detected using InSAR—Palm Desert, Indian Wells, and La Quinta—coincide with or are near areas where ground-water pumping generally caused ground-water levels to decline. Available water-level data indicate that water levels measured in 2005 in many wells in the Palm Desert, Indian Wells, and La Quinta areas were at the lowest levels in their recorded histories (fig. 9B). The coincident areas of the subsidence signals and the declining water levels, and the localized character of the subsidence signals typical of subsidence caused by localized pumping, strongly suggest a relation between subsidence and ground-water-level declines. The stresses imposed by the declining water levels and the significant subsidence at the three areas revealed by InSAR indicate that the water levels may have exceeded the preconsolidation stress, and if so, the subsidence may be permanent.
While ground-water-level data are too sparse to evaluate the short-term effect on vertical land-surface changes shown by the InSAR results, the available ground-water-level and InSAR data can be used to draw some inferences about the affected aquifer system. In Palm Desert (Area 1 in figs. 6, 7, 8), the area of maximum subsidence is concentric. Although water-level data are measured infrequently and are not available within the area of the concentric InSAR signal, data for the surrounding region indicate that water-level declines have been fairly uniform. The concentric subsidence shape suggests that the stress primarily causing the subsidence is near the center of the concentric shape and that sediments constituting the aquifer system in the area are fairly homogenous as the sediments responded uniformly to similar water-level declines in areas surrounding the center of the subsidence bowl.
While ground-water-level data are too sparse to evaluate the short-term effect on vertical land-surface changes shown by the InSAR results, the available ground-water-level and InSAR data can be used to draw some inferences about the affected aquifer system. In Palm Desert (Area 1 in figs. 6, 7, 8), the area of maximum subsidence is concentric. Although water-level data are measured infrequently and are not available within the area of the concentric InSAR signal, data for the surrounding region indicate that water-level declines have been fairly uniform. The concentric subsidence shape suggests that the stress primarily causing the subsidence is near the center of the concentric shape and that sediments constituting the aquifer system in the area are fairly homogenous as the sediments responded uniformly to similar water-level declines in areas surrounding the center of the subsidence bowl.
In Indian Wells, the areas of subsidence form two separate elongated shapes—one parallel to and one perpendicular to the mountain front. The available water-level data indicate fairly uniform water-level declines in the area, suggesting that the aquifer system is heterogeneous in this part of the valley because sediments responded dissimilarly to similar water-level declines, as exhibited by the irregular shapes of the subsidence areas. Additional evidence of aquifer heterogeneity in the area is given by the distinct northwest-southeast linearity shown on the northern margins of the pair of subsiding bowls, suggesting the presence of barriers to ground-water flow or abrupt changes in lithology that control the northern extent of subsidence (Area 2 in fig. 6).
In La Quinta, water-level data are too sparse to relate to the aquifer-system response, but the northwestern part of the subsiding area has an irregular shape that appears to be related to the contact between the alluvial aquifer and the bedrock basement complex that defines the extent of the aquifer system (Area 3 in figs. 6, 7, 8).
U.S. Department of the Interior | U.S. Geological Survey
URL: https://pubs.usgs.gov/sir/2007/5251
Page Contact Information: Publications Team
Page Last Modified: Thursday, 01-Dec-2016 19:55:49 EST
