This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/authorsrights
Author's personal copy
Ecological Engineering 59 (2013) 144–156
Contents lists available at ScienceDirect
Ecological Engineering journal homepage: www.elsevier.com/locate/ecoleng
Delta dynamics: Effects of a major earthquake, tides, and river flows on Ciénega de Santa Clara and the Colorado River Delta, Mexico Steven M. Nelson a,∗ , Eric J. Fielding b , Francisco Zamora-Arroyo c , Karl Flessa d a
6101 NE 102nd Avenue Apt 5, Vancouver, WA 98662, United States Jet Propulsion Laboratory, California Institute of Technology, M/S 300-233, 4800 Oak Grove Drive, Pasadena, CA 91109, United States c Sonoran Institute, 44 E. Broadway Blvd., Suite 350, Tucson, AZ 85701, United States d Department of Geosciences, University of Arizona, Tucson, AZ 85721, United States b
a r t i c l e
i n f o
Article history: Received 17 April 2012 Received in revised form 31 August 2012 Accepted 14 September 2012 Available online 28 November 2012 Keywords: Colorado River Delta Channel change Tidal inundation Earthquake surface deformation
a b s t r a c t The intertidal portion of Mexico’s Colorado River Delta is a dynamic environment subject to complex interactions of tectonic, fluvial, and tidal forces at the head of the Gulf of California. We review the historical interactions of these forces, use sequential satellite images, overflights, ground observations, and interferometric synthetic aperture radar (InSAR) data to study the effects of the 2010 Mw 7.2 El MayorCucapah Earthquake on changing patterns of tidal inundation within the Delta, and assess effects of these changes to the fluvial/hydrological regime of the Colorado River estuary and nearby Ciénega de Santa Clara wetland. The objectives of this study are to highlight for environmental scientists, land managers, and ecological engineers the contribution of tectonic forces in shaping the intertidal Delta environment and to provide information on the effects of the 2010 earthquake which will be of practical value in planning and designing management measures and restoration projects for the estuary and Ciénega. The Colorado River estuary is at present blocked by a tidal sand bar which restricts access by marine species to the upper estuary and obstructs the flow of fresh water into the lower estuary. Located 13 km east of the estuary, the Ciénega is a 6000 ha wetland supported by agricultural drain water from Arizona and Mexico. South of the Ciénega is the Santa Clara Slough, an unvegetated 26,000 ha basin subject to periodic inundation from the northern Gulf’s high amplitude tides, which have historically reached the margins of the Ciénega several times each year. The El Mayor-Cucapah earthquake ruptured the previously unknown Indiviso Fault which extends into the intertidal zone just west of the Ciénega. The Ciénega experienced only minor surface deformation having no direct effects to the wetland. Most of the significant ground movement and surface deformation occurred west of the Indiviso Fault adjacent to the estuary, where portions of the intertidal flats underwent extensive liquefaction, northward coseismic displacement and post-seismic subsidence. These surface deformations changed the pattern of tidal inundation, triggering development of a new system of natural tidal channels and creating conditions favorable for installation of projects to restore connectivity between the upper and lower estuary. The changed pattern of tidal inundation may also have contributed to an observed reduction in the occurrence of tidal flooding along the southwestern margin of the Ciénega following the earthquake. © 2012 Elsevier B.V. All rights reserved.
1. Introduction 1.1. The confluence of powerful tectonic, tidal, and fluvial forces For the past 5 to 6 million years, the Colorado River has flowed into the Gulf of California near the northern end of a great rift depression formed by a system of transform faults and seafloor
∗ Corresponding author. Tel.: +1 360 823 7183. E-mail address:
[email protected] (S.M. Nelson). 0925-8574/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ecoleng.2012.09.004
spreading centers at the boundary of the North American and Pacific tectonic plates (Alles, 2011). The area marks a transition zone between the right lateral movement of the San Andreas Fault system and the spreading movement of the East Pacific Rise, which is wedging the Pacific plate (including the Baja Peninsula and southwestern California) away from the North American plate (Burnett et al., 1997). The river has deposited 2.2–3.4 × 105 km3 of eroded Colorado Plateau sediments in a delta cone which has partially filled the rift depression, isolating its landlocked but mostly sub-sea level northern extension (the Salton Trough) from the southern portion occupied by the Gulf (Dorsey, 2010). The Gulf’s long, narrow form
Author's personal copy
S.M. Nelson et al. / Ecological Engineering 59 (2013) 144–156
145
Fig. 1. Main features of the Colorado River Delta in Mexico showing the Ciénega de Santa Clara, Santa Clara Slough, Cerro Prieto transverse fault (CPF), the newly discovered Indiviso Fault zone (IF) and other transverse faults (OF) that ruptured during the April 2010 Mw 7.2 El Mayor-Cucapah Earthquake. The epicenter was 45 km northwest of the study area.
contributes to high amplitude (up to 8.5 m) tides at the mouth of the Colorado River, resulting in an extensive intertidal plain where the low-gradient (about 0.016 m/km) delta cone approaches the sea (Thompson, 1968). The Ciénega de Santa Clara, a 6000 ha wetland supported by agricultural drain water from Arizona and Mexico, is located 50 km south of the Arizona-Sonora border and 20 km north of the Gulf (Fig. 1). The wetland is situated at the northern edge of the intertidal zone in low ground along the Cerro Prieto transform fault, which marks the eastern margin of the Delta plain and has been considered the principal plate boundary fault in this area (Hauksson et al., 2010). The river’s flow has been anthropogenically manipulated over the last century by implementation of water storage and diversion projects upstream in the Colorado River Basin and irrigation and flood control projects locally in the Mexicali and San Luis valleys. Prior to this manipulation, most of the river’s annual flow of 14.5–20.7 × 109 m3 reached the Delta (Fradkin, 1996), delivering about 160 million metric tons of sediment each year (Van Andel, 1964). In most years since the completion of upstream projects, only a fraction of the 1.9 × 109 m3 of Colorado River flow allocated annually to Mexico has reached the Delta (Fradkin, 1996), although additional flood releases averaging about 5.2 × 109 m3 annually arrived during the period 1983–1998 when upstream reservoirs were full (Glenn et al., 1999). Most river sediments are now trapped in upstream impoundments and no longer reach the Delta (Thompson, 1968). While water management decisions have served mainly to deprive the Delta of water, they have also provided a relatively constant (though never guaranteed) source of water for the Ciénega, which has developed into the largest wetland on the Mexican portion of the Delta since the initiation of agricultural wastewater delivery from Arizona in 1977 (Flessa and García-Hernández, 2007; Glenn et al., 1996; Greenberg and Schlatter, 2012). The Ciénega is located at the northern end of the Santa Clara Slough, a shallow, enclosed 26,000 ha intertidal basin situated between the fault-controlled Gran Desierto Escarpment
Fig. 2. The lower Colorado River Delta study area. CSC: Ciénega de Santa Clara. SCS: Santa Clara Slough MA: Mesa de Andrade GDE: Gran Desierto Escarpment CPF: Cerro Prieto Fault (red dashed line) CRE: Colorado River Estuary SB: Tidal sand bar LB: Fluvial levees/tidal berms IF: Indiviso Fault (black dashed line) EI: El Indiviso EJ: Ejido Johnson.
on the east and natural fluvial levees/tidal berms bordering the Colorado River estuary and Gulf coast on the west and south (Fig. 2). 1.2. Earthquake history The Cerro Prieto Fault in the Ciénega/Slough area has probably ruptured on several occasions in 120 years (Anderson and Bodin, 1987; Felzer and Cao, 2008; Munguia et al., 1988). An 1891 earthquake (M 6.0) caused the collapse of a 100-foot section of bluff at the north end of the Mesa de Andrade (Fig. 2), and opened up three large cracks, each over 450 m in length along the banks of “Salt River” in the fault depression on the east side of the mesa (Strand, 1981). The alluvial plain west of the river (which includes the area of the modern community of El Indiviso) was reported to have been more severely disturbed, with wider and more frequent cracks and trees thrown down in great numbers (Strand, 1981). Sykes (1937) observed extensive structural damage and ground fissures at the Lerdo Colony on the west side of the Mesa de Andrade following a 1903 quake (M 6.6). Ground displacement from a 1934 quake (M 7.0) has been surmised from fresh fault scarps visible in 1935 aerial photographs of the Santa Clara Slough (Biehler et al., 1964;
Author's personal copy
146
S.M. Nelson et al. / Ecological Engineering 59 (2013) 144–156
Kovach et al., 1962). The 1966 El Golfo Earthquake (M 6.3) was centered on the Cerro Prieto Fault 34 km south of the Ciénega, and reportedly caused ground breaking in the Delta area (Ebel et al., 1978). The mainshock of the 1976 Mesa de Andrade Earthquake (M 5.7) consisted of two large events, the first centered about 6 km northwest of the Cienega under the Mesa de Andrade and the second under land now covered by the northwestern part of the Ciénega (Gonzalez et al., 1984; Nava and Brune, 1983). The 1980 Victoria event (M 6.3) was centered along the trace of the Cerro Prieto Fault about 25 km northwest of the Cienega (Anderson and Simons, 1982; Wong et al., 1997). Additional earthquakes exceeding M 5.0 occurred near the northern end of the Cerro Prieto Fault in 1852, 1915, 1955, 1978, 1987, and 2006 (Anderson and Bodin, 1987; Felzer and Cao, 2008). A major (M 7.2) earthquake also occurred in 1892 along the Laguna Salada Fault on the western side of the Sierra Cucapa 75 km northwest of the Ciénega (Hough and Elliott, 2004). The Laguna Salada Fault is thought to be the southern extension of the Elsinore Fault, a major plate boundary fault in Southern California (Mueller and Rockwell, 1995). 1.3. The El Mayor-Cucapah Earthquake The mainshock of the Mw 7.2 El Mayor-Cucapah Earthquake occurred on Easter Sunday, April 4, 2010, producing a nearly continuous fault trace extending 120 km from the northern Sierra Cucapa to the Gulf of California (Fig. 1) (Hauksson et al., 2010). The quake was felt across Baja California and adjoining regions of northwestern Mexico and the southwestern US. It produced extensive liquefaction and ground fracturing in the Mexicali and Imperial valleys which resulted in significant damage to buildings, roads, irrigation canals and agricultural fields and caused widespread social and economic disruption (Brandenberg et al., 2010). Brandenberg et al. (2010), Hauksson et al. (2010), and Wei et al. (2011) have described the complex rupture sequence, which occurred along multiple fault segments which together transmit slip from near the southern end of the Elsinore Fault to the transform plate boundary along the Cerro Prieto fault zone in the northern Gulf. The rupture was initially slow and involved about 8 s of normal faulting along a buried north-trending fault plane near the epicenter. Following a pause of about 6 s the main right lateral strike-slip release propagated rapidly to the northwest with rupture occurring along the Pescadores and Borrego Faults in the Sierra Cucapa Mountains. Simultaneously a less rapid right lateral strike slip propagation progressed southeastward 60 km along the previously unmapped Indiviso Fault beneath the Delta. Altogether the rupture lasted about 45 to 50 s. The rupture is particularly complex because the fault plane segments north of the epicenter dip toward the east, while the fault plane of the Indiviso Fault dips predominantly toward the west (Wei et al., 2011). The block above the fault plane (hanging-wall block), which slipped to the right in relation to the block below the fault plane (footwall block), is thus on the east side of the northern segments and on the western side of the Indiviso Fault. The direction of movement of the respective hanging wall blocks of the northern and southern segments is indicated by the arrows in Fig. 1. Scientific study of the El Mayor-Cucapah Earthquake benefitted from a rapid cooperative response by researchers, institutions, and government agencies on both sides of the international border. Perhaps most notable in these efforts was the rapid acquisition of post-event high resolution airborne light detection and ranging (LIDAR) topographic imagery of nearly all of the 120-km fault rupture for comparison with available pre-event LIDAR data (Oskin et al., 2012). However, post-event LIDAR data acquisition in the intertidal area south of the flood control levee was limited to
areas southeast of the community of Indiviso, including the northwestern portion of the Cienega but excluding intertidal areas to the southwest that experienced surface deformation and subsidence. Rupture features in the Indiviso Fault zone were obscured by Delta sediments that commonly underwent liquefaction at the surface and at depth (Hauksson et al., 2010). Liquefaction was particularly heavy in some of the intertidal areas of the lower Delta (Brandenberg et al., 2010). For these reasons the nature of the rupture event under the intertidal mudflats is less understood than in the northern part of the fault trace where surface rupture was readily visible and extensively studied. 1.4. Santa Clara Slough: the tidal connection The largely unvegetated Santa Clara Slough was formerly connected to tidewater by two outlet channels, the Santa Clara Channel and Shipyard Slough, which became congested with silt in the 1920s after fluvial flow through the Slough was curtailed (Sykes, 1937). Development of a shrimp farm at the southern end of the Slough further obstructed the channels (Glenn et al., 1992). Burnett et al. (1997) reported that the Slough was subject to tidal inundation during spring tides exceeding 17 feet (5.18 m) on the ˜ Puerto Penasco tide calendar. Such tides overtop the natural fluvial levees/tidal berms separating the southwestern side of the Slough from the river estuary (Fig. 2b). Tidal water from the Slough has historically reached the southern end of the Ciénega several times each year. Even when direct tidal flooding does not completely fill the Slough, tidal water may be driven across it toward the Ciénega by southeasterly winds exceeding 15–20 mph (6.7–8.9 m/s−1 ) (Burnett et al., 1997). Tidal intrusion does not appear to have a significant effect on salinity within the vegetated Ciénega, but the presence of peripheral tidal water may limit the expansion of brackish water vegetation to the south (Baeza et al., 2013; Burnett et al., 1997; Flessa and García-Hernández, 2007). In addition to periodic tidal inflow, the Slough receives brackish water inflow from the Ciénega, especially during winter months when delivery of agricultural waste water to the Cienega increases and evapotranspiration decreases (Glenn et al., 2013; Greenberg and Schlatter, 2012). Water leaves the Slough primarily through evaporation (Flessa and García-Hernández, 2007). Surface water presence is highly variable due to the irregular occurrence of tidal overflows and seasonal nature of brackish water inflow and evaporation loss, with more water usually being present during the winter months and following the highest spring tide events. 1.5. Fluvial history Prior to 1909 the main Colorado River channel followed the eastern edge of the Delta (Sykes, 1937), passing just west of the Mesa de Andrade (also known historically as Mesa Arenosa or Colony Mesa) near the modern community of Ejido Johnson. The river had followed this easterly course for at least a half century, and its bed had become aggraded from the buildup of sediments (Sykes, 1937). The river channel was now higher than the surrounding floodplain, increasing the chances that it would break out of the channel during a flood. During high water periods the Santa Clara Slough became an outlet to the Gulf, supporting a network of lagoons and channels that had the appearance of fresh water streams even in their lower intertidal reaches (Sykes, 1937). The lagoons received river overflow at the northern end of the Mesa de Andrade via a distributary channel known variously as the Salt River, Salt Slough, Riito Salado, Santa Clara River, Santa Clara Slough, or simply the “high water channel.” D.T. MacDougal observed river flood waters entering the Santa Clara channel in 1905 (MacDougal and Sykes, 1906). The same
Author's personal copy
S.M. Nelson et al. / Ecological Engineering 59 (2013) 144–156
147
Fig. 4. Colorado River flood waters backed up behind a tidal sand bar (SB) connect the upper estuary and Colorado River/Rio Hardy wetlands (CHW) with the Cienega de Santa Clara (CSC), 1983. Landsat MSS image acquired August 2, 1983.
Fig. 3. The appropriation of the Santa Clara Slough/Cerro Prieto Fault trace by the Colorado River, 1907. Earlier maps prepared by Sykes depict a braided high water channel with numerous lagoons. Note the former main Colorado River channel to the west of Mesa de Andrade and the confluence with Rio Hardy (Hardy’s Colorado). From MacDougal and Sykes (1908).
flood appropriated the Alamo Canal leading to the Imperial Valley, through which the river flowed uncontrolled into the Salton Sink for two years, forming the Salton Sea (Sykes, 1937). By the time the Alamo diversion was closed in 1907 the old river channel had become choked with willow seedlings and partially filled by tidal sandbars in its lower reaches, which together caused most of the flow to divert toward the Santa Clara Slough (MacDougal and Sykes, 1908). Sykes mapped the “New Channel of the Colorado River” in a straight alignment which suggests it followed the Cerro Prieto Fault (Fig. 3). Two years later the river diverted toward the west at another low point further upstream, reducing the old channel system along the eastern side of the Delta to a backwater (Sykes, 1937). Although no longer directly connected to the river, the Santa Clara channel still received river water by way of a maze of sloughs. Aldo and Carl Leopold visited it on a hunting trip in 1922 and found “green lagoons” and abundant wildlife (Leopold, 1968; Leopold and Leopold, 1922). By the late 1920s the lower reaches of the channel had lost their former appearance of a fresh-water stream and become a saltwater tidal channel (Sykes, 1937). The completion of Hoover Dam in 1935 curtailed the floods that had supplied water to this area, causing the wetland to retreat to a small marsh supported by irrigation drainage and local springs (Glenn et al., 1999). The modern Ciénega de Santa Clara had its beginning in 1977 with the first delivery of agricultural drain water from Arizona via the Main Outlet Drain Extension (MODE) bypass drain, which has provided a stable mean input flow of 4.74 m3 s−1 since that time (Greenberg and Schlatter, 2012). For most of the history of the modern wetland, its only connectivity with the Colorado River has been by way of the bypass drain and the Riito Canal which delivers local drain water from Mexico (Glenn et al., 1999; Greenberg and Schlatter, 2012). Normally separated by more than 13 km from the wetlands of the Colorado River estuary and Hardy River to the west, the Ciénega was briefly connected to these wetlands in the early 1980s when river flood waters were temporarily impounded
behind a tidal sand bar obstructing the estuary channel (Fig. 4) (Nelson et al., 2013). Analysis of Landsat scenes from late 1979 through 1985 indicates that this direct connection with the river occurred during two high water periods, the first extending from October 1979 to January 1981, and the second from May 1983 to October 1984. These periods of river flooding temporarily reduced salinity within the developing Cienega (Mexicano et al., submitted for publication). The connection also provided an opportunity for river fish to reach the wetland. Connectivity ended when the river cut a new channel through the tidal sandbar in late 1984, causing the flood waters to recede (Nelson et al., 2013). Additional flood releases during the 1990s kept the estuary channel open, but a new tidal sand bar formed after 2000 (Nelson et al., 2013; Zamora et al., 2013) when fluvial flow to the Delta was curtailed (IBWC, 2012) in response to drought conditions in the Colorado River basin (USGS, 2011). The sand bar reduces connectivity of the upper and lower sections of the estuary, exacerbating impacts to ecosystem services brought about by reduced fluvial flow within both the estuary and upper Gulf of California (Avila-Serrano et al., 2006; Calderon-Aguilera and Flessa, 2009; Galindo-Bect et al., 2000; Galindo-Bect, 2003; Pérez-Arvizu et al., 2009). 1.6. Study objective An understanding of the tectonic, tidal, and fluvial forces acting on the Colorado River Delta is essential to the successful implementation of ecological engineering and restoration projects in this area. The objectives of this study are to highlight for environmental scientists, land managers, and ecological engineers the contribution of tectonic forces in shaping the intertidal Delta environment and to provide information on the effects of the 2010 earthquake which will be of practical value in planning and designing management measures and restoration projects for the estuary and Ciénega. 2. Methods We studied the effects of the Mw 7.2 El Mayor-Cucapah Earthquake on patterns of tidal inundation within the intertidal portion of the Delta with sequential satellite images, overflights, and ground observations; compared the observed patterns with coseismic and post-seismic surface deformation measured using interferometric synthetic aperture radar (InSAR) and pixel tracking analysis, and assessed effects of these changes on the fluvial context of the estuary and Ciénega.
Author's personal copy
148
S.M. Nelson et al. / Ecological Engineering 59 (2013) 144–156
Table 1 List of ASTER and Landsat satellite images used in this study. 1983-8-02 L4 MSS 1997-9-17 Landsat 5 2004-8-03 Landsat 5 2006-7-08 Landsat 5 2006-10-12 Landsat 5 2006-9-10 Landsat 5 2008-4-08 Landsat 5 2008-5-10 Landsat 5 2008-6-11 Landsat 5 2008-7-13 Landsat 5 2008-8-06 Landsat 5 2008-12-20 Landsat 5 2009-8-01 Landsat 5 2009-8-25 Landsat 7 2010-3-29 Landsat 5 2010-4-05 Landsat 5 2010-4-06 Landsat 7 2010-4-14 Landsat 5 2010-4-21 Landsat 5 2010-4-30 Landsat 5 2010-5-07 Landsat 5 2010-5-08 ASTER 2010-5-16 Landsat 5 2010-6-17 Landsat 5
2010-7-19 Landsat 5 2010-8-04 Landsat 5 2010-8-11 Landsat 5 2010-8-12 Landsat 7 2010-8-12 ASTER 2010-8-20 Landsat 5 2010-9-05 Landsat 5 2010-9-13 Landsat 7 2010-10-07 Landsat 5 2010-10-15 Landsat 7 2010-11-08 Landsat 5 2011-3-16 Landsat 5 2011-3-24 Landsat 7 2011-4-17 Landsat 5 2011-5-03 Landsat 5 2011-6-20 Landsat 5 2011-7-22 Landsat 5 2011-8-07 Landsat 5 2011-8-23 Landsat 5 2011-8-31 Landsat 7 2011-9-08 Landsat 5 2011-10-02 Landsat 7 2011-10-26 Landsat 5 2012-2-14 ASTER
2.1. Landsat, ASTER, and MODIS imagery, overflights, and field monitoring visits Satellite imagery was analyzed using elements of analog image interpretation to estimate the areal extent of liquefaction and changes to patterns of tidal inundation. We acquired pre- and post-event Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) and Landsat Multispectral Scanner (MSS), Thematic Mapper (TM) and Enhanced Thematic Mapper (ETM+) satellite images from the U.S.G.S. Earthexplorer website at http://earthexplorer.usgs.gov/. Color composite images of Landsat scenes were prepared using MSS bands 3 (red, 0.63–0.69 m), 4 (NIR, 0.76–0.90 m) and 6 (SWIR, 2.08–2.35 m) and TM/ETM+ bands 1 (blue, 0.45–0.52 m), 4 (NIR, 0.76–0.90 m), and 5 (SWIR, 1.55–1.75 m) to enhance visibility of wet and watercovered surfaces. Level 1T images were selected for analysis; these images are systematically corrected for radiometric and geometric accuracy by the U.S.G.S. EROS data center (Sioux Falls, SD). ASTER color composites were prepared using VNIR Band1 (green, 0.52–0.60 m), VNIR Band2 (red, 0.63–0.69 m) and VNIR Band3N (NIR, 0.76–0.86 m). ASTER Level 1B images are systematically corrected for radiometric and geometric accuracy by the U.S.G.S. Land Processes Distributed Active Archive Center (LP DAAC, Sioux Falls, SD). No atmospheric correction of DN values was attempted for Landsat or ASTER images. ASTER and Landsat images used in the analysis are listed in Table 1. Images not directly referenced in the text provide documentation of changes in the pattern of tidal inundation to support general observations about trends included in the discussion. An analysis of tidal overflows into the Santa Clara Slough during the years 2008–2011 was conducted using georeferenced and geographically sub-setted MODIS 7-2-1 band Aqua and Terra images acquired from NASA’s Earth Orbiting System Data and Information System (EOSDIS), using the USDA Foreign Agricultural Service CAmerica 1 01 subsets available at http://lancemodis. eosdis.nasa.gov/imagery/subsets/?subset=CAmerica 1 01. Images from the first cloud-free day following high spring tide events ˜ predicted to exceed 5.2 m at Puerto Penasco were examined for evidence of tidal overflows into the Santa Clara Slough and tidal advance to the margins of the Cienega de Santa Clara. The first available Landsat image following the tide event was also
Table 2 List of MODIS satellite images used in this study. 2008-01-24 Aqua 2008-04-07 Aqua 2008-05-06 Terra 2008-06-05 Terra 2008-07-04 Aqua 2008-08-02Aqua 2008-08-31 Terra 2008-09-16 Terra 2008-10-16 Terra 2008-11-15 Terra 2008-12-18 Aqua 2009-01-13 Terra 2009-02-11 Aqua 2009-03-11 Terra 2009-06-24 Terra 2009-07-23 Aqua 2009-08-22 Aqua 2009-09-19 Aqua 2009-10-19 Terra
2010-01-02 Terra 2010-02-02 Aqua 2010-02-28 Aqua 2010-03-31 Aqua 2010-07-14 Aqua 2010-08-12 Terra 2010-09-09 Terra 2010-10-09 Aqua 2010-11-07 Terra 2011-02-20 Terra 2011-03-19 Terra 2011-04-19 Aqua 2011-05-19 Aqua 2011-06-16 Terra 3011-08-02 Aqua 2011-08-30 Aqua 2011-09-29 Aqua 2011-10-28 Aqua 2011-11-26 Terra
examined to confirm or clarify the MODIS interpretation. MODIS images documenting evidence of tidal overflows are listed in Table 2. Post-event fixed-wing overflights of the Ciénega, estuary, and intertidal zone were conducted to photographically document changing tidal and fluvial conditions and to obtain reference information for interpretation of satellite images. Flight dates were May 4–5, 2010, June 5 and 14, 2010, October 26, 2010, September 9, 2011, October 18, 2011, and March 13, 2012. Field monitoring visits were conducted on April 25–27, 2010, June 4–6, 2010, November 12–14, 2010, February 18, 2011, March 18–20, 2011, November 11, 2011, and March 8–10, 2012 to photographically document liquefaction, surface rupture, subsidence, and other in situ conditions for use in satellite image interpretation. 2.2. Synthetic aperture radar pixel tracking We used sub-pixel correlation or pixel offset tracking analysis from Fielding et al. (2010) and Wei et al. (2011) to measure the horizontal along-track (roughly N-S) coseismic displacements within the study area (Fig. 8a). This technique cross-correlates patches of the two images to determine the offsets at a number of locations in the images with a precision around 10–30 cm. Large areas of the intertidal zone are incoherent in these images because of frequent tidal flooding and draining, but there is sufficient correlation to document significant ground displacement in some areas. Subpixel correlation between pre- and post- earthquake ALOS PALSAR images from the A211 track was performed on the full-resolution SAR image pairs using the spatial domain cross-correlation program from ROI pac (Pathier et al., 2006). See the Supplementary Information from Wei et al. (2011) for a full methods description. 2.3. Interferometric Synthetic Aperture Radar (InSAR) data analysis We used coseismic interferogram data from Fielding et al. (2010) and Wei et al. (2011) and new post-seismic interferograms to locate areas of measured ground deformation within the study area. Interferograms measure displacements in the line of sight of the radar by phase changes of each pixel between a pair of images. Even more than with sub-pixel correlation, the recurring cycle of tidal flooding and draining causes large incoherent areas in these interferograms, which are sensitive to changes smaller than the 10 m scale of a SAR pixel. However, deformation measured in
Author's personal copy
S.M. Nelson et al. / Ecological Engineering 59 (2013) 144–156 Table 3 ALOS PALSAR pairs used for interferograms.
149
2.5. Tide heights
ALOS path
Date1
Date2
Bperp.a (m)
A211 A211
2010/01/15 2010/4/17
2010/04/17 2010/12/03
663–725 953–1037
a Perpendicular component of baseline at center of swath from top to bottom of interferogram.
coherent areas is often useful in interpreting the extent of deformation in adjacent inundated areas visible in Landsat imagery. Synthetic aperture radar (SAR) interferometry analysis presented here used data acquired by the Japanese Aerospace Exploration Agency (JAXA) Advanced Land Observation Satellite (ALOS), with its phased-array L-band SAR instrument (PALSAR) that has a radar wavelength of 23.5 cm, and by the European Space Agency (ESA) Envisat satellite, with its advanced SAR (ASAR) Cband instrument that has a radar wavelength of 5.6 cm. For the coseismic analysis, we formed an interferogram from ALOS PALSAR scenes on one satellite track (A211) with the post-earthquake scenes acquired 13 days after the earthquake (Table 3) using the JPL/Caltech ROI pac (Rosen et al., 2004) and Stanford SNAPHU (Chen and Zebker, 2000) software. For the post-seismic analysis, we formed interferograms from Envisat ASAR and ALOS PALSAR scenes on four different satellite tracks with beginning and ending dates bracketing a 9 month period following the earthquake (Tables 3 and 4). The data was processed from the PALSAR Level 1.0 and ASAR Level 0 raw data with the JPL/Caltech ROI pac SAR interferometry package (Rosen et al., 2004). ALOS PALSAR data were processed from one ascending (satellite moving northward and radar looking eastward) satellite path as shown in Table 3. All the PALSAR data were acquired with the standard 34.3◦ look angle (at the satellite) that results in line-ofsight (LOS) angles relative to the vertical at the Earth’s surface varying from 36◦ to 41◦ across the radar swath. Envisat ASAR data were processed from three satellite paths, one ascending track and two descending (satellite moving southward and radar looking westward) tracks as shown in Table 4. ASAR scenes were acquired in image mode beam I2 that has LOS angles relative to the vertical at the Earth’s surface varying from 18◦ to 26◦ across the radar swath. See the Supplemental Information from Wei et al. (2011) for a full methods description. 2.4. Ciénega inflow and water level data From January 2010 through July 2011 the Cienega Monitoring Team took bimonthly inflow measurements at the MODE bypass drain and Riito Canal near their discharge points into the Ciénega (Greenberg and Schlatter, 2012). In addition to the inflow measurements, water levels were measured at 20 sites inside the Cienega using water level gauges and automated water level loggers. Water level data from one representative site in the interior of the Ciénega (Site 1) was used in this study to show the relationship of inflow to water level in the months immediately before and after the earthquake.
Table 4 Envisat ASAR pairs used for interferograms. Envisat track
Date1
Date2
Bperp.a (m)
A306 D084 D313
2010/04/13 2010/05/02 2010/04/13
2010/08/31 2010/06/06 2010/08/31
263–284 −98 to −86 313–279
a Perpendicular component of baseline at center of swath from top to bottom of interferogram.
Measured water levels are not available for the estuary or other points in the northern Gulf. Tidal heights cited in this paper are pre˜ dictions for Puerto Penasco, Sonora, calculated using the MAR V1.0 software program developed by the Center for Scientific Research and Higher Education at Ensenada, Baja California (CICESE). Areas inundated by high tide events of similar predicted height were compared to provide a general indication of potential changes in land surface height over time. Aerial extent of tidal inundation was estimated based on the presence of residual water and wet ground visible in the first available image acquired following the spring tide event. Because shallow tidal water is subject to redistribution by strong winds, images were accepted for analysis only if acquired on dates preceded by periods of relatively calm (
