Rapid drought-induced land subsidence and its impact on the California aqueduct

Highlights

• UAVSAR detects rapid subsidence adjacent to 10.5+ km of California Aqueduct.

• Exceptional drought conditions from 2012 to 2016 coincide with subsidence acceleration.

• Permanent aquifer volume storage loss is greater than 7122 m³/day for ~2.7 years.

Abstract

The Central Valley in California is characterized by a semi-arid climate prone to droughts, a variable surface water supply, and immense agricultural areas dependent on groundwater irrigation. The groundwater is stored in a complex aquifer system composed of alternating layers of coarse sediments and fine-grained sediments acting as confining materials. Groundwater fluctuations are coupled with both the elastic and inelastic land surface deformation historically observed in the Central Valley. Surface deformation poses a hazard to the California Aqueduct, which supports Central Valley agriculture and large urban populations in Southern California. The risk of reduced aqueduct efficacy and expensive repairs compels water resource managers to carefully monitor land deformation in the Valley.

A persistent drought in the region began in 2012, intensified in 2014, and was abruptly alleviated by a wet period from Dec-2016 to Feb-2018. NASA's UAVSAR L-band synthetic aperture radar acquired 31 high resolution radar images between May-2013 and Nov-2018. The interferometric phase difference between acquisitions is used to develop a time series of vertical displacement and identify and track a rapidly forming subsidence feature adjacent to the California Aqueduct. The surface area of the feature that subsided 10 cm or more by the end of the time series reaches 4452 ha and a 10.5+ km segment of the aqueduct. This study also incorporates extensometer measurements, precise leveling surveys, Sentinel-1A displacement, concurrent water elevation data, well construction reports, nearby extensometer measurements, aquifer material characterization, and environmental conditions. Spatiotemporal data availability limits the appropriateness of calculations and models able to be performed for different sites along the aqueduct. We aim to offer insight into heterogeneous subsurface properties and mechanics, estimate the permanent loss of aquifer storage volume, and identify additional data that would aid water management.

Introduction

The California Aqueduct conveys water from the San Joaquin-Sacramento river delta, through the San Joaquin Valley, and supplies vast agricultural regions and large cities in Southern California. The valley itself is characterized by a variable surface water supply, a semi-arid climate, and ample agricultural areas that are heavily dependent on groundwater irrigation. The spatiotemporal shortage and surplus conditions of California's water supply are managed by surface and groundwater storage connected by a network of aqueducts, pipelines, and flood control structures (Lassiter, 2015). Anthropogenic aquifer overdraft is the cause of historical land subsidence in the valley (Ireland et al., 1984; Poland and Davis, 1969) and drought induces increases in groundwater pumping that can exacerbate subsidence (Jeanne et al., 2019). Deformation of the land surface disturbs water conveyance infrastructure, leading to expensive repairs and hazards due to deformation-induced stresses (Sneed et al., 2018). Therefore, monitoring land subsidence coupled with groundwater change is a priority for modeling and management of water resources in a changing climate (Faunt et al., 2016; Massoud et al., 2018).

Poroelastic deformation of the land surface due to groundwater flux can be elastic (recoverable), or inelastic (permanent) (Galloway and Burbey, 2011). Elastic behavior manifests as seasonal subsidence and rebound of the land surface and water levels. Water in shallow, coarse unconfined aquifers is removed or replaced with minimal surface response. However, many wells exploit semi-confined or confined aquifer systems, which are deeper and often more complex. Confined aquifer systems can exhibit both elastic and inelastic deformation, because excessive drawdown often leads to slow drainage and permanent compaction of the fine-grained aquitard layers (Green and Wang, 1990; Poland and Davis, 1969). Aquitards host significant amounts of water, but drainage delays can lead to situations where subsidence continues after water levels recover. In general, if water levels do not approach new lows, when pumping abates, subsidence can cease or even reverse (Miller et al., 2017). However, if water levels are drawn to new lows, irreversible compaction of the fine-grained aquitards occurs, subsidence is observed at the surface, future recharge of compacted layers is prevented, and groundwater storage capacity is reduced (Smith et al., 2017).

The complex aquifer systems of the San Joaquin Valley are composed of alternating layers of coarse-grained sediment layers and fine-grained sediments lenses (Bertoldi et al., 1991; Williamson et al., 1989). Also, a low-permeability layer called the Corcoran Clay extends through much of the southern valley and acts as a continuous confining layer (Farrar and Bertoldi, 1988). Deposited in an early Pleistocene lacustrine setting, this diatomaceous silt/clay layer is 60+ m thick and found at varied depths along warped structures (Croft, 1972). Areas featuring Corcoran clay have a semi-confined upper aquifer above the clay layer and a confined aquifer below (Faunt et al., 2009). Laterally beyond the extent of the Corcoran clay, the aquifer system features lenses and interbeds of confining, low hydraulic vertical conductivity material. To characterize the complex heterogenous deposits, a three-dimensional geologic textural model, part of the Central Valley Hydrologic Model, was derived from borehole data to define material aquifer properties (Faunt et al., 2010). Aquifer system conditions are additionally described by water elevation measurements, reports of newly constructed wells, and compaction measurements at extensometer sites.

To detect deformation at broad spatial scales, interferometric synthetic aperture radar (InSAR) relates the coherent phase difference between radar returns of the same ground track at different times to the difference in path-length distances between the SAR antenna and the ground (Masson et al., 1993; Rosen et al., 2000). The fractional wavelength difference translates to a measurement of surface displacement in the line-of-sight direction. National Aeronautics and Space Administration's (NASA) Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR) instrument was designed, built, and operated by the Jet Propulsion Laboratory since 2008. UAVSAR is an L-band (24 cm) instrument mounted on a Gulfstream-3 aircraft and flown at 12,500 m altitude via precise autopilot to maintain the same flight path within 10 m of the planned track. Compared to satellite SARs, UAVSAR has a higher spatial resolution (<2 m instrument ground resolution) and signal-to-noise (SNR) ratio (a factor of 100 increase). These improvements along with spatial averaging to reduce phase noise, increase accuracy and reduce temporal decorrelation, so a greater proportion of the scene produces useful measurements. Previous studies using UAVSAR InSAR results include measurements of fault slip (Donnellan et al., 2014), landslides (Scheingross et al., 2013), and ground movement precursory to catastrophic sinkhole formation (Jones and Blom, 2014).

Recent studies of the Central Valley focus on groundwater storage change (Famiglietti et al., 2011; Scanlon et al., 2012), land subsidence related to the 2007–2009 drought (Ojha et al., 2018), drought severity and resilience (Scanlon et al., 2016; Thomas et al., 2017), deformation during and after the persistent 2012–2016 drought (Ojha et al., 2019; Zhu et al., 2017), as well as the aquifer mechanical properties governing the related poroelastic response (Shirzaei et al., 2019). Our study complements and supplements recent work focusing on the California Aqueduct (Sneed et al., 2018) by identifying and analyzing a rapidly forming, localized subsidence feature that is detailed by airborne interferometry. We report the formation and progression of this subsidence feature surface area, compare it with the surrounding hydrogeologic conditions, estimate associated aquifer parameters if appropriate, and determine the permanent loss of storage volume.