1 Introduction
A foundational paradigm in landslide science is that precipitation triggers landslides. Precipitation promotes slope instability as water flows through the ground and raises the water table, or creates perched water tables, and as a result, increases pore-water pressures, reduces the effective normal stress (normal stress minus pore-water pressure), and reduces the frictional strength of the hillslope (Bogaard & Greco, 2016; Terzaghi, 1951). Once a hillslope fails as a landslide it can accelerate rapidly and fail catastrophically (Iverson et al., 2015; Jibson, 2006; Shugar et al., 2021), move downslope slowly for years to hundreds of years (Mackey et al., 2009; Nereson & Finnegan, 2019; Rutter & Green, 2011), or move slowly for a period of time before stabilizing or failing catastrophically (Agliardi et al., 2020; Handwerger, Huang, et al., 2019; Iverson, 2005; Kilburn & Petley, 2003). These different behavioral modes have important consequences for hazard assessment because fast-moving landslides can move at rates up to tens of meters per second and can easily claim lives (Iverson et al., 2015; Shugar et al., 2021), while slow-moving landslides move at rates of meters per year or less and can damage infrastructure (Lacroix, Handwerger, et al., 2020; Merriam, 1960).
Persistently active slow-moving landslides are well-suited for exploring hydrologic controls on landslide motion because they are relatively easy to monitor (compared to landslides with catastrophic failures), occur in wet and dry environments around the world where water is delivered by rainfall (Bayer et al., 2018; Malet et al., 2002), snowmelt (Coe et al., 2003; Matsuura et al., 2008), or irrigation (Lacroix, Dehecq, et al., 2020; Merriam, 1960), and their motion is closely linked to local groundwater conditions (Corominas et al., 2005; Iverson & Major, 1987; Murphy et al., 2022). Furthermore, the hydrologic controls on slow-moving landslides, via pore pressure changes, are akin to the hydrologic controls on faults (Bhattacharya & Viesca, 2019; Cappa et al., 2019), glaciers (Minchew & Meyer, 2020; Moon et al., 2014), and rock glaciers (Cicoira et al., 2019; Kenner et al., 2017), and therefore investigating these landslides allows us to better understand each system.
Previous investigations on the hydrologic controls on slow-moving landslides have shown that precipitation causes slow-moving landslides to accelerate once the pore-water pressures have increased to sufficient levels in the landslide body and decelerate when the pore-water pressures drop (Finnegan et al., 2021; Iverson & Major, 1987; Malet et al., 2002). Thus, slow-moving landslides can slow down or stop moving during dry periods, and speed up, reactivate, or fail catastrophically during wet periods (Bennett, Roering, et al., 2016; Handwerger, Fielding, et al., 2019; McSaveney & Griffiths, 1987; Nereson & Finnegan, 2019). The hydrologic response of landslides can also be size-dependent where larger and thicker landslides are somewhat less sensitive to daily to annual changes in rainfall compared to smaller and thinner landslides that typically experience greater swings in pore-water pressure (Bennett, Roering, et al., 2016; Handwerger, Fielding, et al., 2019). Indeed, this is an expected consequence of pore water transmission in saturated ground (Iverson and Major, 1987). However, recent work by Finnegan et al. (2021) shows nearly instantaneous pore pressure transmission to depth once the vadose zone of the Oak Ridge landslide, California becomes saturated each year, which suggests vadose zone thickness, rather than total landslide thickness, may be the relevant length scale controlling the landslide response in settings where the surface of a landslide becomes unsaturated, for example in locations with highly seasonal rainfall delivery. Nonetheless, both climate and landslide size may govern the hydrologic sensitivity of landslides.
Satellite-based interferometric synthetic aperture radar (InSAR) data can be analyzed alongside precipitation and groundwater data and used to inventory and monitor landslides with the high spatial and temporal resolution necessary to explore hydrologic controls on landslide motion (Bayer et al., 2018; Cohen-Waeber et al., 2018; Handwerger et al., 2013). The open-access data collected by Copernicus Sentinel-1 A/B satellites, in particular, has revolutionized InSAR studies on landslides (Bayer et al., 2018; Carlà et al., 2019; Handwerger, Huang, et al., 2019; Intrieri et al., 2017; Liu et al., 2021; Raspini et al., 2018), and other ground surface deformation (Cigna & Tapete, 2021; Huang et al., 2017; Lundgren et al., 2020; Strozzi et al., 2020), and has led to the development of automated InSAR processing systems that produce derived higher-level standard products that can be used for scientific research (Buzzanga et al., 2020; Dehls et al., 2019; Jones et al., 2021; Lazecký et al., 2020). These derived standard products will become especially important as the volume of InSAR data continues to grow, making it increasingly challenging to process and download InSAR data for large regions on a personal computer. Furthermore, the recent push to provide open-access standardized InSAR products, along with a suite of tools to analyze these data (e.g., Morishita et al., 2020; Yunjun et al., 2019), increases data accessibility to the broader geoscience community, which will undoubtedly lead to major scientific advances.
In this study we analyze open-access standardized Sentinel-1 interferograms automatically processed by the JPL-Caltech Advanced Rapid Imaging and Analysis (ARIA) Center for Natural Hazards project (Bekaert et al., 2019) to identify and monitor landslides in both wet and dry climates in California, USA. California has a large quantity of active slow-moving landslides and has been a major focus area for landslide investigations for decades (Iverson & Major, 1987; Keefer & Johnson, 1983; Kelsey, 1978; Merriam, 1960). Slow-moving landslides in California exhibit distinct seasonal kinematic patterns (Cohen-Waeber et al., 2018; Finnegan et al., 2021; Handwerger et al., 2013; Iverson & Major, 1987) that are a consequence of the regions Mediterranean climate with mild wet winters and hot dry summers, and multi-year kinematic changes that result from precipitation deficits or surplus (Bennett, Roering, et al., 2016; Booth et al., 2020; Mackey et al., 2009; Nereson & Finnegan, 2019). California also has a large rainfall gradient from north to south and west to east, with parts of northern California receiving > 3000 mm/yr of rainfall and parts of southern California receiving < 200 mm/yr (Figure 1a). There are slow-moving landslides in both wet and dry regions of California, which presents an opportunity to examine how variability in hydroclimatology controls landslide behaviors.