REAL-TIME SUBSURFACE HYDROLOGIC MONITORING FOR IMPROVED LANDSLIDE EARLY WARNING ALONG SEATTLE-EVERETT RAILWAY CORRIDOR

Early warning for rainfall-induced shallow landsliding can help reduce fatalities and economic losses. Although these commonly occurring landslides are typically triggered by subsurface hydrological processes, most established early warning criteria rely exclusively on empirical rainfall thresholds and other indirect rainfall-based proxies for subsurface wetness. We explore the utility of directly accounting for antecedent wetness by using real-time subsurface hydrologic measurements. Our initial efforts build on previous advances in understanding using rainfall thresholds, monitoring, and numerical modeling along the landslide-prone railway corridor between Everett and Seattle, Washington, USA. We propose a modification to an established recent versus antecedent (RA) cumulative rainfall threshold by replacing the 15-day antecedent rainfall component with the average saturation observed over the same timeframe. For this recent-rainfall versus antecedent-saturation (RS) hybrid threshold, we calculate the antecedent saturation using near-real-time telemetered volumetric water content measured by probes installed within a steep vegetated hillslope along the railway. To allow easy interpretation of RS hybrid threshold exceedance and facilitate integration with quantitative precipitation forecasts, we still rely on the same linear functional relation as the RA rainfall-only threshold and also use the same recent 3-day rainfall component. During a two year monitoring period (2015-2017), the new RS hybrid threshold results in improved precision and threat scores relative to the previous RA rainfall threshold, but might predict landslide occurrence even more accurately with alternative threshold functional relations and components. Although the RS hybrid threshold would still benefit from further testing during future landslide seasons, the positive results confirm that subsurface hydrologic measurements can improve antecedent wetness thresholds for early warning of rainfall-induced shallow landsliding.