Global rainfall thresholds for urban landslides

By 2050, 68% of people are projected to live in urban areas. As cities grow into steeper terrain, residents are increasingly exposed to hazards like rainfall-triggered landslides. Landslide early warning systems can help reduce this risk, but so far, few cities have established such systems.  One barrier to doing so is lack of sufficient landslide inventory data to estimate rainfall thresholds for warning. In this talk, I will describe our efforts to estimate a global rainfall threshold for urban landslides and to learn how variable rainfall thresholds are between cities. Using a multi-level regression model and a new compilation of thousands of urban landslides reported in cities worldwide, we find that landslides were triggered under surprisingly similar rainfall conditions in most cities, despite contrasting environmental and economic characteristics. Furthermore, we show that one-third of landslides occurred during moderate storms, not only during extreme rainfall, and that urban landslides occurred at lower rainfall intensities than previously proposed global thresholds. Our results suggest that regional landslide early warning systems should consider differences between rural and urban areas and that cities with limited landslide inventory data may be able to use thresholds from other cities to inform warning efforts.