LAUSR

Landslide susceptibility estimates are essential for reducing the risk posed by landslides to social and economic well‐being. However, estimates of landslide susceptibility depend on reliable landslide inventories whose production requires extensive field or remote sensing efforts. Further, most inventories are not updated through time and thus may not capture the influence of changes in climate and/or land use. Inventories based on citizen reports of landslide occurrence, have the potential to overcome these limitations. Such an inventory can be produced from citizen reports to a 311‐phone and online system, a nationwide database that updates real‐time and records reported landslides location and timing. Whereas this landslide inventory is promising, it has not used for landslide susceptibility analyses and may be associated with spatial uncertainties and reporting biases. In this study we explore the use of 311‐based landslide inventory for landslide susceptibility estimates in Pittsburgh, PA, USA, where landslide risk is among the highest in the nation. We compare the 311‐based inventory to field‐validated inventories through a multi‐pronged approach that combines field validation of 311‐reported landslides, probabilistic analysis of the association between landslides and the underlying topographic and geologic factors, and spatial filtering. Our results show that: (a) approximately 70% of the 311‐reported landslides are associated with an identifiable landslide in the field; (b) the spatial uncertainty of the 311‐reported landslides is 104 ± 25 m; (c) 311‐reported landslides differ from other inventories in that they are primarily associated with proximity to roads, however, field‐correction of 311‐reported landslide locations rectifies this anomaly; (d) a simple spatial filter, scaled by the uncertainty in location as determined from a subset of the 311‐data, can increase the consistency between the 311‐reported inventory and field‐validated inventories. These results suggest that 311‐based landslide inventories can improve susceptibility estimates at a relatively low cost and high temporal resolution.