COMPARISON OF REMOTELY MAPPED LANDSLIDES ACROSS MULTIPLE REGIONS OF PUERTO RICO FOLLOWING HURRICANE MARIA, 2017

Hurricane María caused widespread landsliding throughout Puerto Rico during September 2017. Detailed landslide inventories following the hurricane focused on mountainous regions underlain by igneous and volcaniclastic bedrock. Here, we compare landslides in multiple regions of Puerto Rico: the greater karst region on the northwest side of the island covering ~1425 km², five regions underlain by similar igneous rock and three regions underlain by variable volcaniclastic rock, with all igneous and volcaniclastic study areas covering several square kilometers each due to the higher density of landslides in granodiorite and volcaniclastics compared to the karst. To map landslides triggered by María, we used aerial imagery collected between 9-15 October 2017 (Quantum Spatial, Inc., 2017) and 1-m pre- and post-event lidar (U.S. Geological Survey, 2018, 2020) as a digital base. The map data consist of geographic information system (GIS) files of headscarp points, travel-path lines, source-area polygons, and affected-area polygons. Characteristics of the landslides and their geomorphic settings including curvature and slope are included in attribute tables of the mapped features. Other items included in the attribute tables are major landform type, land use, and precipitation totals. Quantitative attributes (e.g., travel distance, fall height, watershed contributing area) were determined using tools available in ArcMap GIS. Statistical analysis of these attributes showed that limestone units in the karst region experienced landslides on slopes >45°, while other areas experienced peak landslide occurrence within +/-5° of a mean slope angle which varied depending on the geologic unit. Most of the landslides in all regions were less than 1 m thick, and most of these shallow landslides were translational failures, though the karst region also experienced a few rock falls. Dissected landforms of igneous regions displayed cases of runouts as much as 3-8 times longer than that of other areas, and finally, failures on planar surfaces were more common than failures on concave or convex slopes for all study areas. In the future, our inventories should be useful in helping to understand and forecast landslide hazards on the island.