TOPOGRAPHIC CHANGE DETECTION FOR SOME PITTSBURGH, PENNSYLVANIA LANDSLIDES USING DIGITAL ELEVATION MODELS FROM MULTIPLE YEARS: A CASE STUDY IN RIVERVIEW PARK

Southwestern Pennsylvania, and specifically the Pittsburgh area, is well known for landslide activity. Steep slopes, clay-rich rock and colluvial soil, and an old urban area combine to set the stage for instability. USGS mapping in the 1970s created a base inventory and some characterization of features from aerial photo interpretation, identifying recent and prehistoric landslide features. Other workers have since added much detail from field investigations and reporting of slide events. The advent of widespread lidar data in the 2000s allowed a much more detailed look at the ground surface and revealed features previously obscured by vegetation and construction.

Now, multiple sets of lidar data allow a new look at progressive development of several landslides in a small area in Pittsburgh. The test area includes a landslide known to have moved in 2010 and expanded to the adjoining slope in 2018. Several older slides in the area are evident on lidar hillshade images and on a 1977 USGS inventory map. The documented slide abuts Riverview Park, where slides significantly affect the roads, and $1.8 million was allocated for landslide repairs in 2020.

Bare Earth Digital Elevation Models (DEMs) from lidar collected in 2006, 2015 and 2019 were resampled to equivalent cell size (2.5 foot). The newer data sets were then subtracted from the older ones, yielding elevation change rasters. Historical elevation models were also developed from aerial photographs from 1960s and 1970s to extend the time period analysis.

Many Pittsburgh area landslides are known to be reactivations or expansions of earlier landslides, including the example in the study area. Change detection from lidar DEMs clearly shows the known progression of a landslide with documented slope movement in 2010 which expanded 75 m westward in 2018. Change detection further documents slope movement of a nearby, previously undocumented, slide during the period between 2015 and 2019, but most other pre-existing landslides show no evidence of movement during the entire time interval between 2006 and 2019.

These techniques and other data manipulations offer enticing prospects for analysis of historical changes in landslide terrain over time. Such spatial and temporal resolution may improve hazard assessments and be a basis for improved landslide susceptibility mapping.