InSAR-based detection method for mapping and monitoring slow-moving landslides in remote regions with steep and mountainous terrain: An application to Nepal

• Published in: Remote sensing of environment 2020-11, Vol.249, p.111983, Article 111983
• Main Authors: Bekaert, David P.S., Handwerger, Alexander L., Agram, Piyush, Kirschbaum, Dalia B.
• Format: Article
• Language: English
• Subjects: Clustering; Data acquisition; Deep-seated landslide; Deformation; Detection; Drainage basins; Dry season; Earth Resources And Remote Sensing; Earthquake; Earthquakes; Gorkha earthquake; Hazard mitigation; Identification methods; Interferometric synthetic aperture radar; Interferometry; Landslides; Landslides & mudslides; Mapping; Monsoons; Mountains; Rainfall; Remote monitoring; Remote regions; Remote sensing; Satellite data; Satellites; Seasons; Seismic activity; Slow-moving landslide; Spatial filtering; Synthetic aperture radar; Synthetic aperture radar interferometry; Terrain; Time series; Time-series InSAR; Trishuli; Wind
• ISSN: 0034-4257; 1879-0704
• DOI: 10.1016/j.rse.2020.111983

Mapping and monitoring landslides in remote areas with steep and mountainous terrain is logistically challenging, expensive, and time consuming. Yet, in order to mitigate hazards and prevent loss of life in these areas, and to better understand landslide processes, high-resolution measurements of landslide activity are necessary. Satellite-based synthetic aperture radar interferometry (InSAR) provides millimeter-scale measurements of ground surface deformation that can be used to identify and monitor landslides in remote areas where ground-based monitoring techniques are not feasible. Here we present a novel InSAR deformation detection approach, which uses double difference time-series with local and regional spatial filters and pixel clustering methods to identify and monitor slow-moving landslides without making a priori assumptions of the location of landslides. We apply our analysis to freely available Copernicus Sentinel-1 satellite data acquired between 2014 and 2017 centered on the Trishuli River drainage basin in Nepal. We found a minimum of 6 slow-moving landslides that all occur within the Ranimatta lithologic formation (phyllites, metasandstones, metabasics). These landslides have areas ranging from 0.39 to 1.66 km2 and long-term dry-season displacement rates ranging from 2.1 to 8.8 cm/yr. Due to periods of low coherence during the monsoon season (June – September) each year, and following the 25 April 2015 Mw7.8 Gorkha earthquake, our time series analysis is limited to the 2014–2015 and 2016–2017 dry seasons (September–May). We found that each of the landslides displayed slightly higher rates during the 2014 period, likely as a result of higher cumulative rainfall that fell during the 2014 monsoon season. Although we do not have high quality InSAR data to show the landslide evolution directly following the Gorkha earthquake, the similar rates of movement before (2014–2015) and after (2016–2017) Gorkha suggest the earthquake had negligible long-term impact on these landslides. Our findings highlight the potential for region-wide mapping of slow-moving landslides using freely available remote sensing data in remote areas such as Nepal and future work will benefit from expanding our methodology to other regions around the world. •A novel method is developed to detect landslides in mountainous terrain.•InSAR time-series is used to identify and monitor slow-moving landslides.•6 slow-moving landslides in Trishuli, Nepal, unaffected by the Gorkha eart