Rapid landslide identification using synthetic aperture radar amplitude change detection on the Google Earth Engine

Abstract

Abstract. The rapid and accurate mapping of landslides is critical for emergency response, disaster mitigation, and improving our understanding of where landslides occur. Satellite-based synthetic aperture radar (SAR) can be used to identify landslides, often within days after triggering events, because it penetrates clouds, operates day and night, and is regularly acquired worldwide. Although there are many landslide detection methods using SAR, most require downloading a large volume of data to a local system and specialized processing software and training. Here we present a SAR-based amplitude change detection approach designed for those without SAR expertise that uses multi-temporal stacks of freely available data from the Copernicus Sentinel-1 satellites to identify landslides on Google Earth Engine (GEE). We provide strategies that can aid in rapid response and event inventory mapping. We test our GEE-based approach in a ~ 277 km2 area in Hiroshima Prefecture, Japan where ~ 3,800 landslides were triggered by rainfall in July 2018. Our ability to detect landslides improves with the total number of SAR images acquired before and after the landslide event, by combining both ascending and descending acquisition geometry data, and by using topographic data to mask out flat areas unlikely to experience landslides. Importantly, our GEE approach allows the broader hazards and landslide community to utilize these state-of-the-art remote sensing data.

Landslides triggered by the 2015 Mw 6.0 Sabah (Malaysia) earthquake: inventory and ESI-07 intensity assignment

“…The Sabah earthquake and the methodological approach presented in this study highlight some of the challenges commonly encountered when analyzing earthquake-induced landslide inventories. Indeed, the development of a reliable landslide inventory requires the fulfillment of several criteria, either in terms of available images or mapping method-ology (Harp et al, 2011). In this section, I discuss the role of pre-existing landslides, i.e., slope movements already present before the Sabah earthquake (Sect.…”

“…Landslides are one of such environmental effects and may be a significant cause of damage and casualties (Marano et al, 2010;Budimir et al, 2014). Inventories of landslides triggered by earthquakes are crucial for hazard analyses and land planning (Keefer, 1984;Harp et al, 2011;Xu, 2015); currently tens of inventories are available for a variety of territorial and climatic settings (Schmitt et al, 2017;Tanyas et al, 2017). Landslide inventories were usually derived from manual mapping on aerial or satellite images, but in the last few years several efforts have been undertaken to automatically map earthquake-triggered landslides (e.g., Burrows et al, 2020;Handwerger et al, 2020); nevertheless, manually derived inventories are needed to ascertain the validity and accuracy of (semi-)automatic methods.…”

“…Inventories of landslides triggered by earthquakes are crucial for hazard analyses and land planning (Keefer, 1984;Harp et al, 2011;Xu, 2015); currently tens of inventories are available for a variety of territorial and climatic settings (Schmitt et al, 2017;Tanyas et al, 2017). Landslide inventories were usually derived from manual mapping on aerial or satellite images, but in the last few years several efforts have been undertaken to automatically map earthquake-triggered landslides (e.g., Burrows et al, 2020;Handwerger et al, 2020); nevertheless, manually derived inventories are needed to ascertain the validity and accuracy of (semi-)automatic methods. Landslide number, density and areal percentage vary in the affected area and are often analyzed with respect to topography, seismological or geological conditions (e.g., Chang et al, 2021;Ghaedi Vanani et al, 2021;Papathanassiou et al, 2021;Wang et al, 2019;Xu et al, 2014).…”

Real-time shallow landslide hazard assessment on a regional scale

“…3 shows the location of landslides that have occurred at the site during the year 2015~2020. These locations were identified by analysing satellite images on different dates using Google earth engine (e.g., Handwerger et al 2020;Prasetya et al 2021). This study focuses on a particularly well timeconstrained landslide (Landslide A) that occurred on October 31, 2020, at 10:00 am, as highlighted in Fig.…”

Automatic Unsupervised Landslide Detection Method Based on Single High-resolution Optical Image for Emergency Response

“…(9) Handwerger et al used the Google Earth Engine platform and SAR amplitude change detection to locate landslides. (10) (2) Digital elevation model (DEM)-based methods. A DEM describes the topography information of the study area; through analysis of the deformation and displacement of the surface topography, landslide sites can be effectively located.…”