Learnings from rapid response efforts to remotely detect landslides triggered by the August 2021 Nippes earthquake and Tropical Storm Grace in Haiti

Abstract

On August 14, 2021, a Mw 7.2 earthquake struck the Tiburon Peninsula of western Haiti triggering thousands of landslides. Three days after the earthquake on August 17, 2021, Tropical Storm Grace crossed shallow waters offshore of southern Haiti triggering more landslides worsening the situation. In the aftermath of these events, several organizations with disaster response capabilities or programs activated to provide information on the location of landslides to first responders on the ground. Utilizing remote sensing to support rapid response, one organization manually mapped initiation point of landslides and three automatically detected landslides. The 2021 Haiti event also provided a unique opportunity to test different automated landslide detection methods that utilized both SAR and optical data in a rapid response scenario where rapid situational awareness was critical. As the methods used are highly replicable, the main goal of this study is to summarize the landslide rapid response products released by the organizations, detection methods, quantify accuracy and provide guidelines on how some of the shortcomings encountered in this effort might be addressed in the future. To support this validation, a manually mapped polygon-based landslide inventory covering the entire affected area was created and is also released through this effort.

Sentinel-1 SAR-based Globally Distributed Landslide Detection by Deep Neural Networks

“…Consequently, lower scores in this context do not necessarily indicate lower overall performance compared to the models presented in Table 5. The SAR-based rapid assessment tool implemented by Amatya et al (2023) stands out for its direct applicability to new events, making it closely aligned with our research objectives. Hence, we draw a direct comparison with their methodology.…”

“…Other rapid assessment tools, such as those developed by Meena et al (2023) and other methods evaluated by Amatya et al (2023), such as HazMapper (Scheip and Wegmann, 2021), and ALADIM (Deprez et al, 2022), rely on optical data and employ diverse techniques, including object-based image analysis (OBIA) (Blaschke, 2010) and DL segmentation models. These approaches excel at accurately outlining landslide boundaries.…”

“…In the aftermath of such occurrences, it is important to investigate locations of landslides to assess damages to both natural and anthropogenic landscapes. Both Williams et al (2018) and Amatya et al (2023) highlighted the importance of rapid mapping for the 2015 Gorkha and 2021 Haiti MLE coseismic landslides, emphasizing the use of Synthetic Aperture Radar (SAR) data, preferably in an automated pipeline for emergency evacuations. Moreover, a comprehensive understanding of these slope instability processes begins with a spatial assessment of the slope failures for both rainfall (Nocentini et al, 2023;Segoni et al, 2014) and earthquake-induced landslides (Meena and Tavakkoli Piralilou, 2019).…”

Retrieval of Monsoon Landslide Timings With Sentinel‐1 Reveals the Effects of Earthquakes and Extreme Rainfall

“…Unimpeded by cloud cover, Sentinel‐1 resolves landslide timing much better than previous approaches based on optical imagery. Beyond Nepal and monsoon‐affected areas, the method could be applied to any landslide inventory in vegetated environments whose timings are poorly constrained and thus difficult to assign to a specific trigger, for example, landslides triggered during sequences of storms or earthquakes, including post‐seismic rainfall‐triggered landsliding (Amatya et al., 2023; Ferrario, 2019; Jones et al., 2023). Our study thus highlights the value that Sentinel‐1 could have in disentangling the effects of multiple landslide‐triggering events in cloudy conditions.…”

Rapid Mapping of Landslides Induced by Heavy Rainfall in the Emilia-Romagna (Italy) Region in May 2023

“…The inventory was released on 28 June 2023, i.e., 1.5 months after the main triggering event; in this sense, our inventory can be assimilated to rapid response products (e.g., [49]). Some operative choices represent a trade-off between the rapidity of execution and data quality/resolution, for instance, the mapping was realized at a 1:5000 scale and should be consulted at most at the same scale; a side effect is that our mapping is not recommended for higher resolution susceptibility studies, unless a field verification is undertaken.…”