Multi-event assessment of typhoon-triggered landslide susceptibility in the Philippines

Abstract

Abstract. There is a clear need to improve and update landslide susceptibility models across the Philippines. This is challenging, as landslides in this region are frequently triggered by temporally and spatially disparate typhoon events, and it remains unclear whether such spatially and/or temporally distinct typhoon events cause similar landslide responses, i.e. whether the landslide susceptibility for one typhoon event is similar for another. Here, we use logistic regression to develop four landslide susceptibility models based on three typhoon-triggered landslide inventories for the 2009 Typhoon Parma (local name Typhoon Pepeng), the 2018 Typhoon Mangkhut (local name Typhoon Ompong), and the 2019 Typhoon Kammuri (local name Typhoon Tisoy). The 2009 and 2018 inventories were mapped across the same 150 km2 region of Itogon in Benguet Province, whilst the 2019 event was mapped across a 490 km2 region of Abuan in Isabela Province. The four susceptibility models produced are for the 2009, 2018, and 2019 inventories separately, as well as for the 2009 and 2018 inventories combined. Using the area under the receiver operator curve (AUROC) validation, the accuracy of the models is found to be 78 %–82 % for the Itogon models and 65 % for the Abuan model. To assess landslide time dependency, we use the AUROC validation and the Itogon models to quantify the degree to which susceptibility models derived from one event(s) in time can forecast/hindcast the landslides triggered by another. We find that using a susceptibility model for a typhoon event in one year to forecast/hindcast a typhoon in another leads to a 6 %–10 % reduction in model accuracy compared to the accuracy obtained when modelling and validating each event separately. This suggests some degree of time dependency in typhoon-triggered landslides in the Philippines. However, using a susceptibility model for two combined typhoon events (2018 + 2009) to forecast/hindcast each typhoon event separately led to just a 1 %–3 % reduction in model accuracy. This suggests that combined multi-event typhoon-triggered landslide susceptibility models will be more accurate and reliable for the forecasting of future typhoon-triggered landslides. Finally, by undertaking a high-level comparison of the Abuan and Itogon susceptibility models through space, we preliminarily suggest that there may be spatial dependency in typhoon-triggered landslides in the Philippines but that further work into issues of spatial dependency in this region is required.

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.…”

Predicting Deep-Seated Landslide Displacements in Mountains through the Integration of Convolutional Neural Networks and Age of Exploration-Inspired Optimizer

“…Landslide triggers can be attributed to loading, slope geometry, weather conditions, and hydrological conditions (Perkins et al, 2024;Van Natijne et al, 2023;Millán-Arancibia and Lavado-Casimiro, 2023;Jones et al, 2023). Among these, hydrological conditions, especially groundwater levels, have been one of the most critical elements considered in studies related to landslide prediction (see Figure 1).…”

A Novel Estimation of the Composite Hazard of Landslides and Flash Floods Utilizing an Artificial Intelligence Approach

“…In addition, the LASSO algorithm serves the purpose of autonomously identifying the pivotal independent predictor variables essential for effectively classifying the response of the dependent variable [27].…”

“…The findings revealed that the RF algorithm exhibited the highest performance, followed by the ANN algorithm, while SVM and Naive Bayes classifiers demonstrated comparatively poor results. Likewise, Jones et al [27] used logistic regression to develop four landslide susceptibility models based on 3 typhoon-triggered landslide inventories between 2009 and 2019.…”