Extending a ML impact-based forecasting model for typhoons in the Philippines with a rainfall threshold for consecutive landslide events
- 1Delft University of Technology, Stevinweg 1, 2628 CN Delft, the Netherlands
- 2The Netherlands Red Cross 510, Anna Van Saksenlaan 50, 2593 HT Den Haag, the Netherlands
- 3Deltares, Boussinesqweg 1, 2629 HV Delft, the Netherlands
- 4The Netherlands Red Cross 510, Anna Van Saksenlaan 50, 2593 HT Den Haag, the Netherlands
- 5CNR IRPI, via Madonna Alta 126, 06128, Perugia, Italy
- 6Delft University of Technology, Stevinweg 1, 2628 CN Delft, the Netherlands
Anticipatory action requires models that can accurately predict the impact of both the primary hazard and its consecutive events. In the Philippines, typhoons trigger 90% of landslides, causing a lot of fatalities and damage to infrastructure and agriculture. The lack of information on past landslides hampers the development of accurate forecasting models of landslide occurrence and impact. An impact-based forecasting model for typhoons currently operational in the Philippines predicts impact due to the high wind speeds associated with typhoons and includes the possible impact due to landslides only via a static landslide susceptibility map. This study expands the impact-based forecasting model of 510, an initiative of the Netherlands Red Cross, with a dynamic landslide component via hybrid modeling for two typhoon events in the Bicol region in the Philippines.
A hydrometeorological model to forecast landslide occurrences was successfully created, even with the limited data on landslide occurrences and rainfall available. The newly established regional event duration threshold was applied on the case study events with an increased impact boundary of 300 km compared to the typhoon impact boundary of 100 km. The dynamic multi-hazard model showed an improved impact forecast - compared to the model considering solely static input of landslides - both in geographical impact extent and accuracy: the True Positives doubled, whereas the False Negatives reduced by half. A separate landslide forecasting model as an extension of the existing ML model provided additional benefits as the models can be decoupled to optimize the performance and reliability of both models. This study resulted in a prototype of an impact-based multi-hazard or consecutive event model for the Philippines and demonstrated the importance of considering the impact from consecutive hazards.
Keywords: Landslide, typhoon, consecutive hazards, impact-based forecasting, rainfall, machine learning, Philippines
How to cite: Free, R., van den Homberg, M., Sperna Weiland, F., Teklesadik, A., Melillo, M., and Bogaard, T.: Extending a ML impact-based forecasting model for typhoons in the Philippines with a rainfall threshold for consecutive landslide events, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-10474, https://doi.org/10.5194/egusphere-egu22-10474, 2022.
