DETERMINATION OF RAINFALL THRESHOLD TRIGGERING LANDSLIDES; CASE STUDY IN BANDUNG BASIN AND SURROUNDINGS

• Main Author: Sausan, Zainab
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/73820
• Institution: Institut Teknologi Bandung

<p align="justify">Based on data on disaster events owned by BNPB, landslides are the third most common disaster in Indonesia, with 377 events in 2021 (Geoportal Data Disaster Indonesia, 2021). Rain, as one of the triggers for landslides, causes an increase in the number of landslides during the rainy season in Indonesia. Determination of rainfall thresholds is one of the efforts to solve safety problems because it can be used in making a landslide early warning system. Making landslide-triggering rainfall thresholds can be done with an empirical model using data from previous landslide events and rainfall data. The rain data used is the observed rainfall value from the BBWS Citarum rain post and the Bandung BMKG geophysical station. Information on landslide events uses data from PVMBG ESDM and BPBD West Java Province. The threshold is made based on rain intensity, accumulated rain, rain duration, and antecedent rainfall. This study used three assumptions of rain duration, i.e., three hours, six hours, and eight hours for each day. The threshold value is determined by inputting the log 10 – log 10 graph data. Two probability levels are used in determining the threshold for landslide-triggering rainfall, namely 1% and 5%. Analysis of the area under the curve was carried out to determine the quality of the threshold model. Based on 226 points of landslide occurrence in the Bandung Basin and its surroundings in 2011 – 2021, the rain threshold value was obtained with good model quality, namely I = 5.5462D-0.613 with a duration of 3 ? D ? 48 and I = 2.0543D -0.312 with a duration of 6 ? D ? 96, while the rain threshold based on previous rain using 20 days antecedent rainfall is R = 4.6095(A20)-0.131 and R = 5.3224(A30)-0.16 for 30 days antecedent rainfall. A simulation was carried out on the slopes in the research area, namely in Cihanjuang Village, Cimanggung, Sumedang, to determine the accuracy of the threshold. The simulation uses a threshold with a good-quality model based on the AUC value. Based on the simulations carried out on the slopes, simulated avalanche trigger threshold intensity – duration with a probability level of 1% shows more accurate results than a probability level of 5% for both the assumed duration of three and six hours. Threshold I =2.0543D-0.312 can estimate the occurrence of landslide in the Cimanggung landslide case study.