Wang, Kai-ShyrKai-ShyrWangChao, Wei-AnWei-AnChaoYang, Che-MingChe-MingYangLai, Tz-ShinTz-ShinLaiWu, Yih-MinYih-MinWu2025-12-182025-12-182025-04-11https://www.scopus.com/pages/publications/105003283028https://scholars.lib.ntu.edu.tw/handle/123456789/734789Rapid assessment of potential landslides in regions of risk and their associated impacts following an earthquake is needed to provide crucial information to guide emergency responses. After the 2025 M6.4 Dapu earthquake in southern Taiwan, we applied a combined analysis of Newmark displacement and a decision tree model to rapidly identify slopes with a high potential for failure. Then, the coseismic landslide inventory taken in the weeks following the Dapu earthquake was compiled from satellite images in conjunction with field photographs. In total, 45 landslides were identified. According to the available near-real-time earthquake shaking proxies and landslide inventories, most landslides occurred within the 250 gal, 15 cm/s and 0.3 cm contour lines of earthquake shaking associated with peak ground acceleration (PGA), peak ground velocity (PGV) and peak vertical displacement of 3-s P waves (Pd3), respectively. In addition, we found that a peak vertical displacement (Pd) larger than 0.3 cm is a good indicator for identifying coseismic landslides and can provide sufficient lead times (LTs) of 3–5 s for emergency response. Our work can serve as a basis for advancing conventional susceptibility assessment after an earthquake occurrence to achieve a coseismic landslide early warning (CoLEW) system.Coseismic landslideCoseismic landslide early warningLead time[SDGs]SDG13journal article10.1007/s44195-025-00102-z