This study investigates the hydrometeorological thresholds of reservoir-induced landslides using subsurface strain observations from a giant reservoir landslide in the Three Gorges Reservoir (TGR) region, China. The research leverages high-resolution fiber optic sensing nerves installed since February 2021 to measure spatiotemporal strain profiles, which help identify slip zones and potential drivers. The analysis reveals that high-intensity short-duration rainstorms control landslide kinematics. By considering the time lag effect, the study reexamines and quantifies the controls on accelerated movements, showing an immediate response to extreme rainfall with a zero-day shift. A landslide prediction model is developed using the boosting decision tree (BDT) algorithm, incorporating daily rainfall, rainfall intensity, reservoir water level, water level fluctuations, and slip zone strain time series. The results indicate that landslide acceleration is most likely under mid-low water levels (<169,700 m) and large-amount, high-intensity rainfalls (daily rainfall >57.9 mm and rainfall intensity >24.4 mm/h). The model allows for updating hydrometeorological thresholds with the latest monitoring data, providing a practical and reliable pathway for early warning based on subsurface observations, especially in the context of enhanced extreme weather events.This study investigates the hydrometeorological thresholds of reservoir-induced landslides using subsurface strain observations from a giant reservoir landslide in the Three Gorges Reservoir (TGR) region, China. The research leverages high-resolution fiber optic sensing nerves installed since February 2021 to measure spatiotemporal strain profiles, which help identify slip zones and potential drivers. The analysis reveals that high-intensity short-duration rainstorms control landslide kinematics. By considering the time lag effect, the study reexamines and quantifies the controls on accelerated movements, showing an immediate response to extreme rainfall with a zero-day shift. A landslide prediction model is developed using the boosting decision tree (BDT) algorithm, incorporating daily rainfall, rainfall intensity, reservoir water level, water level fluctuations, and slip zone strain time series. The results indicate that landslide acceleration is most likely under mid-low water levels (<169,700 m) and large-amount, high-intensity rainfalls (daily rainfall >57.9 mm and rainfall intensity >24.4 mm/h). The model allows for updating hydrometeorological thresholds with the latest monitoring data, providing a practical and reliable pathway for early warning based on subsurface observations, especially in the context of enhanced extreme weather events.