Using Runoff Hydrograph Model for Early Detecting Landslide Dam Failure

Abstract:

In mountainous areas, many studies explored that during extreme rainfall event induced landslide dams failure and consequently caused the catastrophe damage of lives and infrastructures at the downstream of the watershed. The detection of initial landslide dam formation and the estimation of its occurrence position in the upstream were essential to warn the residents and evacuate in advance. The devices currently used, such as wire sensor, geophone sensor, and infrared cameras (CCD) are classified as the post-event type sensor, which only has a shorter response time, and can just hardly satisfy the requirement which people expect from precaution system.

To predict disasters earlier and increase response time, this study used the change point derived from comparing theoretical and observing runoff hydrograph. Chishan River watershed was selected as a case study. Grid Rational Algorithm for Predicting Hydrograph (GRAPH), the dynamic rainfall - runoff model, was used to calibrate watershed runoff hydrograph parameter. The landslide dams were simulated at difference distances along the river, and the runoff hydrographs were compared. Then, the rainfall data and landslide dam failure during Typhoon Morakot in August 2009 was verified the model.

The result showed that, during the initial landslide dam formation, the observed accumulate and peak flow hydrograph reduced significantly in comparison with theoretical flow. This model can be applied to establish an efficient warning system for debris flow occurring precaution. In addition, this study has been improved for a longer response time by integrating traditional observation system and runoff hydrograph warning systems and can provide to the references of related authorities.