On February 17, 1996, an earthquake occurred northeast of Biak Island, Indonesia, and caused a tsunami.
Interestingly, the southwest side of Biak Island, which was not facing the epicenter, had a higher tsunami
runup than the facing side (Imamura et al., 1997). Previous researchers assumed that the earthquake
triggered submarine landslides. As no one has addressed this phenomenon, this assumption remains an
unsolved hypothesis.
The tsunami arrival time obtained from local people’s eyewitness testimonies was used to
perform backward tsunami raytracing. Considering the raytracing result and multibeam topography, we
found two submarine landslide candidates: one large submarine landslide and a small submarine
landslide were located in the southeast and south of Biak Island, respectively. Since the small submarine
landslide only had a small effect on the land, we only performed a tsunami simulation for the large
submarine landslide. The result showed that the submarine landslide located at 135.62oE and -1.01oS
with a geometry of about 950 m x 5000 m and a thickness of about 50 m seems to explain the observed
runup and arrival time.
Previous researchers made a slip distribution without considering the submarine landslide
event. As a result, their model could not explain the observed runup in the southwest coastal area of Biak
Island. To accommodate this problem, we propose a new model by combining the submarine landslide
model with a modified fault model from the previous researchers. Our new model explains observed
runup heights well; we obtained a geometric mean of 1.00 and a geometric standard deviation of 1.40.
Keywords:
Tsunami simulation, submarine landslide, backward tsunami raytracing, Biak.