Source discriminants for near-field tsunamis

We present numerical simulations of tsunami run-up in the near field for a data set of 72 models of sources, involving both seismic dislocations and landslides. By varying one by one the parameters describing the source and the receiving beach, we are able to separate their individual influence on the amplitude and distribution of run-up, which we characterize by forming several dimensionless parameters, principally the ratio I-2 of its maximum amplitude to its lateral extent along the beach. We find that I-2 remains less than 10(-4) for seismic dislocation sources but is greater than this threshold for all physically realistic models of underwater landslides. Thus, it can be used as a discriminant for the nature of the source of a near-field tsunami. For seismic dislocations, we also consider the ratio I-1 of maximum run-up to seismic slip on the fault plane, and validate numerically the previously suggested 'rule of thumb' that this ratio cannot be much greater than 1. In the case of underwater landslides, we show that the distribution of near-field run-up is primarily controlled by the 2-D 'wall of water' displaced on the ocean surface as an initial condition to the simulation, and is practically independent of its extension in the third dimension, perpendicular to the beach. We apply this approach to nine profiles of tsunami run-up obtained experimentally during recent field surveys, and show that our method successfully identifies the 1998 Papua New Guinea tsunami as having been generated by an underwater landslide, thus confirming the results of shipboard and hydroacoustic investigations. It also strongly suggests a similar mechanism for the generation of the near-field Aleutian tsunami of 1946 April 1.