The April Fool's Tsunami of 1946: Lessons From Sumatra

Submarine landslides are often invoked to explain very large runups because of "Plafker's rule of thumb" (corroborated by Okal &Synolakis), that maximum tsunami runup in the near field does not much exceed maximum slip of the generating earthquake. The huge (30 m) runups along the Aceh coast in 2004 do not require any landslide, however, because earthquake slip approached 30 m. From its tsunami alone, Tanioka &Seno have inferred similar large slip for the eastern Aleutian earthquake of 1 April 1946. But 1946 remains an enigma because the tsunami source region seems too small for the generating earthquake, and the earthquake, supposedly magnitude 8.5, occurred in a region of no strain accumulation. The unambiguous evidence of Sumatra, however, demands that the 1946 event be reexamined. The small source region of the 1946 tsunami depends critically on the one Japanese tide gauge that recorded the tsunami clearly, Hanasaki. But the calibration notations on the Hanasaki record are inconsistent with the identified tsunami arrival time, so the timing remains suspect. The lack of measurable strain is more difficult to dismiss: rupture areas of known great earthquakes (including both Sumatra 2004 and Alaska 1964) all show post-seismic or inter-seismic deformation, making the Unimak segment of the Aleutian Megathrust, and the adjacent Shumagin segment (which also shows no strain accumulation, despite a supposed great earthquake there in 1788), anomalous. A landslide in 1946 was invoked to explain 42 m runup at Scotch Cap, near the western end of Unimak Island. A multibeam survey of the Aleutian slope in 2004 showed that there is no landslide capable of producing such runup. While a landslide on the adjacent unsurveyed slope is possible, it would have to be so large that runup exceeded 50 m along Unimak Bight, the central Unimak coast. No evidence for such huge runup has yet been found, though the tsunami runup exceeded 22 m at Cape Pankof, the eastern extremity of the island. From Scotch Cap to Cape Pankof is 115 km. This compares to the 190 km from Banda Aceh to Meulaboh in Sumatra, the extent of >20 m runups in 2004. We remain uncertain of the origins of the 1946 tsunami. If the source was a landslide, then great ocean- crossing tsunamis can be generated without great earthquakes, which has serious implications for tsunami warning. But if, as comparison with Sumatra now suggests, the source turns out to be purely the earthquake, then we must infer that between major events it is possible for a megathrust to display no deformation at all. A few multibeam tracks would resolve the issue.