On 22 December 2018, a devastating tsunami struck Sunda Strait, Indonesia without warning, leaving 437 dead and thousands injured along the western Java and southern Sumatra coastlines. Synthetic aperture radar and broadband seismic observations demonstrate that a small, <~0.2 km³ landslide on the southwestern flank of the actively erupting volcano Anak Krakatau generated the tsunami. The landslide did not produce strong short-period seismic waves; thus, precursory ground shaking did not provide a tsunami warning. The source of long-period ground motions during the landslide can be represented as a 12° upward-dipping single-force directed northeastward, with peak magnitude of ~6.1 × 10¹¹ N and quasi-sinusoidal time duration of ~70 s. Rapid quantification of a landslide source process by long-period seismic wave inversions for moment-tensor and single-force parameterizations using regional seismic data available within ~8 min can provide a basis for future fast tsunami warnings, as is also the case for tsunami earthquakes.

The 1556 CE Huaxian earthquake resulted in an estimated 830,000 deaths and caused widespread devastation in the Weihe Basin, China. Seismic intensities from historical accounts yield, via magnitude‐intensity relations, a commonly quoted magnitude of 8¼ to 8½. The maximum recorded shaking was confined to a zone close to the Huashan and Weinan faults, which exhibit fresh scarps up to 7–8 m high. Recent palaeoseismic studies have suggested, however, that the Weinan fault has not ruptured at the surface for several thousand years. Furthermore, the 90‐km combined length of the Huashan and Weinan faults is short for an earthquake of magnitude 8¼ to 8½. We present a detailed analysis of the Weinan fault at one well‐preserved site, combining field observations and age constraints from fluvial terraces displaced by faulting, analysis of a high‐resolution digital elevation model (DEM), interpretation of the walls of a quarry that cuts through the fault zone, and from a profile of borehole cores across the fault. We find that the fault ruptured within the last ~900 years and is likely, along with the Huashan segment, to be the causative fault for the 1556 earthquake. The magnitude remains uncertain, withM_w ~ 7.5 being a plausible estimate given the fault length, and no more thanM_w8.0 if we use the maximum estimates of slip. These estimates are considerably smaller than magnitudes estimated from intensities, with importance in estimating the recurrence intervals between destructive earthquakes and hazard across central China.