Selection of the earthquake source used in tsunami models of the 2011 Tohoku event affects the simulated tsunami waveform across the near field. Different earthquake sources, based on inversions of seismic waveforms, tsunami waveforms, and Global Positioning System (GPS) data, give distinguishable patterns of simulated tsunami heights in many locations in Tohoku and at near‐field Deep‐ocean Assessment and Reporting of Tsunamis (DART) buoys. We compared 10 sources proposed by different research groups using the GeoClaw code to simulate the resulting tsunami. Several simulations accurately reproduced observations at simulation sites with high grid resolution. Many earthquake sources produced results within 20% difference from the observations between 38° and 39° N, including realistic inundation on the Sendai plain, reflecting a common reliance on large initial seafloor uplift around 38° N (±0.5°), 143.25° E (±0.75°). As might be expected, DART data was better reproduced by sources created by inversion techniques that incorporated DART data in the inversion. Most of the earthquake sources tested at sites with high grid resolution were unable to reproduce the magnitude of runup north of 39° N, indicating that an additional source of tsunamigenic energy, not present in most source models, is needed to explain these observations.

Online Material: Figures of buoy, inundation and run‐up simulations and observations, associated data sets, and simulation code.