Ground motions of the South Napa earthquake (24 August 2014; M 6.0) were recorded at 19 stations within 20 km and 292 stations within 100 km of the rupture surface trace, generating peak ground motions in excess of 50%g and 50  cm/s in and near Napa Valley. This large dataset allows us to compare the ground motion from the earthquake to existing ground‐motion prediction equations (GMPEs) in considerable detail.

Using the ground‐motion data compiled and reported by ShakeMap (Wald et al., 2000), we examine the peak ground acceleration (PGA) and peak ground velocity (PGV), as well as the pseudospectral acceleration (PSA) at periods of 0.3, 1.0, and 3.0 s. At the higher frequencies, especially PGA, data recorded at close distances (within ∼20  km) are very consistent with the GMPEs, implying a stress drop for this event similar to the median for California, that is, 5 MPa (Baltay and Hanks, 2014). At all frequencies, the attenuation with distance is stronger than the GMPEs would predict, which suggests the attenuation in the Napa and San Francisco Bay delta region is stronger than the average attenuation in California. The spatial plot of the ground‐motion residuals is positive to the north, in both Napa and Sonoma Valleys, consistent with increases in amplitude expected from both the directivity and basin effects. More interestingly, perhaps, there is strong ground motion to the south in the along‐strike direction, particularly for PSA at 1.0 s. These strongly positive residuals align with an older, Quaternary fault structure associated with the Franklin or Southampton fault, potentially indicating a fault‐zone‐guided wave.