Coseismic groundwater-level changes induced by earthquakes have been reported for thousands of years. The M8.0 Wenchuan earthquake caused coseismic groundwater-level responses across the Chinese mainland. Three types of changes were recorded in 197 monitoring wells: coseismic oscillations ranging in amplitude from 0.004 to 1.1m, immediate coseismic step changes ranging from 0.0039 to 9.188 m, and more gradual postseismic changes ranging from 0.014 to 1.087 m. Hydrogeological and tectonic settings are dominant factors in determining the coseismic response. Wells located near the boundaries of active blocks can have large coseismic responses even when the epicenter distance is large. Both the sign and amplitude of water-level changes are random at the continental scale. Poroelastic response to the coseismic static strain cannot explain most of the coseismic changes. Permeability changes caused by stress changes may explain the large variability of the coseismic response amplitude.

We analyze the co-seismic groundwater level responses to four great earthquakes recorded by China's network of groundwater monitoring wells. The large number of operational wells (164 wells for the 2007 Mw 8.5 Sumatra earthquake, 245 wells for the Mw 7.9 Wenchuan earthquake, 228 wells for the Mw 9.0 Tohoku earthquake and 223 wells for 2012 Mw 8.6 Sumatra earthquake) and co-seismic responses provide an opportunity to test hypotheses on mechanisms for co-seismic water level changes. Overall, the co-seismic water level responses are complex over large spatial scales, and there is great variability both in the sign and amplitude of water level responses in the data set. As shown in previous studies, permeability change, rather than static strain, is a more plausible mechanism to explain most of the co-seismic responses. However, we find through tidal analysis of water level responses to solid Earth tide that only one third of these wells that showed a sustained post-seismic response can be explained by earthquake-induced permeability change in aquifers, and these wells had sustained (>30 days) water level changes. Wells that did not show sustained changes are more likely affected by permeability changes only immediately adjacent to the wellbore.