Common cyclicities in seismicity and water level fluctuations at the Charlevoix seismic zone of the St. Lawrence River

TR Number



Journal Title

Journal ISSN

Volume Title


Virginia Tech


Intraplate seismicity has no widely accepted explanation for its origin. The Hydroseismicity hypothesis, developed by Costain and co-workers, suggests that natural increases in hydraulic head, caused by transient increases in water table elevation, can be transmitted to hypocentral depths (10-25 km) in a fractured, prestressed, near-failure crust, and along with long term hydrolitic weakening of rocks, contribute to the triggering of earthquakes. In this study, the temporal characteristics of seismicity and water level fluctuations at the Charlevoix seismic zone on the Saint Lawrence river are investigated to provide a test of the Hydroseismicity hypothesis.

To characterize the temporal release of seismic energy, two measures of seismic activity are considered for the available 200-year record of seismicity: the strain factor (magnitude dependent), and the number of events per unit time (magnitude independent). Residual analysis, applied to the strain factor time series, indicates a cyclical variation of the seismicity with long term periods ranging from 65 to 70 years. Fourier spectral analysis, applied on both strain factor and number of events time series, indicates the presence of short term cyclicities of seismic energy release at 13 to 14 year periods, along with longer-term cyclicities of approximately 55 to 70 year periods. Fisher’s periodogram ordinate test, applied to the spectral analyses results to determine the significance of the largest periodogram ordinate with respect to the average, tested the 13.4-year periodogram ordinate of the number of events time series significant at the 10% level, whereas the 13.4-year periodogram ordinate of the strain factor time series failed to test statistically significant.

Analysis of the residual water level 70 year-long time series indicates a cyclical process with quarter cycles of approximately 20 years, suggesting a complete cycle of 70 to possibly 80 years. Fourier spectral analysis of the water level data set indicates cyclicities with periods of 1, 23, 14, and 8 years. The Grenander-Rosenblatt method (Priestley, 1981), applied to the spectral analysis results, tests the statistical significance of the largest periodogram ordinates, and showed the 1, 23, 14, and 8-year periodogram ordinates to be significant at the 1% level.

Cyclicities present in the water level and seismicity time series were investigated for temporal (time lag) relationships using Group Delay analysis, a procedure that tests for time relationships at selected bandwidths. The maximum value of the Group Delay function occurs at the 14 year period with a time lag of + 2.3 years, indicating that the water levels lead the seismicity time series. The crustal diffusivity values estimated from the Group Delay analysis of the 14 year center period, range between 0.3 m²/sec and 2.7 m²/sec for a depth interval from 7 to 20 km. Direct correlation between depth and crustal diffusivity is not justified with the data at hand; however, the range of estimated diffusivities is within the range of published values of diffusivities for the crust (0.1 to 100m²/sec).

This study allows for a possible causal relationship between repetitive mechanical effects of pore pressure transients and seismicity in the Charlevoix seismic zone. The temporal behavior of the physical processes studied is in general agreement with the Hydroseismicity model.