Climatic changes, streamflow, and long-term forecasting of intraplate seismicity

Abstract

Regions of intraplate seismicity in the eastern United States are spatially isolated areas of persistent, diffuse earthquake activity. There is no widely accepted explanation for the origin of these earthquakes. We have suggested that climate plays a key role in triggering such intraplate seismicity. Long-term increases and decreases in rainfall cause periodic regional and temporal variations in the elevation of the water table that, by pore pressure diffusion, result in small changes in fluid pressure at any given depth in the crust. In a fractured, hydraulically permeable crust, the depth of penetration of this pore pressure diffusion can be as deep as the brittle-ductile transition (15–18 km). In a seismogenic crust, stress corrosion and fatigue of rock asperities might be more important than purely mechanical effects due to small changes in hydrostatic fluid pressure; however, because any chemical effects are quasi-static, the temporal characteristics of the triggering process might ultimately be determined by the mechanical process, resulting in a “hydraulically induced” seismicity trigger that acts somewhere along paths of pore pressure diffusion. We called this model of intraplate earthquake generation “hydroseismicity”. Because streamflow is related to the regional and temporal morphology of the water table, we searched for and report here several attempts to find links between climatic changes, streamflow, and intraplate seismicity. Our results fall broadly into four categories: (1) temporal correlations of streamflow with the earthquake strain factor; (2) spectral analyses of flow and seismicity and the identification of common spectral peaks; (3) numerical modeling to estimate fluctuations in pore pressure at hypocentral depths; and (4) climatic driving mechanisms (e.g., sunspot cycles) that might substantiate a climate-earthquake link. We observe a statistically significant peak in the Fourier spectrum of surface streamflow for the seismic zones bisected by the Mississippi River, Illinois, and James River, Virginia, in the period range of 11–13 years that might be associated with sunspot activity. In addition, there is positive correlation between periods of above average values of the standard deviation of streamflow time series and periods of seismicity in the central Virginia seismic zone. Many aspects of the weather appear to be modulated by a 20-year cycle. We observe a similar periodicity (18–20 years) in seismicity in the central Virginia seismic zone. A good agreement is observed when a streamflow time series is superimposed on the record of the earthquake strain factor if a value of 50 km 2/year is assumed for crustal hydraulic diffusivity. In the central Virginia seismic zone, it is found that the number of earthquakes versus depth, ψ, is directly proportional to pressure fluctuations at the depth ψ. In addition, the fractal dimension determined from downward-continued streamflow is approximately the same as the fractal dimension of intraplate seismicity. Furthermore, using the Gutenberg-Richter relation and assuming that the earthquake data sets in the New Madrid and central Virginia seismic zones are complete for all magnitudes m ⩾ 2, the ratio of the number of earthquakes occurring per year in the New Madrid zone to the central Virginia zone is about 40. The ratio of the standard deviations of downward-continued Mississippi River streamflow (at Thebes, Illinois) to the James River streamflow is also about 40. One interpretation of this common ratio is that the number of intraplate earthquakes generated in a seismogenic crust is directly proportional to the standard deviation of vertical variations in the elevation of the water table. If the hydroseismicity hypothesis is correct, then long-term variations in streamflow can be used to forecast long-term statistical variations in intraplate seismic activity.