This open access book explores the interactions between water and earthquakes, including recent concerns about induced seismicity. It further highlights that a better understanding of the response of ...the water system to disturbances such as earthquakes is needed to safeguard water resources, to shield underground waste repositories, and to mitigate groundwater contamination. Although the effects of earthquakes on streams and groundwater have been reported for thousands of years, this field has only blossomed into an active area of research in the last twenty years after quantitative and continuous documentation of field data became available. This volume gathers the important advances that have been made in the field over the past decade, which to date have been scattered in the form of research articles in various scientific journals.
Hydrogeological properties can change in response to large crustal earthquakes. In particular, permeability can increase leading to coseismic changes in groundwater level and flow. These processes, ...however, have not been well-characterized at regional scales because of the lack of datasets to describe water provenances before and after earthquakes. Here we use a large data set of water stable isotope ratios (n = 1150) to show that newly formed rupture systems crosscut surrounding mountain aquifers, leading to water release that causes groundwater levels to rise (~11 m) in down-gradient aquifers after the 2016 M
7.0 Kumamoto earthquake. Neither vertical infiltration of soil water nor the upwelling of deep fluids was the major cause of the observed water level rise. As the Kumamoto setting is representative of volcanic aquifer systems at convergent margins where seismotectonic activity is common, our observations and proposed model should apply more broadly.
Detecting whether and how river discharge responds to strong earthquake shaking can be time‐consuming and prone to operator bias when checking hydrographs from hundreds of gauging stations. We use ...Bayesian piecewise regression models to show that up to a fifth of all gauging stations across Chile had their largest change in daily streamflow trend on the day of the Mw 8.8 Maule earthquake in 2010. These stations cluster distinctly in the near field though the number of detected streamflow changes varies with model complexity and length of time window considered. Credible seismic streamflow changes at several stations were the highest detectable in eight months, with an increased variance of discharge surpassing the variance of discharge following rainstorms. We conclude that Bayesian piecewise regression sheds new and unbiased insights on the duration, trend, and variance of streamflow response to strong earthquakes, and on how this response compares to that following rainstorms.
Plain Language Summary
Strong earthquakes can abruptly change the discharge or rivers. Detecting this streamflow response mostly depends on visually checking data from gauging stations, and may be prone to bias. We test an alternative approach that estimates the most probable date of the largest change in daily river discharge. We test the approach with data from over 200 Chilean stations to see which responded to the 2010 magnitude 8.8 Maule earthquake. We find that dozens of stations near the epicenter had their largest jumps in streamflow on the day of the earthquake. This response lingered for up to four months at some stations, also raising the variability of discharge much above that following rainstorms. These findings are consistent with the idea of an earthquake‐driven increase in vertical permeability underground, and show how data‐driven models can usefully augment visual checks of river data.
Key Points
We offer an objective, consistent, and reproducible way of detecting streamflow response to earthquakes
Bayesian piecewise regression reveals credible and previously unrecognized increases in post‐seismic discharge variability
The variance in thus detected post‐seismic streamflow changes exceeds by far the discharge variability following rainstorms
Shaking‐induced changes in permeability have been invoked to explain a wide range of hydrologic responses to earthquakes. We measure the evolution of permeability in fractured sandstone in response ...to repeated shaking under undrained conditions. The frequency and amplitude of the imposed shaking are similar to those from earthquakes at distances of ∼ a fault length from the ruptured fault. To assess the role of mobile particles, we also add silt‐sized particles to the fractures. We find that, in general, permeability decreases after shaking. The samples with added particles show larger changes in permeability. Shaking‐induced transport of particles to block narrow fracture apertures can explain some of the observations. We also find that decreases in permeability are often accompanied by a contraction of sample indicating that fracture apertures sometimes decrease, presumably by mobilizing particles and removing asperities that hold fractures open.
Oscillations in stress, such as those created by earthquakes, can increase permeability and fluid mobility in geologic media. In natural systems, strain amplitudes as small as 10−6 can increase ...discharge in streams and springs, change the water level in wells, and enhance production from petroleum reservoirs. Enhanced permeability typically recovers to prestimulated values over a period of months to years. Mechanisms that can change permeability at such small stresses include unblocking pores, either by breaking up permeability‐limiting colloidal deposits or by mobilizing droplets and bubbles trapped in pores by capillary forces. The recovery time over which permeability returns to the prestimulated value is governed by the time to reblock pores, or for geochemical processes to seal pores. Monitoring permeability in geothermal systems where there is abundant seismicity, and the response of flow to local and regional earthquakes, would help test some of the proposed mechanisms and identify controls on permeability and its evolution.
Key Points
Distant earthquakes affect hydrological processes
Time varying stresses can change fluid mobility by dislodging particles and bubbles
Monitoring geothermal systems with abundant seismicity may yield new insights
Explosive volcanic eruption requires that magma fragments into discrete parcels. Silicic magma can fragment through brittle failure or other processes that depend on the viscoelasticity of the melt. ...Owing to the low viscosity of basaltic magmas, however, the fragmentation mechanism must be different and will be governed by fluid mechanics alone. We perform a series of decompression experiments on bubbly Newtonian fluids with viscosities and surface tensions similar to those of basaltic magmas. For sufficiently rapid expansion, the bubbly fluid expands continuously, eventually tearing into several pieces. We find that the fragmentation threshold is governed by a critical Reynolds number of ∼
1, indicating that it is the inertia of the expanding fluid that drives the continued expansion and ultimate breakup into discrete parcels. Experiments in which the fluid does not fragment allow us to determine the gas permeability of the bubbly fluid as the bubbles expand. Permeability remains small until the volume fraction of bubbles exceeds about 70%. We scale the results of the laboratory experiments to basaltic eruptions and find that the predicted fragmentation threshold is consistent with the exit velocities that characterize effusive and explosive eruptions. Our experimental results suggest that the mechanism for fragmentation of low viscosity basaltic magma is fundamentally different from that of high-viscosity silicic magma, and that magma with low viscosities can fragment easily.
Geofluids (2010) 10, 206–216
Hydrologic responses to earthquakes, including liquefaction, changes in stream and spring discharge, changes in the properties of groundwater such as geochemistry, ...temperature and turbidity, changes in the water level in wells, and the eruption of mud volcanoes, have been documented for thousands of years. Except for some water‐level changes in the near field which can be explained by poroelastic responses to static stress changes, most hydrologic responses, both within and beyond the near field, can only be explained by the dynamic responses associated with seismic waves. For these responses, the seismic energy density e may be used as a general metric to relate and compare the various hydrologic responses. We show that liquefaction, eruption of mud volcanoes and increases in streamflow are bounded by e ∼ 10−1 J m−3; temperature changes in hot springs are bounded by e ∼ 10−2 J m−3; most sustained groundwater changes are bounded by e ∼ 10−3 J m−3; geysers and triggered seismicity may respond to seismic energy density as small as 10−3 and 10−4 J m−3, respectively. Comparing the threshold energy densities with published laboratory measurements, we show that undrained consolidation induced by dynamic stresses can explain liquefaction only in the near field, but not beyond the near field. We propose that in the intermediate field and far field, most responses are triggered by changes in permeability that in turn are a response to the cyclic deformation and oscillatory fluid flow. Published laboratory measurements confirm that changes in flow and time‐varying stresses can change permeability, inducing both increases and decreases. Field measurements in wells also indicate that permeability can be changed by earthquakes in the intermediate field and far field. Further work, in particular field monitoring and measurements, are needed to assess the generality of permeability changes in explaining far‐field hydrologic responses to earthquakes.
Earthquakes can trigger the eruption of mud. We use eruptions in Azerbaijan, Italy, Romania, Japan, Andaman Islands, Pakistan, Taiwan, Indonesia, and California to probe the nature of stress changes ...that induce new eruptions and modulate ongoing eruptions. Dynamic stresses produced by earthquakes are usually inferred to be the dominant triggering mechanism; however static stress changes acting on the feeder systems of mud volcanoes may also play a role. In Azerbaijan, eruptions within 2–10 fault lengths from the epicenter are favored in the year following earthquakes where the static stress changes cause compression of the mud source and unclamp feeder dikes. In Romania, Taiwan, and some Italian sites, increased activity is also favored where the static stress changes act to unclamp feeder dikes, but responses occur within days. The eruption in the Andaman Islands, and those of the Niikappu mud volcanoes, Japan are better correlated with amplitude of dynamic stresses produced by seismic waves. Similarly, a new island that emerged off the coast of Pakistan in 2013 was likely triggered by dynamic stresses, enhanced by directivity. At the southern end of the Salton Sea, California earthquakes increase the gas flux at small mud volcanoes. Responses are best correlated with dynamic stresses. The comparison of responses in these nine settings indicates that dynamic stresses are most often correlated with triggering, although permanent stress changes as small as, and possibly smaller than, 0.1bar may be sufficient to also influence eruptions. Unclamping stresses with magnitude similar to Earth tides (0.01bar) persist over time and may play a role in triggering delayed responses. Unclamping stresses may be important contributors to short-term triggering only if they exceed 0.1–1bar.
•Near- and far-field earthquakes can trigger the eruption of mud volcanoes.•Response of mud volcanoes to earthquakes can occur in the short- and long term.•Both static and dynamic stresses influence eruptions.
Laboratory density currents comprising warm talc powder turbulently suspended in air simulate many aspects of dilute pyroclastic density currents (PDCs) and demonstrate links between bulk current ...behavior, sedimentation, and turbulent structures. The densimetric and thermal Richardson, Froude, Stokes, and settling numbers match those of natural PDCs as does the ratio of thermal to kinetic energy density. The experimental currents have lower bulk Reynolds numbers than natural PDCs, but the experiments are fully turbulent. Consequently, the experiments are dynamically similar to the dilute portions of some natural currents. In general, currents traverse the floor of the experimental tank, sedimenting particles and turbulently entraining, heating, and thermally expanding air until all particle sediments or the currents become buoyant and lift off to form coignimbrite plumes. When plumes form, currents often undergo local flow reversals. Current runout distance and liftoff position decrease with increasing densimetric Richardson number and thermal energy density. As those parameters increase, total sedimentation decreases such that >50% of initial current mass commonly fractionates into the plumes, in agreement with some observations of recent volcanic eruptions. Sedimentation profiles are best described by an entraining sedimentation model rather than the exponential fit resulting from non-entraining box models. Time series analysis shows that sedimentation is not a constant rate process in the experiments, but rather occurs as series of sedimentation–erosion couplets that propagate across the tank floor tracking current motion and behavior. During buoyant liftoff, sedimentation beneath the rising plumes often becomes less organized. Auto-correlation analysis of time series of particle concentration is used to characterize the turbulent structures of the currents and indicates that currents quickly partition into a slow-moving upper portion and faster, more concentrated, lower portion. Air entrainment occurs within the upper region. Turbulent structures within the lower region track sedimentation–erosion waves and indicate that eddies control deposition. Importantly, both eddies and sedimentation waves track reversals in flow direction that occur following buoyant liftoff. Further, these results suggest that individual laminations within PDC deposits may record passage of single eddies, thus the duration of individual PDCs may be estimated as the product of the number of laminations and the current's turbulent timescale.
► We model dilute pyroclastic density currents using hot talc powder suspended in air. ► Runout and liftoff distances are controlled by Ri and thermal energy. ► Coignimbrite mass fractionation is controlled by thermal energy. ► Sedimentation occurs as depositional–erosional couplets tracking turbulent eddies. ► Flow reversals occur during buoyant plume liftoff.