Seismic Velocity Changes Detected By The Noise-based Surface And Reflected Waves

Traditional ambient noise theory suggests that the coda wave in the empirical Green’s function has a longer and more complex path than the direct wave,so it is more sensitive to changes in the underground medium.Therefore,the coda wave has been widely used to detect the weak seismic velocity changes caused by different factors(earthquakes,volcanoes,landslides,etc.)in the past 20 years.However,the coda is mainly composed of the scattered wave,whose overly complex paths makes it difficult to establish as to how the velocity change depends on depth.How to understand these complex geophysical processes and obtain seismic velocity variations at different depths? Based on the basic theory of ambient noise cross-correlation and auto-correlation,this paper attempts to use surface wave and high-frequency reflected wave to detect changes associated to the volcanic activities,rock slopes,and glacier,respectively.The new understandings are listed in the blow:This thesis used volcanic tremor(1–2.5 s)and oceanic noise(larger than 3 s)as passive sources to recover the seismic signals between seismic stations in Hawaii Island using the ambient noise cross-correlation method.By calculating the time delay of Rayleigh surface waves at different periods,this paper obtained the wave velocity changes at different depths before and after the eruption of Kilauea volcano in 2018.Based on the simulation results of the seismic velocity changes in multi-periods,our results have showed that,the velocity of the narrow magma dike(at a depth of 1-2 km)below the East Rift Zone decreased by at least 5% in the co-eruption time,it might be associated to the inflation of the magma.Previous study calculated the damage model before the eruption of Kilauea volcano in 2018,and reported that the seismic velocity should be reduced by 1%-1.5% from March 1 to May 4).However,limited by the spatial resolution of the coda wave,they only observed velocity increase of the surrounding rock due to magma expansion before the eruption.This thesis obtained the velocity changes at different depths and revealed more details before the eruption.Our result show that the velocity of the magma channel decreased of 1% before the eruption,but the wave velocity increased by 0.2-1% in the surrounding areas at depth of 0-1 km and 2-8 km.At the same time,losts of micro-earthquakes occurred in the magma channel,and GPS observations also showed that the ground uplifted about 20 cm.Our observations suggest that magma intrusion may open the magma dike at 1-2km depth but induce higher compressive stress in the surrounding areas.According to the Granular model,this thesis estimate that a 1% increase in velocity before eruption at depths of 2-8 km would require a stress change of 3.6-14.4 MPa,which is also close to the estimation by previous studies that magma intrusion in the Hawaiian region。The seismic signal recovered by the ambient noise autocorrelation is mainly composed of the reflected waves below and around the seismic station.This paper finds that the high-frequency signal in the time window of [0,0.1 s] can be used to extract seismic changes in the shallow layer(0-25 m),with the highest time resolution of up to 10 min.By waveform smilating and comparing the measurement results in different components,different frequencies and different stations,our method is proven to be is stable and effective.This thesis applied this method to detect a rocky landslide in Changning,Sichuan,and it is found that robust hourly velocity variations that correlated with air temperature(with an 11-hour time delay),atmospheric pressure,and rainfall.This implies that elastic strains of these three factors(1 × 10 to 4 × 10)control the accumulated diurnal dv/v fluctuations of ～6% near the surface.(1)This thesis found that there is an obvious positive correlation between the air temperature and the seismic change of the shallow layer of the landslide body,with a time delay of 11 hours.However,petrophysical experiments have shown that increasing the temperature causes the velocity decrease.How could this discrepancy be explained? The total thermo-strain contains the thermobody strain and thermoelastic strain.The velocity reduction observed in laboratory experiments can be explained by the heating-induced thermo-expansion,which mainly occurs at a very shallow depth of 0-0.4 m for the daily change and at 0-2 m for the annual change on the Earth’s surface.The daily velocity changes(deeper than 0.4 m)value positively correlated with temperature,which was mainly caused by the thermoelastic effect(Ben-Zion and Allam,2013).Because the daily cyclical temperature fluctuation produces “daily thermoelastic wave” and spreads several kilometers deep,it increases and decreases seismic wave velocities by compression and expansion in the rocks at those depths.The time delay between temperature and velocity change is related to the thickness of the diurnal thermosphere.Through calculations,this thesis finds that the delay of 11 hours corresponds to its thickness of 0.36 m,which is also in line with the previous estimate of 0.4 m.(2)Another major factor affecting the seismic wave velocity of elastic media is the effective pressure change of the instantaneous interaction,which includes the velocity increase caused by normal stress and the velocity decrease caused by the change of pore pressure.The results of this paper show that the changes in the properties of rock fractures in shallow layers are not only affected by thermoelastic effects,but also related to changes in atmospheric pressure.When the atmospheric pressure increases by 2 × 10 ,it can cause the elastic strain change of 2～3.2 × 10,which is similar to the temperature-induced strain range of 2～3.2 × 10.(3)The simulation results also show that the pore pressure and pore elastic strain caused by rainfall are mainly affected by the depth of the pore elastic layer.The seismic velocity at the bottom of the slope decreased by about 6%during the rainfall period(about 10 mm),but it is limit for rock slope.This thesis suggest that this phenomenon can be explained by the structure of the landslide.In general,the sediments material will accumulate at the bottom of the landslide body,while the rock slope body is mainly dominated by bedrock.(4)After the two earthquakes(M4.0 and M4.5),the seismic velocity of the rock slope decreased by 2%and 4%,respectively,and then recovered rapidly within a few days.The physical mechanism of the recovery time mainly depends on the damage magnitude,and the recovery process of the larger fracture scales requires longer time scales.The study area of this thesis is far away from the epicenters of the two earthquakes,which may reflect the shallow basement damage of rock slopes accompanied by the sloshing and subsequent healing caused by strong ground.It is limited to very shallow depth,and the scales can be very small,so recover quickly within a few days.Increasing temperatures are known to cause melting and accelerated sliding of glaciers,but the detailed responses of the microstructures inside glaciers to this complex processing are still not fully understood.Here,this thesis uses seismic noise autocorrelation to extract hourly velocity changes(dv/v)from a dense seismic array of97 seismic stations deployed on Argentière glacier(French Alps)during the 2018 early spring melting season.(1)The results show that the velocity the Argentière Glacier decreases sharply with the increase of solar radiation,which may be related to opening micro-cracks caused by the thermal expansion of the glacier.Measurement results of different stations also show that this effect mainly occurs in the seismic gap region in the central part of the glacier,and the region with low correlation between velocity changes and radiation values may be related to the large ice crevasse near the glacier sides.When thermal expansion occurs inside the glacier,these ice crevasse are simultaneously subjected to the expansion stress of the glacier and the resisting stress of the surrounding mountains,which may increase the effective stress and make it difficult to open.After thermal expansion,the velocity changes of the glacier recovered rapidly within a few hours,and the recovery curves were consistent with the damage recovery mechanism of the subsurface medium observed in the laboratory and in the field.By fitting the characteristic recovery time,this paper finds that as the maximum velocity decline increases,the glacier takes more time to recover.This paper suggests that it is possible to estimate the damage degree of the subsurface medium by the characteristic time.(2)In the P2 stage(accelerated sliding),the velocity change of the Argentière Glacier was significantly negatively correlated with the acceleration(CC=-0.73),that is,the seismic velocity decreased associated to the glacier accelerated sliding,and vice versa.This phenomenon may be related to the inertial stress generated by the non-uniform sliding of the glacier,the stress changes caused by glacial non-uniform slide is estimated by-4 × 10 m/h,corresponding to stress change of 1.35 MPa.In addition,the region with low correlation between the velocity change and glacier acceleration at different stations corresponds well to the occurrence region of the basement stick-slip event.Previous studies have shown that the occurrence of a stick-slip event must satisfy the effective stress N=Pi-Pw(Pi and Pw represent the pressure of ice and water,respectively)exceeding the critical value.Specifically,if the effective pressure is too low,friction reduction will never become enough to generate stick-slip seismicity.Therefore,this thesis speculates that the micro-cracks at the front of the glacier may be in a closed state under the action of strong stress,so the influence of the sliding stress is limited in this region.(3)During the P3 stage(rapid melting period),the average velocity changes of Argentière Glacier decreased by 4%,which can be explained by the mechanism of pore water pressure increase,which can reduce the effective stress and thus lead to the seismic velocity decrease.The low correlation between velocity changes and subglacial drainage in the left front edge of the Argentière Glacier may be related to the Moraine material with low seismic wave velocity.Moraine refers to deposits with high porosity composed of debris transported by glaciers in different forms.Pressurized water inside the glacier can easily penetrate these high porosity materials to the bottom of the glacier,so the groundwater table in these areas.Both height and pore pressure drop sharply at the same time resulting in a low correlation with glacier drainage,suggesting that the noise autocorrelation method can capture the hydrological structure of glacier with high spatial and temporal resolution.