Discrimination of earthquakes and explosions using multi-fractal singularity spectrums properties

A new method of discrimination of seismic records from earthquakes and explosions is proposed which is based on using properties of their multi-fractal singularity spectrums. The efficiency of the method is illustrated by analysing seismic records from the region of Aswan Dam in Egypt. Current pattern of seismicity in the Upper Egypt is composed by three types of records: tectonic earthquakes, reservoir-induced earthquakes and seismic events generated by quarry blasts. To discriminate quarry blasts from earthquakes of both natures, multi-fractal analysis of records were used. Singularity spectrum support width and multi-fractal generalised Hurst exponents were calculated for all seismic events in the selected data set from a given area. The linear Bayesian discriminator using these characteristics of seismic records provides correct classification for 93 % of earthquakes records and for 99 % of quarry blasts records.     

Precursory groundwater level changes in the period of activation of the weak intraplate seismic activity on the NE margin of the Bohemian Massif (central Europe) in 2005

tWe analyse continuous measurements of groundwater level in two deep wells VS-3 and V-28 at the experimental hydro-meteorological station situated on the NE margin of the Bohemian Massif, central Europe, characterized by the weak intraplate seismic activity. The aim of our study is to examine the relationships between changes in the groundwater level and earthquake occurrence. Based on the tidal and barometric response of the water level, we estimated selected elastic parameters of the observed aquifers: the shear modulus G, the Skempton ratio B, the drained matrix compressibility β and the undrained compressibility β_u. Using these parameters and assuming the homogeneous poroelastic material, we derived the sensitivity of the wells to the crustal volume strain. During the observation period from November 1998 to December 2005 we detected in the VS-3 well two pre-seismic steps, related to August 10, 2005 (M = 2.4) and October 25, 2005 (M = 3.3) earthquakes. Amplitudes of the recorded precursory changes (+6 cm and +15 cm) are several times higher than the values predicted from the theoretical precursory crustal strain and the strain sensitivity of the well. Therefore, we presume that the observed pre-seismic water level steps can be attributed to heterogeneity of poroelastic material. We consequently propose the hypothesis of the origin of precursory events based on the presumption of a sensitive site, at which the well is situated.     

Microseismic noise in the low frequency range (Periods of 1–300 min): Properties and possible prognostic features

The paper generalizes the experience accumulated in studies of microseismic noise in the period range from 1 to 300 min observed during time intervals preceding a few strong earthquakes. This frequency range is the least studied and occupies an intermediate position between low frequency seismology and investigations of slow geophysical processes. The range includes oscillations induced by atmospheric and oceanic processes and various modes of the Earth’s free oscillations excited by very strong earthquakes. The main attention in the paper is given to the background behavior of microseisms, which contains continuous present arrivals from near weak and far strong and moderate earthquakes. The paper focuses on the examination of synchronization effects arising in joint multivariate analysis of information from several stations with estimation of multifractal spectra of singularity and multidimensional spectral measures of coherent behavior of singularity spectral parameters. The problem of using the synchronization effects of microseismic background in the search for new precursors of strong earthquakes is discussed.     

Asymmetrical pulses, the periodicity and synchronization of low frequency microseisms

Seismic records of eight IRIS broadband stations were analyzed at distances of 70 to 7160 km from the magnitude 8.3 Hokkaido earthquake of September 25, 2003. The stations situated in the subduction zone recorded asymmetrical microseismic pulses lasting 3–10 min a few days before the earthquake. No such pulses were observed in the records of the stations situated outside the subduction zone. Similar pulses were also recorded before the magnitude 7.8 Kronotskii, Kamchatka earthquake of 1997. The pulses are hypothesized to have been caused by creeping movements. Synchronous oscillations of microseismic noise with periods of 1–3 h were recorded as far as 3000 km from the Hokkaido earthquake a few days before it occurred. The noise coherence measure increased for stations closer to the epicenter. The question of the source of this coherence remains open. These effects belong to the class of those occurring in dissipative metastable systems; parts of the terrestrial lithosphere during the precursory periods of seismic catastrophes seem to be such systems.     

Entropy measure of stepwise component in GPS time series

A new method for estimating the stepwise component in the time series is suggested. The method is based on the application of a pseudo-derivative. The advantage of this method lies in the simplicity of its practical implementation compared to the more common methods for identifying the peculiarities in the time series against the noise. The need for automatic detection of the jumps in the noised signal and for introducing a quantitative measure of a stepwise behavior of the signal arises in the problems of the GPS time series analysis. The interest in the jumps in the mean level of the GPS signal is associated with the fact that they may reflect the typical earthquakes or the so-called silent earthquakes. In this paper, we offer the criteria for quantifying the degree of the stepwise behavior of the noised time series. These criteria are based on calculating the entropy for the auxiliary series of averaged stepwise approximations, which are constructed with the use of pseudo-derivatives.     

Mapping the properties of low-frequency microseisms for seismic hazard assessment

The structure of low-frequency seismic noise in the range of periods from 2 min to 500 min is studied from the data of continuous seismic monitoring at 77 seismic stations of the F-net broadband network in Japan from the beginning of 1997 to May 15, 2012. A new statistical characteristic of seismic noise is suggested, namely, the minimal normalized entropy En of the distribution of squared orthogonal wavelet coefficients. This parameter of seismic noise is analyzed in conjunction with the multifractal statistics—the support width of the singularity spectrum, Δα, and the generalized Hurst exponent, α*, which were extensively used by the author in the previous works for analyzing the low-frequency seismic noise. The method for constructing the maps of spatial distribution of Δα, α*, En, and their aggregated normalized value over the time windows with a given length is proposed. The maps are constructed by averaging the succession of the elementary charts, each of which corresponds to a day of observations. It is shown that, for the islands of Japan, the reduction in Δα and α* and the increase in En outline the area of the forthcoming mega earthquake of March 11, 2011, with M = 9 (Tohoku earthquake). According to the analysis of about a year’s worth of data after this event, the region south of Tokyo (Nankai trough) is still dominated by decreased Δα and α* and increased En. This gives grounds to hypothesize that this region remains at a high level of seismic threat since the accumulated stresses were incompletely released by the Tohoku earthquake. Drawing an analogy to the behavior of the coefficient of correlation between Δα and α*, we may suppose that there is an increased probability of a strong earthquake occurring in the second half of 2013 or the first half of 2014. Constructing the averaged maps of the distributions of seismic noise parameters and their aggregated value in a moving time window is suggested as a new method for dynamical assessment of seismic hazards.     

Coherence between the fields of low-frequency seismic noise in Japan and California

The relationship is considered between the statistics of the field of low-frequency seismic noise which was synchronously recorded by two broadband seismic networks in Japan (78 stations) and California (81 stations). The analysis is based on the data for seven years of observations (2008–2014). For each network, the daily time series of the median values are constructed for five parameters of seismic noise: kurtosis (excess), minimal normalized entropy of the distribution of the squared wavelet coefficients, generalized Hurst exponent, support width of the singularity spectrum, and index of linear predictability. The median values for each parameter were calculated on a daily basis over all the stations of the networks and resulted in a time series containing 2557 data points of the integral characteristics of the noise with a daily time step. The use of the median values of the noise parameters avoids considering the effects of the gaps in recording by individual stations and provides the continuous time series as the integral characteristic of the whole network. Next, for each network, the aggregate signals were calculated for the obtained five-variate time series. By construction, the aggregate signal is a scalar signal which maximally accumulates the most general variations that are simultaneously present in all the analyzed signals and, at the same time, rejects the components that are only characteristic of a single process. The final step of the analysis consists in estimating the evolution of the quadratic spectrum in the moving time window with a length of one year. It is shown that during the considered interval of the observations, the coherence is characterized by the increasing linear trend, which independently supports the previous conclusion about the enhancement of the synchronization between the parameters of the global seismic noise.     

Large-scale aseismic creep in the areas of the strong earthquakes revealed from the GRACE data on the time variations of the Earth’s gravity field

The refinement of the accuracy and resolution of the monthly global gravity field models from the GRACE satellite mission, together with the accumulation of more than a decade-long series of these models, enabled us to reveal the processes that occur in the regions of large (M_w≥8) earthquakes that have not been studied previously. The previous research into the time variations of the gravity field in the regions of the giant earthquakes, such as the seismic catastrophes in Sumatra (2004) and Chile (2010), and the Tohoku mega earthquake in Japan (2011), covered the coseismic gravity jump followed by the long postseismic changes reaching almost the same amplitude. The coseismic gravity jumps resulting from the lower-magnitude events are almost unnoticeable. However, we have established a long steady growth of gravity anomalies after a number of such earthquakes. For instance, in the regions of the subduction earthquakes, the growth of the positive gravity anomaly above the oceanic trench was revealed after two events with magnitudes M_w=8.5 in the Sumatra region (the Nias earthquake of March 2005 and the Bengkulu event of September 2007 near the southern termination of Sumatra Island), after the earthquake with M_w=8.5 on Hokkaido in September 2007, a doublet Simushir earthquake with the magnitudes M_w = 8.3 and 8.1 in the Kuriles in November 2006 and January 2007, and after the earthquake off the Samoa Island in September 2009 (M_w=8.1). The steady changes in the gravity field have also been recorded after the earthquake in the Sichuan region (May 2008, M_w = 8.0) and after the doublet event with magnitudes 8.6 and 8.2, which occurred in the Wharton Basin of the Indian Ocean on April 11, 2012. The detailed analysis of the growth of the positive anomaly in gravity after the Simushir earthquake of November 2006 is presented. The growth started a few months after the event synchronously with the seismic activation on the downdip extension of the coseismically ruptured fault plane zone. The data demonstrating the increasing depth of the aftershocks since March 2007 and the approximately simultaneous change in the direction and average velocity of the horizontal surface displacements at the sites of the regional GPS network indicate that this earthquake induced postseismic displacements in a huge area extending to depths below 100 km. The total displacement since the beginning of the growth of the gravity anomaly up to July 2012 is estimated at 3.0 m in the upper part of the plate’s contact and 1.5 m in the lower part up to a depth of 100 km. With allowance for the size of the region captured by the deformations, the released total energy is equivalent to the earthquake with the magnitude M_w = 8.5. In our opinion, the growth of the gravity anomaly in these regions indicates a large-scale aseismic creep over the areas much more extensive than the source zone of the earthquake. These processes have not been previously revealed by the ground-based techniques. Hence, the time series of the GRACE gravity models are an important source of the new data about the locations and evolution of the locked segments of the subduction zones and their seismic potential.