Limits...
Seismic tomography of the area of the 2010 Beni-Ilmane earthquake sequence, north-central Algeria.

Abacha I, Koulakov I, Semmane F, Yelles-Chaouche AK - Springerplus (2014)

Bottom Line: This sequence, which lasted several months, was triggered by conjugate E-W reverse and N-S dextral faulting.These high values may indicate high fluid contents in the aftershock area.These fluids could have been released from deeper levels by fault movements during earthquakes and migrated rapidly upwards.

View Article: PubMed Central - PubMed

Affiliation: Centre de Recherche en Astronomie, Astrophysique et Geophysique, BP. 63, Bouzareah, Alger, Algeria.

ABSTRACT
The region of Beni-Ilmane (District of M'sila, north-central Algeria) was the site of an earthquake sequence that started on 14 May 2010. This sequence, which lasted several months, was triggered by conjugate E-W reverse and N-S dextral faulting. To image the crustal structure of these active faults, we used a set of 1406 well located aftershocks events and applied the local tomography software (LOTOS) algorithm, which includes absolute source location, optimization of the initial 1D velocity model, and iterative tomographic inversion for 3D seismic P- and S-wave velocities (and the Vp/Vs ratio), and source parameters. The patterns of P-wave low-velocity anomalies correspond to the alignments of faults determined from geological evidence, and the P-wave high-velocity anomalies may represent rigid blocks of the upper crust that are not deformed by regional stresses. The S-wave low-velocity anomalies coincide with the aftershock area, where relatively high values of Vp/Vs ratio (1.78) are observed compared with values in the surrounding areas (1.62-1.66). These high values may indicate high fluid contents in the aftershock area. These fluids could have been released from deeper levels by fault movements during earthquakes and migrated rapidly upwards. This hypothesis is supported by vertical sections across the study area show that the major Vp/Vs anomalies are located above the seismicity clusters.

No MeSH data available.


Related in: MedlinePlus

Examples of reconstructions of four different synthetic models defined along sections 1 and 2 with the position indicated in Figure6. Red stars are epicenters of the three main shocks
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4230316&req=5

Fig5: Examples of reconstructions of four different synthetic models defined along sections 1 and 2 with the position indicated in Figure6. Red stars are epicenters of the three main shocks

Mentions: To assess the resolution of the results, we performed a series of synthetic tests for which we simulated the conditions of the observed data inversion. In particular, we used identical source–receiver configurations as in the real data experiment. The synthetic travel times were computed using a 3D ray tracing algorithm based on the bending approach. After computing the synthetic data, we ignored the coordinates of the “true” sources, and performed a full calculation workflow, including the source locations, as in the real data case. Thus, in this modeling, we faced the problem of a trade-off between source and velocity parameters, both of which affect the results of passive-source tomography.Figure 3 presents the results of inversion for two checkerboard models composed of 3 × 3 (model “Board 1”) and 2 × 2 (model “Board 2”) patterns. The sizes of anomalies for these two models are 2 km and 3 km, and the spacings are 1 km and 2 km, respectively. The amplitudes of anomalies in all cases were ±5%. In these cases, the anomalies are unlimited with depth. The results of the checkerboard reconstruction show that in the focal area (Figure 3, center of the map), where most earthquakes are located, the boundaries between anomalies in the 2 × 2 case can be clearly discerned. The outer limits of the anomalies are strongly smeared due to the dominant orientations of rays. The test with the 3 × 3 model shows that the 2-km size anomaly can be resolved only in the center of the aftershock area, but the outer anomalies are strongly biased.To further investigate the effect of smearing, we performed 12 different tests with single synthetic anomalies of 3 × 3 km lateral size placed in different parts of the study region. In all these cases, the amplitude of anomalies was 5%, and the anomalies were unlimited with depth. Examples of reconstructions for six of these models are presented in Figure 4. For the case of locating the pattern in the aftershock area (model “Single_1”), the reconstruction reveals the limits of the anomaly correctly. However, for all other cases of placing the patterns outside the central area, the reconstructed anomalies are strongly smeared in the radial directions. This smearing effect should be taken into account when considering the results of data inversion.It is well known than the vertical resolution of passive tomography is usually poorer than the horizontal resolution, which can be explained by the effect of the trade-off between seismic velocity and source parameters. This effect is demonstrated in a series of tests presented in Figure 5. In this case, we defined several synthetic patterns across two vertical sections used for presenting the main results (Figure 6). All these eight plots represent different synthetic models, four for each of two sections; thickness of the anomaly in the direction across the section is 5 km. Here we present the reconstruction results for the P-anomalies. We analyzed different depth intervals of the anomalies, and the purpose of the test was to check whether the existing observation system is able to resolve any differences between these models. In fact, the reconstruction results shown in Figure 5 display a very strong vertical smearing of anomalies, which does not allow the upper and lower limits of the synthetic anomalies to be resolved. These tests show that care is needed when interpreting any vertical changes of structures in real data results.The results of tomographic inversion, which include P- and S-wave velocity anomalies, the Vp/Vs ratio, and the locations of sources, are presented in horizontal and vertical sections in Figures 6 and7, respectively. We show the anomalies only in areas located at distances of less than 4 km from a nearest node of the parameterization grid which was set according to the ray density. Some anomalies in areas outside the network are related to the ray paths travelling to the remote stations of the Algerian network which were also involved in this study (Figure 8). Note that these outside patterns might be strongly smeared and thus should be considered with prudence. Both the P- and S-wave velocity structures appear to differ in shape from those in our model. The P-wave velocity structures are represented by contrasting patterns with radial orientations. However, as was shown by the synthetic tests, these shapes could be artifacts due to smearing outside the aftershock area. The S-wave velocity model is represented primarily by a single low-velocity anomaly in the central part of the study area that is surrounded by high-velocity anomalies. Interestingly, the amplitudes of the S-wave velocity anomalies are around 5%, which is lower than the value for the P-wave velocity anomalies (7%–9%). The S-wave velocity model shows a much larger reduction in variance and a smaller rms of remnant residuals after inversion than does the P-wave velocity model. Thus, the lower amplitude in the S-wave velocity model cannot be due to poorer quality of the S-wave data and consequent stronger damping in the inversion. On the basis of several inversion trials, we conclude that the obtained amplitudes are realistic. The Vp/Vs ratio model is obtained from the division of the resulting P- and S-wave velocities. This model displays a clear feature of high Vp/Vs ratio values, up to 1.78, in the area of the fault. In the surrounding areas, values are generally low (Vp/Vs =1.62–1.66).The resulting images of horizontal and vertical sections do not show significant variation in structure with depth. However, as was shown by the synthetic tests in Figure 5, the existing data do not allow a robust resolution of the vertical variation.Figure 3


Seismic tomography of the area of the 2010 Beni-Ilmane earthquake sequence, north-central Algeria.

Abacha I, Koulakov I, Semmane F, Yelles-Chaouche AK - Springerplus (2014)

Examples of reconstructions of four different synthetic models defined along sections 1 and 2 with the position indicated in Figure6. Red stars are epicenters of the three main shocks
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4230316&req=5

Fig5: Examples of reconstructions of four different synthetic models defined along sections 1 and 2 with the position indicated in Figure6. Red stars are epicenters of the three main shocks
Mentions: To assess the resolution of the results, we performed a series of synthetic tests for which we simulated the conditions of the observed data inversion. In particular, we used identical source–receiver configurations as in the real data experiment. The synthetic travel times were computed using a 3D ray tracing algorithm based on the bending approach. After computing the synthetic data, we ignored the coordinates of the “true” sources, and performed a full calculation workflow, including the source locations, as in the real data case. Thus, in this modeling, we faced the problem of a trade-off between source and velocity parameters, both of which affect the results of passive-source tomography.Figure 3 presents the results of inversion for two checkerboard models composed of 3 × 3 (model “Board 1”) and 2 × 2 (model “Board 2”) patterns. The sizes of anomalies for these two models are 2 km and 3 km, and the spacings are 1 km and 2 km, respectively. The amplitudes of anomalies in all cases were ±5%. In these cases, the anomalies are unlimited with depth. The results of the checkerboard reconstruction show that in the focal area (Figure 3, center of the map), where most earthquakes are located, the boundaries between anomalies in the 2 × 2 case can be clearly discerned. The outer limits of the anomalies are strongly smeared due to the dominant orientations of rays. The test with the 3 × 3 model shows that the 2-km size anomaly can be resolved only in the center of the aftershock area, but the outer anomalies are strongly biased.To further investigate the effect of smearing, we performed 12 different tests with single synthetic anomalies of 3 × 3 km lateral size placed in different parts of the study region. In all these cases, the amplitude of anomalies was 5%, and the anomalies were unlimited with depth. Examples of reconstructions for six of these models are presented in Figure 4. For the case of locating the pattern in the aftershock area (model “Single_1”), the reconstruction reveals the limits of the anomaly correctly. However, for all other cases of placing the patterns outside the central area, the reconstructed anomalies are strongly smeared in the radial directions. This smearing effect should be taken into account when considering the results of data inversion.It is well known than the vertical resolution of passive tomography is usually poorer than the horizontal resolution, which can be explained by the effect of the trade-off between seismic velocity and source parameters. This effect is demonstrated in a series of tests presented in Figure 5. In this case, we defined several synthetic patterns across two vertical sections used for presenting the main results (Figure 6). All these eight plots represent different synthetic models, four for each of two sections; thickness of the anomaly in the direction across the section is 5 km. Here we present the reconstruction results for the P-anomalies. We analyzed different depth intervals of the anomalies, and the purpose of the test was to check whether the existing observation system is able to resolve any differences between these models. In fact, the reconstruction results shown in Figure 5 display a very strong vertical smearing of anomalies, which does not allow the upper and lower limits of the synthetic anomalies to be resolved. These tests show that care is needed when interpreting any vertical changes of structures in real data results.The results of tomographic inversion, which include P- and S-wave velocity anomalies, the Vp/Vs ratio, and the locations of sources, are presented in horizontal and vertical sections in Figures 6 and7, respectively. We show the anomalies only in areas located at distances of less than 4 km from a nearest node of the parameterization grid which was set according to the ray density. Some anomalies in areas outside the network are related to the ray paths travelling to the remote stations of the Algerian network which were also involved in this study (Figure 8). Note that these outside patterns might be strongly smeared and thus should be considered with prudence. Both the P- and S-wave velocity structures appear to differ in shape from those in our model. The P-wave velocity structures are represented by contrasting patterns with radial orientations. However, as was shown by the synthetic tests, these shapes could be artifacts due to smearing outside the aftershock area. The S-wave velocity model is represented primarily by a single low-velocity anomaly in the central part of the study area that is surrounded by high-velocity anomalies. Interestingly, the amplitudes of the S-wave velocity anomalies are around 5%, which is lower than the value for the P-wave velocity anomalies (7%–9%). The S-wave velocity model shows a much larger reduction in variance and a smaller rms of remnant residuals after inversion than does the P-wave velocity model. Thus, the lower amplitude in the S-wave velocity model cannot be due to poorer quality of the S-wave data and consequent stronger damping in the inversion. On the basis of several inversion trials, we conclude that the obtained amplitudes are realistic. The Vp/Vs ratio model is obtained from the division of the resulting P- and S-wave velocities. This model displays a clear feature of high Vp/Vs ratio values, up to 1.78, in the area of the fault. In the surrounding areas, values are generally low (Vp/Vs =1.62–1.66).The resulting images of horizontal and vertical sections do not show significant variation in structure with depth. However, as was shown by the synthetic tests in Figure 5, the existing data do not allow a robust resolution of the vertical variation.Figure 3

Bottom Line: This sequence, which lasted several months, was triggered by conjugate E-W reverse and N-S dextral faulting.These high values may indicate high fluid contents in the aftershock area.These fluids could have been released from deeper levels by fault movements during earthquakes and migrated rapidly upwards.

View Article: PubMed Central - PubMed

Affiliation: Centre de Recherche en Astronomie, Astrophysique et Geophysique, BP. 63, Bouzareah, Alger, Algeria.

ABSTRACT
The region of Beni-Ilmane (District of M'sila, north-central Algeria) was the site of an earthquake sequence that started on 14 May 2010. This sequence, which lasted several months, was triggered by conjugate E-W reverse and N-S dextral faulting. To image the crustal structure of these active faults, we used a set of 1406 well located aftershocks events and applied the local tomography software (LOTOS) algorithm, which includes absolute source location, optimization of the initial 1D velocity model, and iterative tomographic inversion for 3D seismic P- and S-wave velocities (and the Vp/Vs ratio), and source parameters. The patterns of P-wave low-velocity anomalies correspond to the alignments of faults determined from geological evidence, and the P-wave high-velocity anomalies may represent rigid blocks of the upper crust that are not deformed by regional stresses. The S-wave low-velocity anomalies coincide with the aftershock area, where relatively high values of Vp/Vs ratio (1.78) are observed compared with values in the surrounding areas (1.62-1.66). These high values may indicate high fluid contents in the aftershock area. These fluids could have been released from deeper levels by fault movements during earthquakes and migrated rapidly upwards. This hypothesis is supported by vertical sections across the study area show that the major Vp/Vs anomalies are located above the seismicity clusters.

No MeSH data available.


Related in: MedlinePlus