Acta Geophysica最新文献

In this paper, a method is presented to classify volcano activity into three classes, namely quiet, strombolian, and paroxysm. The method is based on training a six-layered deep neural network (DNN) model using these signals as inputs (features): time series of the number of distances of infrasonic events, radar backscatter power, RMS of tremor in five stations close to craters of the volcano, tilt derivative, and seismic tremor source depth. The method was tested on the data related to a period of five years, and the results were concluded using indexes of precision, recall, F₁ score, and Cohen's Kappa coefficient were calculated to evaluate the qualification of the classification. Also, the results were compared to Bayesian network (BN), K-nearest neighbors (KNN), and decision tree (DT) methods. Decision learning trees and KNN are popular machine learning algorithms belonging to the class of supervised learning algorithms. They mimic the human level thinking and, differing from neural networks, are not black box models. The comparisons reveal the proposed method, especially in classifying both strombolian and paroxysm classes. This advantage makes the presented method a more reliable tool for practical use in the volcano monitoring control rooms.

The characterization of the various rock types through petrophysical data analysis is essential for comprehending geological processes and enhancing the efficacy of geophysical approaches aimed at mineralization zones. In the present study, a Fuzzy C-Means (FCM) clustering algorithm was employed to automatically classify lithounits within the western sector of the North Singhbhum Mobile Belt based on the petrophysical properties. Laboratory measurements of 326 rock samples from the study area show a wide range of density (~2350–3150 kg/m³) and magnetic susceptibility (10⁻⁵ SI to 10⁻¹ SI) values. Further FCM analysis reveals three distinct clusters: (i) cluster 1 displays high density and low magnetic susceptibility responses and comprises majorly metabasic, phyllite, and mica schist rocks, (ii) cluster 2 shows low density and low magnetic susceptibility characteristics and contains mainly metasedimentary rocks (phyllite, quartzite, and mica schist) and (iii) cluster 3 also primarily encompasses metasedimentary rocks, but it displays the low density and high magnetic susceptibility characteristics. Overlap of rock types in different clusters probably indicates the influence of secondary geological processes on the petrophysical measurements such as metamorphism, alteration, and weathering, which is also supported by the petrographical studies. Overall, the present study demonstrates the potential utility of the FCM algorithm for automatic lithology classification and inferring the associated geological processes from the petrophysical measurements. Furthermore, the correlation between the geophysical and petrophysical clusters highlights the role of petrophysical information in the automatic geological/mineral mapping. However, the complexity in cluster attributes on a detailed scale suggests that future studies in the NSMB should focus on comprehensive multi-parameter petrophysical and geochemical measurements. This approach will help in developing better strategies for 3D geophysical data inversion and resolve the complexities in petrophysical data interpretation.

Normal moveout correction is an essential part of seismic data processing. The accuracy of the result of traditional normal moveout correction methods depends largely on the accuracy of the picked normal moveout correction velocity, which has severe stretching at shallow layers and far-offset distances. However, the problem is usually solved by “mute,” leading to a low stacking number at far offset and a short illumination aperture for exploration. Therefore, a non-stretching normal moveout correction method based on extrapolation interferometry is proposed in this paper. While solving the problem of stretching, it further increases the effective extension length of seismic exploration and improves the coverage number of far-offset reflection points through the conversion between primary and multiple waves. Meanwhile, the introduction of high-order accumulation improves the application range of the method and overcomes the influence of coherent Gaussian noise. In this paper, the method is tested on synthetic data with different noise and applied to two field data. These applications in different data show that the proposed method is a purely data-driven method. The proposed method in this paper does not depend on the accuracy of the velocity picking. It not only achieves non-stretching moveout correction, but also effectively suppresses the effects of random and coherent Gaussian noise.

This study addresses the critical issue of urban flooding caused by stormwater network overflow, necessitating unified and efficient management measures to handle increasing water volumes and the effects of climate change. The proposed approach aims to improve the precision and efficiency of overflow rate predictions by investigating advanced machine learning algorithms, specifically ensemble methods such as gradient boosting and random forest algorithms. The main contribution lies in introducing the SWN-ML approach, which integrates hydraulic simulations using MIKE + with machine learning to predict average overflow rates for various rainfall durations and return periods. Mike + model was calibrated for the only available observed data of water depth at the outlet point during the storm event of February 4, 2019. The datasets for model calibration used in ML models consisted of many input variables such as peak flow, max depth, length, slope, roughness, and diameter and average overflow rate as output variable. Experimental results show that these methods are effective under a variety of scenarios, with the ensemble methods consistently outperforming classical machine learning models. For example, the models exhibit similar performance metrics with an MSE of 0.023, RMSE of 0.15, and MAE of 0.101 for a 2-h rainfall duration and a 10-year return period. Correlation analysis further confirms the strong correlation between ensemble method predictions and MIKE + simulated models, with values ranging between 0.72 and 0.80, indicating their effectiveness in capturing stormwater network dynamics. These results validate the utility of ensemble learning models in predicting overflow rates in flood-prone urban areas. The study highlights the potential of ensemble learning models in forecasting overflow rates, offering valuable insights for the development of early warning systems and flood mitigation strategies.

The Zagros collision zone, located in the southwest of Iran, is experiencing an immoderately large number of seismic hazards caused by the convergence between the two Arabia and microplate of central Iran. The coda Q has been widely used as a vital parameter to investigate the different tectonic features as well as seismic risk assessments. In this study, we have analyzed the spatial variation of coda wave attenuation in the Zagros region, to evaluate different geological features affecting the seismic wave’s propagation. Our dataset comprises 87,295 coda records of about 6421 local earthquakes, with magnitude greater than three recorded by 36 seismic stations in the period of 2006–2020. We have applied a very simple (Q_{c}^{{}}) regionalization method to mapping spatial distribution of (Q_{c}^{{}}) in Zagros area. The spatial distributions of coda have a positive correlation with the tectonically and lithology of the interested area. According to the results, three primary elements have been suggested as major controlling factors of variation of seismic Coda waves in different parts of the Zagros area. These factors include: (1) intra-crustal relamination process (crustal channeling), (2) 12 km thickness of sediment-filled by fluid (oil and gas) and (3) Hormoz salt (salt domes). Our results of coda wave attenuation, coupled with the findings from 3D velocity tomography which revealed significant velocity variations across the Main Zagros Reverse Fault (MZRF), particularly toward the Sanandaj-Sirjan zone, suggests a potential influence of the fault zone on seismic wave propagation characteristics.

In this study, the seismic wave attenuation beneath the Southern Granulite Terrain (SGT), India has been investigated using the coda waves from 22 local earthquakes (2.0 ≤ M_L ≤ 3.6) recorded by 21 broadband seismic stations. The dependence of the attenuation ‘Qc’ on frequency was extracted using the single backscattering model at central frequencies 1.5, 2, 2.5, 4, 5, 8, 10, 12 and 16 Hz. Different lapse time windows, from 10 to 90 s with an interval of 10 s, were used to test the lapse time dependency. The Qc value usually ranges from 150 to 1000 within the frequency range of 2–8 Hz. However, at higher frequencies (12–16 Hz), Qc value variations range from 1500 to 2500. Estimated Qc values are comparatively lower at stations ELP and MVT across frequencies and these lower values are aligned with the previous seismic activity in the surrounding area. This correlation is further supported by the presence of shear zones, lineament, and other northwest-oriented faults, indicating a pronounced level of heterogeneity and complexity in the crust, ultimately contributing to the observed lower Qc values. The estimated Qo values range from 150 to 350, and N values range from 0.70 to 0.95. The observed Qo and N values are slightly lower than those of the other stable continental regions. The significant spatial variation in Qo values observed within the study region may be attributed to the potential existence of pore fluids, as supported by the reported Vp/Vs ratio and shear wave velocity models, which introduce heterogeneities within the crust. Taking into account all other existing studies, along with the current findings, it can be suggested that both scattering (g) and intrinsic attenuation (Qi) factors contribute to the observed attenuation of the study region. The estimated intrinsic and scattering factors align well with the global attenuation model. In the absence of detailed body wave attenuation studies in this region, the frequency-dependent Q relationships developed here are useful for the estimation of earthquake source parameters of the region. These relations may be used for the simulation of earthquake-strong ground motions, which are required for the estimation of seismic hazards and geotechnical and retrofitting analysis of critical structures in the region.

Foehn wind occurrence has generated great interest among researchers because of the destructive power and impact on the local climate. Based on anemometric data provided by a high-mountain station on Kasprowy Wierch in the Polish Tatra Mountains, the characteristics of the potential occurrence of foehn wind (referred to as halny in the Polish Tatras) are presented, including its speed and duration, as well as the frequency of occurrence on a multiannual, annual and daily basis. Halny winds occur most frequently in the cold period of the year (Oct–Feb), with the frequency peaking in November, and sporadically in the summer. The occurrence of foehn winds is strongly dependent on the synoptic situation. Therefore, the main aim of the study was to identify the circulation conditions conducive to their occurrence on the Polish side of the Tatra Mountains. Circulation conditions responsible for foehn formation were analysed using gridded sea level pressure (SLP) data from the NCEP-DOE (National Centers for Environmental Prediction-Department of Energy) reanalyses. The occurrence of foehn wind in the Tatra Mountains is associated with a low pressure system over north-western Europe, and above normal pressure over south-eastern Europe. The location and intensity of the centres of atmospheric influence on foehn days can vary, as indicated by the three types of pressure systems favouring the occurrence of the phenomenon, distinguished by the hierarchical grouping method. In type 1, the cyclonic centre spreads over northern Europe, in type 2 over western Europe and in type 3 over north-western Europe. In types 1 and 3, the air masses come from the south-west, and in type 2 more from the south. Type 3 is characterised by the greatest horizontal pressure gradients among the three circulation types and with the greatest SLP anomalies.

The incremental impacts of climate change on elements within the water cycle are a growing concern. Intricate karst aquifers have received limited attention concerning climate change, especially those with sparse data. Additionally, snow cover has been overlooked in simulating karst spring discharge rates. This study aims to assess climate change effects in a data-scarce karst anticline, specifically Khorramabad, Iran, focusing on temperature, precipitation, snow cover, and Kio spring flows. Utilizing two shared socioeconomic pathways (SSPs), namely SSP2-4.5 and SSP5-8.5, extracted from the CMIP6 dataset for the base period (1991–2018) and future periods (2021–2040 and 2041–2060), the research employs Landsat data and artificial neural networks (ANNs) for snow cover and spring discharge computation, respectively. ANNs are trained using the training and verification periods of 1991–2010 and 2011–2018, respectively. Results indicate projected increases in temperature, between + 1.21 °C (2021–2040 under SSP245) and + 2.93 °C (2041–2060 under SSP585), and precipitation, from + 2.91 mm/month (2041–2060 under SSP585) to + 4.86 mm/month (2021–2040 under SSP585). The ANN models satisfactorily simulate spring discharge and snow cover, predicting a decrease in snow cover between − 4 km²/month (2021–2040 under SSP245) and − 11.4 km²/month (2041–2060 under SSP585). Spring discharges are anticipated to increase from + 28.5 l/s (2021–2040 under SSP245) to + 57 l/s (2041–2060 under SSP585) and from + 12.1 l/s (2021–2040 under SSP585) to + 36.1 l/s (2041–2060 under SSP245), with and without snow cover as an input, respectively. These findings emphasize the importance of considering these changes for the sustainability of karst groundwater in the future.

Drought, which is defined as a decrease in average rainfall amounts, is one of the most insidious natural disasters. When it starts, people may not be aware of it, which is why droughts are difficult to monitor. Scientists have long been working to predict and monitor droughts. For this purpose, they have developed many methods, such as drought indices, one of which is the Standardized Precipitation Index (SPI). In this study, SPI was used to detect droughts, and machine learning algorithms, including support vector machines (SVM), artificial neural networks, random forest, and decision tree, were used to predict droughts. In addition, 3 different statistical criteria, which are correlation coefficient (r), root mean square error (RMSE), and Nash–Sutcliffe efficiency (NSE), were used to investigate model performance values. The wavelet transform (WT) was also applied to improve model performance. One of the areas most impacted by droughts in Turkey is the Konya Closed Basin, which is geographically positioned in the center of the country and is among the top grain-producing regions in Turkey. The Apa Dam is one of the most significant water resources in the area. It provides water to many fertile fields in its vicinity and is affected by droughts which is why it was selected as a study area. Meteorological data, such as monthly precipitation, that could represent the region were obtained between 1955 and 2020 from the general directorate of state water works and the general directorate of meteorology. According to the findings, the M04 model, whose input structure was developed using SPI, various time steps, data delayed up to 5 months, and monthly precipitation data from the preceding month (time t − 1), produced the best results out of all the models examined using machine learning algorithms. Among machine learning algorithms, SVM has achieved the most successful results not only before applying WT but also after WT. The best results were obtained from M04, in which SVM with WT was used (NSE = 0.9942, RMSE = 0.0764, R = 0.9971).

Single-channel sparker seismic reflection systems are currently preferred in offshore geo-engineering studies due to their cost-effectiveness, ease of use in shallow areas, their high-resolution data, and straightforward data processing. However, the distinctive characteristics of sparker data introduce specific challenges in the processing of single-channel seismic datasets. These include (i) unavailability of the stacking process for single-channel seismic data, (ii) inability to derive subsurface velocity distribution from single-channel seismic profiles, (iii) limitations imposed by ghost reflections and bubble effects as well as random noise amplitudes, and (iv) the suitability of only predictive deconvolution for suppressing multiple reflections. Applications demonstrate that the inability to apply the stacking process to single-channel seismic data poses a significant challenge in suppressing both random and coherent noise, and increasing the signal-to-noise (S/N) ratio. The F-X prediction filter has proven highly effective in mitigating random noise in sparker data. Appropriate determination of operator length and prediction lag parameters allows predictive deconvolution to effectively suppress multiple reflections, despite some residual multiple amplitudes in the output. Spiking deconvolution significantly eliminates ghost reflections and bubble effects, enhancing temporal resolution by eliminating the ringy appearance of the input signal. Trace mixing is a crucial data processing step for enhancing sparker data resolution. The method can be applied as weighted mix for random noise suppression or as trimmed mix for suppressing high-amplitude spike-like noises. This study incorporates a comprehensive analysis of the various noise components embedded in sparker seismic data. It delineates the processing flow and parameters utilized to effectively mitigate these specific noise types.