Waveform Clustering Research Articles

Acquisition of acoustic emission precursor information for rock masses with a single joint based on clustering-convolutional neural network method

The method for precursor information acquisition based on acoustic emission (AE) data for jointed rock masses is of significant importance for the early warning of dynamic disasters in underground engineering. A clustering-convolutional neural network (CNN) method is proposed, which comprises a clustering component and a CNN component. A series of uniaxial compression tests were conducted on granite specimens containing a persistent sawtooth joint, with different strain rates (10−5–10−2 s−1) and joint inclination angles (0°–50°). The results demonstrate that traditional precursory indicators based on full waveforms are effective for obtaining precursor information of the intact rock failure. However, these indicators are not universally applicable to the failure of rock masses with a single joint. The clustering-CNN method has the potential to be applied to obtain precursor information for all three failure modes (Modes I, II and III). Following the waveform clustering analysis, the effective waveforms exhibit a low main frequency, as well as high energy, ringing count, and rise time. Furthermore, the clustering method and the precursory indicators influence the acquisition of final precursor information. The Birch hierarchical clustering method and the S value precursory indicator can help to obtain more accurate results. The findings of this study may contribute to the development of warning methods for underground engineering across faults.

Prediction of Lithium Oilfield Brines Based on Seismic Data: A Case Study from L Area, Northeastern Sichuan Basin, China

Lithium is an important mineral resource and a critical element in the production of lithium batteries, which are currently in high demand. Oilfield brine has significant value as a raw material for lithium extraction. However, it is often considered a byproduct of oil and gas production and is either abandoned or reinjected underground. Exploration and development of oilfield brines can enhance the economic benefits of oilfields and avoid wasting resources. Current methods for predicting brine distribution rely on geological genetic analysis, which results in low accuracy and reliability. To address this issue, we propose a workflow for lithium brine prediction that uses seismic and logging data. We introduced waveform clustering control and used the mapping relationship between seismic waveforms and well-logging curves to predict high-quality reservoirs based on the electrical and physical properties of lithium brine reservoirs. In this workflow, the seismic waveforms were first clustered using singular value decomposition. The sample sets of well-logging properties were established for the target location. The target properties were divided into high- and low-frequency components and predicted separately. The predicted results of the high-quality reservoirs in the study area were verified using elemental content test results to demonstrate the effectiveness of the method. Our study indicates that well-logging property prediction constrained by waveform clustering can predict lithium brines in a carbonate reservoir.

A cardiologist-like computer-aided interpretation framework to improve arrhythmia diagnosis from imbalanced training datasets

Arrhythmias can pose a significant threat to cardiac health, potentially leading to serious consequences such as stroke, heart failure, cardiac arrest, shock, and sudden death. In computer-aided electrocardiogram interpretation systems, the inclusion of certain classes of arrhythmias, which we term "aggressive" or "bullying," can lead to the underdiagnosis of other "vulnerable" classes. To address this issue, a method for arrhythmia diagnosis is proposed in this study. This method combines morphological-characteristic-based waveform clustering with Bayesian theory, drawing inspiration from the diagnostic reasoning of experienced cardiologists. The proposed method achieved optimal performance in macro-recall and macro-precision through hyperparameter optimization, including spliced heartbeats and clusters. In addition, with increasing bullying by aggressive arrhythmias, our model obtained the highest average recall and the lowest average drop in recall on the nine vulnerable arrhythmias. Furthermore, the maximum cluster characteristics were found to be consistent with established arrhythmia diagnostic criteria, lending interpretability to the proposed method.

A new bandwidth-extended harmonic-tuned Class-E/F3 power amplifier based on continuous mode design philosophy

For Class-E/F3 power amplifiers (PAs) working in the radio frequency (RF) band, the freedom of broadband design is limited because the original design space is a single point. In this paper, a new bandwidth-extended Class-E/F3 PA is proposed, which is based on finite harmonic tuning and continuous mode design philosophy. In the theoretical derivation, the high-efficiency frequency domain equations are solved in conjunction with the inherent conditions of the transistor, resulting in a waveform cluster that can expand the design space while keeping the efficiency constant. This approach eliminates the need for tedious circuit derivation and is able to allow greater design freedom. With the extended Class-E/F3 design space, a broadband PA using GaN HEMT CGH40010F operating in the bandwidth of 1.9–3.3 GHz is designed and tested. The experimental results show that the designed PA demonstrates 65% to 74.8% drain efficiency (DE) and 40.1 to 41.7dBm output power over the target bandwidth.

Probing environmental and tectonic changes underneath Mexico City with the urban seismic field

Abstract. The sediments underneath Mexico City have unique mechanical properties that give rise to strong site effects. We investigated temporal changes in the seismic velocity at strong-motion and broadband seismic stations throughout Mexico City, including sites with different geologic characteristics ranging from city center locations situated on lacustrine clay to hillside locations on volcanic bedrock. We used autocorrelations of urban seismic noise, enhanced by waveform clustering, to extract subtle seismic velocity changes by coda wave interferometry. We observed and modeled seasonal, co- and post-seismic changes, as well as a long-term linear trend in seismic velocity. Seasonal variations can be explained by self-consistent models of thermoelastic and poroelastic changes in the subsurface shear wave velocity. Overall, sites on lacustrine clay-rich sediments appear to be more sensitive to seasonal surface temperature changes, whereas sites on alluvial and volcaniclastic sediments and on bedrock are sensitive to precipitation. The 2017 Mw 7.1 Puebla and 2020 Mw 7.4 Oaxaca earthquakes both caused a clear drop in seismic velocity, followed by a time-logarithmic recovery that may still be ongoing for the 2017 event at several sites or that may remain incomplete. The slope of the linear trend in seismic velocity is correlated with the downward vertical displacement of the ground measured by interferometric synthetic aperture radar, suggesting a causative relationship and supporting earlier studies on changes in the resonance frequency of sites in the Mexico City basin due to groundwater extraction. Our findings show how sensitively shallow seismic velocity and, in consequence, site effects react to environmental, tectonic and anthropogenic processes. They also demonstrate that urban strong-motion stations provide useful data for coda wave monitoring given sufficiently high-amplitude urban seismic noise.

Utilizing 2D seismic forward modeling to constrain the seismic response and plane distribution of grain shoal reservoir in the northern slope of Central Sichuan Paleo-uplift, Sichuan Basin

Carbonate grain shoal reservoirs are well developed in the first member of the Longwangmiao Formation on the northern slope of the Central Sichuan Paleo-uplift. Influenced by the stratal thickness, reservoir heterogeneity, and development patterns, it is difficult to predict reservoirs by using the seismic data present. In this contribution, multiple sets of seismic models were established, and then the seismic forward modeling method was used to constrain the seismic response of the reservoirs. The results show: (1) When reservoirs are not developed in the Longwangmiao Formation, the top interface of the Longwangmiao Formation corresponds to 1/8 λ above the central lobe of the wave peak while the bottom interface corresponds to the central lobe of the wave trough; (2) If a single set of reservoirs is developed in the first member of the Longwangmiao Formation, the waveform of the top interface will be constant, but the bottom trough will shift upward and the frequency will decrease. Additionally, the scale of trough upward movement and frequency reduction is negatively correlated with reservoir velocity and positively correlated with reservoir thickness. (3) If multiple sets of reservoirs are developed in the first member of the Longwangmiao Formation, the waveforms of the top and bottom interface are consistent with that of a single reservoir which has the same total thickness. Similarly, the corresponding relationship between the wave trough variation and reservoir velocity or thickness remains constant. (4) When reservoirs are both developed in the first and second members of the Longwangmiao Formation, the top and bottom waveforms will vary regularly with the difference in reservoir velocity between them. If the reservoir velocity in the first member is less than that of the second member, the top peak is constant while the bottom trough shifts upward, and the frequency decreases. On the contrary, if the reservoir velocity in the second member is lower than that of the first member, the bottom trough is constant while the top peak moves downward, and the frequency increases. Finally, using the waveform clustering method, the reservoir is divided into three grades, namely high-, medium-, and poor-quality. Among them, high-quality reservoirs are mainly distributed in the southeast and northeast of the study area which is the key area for oil and gas exploration in the next step.

Geometry, formation mechanisms, and hydrocarbon exploration significance of Pishan Caldera, Tarim Basin, northwest China

The discovery of an oil reservoir in Pishan Caldera of the Tarim Basin, northwest China, highlights the hydrocarbon exploration potential of large calderas. We used 3 D seismic interpretations, waveform clustering analysis, and seismic profiles to investigate the origin and evolution of Pishan Caldera, along with the mechanisms of hydrocarbon accumulation, and its significance for future oil and gas exploration. Pishan Caldera is located at depths of >7000 m, has an elliptical shape, extends over an area of 25 × 20 km, and is surrounded by ring faults. Pishan Caldera were the result of magma withdrawal and later magma intrusion. Volcanic eruptions, caldera collapse, and reactivation of Pishan Caldera resulted in Cretaceous oil migration and accumulation in Cretaceous breccia reservoirs. Ring faults, dense fracture networks, and lava tubes associated with the caldera acted as conduits for upward hydrocarbon migration. We propose a generalized model for hydrocarbon reservoirs related to the caldera, which includes five major reservoir types: drape anticline, fault block, volcanic-trapped, laterally sealed and lava tube type. Further efforts should be made to identify deep caldera structures that may be targets for hydrocarbon exploration.

Two types of clinical ictal direct current shifts in invasive EEG of intractable focal epilepsy identified by waveform cluster analysis

ObjectiveTo determine clinically ictal direct current (DC) shifts that can be identified by a time constant (TC) of 2 s and to delineate different types of DC shifts by different attenuation patterns between TC of 10 s and 2 s. MethodsTwenty-one patients who underwent subdural electrode implantation for epilepsy surgery were investigated. For habitual seizures, we compared (1) the peak amplitude and (2) peak latency of the earliest ictal DC shifts between TC of 10 s and 2 s. Cluster and logistic regression analyses were performed based on the attenuation rate of amplitude and peak latency with TC 10 s. ResultsIctal DC shifts in 120 seizures were analyzed; 89.1% of which were appropriately depicted even by a TC of 2 s. Cluster and logistic regression analyses revealed two types of ictal DC shift. Namely, a rapid development pattern was defined as the ictal DC shifts with a shorter peak latency and they also showed smaller attenuation rate of amplitude (73/120 seizures). Slow development pattern was defined as the ictal DC shifts with crosscurrent of a rapid development pattern, i.e., a longer peak latency and larger attenuation rate of amplitude (47/120 seizures). Focal cortical dysplasia (FCD) 1A tended to show a rapid development pattern (22/29 seizures) and FCD2A tended to show a slow development pattern (13 /18 seizures), indicating there might be some correlations between two types of ictal DC shift and certain pathologies. ConclusionsIctal DC shifts, especially rapid development pattern, can be recorded and identified by the AC amplifiers of TC of 2 s which is widely used in many institutes compared to that of TC of 10 s. Two types of ictal DC shifts were identified with possibility of corresponding pathology. SignificanceIctal DC shifts can be distinguished by their attenuation patterns.

Seismic-Geological Integrated Study on Sedimentary Evolution and Peat Accumulation Regularity of the Shanxi Formation in Xinjing Mining Area, Qinshui Basin

Accurate identification of the lithofacies and sedimentary facies of coal-bearing series is significant in the study of peat accumulation, coal thickness variation and coal-measured unconventional gas. This research integrated core, logging and 3D seismic data to conduct a comprehensive seismic–geological study on the sedimentary evolution characteristics and peat accumulation regularity of the Shanxi Formation in the Xinjing mining area of the Qinshui Basin. Firstly, the high-resolution sequence interface was identified, and the isochronous stratigraphic framework of the coal-bearing series was constructed. Then, the temporal and spatial evolution of sedimentary filling and sedimentary facies was dynamically analyzed using waveform clustering, phase rotation, stratal slice and frequency–division amplitude fusion methods. The results show that the Shanxi Formation in the study area can be divided into one third-order sequence and two fourth-order sequences. It developed a river-dominated deltaic system, mainly with delta plain deposits, and underwent a constructive–abandoned–constructive development stage. The locally distributed No. 6 coal seam was formed in a backswamp environment with distribution constrained by the distributary channels. The delta was abandoned at the later stage of the SS1 sequence, and the peat accumulation rate was balanced with the growth rate of the accommodation, forming a large-area distributed No. 3 thick coal seam. During the formation of the SS2 sequence, the No. 3 coal seam was locally thinned by epigenetic erosion of the river, and the thin coal belt caused by erosion is controlled by the location of the distributary channels and their extension direction. This study can provide a reference for the research on the distribution of thin sand bodies, sedimentary evolution and peat accumulation regularity in the coal-bearing series under the marine–continental transitional environment.

Research on microseismic event imaging based on waveform clustering analysis

Research on microseismic event imaging based on waveform clustering analysis

Adaptive Gaussian mixture model and convolution autoencoder clustering for unsupervised seismic waveform analysis

Seismic facies analysis can effectively estimate reservoir properties, and seismic waveform clustering is a useful tool for facies analysis. We have developed a deep-learning-based clustering approach called the modified deep convolutional embedded clustering with adaptive Gaussian mixture model (AGMM-MDCEC) for seismic waveform clustering. Trainable feature extraction and clustering layers in AGMM-MDCEC are implemented using neural networks. We fuse the two independent processes of feature extraction and clustering, such that the extracted features are modified simultaneously with the results of clustering. We use a convolutional autoencoder for extracting features from seismic data and to reduce data redundancy in the algorithm. At the same time, weights of clustering network are fine-tuned through iteration to obtain state-of-the-art clustering results. We apply our new classification algorithm to a data volume acquired in western China to map architectural elements of a complex fluvial depositional system. Our method obtains superior results over those provided by traditional K-means, Gaussian mixture model, and some machine-learning methods, and it improves the mapping of the extent of the distributary system.

A Hierarchical Clustering Method of SOM Based on DTW Distance for Variable-Length Seismic Waveform

In seismic facies analysis, waveform clustering of self-organizing map (SOM) usually classifies waveforms truncated by horizon with a fixed time window. However, the fixed time window is not an appropriate method for the situation in which the thickness of the target layer varies in the horizontal direction. In order to adapt SOM to this situation and simultaneously preserve the advantages of SOM, we replace the vanilla SOM measure with dynamic time warp (DTW) distance. Compared with Euclidean distance, DTW distance requires more computational amount about path searching and weight parameter updating in SOM. Therefore, considering the redundancy of seismic data, we introduce a hierarchical clustering strategy, which uses cluster-based stratified sampling and hierarchical mapping to reduce the computational cost of training and prediction. In the visualization section, by combining with hue, saturation, value (HSV) coloring and hierarchical mapping, the method can quickly display the lateral distribution of stratigraphic. The proposed method is validated by a mound shape model in a simulation test. Finally, the method was successfully applied to the field data from a reef-bank reservoir. The experimental and application results show that, when the top and bottom interfaces of the destination layer can be clearly identified, the proposed method can effectively cluster the waveforms with variable length, and the computational cost is acceptable.

Feature Selection and Mislabeled Waveform Correction for Water–Land Discrimination Using Airborne Infrared Laser

The discrimination of water–land waveforms is a critical step in the processing of airborne topobathy LiDAR data. Waveform features, such as the amplitudes of the infrared (IR) laser waveforms of airborne LiDAR, have been used in identifying water–land interfaces in coastal waters through waveform clustering. However, water–land discrimination using other IR waveform features, such as full width at half maximum, area, width, and combinations of different features, has not been evaluated and compared with other methods. Furthermore, false alarms often occur when water–land discrimination in coastal areas is conducted using IR laser waveforms because of environmental factors. This study provides an optimal feature for water–land discrimination using an IR laser by comparing the performance of different waveform features and proposes a dual-clustering method integrating K-means and density-based spatial clustering applications with noise algorithms to improve the accuracy of water–land discrimination through the clustering of waveform features and positions of IR laser spot centers. The proposed method is used for practical measurement with Optech Coastal Zone Mapping and Imaging LiDAR. Results show that waveform amplitude is the optimal feature for water–land discrimination using IR laser waveforms among the researched features. The proposed dual-clustering method can correct mislabeled water or land waveforms and reduce the number of mislabeled waveforms by 48% with respect to the number obtained through traditional K-means clustering. Water–land discrimination using IR waveform amplitude and the proposed dual-clustering method can reach an overall accuracy of 99.730%. The amplitudes of IR laser waveform and the proposed dual-clustering method are recommended for water–land discrimination in coastal and inland waters because of the high accuracy, resolution, and automation of the methods.

Automated Classification of Whole Body Plethysmography Waveforms to Quantify Breathing Patterns

Whole body plethysmography (WBP) monitors respiratory rate and depth but conventional analysis fails to capture the diversity of waveforms. Our first purpose was to develop a waveform cluster analysis method for quantifying dynamic changes in respiratory waveforms. WBP data, from adult Sprague-Dawley rats, were sorted into time domains and principle component analysis was used for hierarchical clustering. The clustering method effectively sorted waveforms into categories including sniffing, tidal breaths of varying duration, and augmented breaths (sighs). We next used this clustering method to quantify breathing after opioid (fentanyl) overdose and treatment with ampakine CX1942, an allosteric modulator of AMPA receptors. Fentanyl caused the expected decrease in breathing, but our cluster analysis revealed changes in the temporal appearance of inspiratory efforts. Ampakine CX1942 treatment shifted respiratory waveforms toward baseline values. We conclude that this method allows for rapid assessment of breathing patterns across extended data recordings. Expanding analyses to include larger portions of recorded WBP data may provide insight on how breathing is affected by disease or therapy.

The Imbricated Foreshock and Aftershock Activities of the Balsorano (Italy) Mw 4.4 Normal Fault Earthquake and Implications for Earthquake Initiation

AbstractForeshocks in the form of microseismicity are among the most powerful tools to study the physical processes that occur before main earthquakes. However, their detection and precise characterization is still sparse, especially for small-to-moderate-size earthquakes (Mw<6). We present here a detailed foreshock analysis for the 7 November 2019, Balsorano, Italy, normal fault earthquake (Mw 4.4). To improve the detection of the microseismicity before and after the mainshock, we use six three-component broadband receivers at distances of less than 75 km from the targeted seismicity, through template matching. To improve the understanding of the physical mechanism(s) behind the earthquake initiation process, as well as other accompanying phenomena, we also detail the spatiotemporal evolution of the sequence associated with this medium-sized earthquake, using waveform clustering and hypocenter relocation. Clear differences between foreshocks and aftershocks are revealed by this analysis. Moreover, five distinct spatiotemporal patterns associated with the different seismic activities are revealed. The observed spatiotemporal behavior shown by the foreshocks highlights a complex initiation process, which apparently starts on an adjacent unmapped antithetic fault. Finally, the aftershock activity comprises four different clusters with distinct spatiotemporal patterns, which suggests that the different clusters in this sequence have distinct triggering mechanisms.

An Improved Water-Land Discriminator Using Laser Waveform Amplitudes and Point Cloud Elevations of Airborne LIDAR

Zhao, X.; Wang, X.; Zhao, J., and Zhou, F., 2021. An improved water-land discriminator using laser waveform amplitudes and point cloud elevations of airborne LIDAR. Journal of Coastal Research, 37(6), 1158–1172. Coconut Creek (Florida), ISSN 0749-0208. Laser waveforms or point clouds of airborne LIDAR are used to distinguish water and land. However, false alarms often occur in coastal areas with complex environments when laser waveforms are used. Elevations of three-dimensional (3D) point clouds can be used to help discriminate ocean and land but fail to identify inland waters. An improved water-land discriminator that uses laser waveform amplitudes and point cloud elevations is proposed in this study. First, 3D point cloud elevations derived using infrared (IR) laser are used as features to conduct ocean-land discrimination with fuzzy c-means clustering. Second, amplitudes of IR laser waveforms are used to identify inland waters from land derived by ocean-land discrimination. The proposed method is applied to data collected using Optech coastal zone mapping and imaging LIDAR (CZMIL). Land boundary derived from digital orthophoto map is used as a reference to evaluate different water-land discriminators. Results showed that consistency of the water-land interface derived using the improved water-land discriminator is higher with the reference interface than that derived by traditional waveform saturation, waveform clustering, and 3D point cloud methods. Overall accuracy of water-land points discriminated via waveform saturation, waveform clustering, 3D point cloud, and improved water-land discriminator is 93.80%, 93.84%, 95.66%, and 99.27%, respectively. High accuracy of the water-land interface and points indicates the effectiveness of the improved water-land discriminator for Optech CZMIL.

A waveform clustering method for damage mode identification for composite laminates under hygrothermal environment

This work performs acoustic emission monitoring of damage behaviors in carbon fiber/ epoxy composite laminates during tensile loads under hygrothermal environment. First, a waveform-based clustering method is introduced by employing the wavelet packet decomposition (WPD), the Shannon’s entropy, the cosine similarity and the clustering algorithm. Second, based on the clustering results, the characteristic features of various damage modes are derived. Finally, further investigation is carried out to explore the effects of hygrothermal aging on the damage source localization, damage initiation and evolution. From the theoretical analysis, several conclusions are obtained: 1. Fiber breakage behaves as a combination of matrix cracking, fiber/matrix interface debonding and delamination in the space of extracted features. 2. The frequency response of each damage mode is demonstrated to be degraded after hygrothermal aging. The principal features of fiber/matrix interface debonding and delamination are found to be significantly affected by hygrothermal aging as compared to other two damage modes. 3. The distortion of the principal feature of each micro-damage mode is found to be negligible with the damage accumulation, indicating the robustness of the method.

Insight on the Crustal Stress State in Faial and Pico Islands (Azores), from Analysis of Aftershocks of the 1998 Earthquake

The seismicity in the Azores plateau can be of volcanic or tectonic origin, the latter usually more destructive, especially in the central group of islands. In this study we aimed to obtain crustal stress indicators from seismic data collected in July 1998, following the July 9th M5.8 earthquake that struck near Faial and Pico islands. The maximum horizontal stress (SHmax) azimuth was obtained from two methods. For the seismogenic volume, the SHmax was obtained from single and composite focal mechanisms, whereas the state of stress beneath the seismic network was derived from shear-wave splitting analysis of local earthquakes. The composite focal mechanisms were obtained from a waveform clustering method applied to identify “identical” earthquakes. The seismic anisotropy analysis could be applied to only six stations, the polarization of the first S arrival being stable, and with a secondary perpendicular pulse arriving 0.6–0.24 s later. The focal mechanisms provided a NW–SE SHmax with azimuth range of 96°–114°, and an associated transition of normal-fault to strike-slip dominant regime consistent with the estimated regional field of 110°–120°, and coherent with movement along three fault systems. Inland the SHmax azimuths present greater variability. In Faial, the stations located north of the Caldeira present a stable azimuth of 151°–153° and are consistent with the effect of a NNW–SSE fault system present in the islands, rotating to 50°–77° in the eastern sector of the Pedro Miguel Graben, probably due to a complex interaction of the WSW–ENE dominant fault system with old eroded volcanic complexes that segmented the island into several blocks. In Pico island, the SHmax direction azimuth ranges from 93° to 135°, in a radial pattern around the island's main stratovolcano, overlapping the regional stress field.

Gabapentin and memantine increases randomness of oscillatory waveform in ocular palatal tremor.

Syndrome of oculopalatal tremor (OPT) causes pendular nystagmus of the eyes and its disabling consequence on the visual system. Classic pharmacotherapeutic studies revealed reduction in the eye velocity of the oscillatory waveforms. Subjective improvement in vision, however, remains out of proportionately low. Elegant models depicting quasi-sinusoidal coarse oscillations of the eyes highlighted two distinct oscillators; one at the inferior olive causing primary 2Hz oscillations, while the second, independent oscillator, at the cerebellum adding the randomness to the waveform. Here we examined whether pharmacotherapy affects the randomness of the oscillatory waveform. Horizontal, vertical, and torsional angular eye positions were measured independently from both eyes as six subjects with OPT directed gaze toward a straight-ahead target. The measurements were performed before administration of alpha-2-delta calcium channel blocker (gabapentin) or NMDA receptor antagonist (memantine) and after the subjects were treated with each of these drugs for at least 8days. Amplitude and velocity of eye oscillations were reduced by gabapentin and memantine, but there was an increase in the waveform randomness. We found that the increase in randomness was proportionate to the amount of reduction in the waveform velocity or amplitude. Hierarchical clustering revealed distinct patterns of oscillatory waveforms, with each subject belonging to a specific cluster group. The pharmacotherapy changed the waveform clustering pattern of the waveform in each subject. We conclude that in addition to incomplete resolution of the oscillation intensity, increased randomness could be one of the reasons why there is not enough clinical difference in the patients' visual quality.

A DTW distance-based seismic waveform clustering method for layers of varying thickness

Seismic waveform clustering is a useful technique for lithologic identification and reservoir characterization. The current seismic waveform clustering algorithms are predominantly based on a fixed time window, which is applicable for layers of stable thickness. When a layer exhibits variable thickness in the seismic response, a fixed time window cannot provide comprehensive geologic information for the target interval. Therefore, we propose a novel approach for a waveform clustering workflow based on a variable time window to enable broader applications. The dynamic time warping (DTW) distance is first introduced to effectively measure the similarities between seismic waveforms with various lengths. We develop a DTW distance-based clustering algorithm to extract centroids, and we then determine the class of all seismic traces according to the DTW distances from centroids. To greatly reduce the computational complexity in seismic data application, we propose a superpixel-based seismic data thinning approach. We further propose an integrated workflow that can be applied to practical seismic data by incorporating the DTW distance-based clustering and seismic data thinning algorithms. We evaluated the performance by applying the proposed workflow to synthetic seismograms and seismic survey data. Compared with the the traditional waveform clustering method, the synthetic seismogram results demonstrate the enhanced capability of the proposed workflow to detect boundaries of different lithologies or lithologic associations with variable thickness. Results from a practical application show that the planar map of seismic waveform clustering obtained by the proposed workflow correlates well with the geological characteristics of wells in terms of reservoir thickness.