Gisola: A High-Performance Computing Application for Real-Time Moment Tensor Inversion

Automatic moment tensor (MT) determination is essential for real-time seismological applications. In this article, Gisola, a highly evolved software for MT determination, oriented toward high-performance computing, is presented. The program employs enhanced algorithms for waveform data selection via quality metrics, such as signal-to-noise ratio, waveform clipping, data and metadata inconsistency, long-period disturbances, and station evaluation based on power spectral density measurements in parallel execution. The inversion code, derived from ISOLated Asperities—an extensively used manual MT retrieval utility—has been improved by exploiting the performance efficiency of multiprocessing on the CPU and GPU. Gisola offers the ability for a 4D spatiotemporal adjustable MT grid search and multiple data resources interconnection to the International Federation of Digital Seismograph Networks Web Services (FDSNWS), the SeedLink protocol, and the SeisComP Data Structure standard. The new software publishes its results in various formats such as QuakeML and SC3ML, includes a website suite for MT solutions review, an e-mail notification system, and an integrated FDSNWS-event for MT solutions distribution. Moreover, it supports the ability to apply user-defined scripts, such as dispatching the MT solution to SeisComP. The operator has full control of all calculation aspects with an extensive and adjustable configuration. MT’s quality performance, for 531 manual MT solutions in Greece between 2012 and 2021, was measured and proved to be highly efficient.

INSTANCE – the Italian seismic dataset for machine learning

The Italian earthquake waveform data are collected here in a dataset suited for machine learning analysis (ML) applications. The dataset consists of nearly 1.2 million three-component (3C) waveform traces from about 50 000 earthquakes and more than 130 000 noise 3C waveform traces, for a total of about 43 000 h of data and an average of 21 3C traces provided per event. The earthquake list is based on the Italian Seismic Bulletin (http://terremoti.ingv.it/bsi, last access: 15 February 2020​​​​​​​) of the Istituto Nazionale di Geofisica e Vulcanologia between January 2005 and January 2020, and it includes events in the magnitude range between 0.0 and 6.5. The waveform data have been recorded primarily by the Italian National Seismic Network (network code IV) and include both weak- (HH, EH channels) and strong-motion (HN channels) recordings. All the waveform traces have a length of 120 s, are sampled at 100 Hz, and are provided both in counts and ground motion physical units after deconvolution of the instrument transfer functions. The waveform dataset is accompanied by metadata consisting of more than 100 parameters providing comprehensive information on the earthquake source, the recording stations, the trace features, and other derived quantities. This rich set of metadata allows the users to target the data selection for their own purposes. Much of these metadata can be used as labels in ML analysis or for other studies. The dataset, assembled in HDF5 format, is available at http://doi.org/10.13127/instance (Michelini et al., 2021).

Development Of A Machine Learning Model Using Electrocardiogram Signals To Improve Acute Pulmonary Embolism Screening

Aims Clinical scoring systems for pulmonary embolism (PE) screening have low specificity and contribute to CT pulmonary angiogram (CTPA) overuse. We assessed whether deep learning models using an existing and routinely collected data modality, electrocardiogram (ECG) waveforms, can increase specificity for PE detection. Methods and Results We create a retrospective cohort of 21,183 patients at moderate- to high-suspicion of PE and associate 23,793 CTPAs (10.0% PE-positive) with 320,746 ECGs and encounter-level clinical data (demographics, comorbidities, vital signs, and labs). We develop three machine learning models to predict PE likelihood: an ECG model using only ECG waveform data, an EHR model using tabular clinical data, and a Fusion model integrating clinical data and an embedded representation of the ECG waveform. We find that a Fusion model (area under receiver-operating characteristic [AUROC] 0.81 ± 0.01) outperforms both the ECG model (AUROC 0.59 ± 0.01) and EHR model (AUROC 0.65 ± 0.01). On a sample of 100 patients from the test set, the Fusion model also achieves greater specificity (0.18) and performance (AUROC 0.84 ± 0.01) than four commonly evaluated clinical scores: Wells' Criteria, Revised Geneva Score, Pulmonary Embolism Rule-Out Criteria, and 4-Level Pulmonary Embolism Clinical Probability Score (AUROC 0.50-0.58, specificity 0.00-0.05). The model is superior to these scores on feature sensitivity analyses (AUROC 0.66 to 0.84) and achieves comparable performance across sex (AUROC 0.81) and racial/ethnic (AUROC 0.77 to 0.84) subgroups. Conclusion Synergistic deep learning of electrocardiogram waveforms with traditional clinical variables can increase the specificity of PE detection in patients at least at moderate suspicion for PE.

The CIELO Seismic Experiment

The Crust and lithosphere Investigation of the Easternmost expression of the Laramide Orogeny was a two-year deployment of 24 broadband, compact posthole seismometers in a linear array across the eastern half of the Wyoming craton. The experiment was designed to image the crust and upper mantle of the region to better understand the evolution of the cratonic lithosphere. In this article, we describe the motivation and objectives of the experiment; summarize the station design and installation; provide a detailed accounting of data completeness and quality, including issues related to sensor orientation and ambient noise; and show examples of collected waveform data from a local earthquake, a local mine blast, and a teleseismic event. We observe a range of seasonal variations in the long-period noise on the horizontal components (15–20 dB) at some stations that likely reflect the range of soil types across the experiment. In addition, coal mining in the Powder River basin creates high levels of short-period noise at some stations. Preliminary results from Ps receiver function analysis, shear-wave splitting analysis, and averaged P-wave delay times are also included in this report, as is a brief description of education and outreach activities completed during the experiment.

Stochastic Simulation of Strong Ground Motions From Two M > 5 Uttarakhand Earthquakes

Many studies based on the geodetic data and statistical analysis of seismicity have pointed out that sufficient amount of stress accumulated in the Himalayan plate boundary may host a big earthquake. Consequently, high seismic activities and infrastructural developments in the major cities around Himalayan regions are always of major concern. The ground motion parameter estimation plays a vital role in the near real time evaluation of potentially damaged areas and helps in mitigating the seismic hazard. Therefore, keeping in mind the importance of estimation of ground motion parameters, we targeted two moderate-size earthquakes that occurred recently within a gap of 10 months in Uttarakhand region with M > 5.0 on 06/02/2017 and 06/12/2017. The ground motions are simulated by adopting a stochastic modeling technique. The source is assumed as ω−2, a circular point source (Brune’s model). The average value of reported anelastic attenuation from various studies, the quality factor, Qs = 130.4*(f0.996), and stress drop values obtained through iterative procedure are considered for simulations. The stochastic spectra are generated between 0.1 and 10 Hz of frequency range. The site effect is also estimated by using the H/V method in the same frequency range. The synthetic spectra are compared with the observed Fourier amplitude spectra obtained from the recorded waveform data and converted back to the time histories. The stochastic time histories are compared with the observed waveforms and discussed in terms of amplitude (PGA). The simulated and observed response spectra at different structural periods are also discussed. The mismatch between the observed and simulated PGA values along with the GMPE existing for shallow crustal earthquakes is also discussed in the present work.

Seismic Moment Tensor Solutions from GeoNet Data to Provide a Moment Magnitude Scale for New Zealand

<p>The aim of this project is to enable accurate earthquake magnitudes (moment magnitude, MW) to be calculated routinely and in near real-time for New Zealand earthquakes. This would be done by inversion of waveform data to obtain seismic moment tensors. Seismic moment tensors also provide information on fault-type. I use a well-established seismic moment tensor inversion method, the Time-Domain [seismic] Moment Tensor Inversion algorithm (TDMT_INVC) and apply it to GeoNet broadband waveform data to generate moment tensor solutions for New Zealand earthquakes. Some modifications to this software were made. A velocity model can now be automatically used to calculate Green's functions without having a pseudolayer boundary at the source depth. Green's functions can be calculated for multiple depths in a single step, and data are detrended and a suitable data window is selected. The seismic moment tensor solution that has either the maximum variance reduction or the maximum double-couple component is automatically selected for each depth. Seismic moment tensors were calculated for 24 New Zealand earthquakes from 2000 to 2005. The Global CMT project has calculated CMT solutions for 22 of these, and the Global CMT project solutions are compared to the solutions obtained in this project to test the accuracy of the solutions obtained using the TDMT_INVC code. The moment magnitude values are close to the Global CMT values for all earthquakes. The focal mechanisms could only be determined for a few of the earthquakes studied. The value of the moment magnitude appears to be less sensitive to the velocity model and earthquake location (epicentre and depth) than the focal mechanism. Distinguishing legitimate seismic signal from background seismic noise is likely to be the biggest problem in routine inversions.</p>

Seismic Moment Tensor Solutions from GeoNet Data to Provide a Moment Magnitude Scale for New Zealand

<p>The aim of this project is to enable accurate earthquake magnitudes (moment magnitude, MW) to be calculated routinely and in near real-time for New Zealand earthquakes. This would be done by inversion of waveform data to obtain seismic moment tensors. Seismic moment tensors also provide information on fault-type. I use a well-established seismic moment tensor inversion method, the Time-Domain [seismic] Moment Tensor Inversion algorithm (TDMT_INVC) and apply it to GeoNet broadband waveform data to generate moment tensor solutions for New Zealand earthquakes. Some modifications to this software were made. A velocity model can now be automatically used to calculate Green's functions without having a pseudolayer boundary at the source depth. Green's functions can be calculated for multiple depths in a single step, and data are detrended and a suitable data window is selected. The seismic moment tensor solution that has either the maximum variance reduction or the maximum double-couple component is automatically selected for each depth. Seismic moment tensors were calculated for 24 New Zealand earthquakes from 2000 to 2005. The Global CMT project has calculated CMT solutions for 22 of these, and the Global CMT project solutions are compared to the solutions obtained in this project to test the accuracy of the solutions obtained using the TDMT_INVC code. The moment magnitude values are close to the Global CMT values for all earthquakes. The focal mechanisms could only be determined for a few of the earthquakes studied. The value of the moment magnitude appears to be less sensitive to the velocity model and earthquake location (epicentre and depth) than the focal mechanism. Distinguishing legitimate seismic signal from background seismic noise is likely to be the biggest problem in routine inversions.</p>

User perceptions of different monitor modalities during high-fidelity simulation: Visual-Patient-avatar, Split Screen and Conventional - a qualitative analysis (Preprint)

BACKGROUND Patient safety during anaesthesia is crucially dependent on the monitoring of vital signs. However, the values obtained must also be perceived and correctly classified by the attending care providers. To facilitate these processes, we developed Visual-Patient-avatar- an animated virtual model of the monitored patient, which innovatively presents numerical and waveform data following user-centred design principles. After a high-fidelity simulation study, we analysed participants' perceptions of three different monitor modalities, including this new technique. OBJECTIVE After a high-fidelity simulation study, we analysed participants' perceptions of three different monitor modalities, including this new technique. METHODS This study was a researcher-initiated, single-centre, qualitative study. We asked 92 care providers right after finishing three simulated emergency scenarios about their positive and negative opinions concerning the different monitor modalities. Following qualitative research methods, we processed the field notes obtained and derived main categories and corresponding subthemes. RESULTS We gained a total of 307 statements. Visual-Patient-avatar was the most occurring term in both positive and negative responses. We identified three main categories and divided them into eleven positive and negative subthemes. In assigning the statements to one of the topics, we achieved substantial inter-rater reliability. Most of the statements concerned the design and usability features of the avatar, respectively, the Split Screen mode. CONCLUSIONS This study qualitatively reviewed the clinical applicability of the Visual-Patient-avatar technique in a high-fidelity simulation study and revealed strengths and limitations of the avatar only und Split Screen modality. We received valuable suggestions for improving the design. The requirement of training before clinical implementation was reinforced. The responses regarding the Split Screen suggested that this symbiotic modality generates improved situation awareness combined with numerical data and accurate curves.