Scribd Logo
Upload

Welcome to Scribd!

  • 6 views

    Eqorrscan Resume

    Uploaded by

    clara janie frica
    RT-EQcorrscan uses matched filtering to detect seismic events by comparing continuous seismic data to predefined template waveforms and computing cross-correlations between them. It stacks the cross-correlations to generate network sums and detects events using a fixed threshold or one based on the median absolute deviation of the sums. This allows it to detect overlapping events and reduces sensitivity to amplitude variations, making it suitable for aftershock detection. The limitations of RT-EQcorrscan for near-real-time detection include latency under 2 minutes typically, sensitivity to amplitude variations, increased computational load with more channels and templates, and the need to limit templates for large networks to maintain near-real-time performance. Hardware configuration and template threshold selection should

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd

    Eqorrscan Resume

    Uploaded by

    clara janie frica
    6 views2 pages
    RT-EQcorrscan uses matched filtering to detect seismic events by comparing continuous seismic data to predefined template waveforms and computing cross-correlations between them. It stacks the cross-correlations to generate network sums and detects events using a fixed threshold or one based on the median absolute deviation of the sums. This allows it to detect overlapping events and reduces sensitivity to amplitude variations, making it suitable for aftershock detection. The limitations of RT-EQcorrscan for near-real-time detection include latency under 2 minutes typically, sensitivity to amplitude variations, increased computational load with more channels and templates, and the need to limit templates for large networks to maintain near-real-time performance. Hardware configuration and template threshold selection should

    Copyright

    © © All Rights Reserved

    Available Formats

    DOCX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    RT-EQcorrscan uses matched filtering to detect seismic events by comparing continuous seismic data to predefined template waveforms and computing cross-correlations between them. It stacks the cross-correlations to generate network sums and detects events using a fixed threshold or one based on the median absolute deviation of the sums. This allows it to detect overlapping events and reduces sensitivity to amplitude variations, making it suitable for aftershock detection. The limitations of RT-EQcorrscan for near-real-time detection include latency under 2 minutes typically, sensitivity to amplitude variations, increased computational load with more channels and templates, and the need to limit templates for large networks to maintain near-real-time performance. Hardware configuration and template threshold selection should

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    6 views2 pages

    Eqorrscan Resume

    Uploaded by

    clara janie frica
    RT-EQcorrscan uses matched filtering to detect seismic events by comparing continuous seismic data to predefined template waveforms and computing cross-correlations between them. It stacks the cross-correlations to generate network sums and detects events using a fixed threshold or one based on the median absolute deviation of the sums. This allows it to detect overlapping events and reduces sensitivity to amplitude variations, making it suitable for aftershock detection. The limitations of RT-EQcorrscan for near-real-time detection include latency under 2 minutes typically, sensitivity to amplitude variations, increased computational load with more channels and templates, and the need to limit templates for large networks to maintain near-real-time performance. Hardware configuration and template threshold selection should

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    of 2

    RT-EQcorrscan uses matched-filtering methods to detect seismic events.

    Matched-filtering is a
    template-matching technique that compares continuous seismic data with pre-defined template
    waveforms to identify similar events. The software computes fully normalized cross-correlations
    between the data and template waveforms, using frequency-domain cross-correlation chunking for
    memory efficiency. The cross-correlations are aligned based on template moveouts and stacked to
    generate network cross-correlation sums. Detection is then performed on these sums using either a
    fixed threshold or a threshold based on a multiple of the median absolute deviation (MAD) of the
    sum. This approach allows RT-EQcorrscan to detect earthquakes with overlapping waveforms and
    reduces sensitivity to variations in background seismic amplitudes, making it well-suited for
    detecting and characterizing aftershock sequences [1][4].

    The limitations and computational load considerations of using RT-EQcorrscan for near-real-time
    earthquake detection include:

    1. Latency: The overall latency of the RT-EQcorrscan system is the sum of multiple latencies, including
    data latency and the detection interval specified by the user. The detection interval is adjusted
    automatically to be at least the processing time for the template set in use. For sets of 1000-4000
    templates running on systems with an 8-12 thread CPU, the latency of the entire system is typically
    less than 2 minutes [2].

    2. Amplitude Variations: Cross-correlations used in matched-filtering are sensitive to amplitude


    variations. This can be problematic when data arrive in small chunks with strong variations in
    amplitude, such as in aftershock sequences. While this problem can be overcome, the benefits of
    real-time cross-correlations are reduced compared to a delayed approach [2].

    3. Computational Load: The computational load of RT-EQcorrscan depends on the number of


    channels and templates used. Processing time and peak memory use increase with the number of
    channels and templates. For example, running 403 templates on a quad-core Intel i7 processor took
    approximately 30 seconds for each detection step and used less than 2 GB of RAM. Running a similar
    process with 1000 templates consumed less than 3 GB of RAM [5].

    4. Hardware Considerations: The computational efficiency of RT-EQcorrscan can be improved by


    increasing the number of available CPU cores. More CPU cores can accelerate the correlations, but
    there is a limit to the performance improvement. For large networks, the number of templates used
    should be limited to avoid exceeding the near-real-time limit [1].

    These limitations and computational load considerations should be taken into account when using
    RT-EQcorrscan for near-real-time earthquake detection to ensure efficient and accurate results.
    Pertimbangan ambang batas template dalam RT-EQcorrscan dapat bervariasi tergantung pada
    preferensi pengguna dan karakteristik data seismik yang sedang diproses. Ambang batas template
    dapat ditentukan berdasarkan nilai ambang tetap atau berdasarkan kelipatan deviasi mutlak median
    (MAD) dari jumlah cross-korelasi.

    Penggunaan ambang batas tetap memungkinkan pengguna untuk secara manual menentukan nilai
    ambang yang sesuai dengan data yang sedang diproses. Sebagai contoh, dalam satu studi, ambang
    batas tetap diatur pada rata-rata korelasi tunggal saluran sebesar 0,3 di setidaknya lima stasiun [1].

    Alternatifnya, pengguna juga dapat menggunakan ambang batas berdasarkan kelipatan MAD. MAD
    adalah statistik yang tahan terhadap nilai ekstrem, dan dapat digunakan untuk mengatasi variasi
    amplitudo yang signifikan dalam korelasi pendek. Namun, perlu diingat bahwa distribusi nilai MAD
    dapat bervariasi secara signifikan tergantung pada panjang korelogram, terutama selama urutan
    dengan laju tinggi. Oleh karena itu, perlu berhati-hati dalam menggunakan ambang batas MAD untuk
    deteksi near-real-time [3].

    Pilihan ambang batas template ini dapat disesuaikan dengan kebutuhan pengguna dan karakteristik
    data seismik yang sedang diproses.

    You might also like