Automatic Denoising of Seismograms using Fingerprints: Algorithms, Properties, Limitations

Capa

Citar

Texto integral

Resumo

This article describes the new technique for automatically preparing a noisy seismic record for further analysis using expert information systems. The technique is based on seismogram fingerprints, which, due to their concise but informative pattern, allow the development of a reliable algorithm for finding important noise characteristics. The search for the optimal cutoff frequency for a high-pass filter is especially important under conditions of partial overlap of the signal and noise spectra at a high intensity of the latter. It is precisely this difficult case that this study aims to address. The article analyzes the developed methodology on the example of several hundred registrations of regional earthquakes and explosions. The analysis showed that reliable results can be achieved in more than 90% of cases. In addition to all the problems and limitations of the method, which are an extension of its capabilities, are mentioned. Appendix to the article contains detailed description of the algorithm underlying the method.

Sobre autores

K. Silkin

Geophysical Survey of the RAS

Autor responsável pela correspondência
Email: const.silkin@ya.ru
Rússia, Lenina ave., 189, Obninsk, 249035

Bibliografia

  1. Адушкин В.В., Спивак А.А. Влияние трассы на затухание сейсмического сигнала от короткозамедленных карьерных взрывов // Динамические процессы в геосферах. 2013. № 4. С. 118–126.
  2. Асминг В.Э., Баранов С.В. Использование спектрально- и частотно-временных вариаций сейсмических сигналов для различия землетрясений и взрывов в Евро-арктическом регионе // Современные методы обработки и интерпретации сейсмологических данных. Материалы Международной сейсмологической школы. Обнинск: ГС РАН, 2006. С. 23–27.
  3. Асминг В.Э., Кременецкая Е.О., Виноградов Ю.А. Евтюгина З.А. Использование критериев идентификации взрывов и землетрясений для уточнения оценки сейсмической опасности региона // Вестник МГТУ. Труды Мурманского государственного технического университета. 2010. Т. 13. № 4–2. С. 998–1007. EDN NQZHFN
  4. Асминг В.Э., Федоров А.В. Возможности применения автоматического детектора-локатора сейсмических событий по одиночной станции для детальных сейсмологических наблюдений // Сейсмические приборы. 2014. Т. 50. № 3. С. 19–29. EDN SLRRSN
  5. Баврина А.П., Борисов И.Б. Современные правила применения корреляционного анализа // Медицинский альманах. 2021. № 3(68). С. 70–79. EDN TPSSIX
  6. Попов С.Е., Замараев Р.Ю. Веб-сервис классификации сейсмических событий на базе системы распределенных вычислений Apache Spark // Обработка пространственных данных в задачах мониторинга природных и антропогенных процессов (SDM-2019) // Сборник трудов Всероссийской конференции с международным участием, Бердск, 26–30 августа 2019 года. Бердск: ИВТ СО РАН, 2019. С. 440–450. EDN ESWFQV
  7. Силкин К.Ю. Использование вейвлет-анализа как дополнительного критерия при идентификации природы сейсмического события // Структура, вещественный состав, свойства, современная геодинамика и сейсмичность платформенных территорий и сопредельных регионов: материалы XXII Всероссийской с международным участием научно-практической Щукинской конференции (Воронеж, 22–25 сентября 2020 г.) / Под ред. Л.И. Надежка, Т.Б. Силкиной. Воронеж: ВГУ, 2020. С. 328–334. EDN VHXKWM
  8. Силкин К.Ю. Новые эвристики на основе вейвлет-анализа записи одиночного датчика для распознавания землетрясений и взрывов // Сейсмические приборы. 2022а. Т. 58. № 3. С. 5–24. https://doi.org/10.21455/si2022.3-1 EDN HFBDZH
  9. Силкин К.Ю. Двоичные отпечатки результатов вейвлет-анализа записи сейсмического события как инструмент формирования его компактного образа для целей нейросетевого распознавания // Российский сейсмологический журнал. 2022б. Т. 4. № 4. C. 42–55. https://doi.org/10.35540/2686-7907.2022.4.03 EDN EMMMDC
  10. Силкин К.Ю. Двоичные отпечатки сейсмической записи как средство для автоматической фильтрации низкочастотного шума // Сейсмические приборы. 2023. Т. 59. № 2. С. 26–45. https://doi.org/10.21455/si2023.2-3 EDN BYNTAQ
  11. Тур А.И., Кокоулин А.Н., Южаков А.А. Возможности применение алгоритмов нечеткого поиска и нейронных сетей в технологии Fingerprint // Нейрокомпьютеры: разработка, применение. 2017. № 6. С. 45–48. EDN ZICARH
  12. Фортуна-Сервантес Х.М., Рамирес-Торрес М.Т., Мартинес-Карранса Х., Мургуиа-Ибарра Х.С., Ме-хиа-Карлос М. Обнаружение объектов в аэронавигации с использованием вейвлет-преобразования и сверхточных нейронных сетей: первый подход // Труды Института системного программирования РАН. 2021. Т. 33. № 2. С. 149–162. https://doi.org/10.15514/ISPRAS-2020-33(2)-9 EDN MYCXZQ
  13. Abma R., Claerbout J. Lateral prediction for noise attenuation by t-x and f-x techniques // Geophysics. 1995. V. 60. № 6. P. 1887–1896. https://doi.org/10.1190/1.1443920
  14. Baluja Sh., Covell M. Waveprint: Efficient wavelet-based audio fingerprinting // Pattern Recognition. 2008. V. 41. № 11. P. 3467–3480. https://doi.org/10.1016/j.patcog.2008.05.006
  15. Bonar D., Sacchi M. Denoising seismic data using the nonlocal means algorithm // Geophysics. 2012. V. 77. № 1. P. A5–A8. https://doi.org/10.1190/geo2011-0235.1
  16. Bergen K.J., Beroza G.C. Earthquake fingerprints: Extracting waveform features for similarity-based earthquake detection // Pure and Applied Geophysics. 2019. V. 176. P. 1037–1059. https://doi.org/10.1007/s00024-018-1995-6
  17. Cano P., Batlle E., Kalker T., Haitsma J. A Review of Audio Fingerprinting // Journal of VLSI Signal Processing Systems for Signal, Image and Video Technology. 2005. V. 41. № 3. P. 271–284.
  18. Chen Z., Stewart R. A multi-window algorithm for real-time automatic detection and picking of P-phases of microseismic events // CREWES Research Report. 2006. V. 18. P. 1–9.
  19. Daubechies I., Lu J., Wu H.-T. Synchrosqueezed Wavelet Transforms: An Empirical Mode Decomposition-like Tool // Appl. Comput. Harmon. Analysis. 2011. V. 30. № 2. P. 243–261. https://doi.org/10.1016/j.acha.2010.08.002
  20. Ferrari G.M. Un saber “sudamericano”. La dactiloscopia en el Congreso Científico Latinoamericano, 1901–1909 // Historia Crítica. 2016. № 60. P. 81–101. https://doi.org/10.7440/histcrit60.2016.05
  21. Fragoulis D., Rousopoulos G., Panagopoulos T., Alexiou C., Papaodysseus C. On the automated recognition of seriously distorted musical recordings // IEEE Transactions on Signal Processing. 2001. V. 49. № 4. P. 898–908.
  22. Iatsenko D., McClintock P.V.E., Stefanovska A. Extraction of instantaneous frequencies from ridges in time-frequency representations of signals // Signal Processing. 2016. V. 125. P. 290–303. https://doi.org/10.1016/j.sigpro.2016.01.024
  23. Jensen A., la Cour-Harbo A. Ripples in mathematics: The Discrete wavelet transform. Berlin, Germany: Springer Science & Business Media Publ., 2011. 246 p. https://doi.org/10.1007/978-3-642-56702-5
  24. Haitsma J., Kalker T. A Highly robust audio fingerprinting system: Proceedings of ISMIR 2002, 3rd International Conference on Music Information Retrieval. Paris, France, 2002. P. 107–115.
  25. Hussain I., Fadhil R., Abdulsatar N.M. An image compression using block truncation coding with two types of wavelet transform (HAAR & DWT) // Test Engineering and Management. 2020. № 83. P. 16707–16713.
  26. Herrera R., Tary J.B., Baan M., Eaton D. Body Wave Separation in the Time-Frequency Domain // IEEE Geoscience and Remote Sensing Letters. 2014. № 12. P. 364–368. https://doi.org/10.1109/LGRS.2014.2342033
  27. Mallat S. A wavelet tour of signal processing: the sparse way. Cambridge, UK: Academic Press, 2008. 109 p.
  28. Mallat S.G. A Theory of multiresolution signal decomposition: The wavelet representation // IEEE Transactions on Pattern Analysis and Machine Intelligence. 1989. № 11. P. 674–693. https://doi.org/10.1109/34.192463
  29. Mousavi S.M. Microseismic monitoring and denoising / Diss. … Doctor. of Philosophy. Memphis, 2017. 188 p. https://doi.org/10.13140/RG.2.2.16996.27529
  30. Mousavi S.M., Langston C.A. Hybrid Seismic Denoising Using Higher-Order Statistics and Improved Wavelet Block Thresholding // Bull. Seismol. Soc. Am. 2016. V. 106. P. 1380–1393. https://doi.org/10.1785/0120150345
  31. Mousavi S.M., Langston Ch., Horton S. Automatic denoising and detection of microseismic events using the synchrosqueezing // SEG Technical Program Expanded Abstracts. 2016a. P. 825–829. https://doi.org/10.1190/segam2016-13262052.1
  32. Mousavi S.M., Langston Ch., Horton S. Automatic microseismic denoising and onset detection using the synchrosqueezed continuous wavelet transform // Geophysics. 2016b. V. 81. № 4. P. V341–V355. https://doi.org/10.1190/geo2015-0598.1
  33. Ranjan R., Kumar P., Naik K., Singh V.K. The HAAR-the JPEG based image compression technique using singular values decomposition // 2022 2nd International Conference on Emerging Frontiers in Electrical and Electronic Technologies (ICEFEET). Patna, India. 2022. P. 1–6. https://doi.org/10.1109/ICEFEET51821.2022.9848400
  34. Shi Y., Zhang D., Ji H., Dai R. Application of Synchrosqueezed Wavelet Transform in Microseismic Monitoring of Mines // IOP Conference Series: Earth and Environmental Science. 2019. № 384. Art. ID 012075. https://doi.org/10.1088/1755-1315/384/1/012075
  35. Tary J.B., Herrera R.H., van der Baan M. Analysis of time-varying signals using continuous wavelet and synchro squeezed transforms // Philos. Trans. R. Soc. A: Math., Phys. Eng. Sci. 2018. V. 376. № 2126. Art. ID 20170254. https://doi.org/10.1098/rsta.2017.0254
  36. Wang D.-w., Li Y.-j., Zhang K., Xu H.-m. An Adaptive Time-Frequency Filtering Method for Nonstationary Signals Based on the Generalized S-Transform // Optoelectron. Lett. 2010. V. 6. № 2. P. 133–136. https://doi.org/10.1007/s11801-010-9250-0
  37. Wang A.L.-Ch. An industrial strength audio search algorithm // Proceedings of ISMIR 2003, 4th International Conference on Music Information Retrieval. Baltimore, Maryland, USA, 2003. P. 7–13.
  38. Yoon C.E., O’Reilly O., Bergen K.J., Beroza G.C. Earthquake detection through computationally efficient similarity search // Science Advances. 2015. V. 1. № 11. P. 1–13. https://doi.org/10.1126/sciadv.1501057
  39. Yu G., Yu M., Xu Ch. Synchroextracting Transform // IEEE Transactions on Industrial Electronics. 2017. V. 64. № 10. P. 8042–8054. https://doi.org/10.1109/TIE.2017.2696503
  40. Zeng Z., Lu T., Han P., Zhang D., Yang X.-H., Shi Y., Chang Y., Zhang J., Dai R., Ji H. Microseismic data denoising in the sychrosqueezed domain by integrating the wavelet coefficient thresholding and pixel connectivity // Geophysical Journal International. 2022. V. 232. № 2. P. 1113–1128. https://doi.org/10.1093/gji/ggac378
  41. Zhang D., Zeng Z., Shi Y., Chang Y., Dai R., Ji H., Han P. An Effective Denoising Method Based on Cumulative Distribution Function Thresholding and its Application in the Microseismic Signal of a Metal Mine with High Sampling Rate (6 kHz) // Frontiers in Earth Science, 2022. V. 10. № 7. Art. ID 933284. https://doi.org/10.3389/feart.2022.933284

Arquivos suplementares

Arquivos suplementares
Ação
1. JATS XML
Baixar
2. Fig. 1. Distribution over the epicenter distance of the number of analyzed events depending on the type of source.

Baixar (102KB)
Metadados
3. Fig. 2. The main elements of binary fingerprint technology. a – recording of the BHZ channel of the LVZ station of the earthquake on April 1, 2020 at 04h44m27c: green and turquoise lines are the marks of the entry of P and S waves; b – an energogram according to the recording waveletogram, the time axis is relative, the color scale is the logarithm of the amplitude values, the arrow “S” shows the superimposed amplitude spectrum of the recording, the horizontal axis the values of which are red at the top, the vertical axis corresponds to the axis of the energogram, the arrow with the designation “fB – fN” marks the expert interval of the permissible limits of the background boundary (horizontal blue dotted lines) and points to its middle; c – binary prints of the energogram; d, e – redundant binary prints (d – sliding wavelet spectrum, horizontal the axis is the frequency of the wavelet decomposition, the vertical axis is the position of the scanning frequency window, the color scale of values is the conditional density of prints, the arrows “LP” and “PP” mark the left and right bands for calculating the sections of the sliding screen

Baixar (721KB)
Metadados
4. Fig. 3. Histogram of the width distribution of the expert frequency range containing the permissible value of the upper noise limit for optimal filtering.

Baixar (75KB)
Metadados
5. Fig. 4. Histograms of the distribution of the quality function of the results of the binary fingerprint algorithm divided into classes. The symbols contain the types of calculated quality estimates for: PRE–binary prints (simple); IDO SV From the N – right profile of the sliding wavelet spectrum; IDO SV From the L – similarly to the left profile; IDO SV O – profile of the sliding wavelet envelope; IDO General – the final generalization of the results from different sides of the IDO.

Baixar (87KB)
Metadados
6. Fig. 5. The scattering diagram of the quality values of the results of simple binary prints (UP to) and the final generalized redundant binary prints (IDO).

Baixar (320KB)
Metadados
7. Fig. 6. Histograms of the distribution of the quality values of the solution results for simple prints (a) and redundant (b) at different ratios of the energy of the useful signal and low-frequency noise (gradations in symbols) divided into quality classes. The quality classes correspond to Fig. 3.

Baixar (120KB)
Metadados
8. Fig. 7. Illustration of the technique for determining the characteristic spectral parameters of background recording. a is a waveletogram of recording before the entry of the P-wave, the color scale is the amplitude in microns/s; b is the average wavelet spectrum, the vertical axis coincides with the waveletogram (1 is a graph of the spectrum, 2 is an approximating polynomial for the main maximum of the spectrum, 3 is the mark of the 1 percent level, 4 is the frequency of the beginning of attenuation the predominant spectral component of the background).

Baixar (540KB)
Metadados
9. Figure 8. An example of a profile graph for redundant binary prints. The vertical axis is the conditional density of prints. 1 – the profile graph; 2 – the extremes of the graph that were removed at the stage of eliminating minor features; 3 – the extremes left; 4 – the point of the strongest negative gradient; 5 – the midpoint between the strongest extremes; 6 – the point of the first minimum (of those left) after the strongest gradient.

Baixar (187KB)
Metadados

Declaração de direitos autorais © Russian Academy of Sciences, 2024