Overview ta3520 Introduction to seismics

Transcription

1 Overview ta3520 Introduction to seismics Fourier Analysis Basic principles of the Seismic Method Interpretation of Raw Seismic Records Seismic Instrumentation Processing of Seismic Reflection Data Vertical Seismic Profiles Practical: Processing practical (with MATLAB)

2 Convolutional model of seismic data In time domain, output is convolution of input and impulse responses where X(t) = S(t) * G(t) * R(t) * A(t) X(t) = seismogram S(t) = source signal/wavelet G(t) = impulse response of earth R(t) = impulse response of receiver A(t) = impulse response of recording-instrument

3 Convolutional model of seismic data In frequency domain, output is multiplication of spectra: where X(ω) = S(ω) G(ω) R(ω) A(ω) X(ω) = seismogram S(ω) = source signal/wavelet G(ω) = transfer function of earth R(ω) = transfer function of receiver A(ω) = transfer function of recording-instrument (transfer function = spectrum of impulse response)

4 Convolutional model of seismic data In time domain, output is convolution of input and impulse responses where X(t) = S(t) * G(t) * R(t) * A(t) X(t) = seismogram S(t) = source signal/wavelet G(t) = impulse response of earth R(t) = impulse response of receiver A(t) = impulse response of recording-instrument

5 Seismic Instrumentation Seismic sources: Airguns VibroSeis Dynamite Seismic detectors: Geophones Hydrophones Seismic recording systems

6 Seismic at sea

7 Seismic at sea

8 Seismic at sea

9 Seismic at sea

10 Seismic source at sea: Airgun

11 Seismic source at sea: Airgun

12 Seismic source at sea: Airgun

13 Airgun: mechanical behaviour

14 Source signals: airgun

15 Source signals S(t): Vibrator source

16 Seismic source on land: VibroSeis

17 Seismic source on land: VibroSeis

18 Seismic source on land: VibroSeis

19 Seismic source on land: VibroSeis

20 VibroSeis: simple mechanical model

21 VibroSeis: mechanical model

22 Source signals S(t): Vibrator source Time domain frequency domain

23 Source signal: δ-pulse Time domain Frequency domain amplitude δ-pulse 1 0 frequency 0 time phase 0 0 frequency

24 Source signal: band-limited δ-pulse Time domain Frequency domain band-limited δ-pulse amplitude 1 frequency 0 time phase 0 frequency

25 Source signal: band-limited sweep (=VibroSeis) Time domain band-limited sweep Frequency domain amplitude 1 0 frequency phase 0 time 0 frequency

26 Source signal: shifted δ-pulse Time domain Frequency domain: exp(-2π i f T) amplitude δ-pulse 1 0 frequency 0 T time Phase = 2 π f T 0 frequency

27 Source signals: Vibrator source For sweep: Higher frequencies, later in time So: 2 π f T non-linear (more quadratic)

28 Source signals: Vibrator source Undoing effect source signal (phase and amplitude): deconvolution Notice that numerator of stabilized deconvolution is correlation

29 Source signals: Vibrator source

30 Seismic source on land: dynamite

31 Dynamite

32 Dynamite

33 Dynamite: model

34 Source signals S(t): Dynamite

35 Source signal: symmetrical signal Time domain Frequency domain symmetrical pulse amplitude 1 frequency 0 time phase 0 frequency

36 symmetrical signal in time = spectrum is purely real, so zero phase

37 Source signal: causal signal (with causal inverse) Causal signal = Amplitude zero before zero time Time domain Causal pulse (with causal inverse) 0 time

38 Source signal: causal signal (with causal inverse) Time domain Frequency domain Causal pulse (with causal inverse) amplitude 1 frequency 0 time phase 0 frequency

39 causal pulse with causal inverse in time = phase spectrum is minimally going through 2π, so minimum-phase Minimum-phase pulse has most of its energy in the beginning

40 Source signals S(t): Dynamite Dynamite signal seen as minimum-phase signal

41 Dynamite: spectrum 2 amplitude Phase (rad) 0-2 Frequency (Hz) Frequency (Hz)

42 Convolutional model of seismic data In time domain, output is convolution of input and impulse responses where X(t) = S(t) * G(t) * R(t) * A(t) X(t) = seismogram S(t) = source signal/wavelet G(t) = impulse response of earth R(t) = impulse response of receiver A(t) = impulse response of recording-instrument

43 Impulse response of earth G(t) Desired for processing Still: undesired events need to be removed

44 Convolutional model of seismic data In time domain, output is convolution of input and impulse responses where X(t) = S(t) * G(t) * R(t) * A(t) X(t) = seismogram S(t) = source signal/wavelet G(t) = impulse response of earth R(t) = impulse response of receiver A(t) = impulse response of recording-instrument

45 Seismic detector on land: geophone (velocity sensor)

46 Seismic detector on land: geophone

47 Geophone

48 Spectrum of geophone R(ω)

49 Spectrum of geophone R(ω)

50 Spectrum of geophone R(ω)

51 Seismic detector at sea: hydrophone (pressure sensor)

52 Hydrophone (pressure sensor)

53 Hydrophones (pressure sensors)

54 Hydrophone model: piezo-electricity

55 Hydrophone: piezo-electric

56 Hydrophone: piezo-electric

57 Spectrum of hydrophone R(ω)

58 Spectrum of hydrophone R(ω)

59 Spectrum of hydrophone R(ω)

60 Convolutional model of seismic data In time domain, output is convolution of input and impulse responses where X(t) = S(t) * G(t) * R(t) * A(t) X(t) = seismogram S(t) = source signal/wavelet G(t) = impulse response of earth R(t) = impulse response of receiver A(t) = impulse response of recording-instrument

61 On-board QC

62 On-board QC

63 Storage: IBM 3592 tapes (right-hand corner above)

64 Seismic recording systems Main tasks: Convert Analog signals to Digital signals Store data

65 Recording Instrument Sample data correctly: Nyquist is determined by setting time-sampling interval Δt: f Nyquist = 1 / (2 Δt) Then: Cut high frequencies such that above f Nyquist analog signal is damped below noise level

66 Recording instrument A(ω): high-cut filter Frequency domain -1/(2Δt) amplitude 1 frequency 1/(2Δt) High-cut filter = Anti-alias filter phase -1/(2Δt) 0 frequency 1/(2Δt)

67 Total response of instrumentation In frequency domain, output is multiplication of spectra: where X(ω) = S(ω) G(ω) R(ω) A(ω) X(ω) = seismogram S(ω) = source signal/wavelet G(ω) = transfer function of earth R(ω) = transfer function of receiver A(ω) = transfer function of recording-instrument (transfer function = spectrum of impulse response)

68 Total response of instrumentation In frequency domain, output is multiplication of spectra: X(ω) = S(ω) G(ω) R(ω) A(ω)

69 amplitude phase Total response of instrumentation