A comparison between airguns and explosives as wide-angle seismic sources

A comparison between airguns and explosives aswide-angle seismic sources1

R.K. Staples,2 R.W. Hobbs3 and R.S. White2

Abstract

The relative merits of a 48-gun, 9324 cu. in. (153 litre) airgun array and a 200 kgexplosive source are considered for the purposes of long-range (0–400 km) refractionseismic work, with particular reference to traveltime modelling. Theoretical sourcecalculations indicate that in the frequency range 2.5–12.0 Hz, the airgun source willproduce an RMS pressure , 8% of that produced by the explosive source and an initialburst pressure ,17% of that produced by the explosive source. Observed data supportthese calculations at short ranges and illustrate the greater attenuation of the airgunsignal with range due to its lack of very low frequency (< 5 Hz) content. At short offsets,the airgun array provides a preferable seismic source to the explosives, due to denselyspaced shots and a consistent waveform resulting in excellent trace-to-trace coherence.With increasing offsets, it may be necessary to stack the airgun data to enhance itssignal-to-noise ratio: here we use a 4-fold stack. Large explosive shots, although morepowerful, produce a less consistent waveform and are more widely spaced due tooperational constraints. The offset at which airguns provide a preferable source isdependent on the ambient noise. This practical comparison of real sources demon-strates that, even without advanced processing, a well-tuned airgun array may providea preferable source to explosives at offsets up to 160 km, under favourableexperimental conditions.

Introduction

Wide-angle refraction surveys have historically been acquired using sparse but largeexplosive sources. For very large offset arrivals (i.e. offsets greater than about 150 km),explosives remain the only certain method of producing a sufficiently large signal to bedetectable above the noise. However, the capability to deploy larger airgun arrays,together with careful design to produce sources rich in low-frequency energy, allowsairgun sources to be used at increasingly large offsets. The advantages of using airgunsrather than explosives include environmental and safety considerations, in addition tothe much closer spatial sampling that can be achieved with airguns. For example,

q 1999 European Association of Geoscientists & Engineers 313

Geophysical Prospecting, 1999, 47, 313–339

1 Received June 1997, revision accepted October 1998.2 Bullard Laboratories, University of Cambridge, Madingley Road, Cambridge, CB3 0EZ, UK.3 British Institutions Reflection Profiling Syndicate (BIRPS), Madingley Road, Cambridge CB3 0EZ, UK.

airguns can readily be fired at 50 m intervals along a profile while the shooting ship isunderway. With a 200 kg explosive charge, it takes a minimum of several minutes toprepare the charge for firing, and it is necessary for the firing ship to be several hundredmetres away from the charge when it explodes. So a 50 m shot spacing can only beachieved by the shooting ship circling round to return to each successive shot point,rather than while underway along a profile. With closely spaced airgun data it is feasibleto migrate the wide-angle data so as to produce images of the deep crustal structure, ina way which is difficult using the sparse data acquired using explosives.

Typical offsets that are used in seismic surveys range from a few kilometres forstudies of the shallow sedimentary layers to several hundred kilometres when the baseof the crust or the upper mantle is the target depth. Single-ship seismic surveys haveused offsets governed by the length of the streamer: in the 1960s this was generally2.4 km, increasing steadily through the years until 6 km is the norm at present forsingle-ship surveys, and under special circumstances up to 12 km streamers have beendeployed. These offsets provide good velocity resolution on the uppermost fewkilometres of sediment in typical sedimentary basin settings, but little velocity infor-mation at greater depths (McBride et al. 1993).

For more problematic areas, such as imaging through basalt layers where they overliesediment, larger offsets have been achieved by using two-ship synthetic aperturetechniques. For example, White et al. (1999) used three passes of two seismic shipsto achieve offsets of 0–38 km with normal 2D industry seismic vessels. This providesgood velocity control down to the mid-crust (about 10 km depth).

In order to investigate the velocity structure at greater depths, larger offsets arerequired. The shallow tectonic structure in a sedimentary basin, which may be of greatinterest for hydrocarbon exploration, is controlled largely by deformation at greaterdepths; therefore, whole-crustal studies are of increasing importance as an explorationtool. In order to constrain the velocity structure down to 20–30 km depth, offsets of150 km or more are required. As we show, although explosive sources have generallybeen required for such large offsets, it is now possible to use large airgun arrays withcareful design for low-frequency content in order to record good seismic data at largeoffsets.We discuss here a comparison of airgun and explosive data shot along an identicalprofile as part of the Faroe–Iceland Ridge Experiment (FIRE). The airgun source wasa large, 48-gun array, while the explosives were a series of 200 kg charges firedunderwater. Arrivals from both sources were recorded at distant 3-componentseismometer stations on land.

FIRE: The Faroe–Iceland Ridge Experiment

The FIRE data were acquired by the British Institutions Reflection Profiling Syndicate(BIRPS) as an integrated normal-incidence and wide-angle seismic profile designed toobtain information on the history of the North Atlantic and the Iceland plume, from thetime of continental break-up to the present day. The results of traveltime and synthetic

314 R.K. Staples, R.W. Hobbs and R.S. White

q 1999 European Association of Geoscientists & Engineers, Geophysical Prospecting, 47, 313–339

seismogram modelling of wide-angle arrivals are described by White et al. (1996),Staples et al. (1997), Brandsdottir et al. (1997) and Richardson et al. (1998). Thispaper compares marine airgun and explosive seismic sources for long-range (up to400 km offset) wide-angle seismic work, with particular reference to traveltimemodelling.

The sources were fired above the Iceland Shelf and the Faroe–Iceland Ridge (waterdepths 50–550 m) and were recorded by seismometers in Iceland (Fig. 1). Theseismometer stations used in this study are station 010 at the Vattarnes lighthouse onthe coast at the mouth of Reydarfjordur (64856.20 N, 13841.10 W) and station 410 onthe eastern slopes of the mountain Stora-Svalbard (65815.90 N, 15824.00 W),approximately 90 km inland (Fig. 1). During the shooting period (airguns: 2nd–5thAugust 1994, explosives: 2nd–7th August 1994), the wind did not exceed force 5 andthe sea-swell was never higher than 2 m. The calm weather contributed to exceptionallyquiet recording conditions and some instruments recorded arrivals to beyond 400 kmoffset.

Airgun and explosive seismic sources 315


18˚

18˚

16˚

16˚

14˚

14˚

12˚

12˚

10˚

10˚

8˚

8˚

6˚

6˚

61˚ 61˚

62˚ 62˚

63˚ 63˚

64˚ 64˚

65˚ 65˚

66˚ 66˚

18˚

18˚

16˚

16˚

14˚

14˚

12˚

12˚

10˚

10˚

8˚

8˚

6˚

6˚

61˚ 61˚

62˚ 62˚

63˚ 63˚

64˚ 64˚

65˚ 65˚

66˚ 66˚

010

410

ICELAND

FAROE ISLANDS

FIRE shooting line500

1000

1500

2000

18˚

18˚

16˚

16˚

14˚

14˚

12˚

12˚

10˚

10˚

8˚

8˚

6˚

6˚

61˚ 61˚

62˚ 62˚

63˚ 63˚

64˚ 64˚

65˚ 65˚

66˚ 66˚

Figure 1. Map of project layout and bathymetry with isobaths at 250 m intervals. The solid lineshows the airgun profile, while the small filled circles (coincident over most of the profile) showthe locations of the explosive shots. The receiver stations in Iceland used in this study are shownby solid squares.

The airgun source was a 48-gun, 9324 cu. in. (153 litre) array fired from the MVGeco Echo. The array was 18.5 m long and 50 m wide, consisting of six subarrays ofeight airguns each (Fig. 2). The airgun source was designed to be able to collect bothreflection and refraction data at the same time. The reflection work required that therewas no notch within the frequency band 8–40 Hz, so the source used a combination of



Figure 2. Schematic layout of airgun array deployed from MV Geco Echo. Individual gunvolumes are shown in cubic inches.

different airgun chamber sizes and was towed at 10 m depth. The low-frequencyresponse of the array was augmented by the use of six 3-gun clusters at the front of eachsubarray. This resulted in a source rich in low frequencies for seismic refraction work.Individual airgun chamber sizes ranged from 30 to 500 cu. in. (0.5–8.2 litre) and wereoperated at 2000 p.s.i. (140 bar). The source was fired at 75 m intervals.

The explosive sources were 200 kg charges of Powergel E700 fired from the RRSChallenger. The shot detonator was activated by a chemical fuse, cut to a length of3.4 m to ensure detonation at 100–150 m depth after a ‘flight-time’ of 90–120 s. Theexplosive charges were used only for refraction work and therefore a low-frequencysource was chosen. Detonation below 100 m enhances the low-frequency content ofthe source by allowing constructive interference between the downgoing wave and thephase-inverted reflection from the sea-surface, and also prevents the significant loss ofenergy which would result if the explosives were to be detonated at a sufficientlyshallow depth for them to blow out through the sea-surface. Charges were fired at, 7 km intervals.

Theoretical source calculations

Early experimental and theoretical studies of underwater explosive shots wereundertaken by Arons (1948, 1954) and Weston (1960), and are summarized byHelmburger (1967) and Staples (1997). The source function takes the form

PðtÞ ¼ K�P0 exp

�¹tt0

�þ P1 exp

�¹ | t ¹ T1|

t1

�þ P2 exp

�¹ | t ¹ T1 ¹ T2|

t2

��; ð1Þ

where P(t) is the pressure function and t is the time after the initial explosion. Themagnitude, time delay and decay rate of each term is dependent on the charge size,burn rate and detonation depth, as described in the references above. The first termrepresents the initial burst and the second and third terms represent the first andsecond bubble pulses, respectively. KP0, KP1 and KP2 are the peak pressuresassociated with each phase, and T1 and T2 are the delays associated with the bubblepulses P1 and P2, respectively.

The initial burst is a shock wave at short ranges, and for a 200 kg explosive shot theinitial burst does not start to demonstrate normal acoustic properties until a range of atleast 450 m (Arons 1954). For a Powergel charge of 200 kg, a detonation depth of125 m and a range of 450 m, P(t) in bar-metres is given by

PðtÞ ¼ 32:2�55:1 exp

�¹t

0:00141

�þ 17:49 exp

�¹ | t ¹ 0:207|

0:00590

�þ 3:71 exp

�¹ | t ¹ 0:356|

0:0113

��: ð2Þ

The full derivation of this equation is given by Staples (1997). The pressure onset atranges greater than 450 m is much slower than the shock wave represented by thisequation and illustrated in Fig. 3.



A water-surface ghost is present 0.167 s later and is calculated assuming a reflectioncoefficient of ¹1. The theoretical explosive source spectrum exhibits several strongpeaks at approximately 3, 9, 15 and 21 Hz, with notches at 6, 12 and 18 Hz. Thefrequency content of the source is controlled almost entirely by the delay to the surfaceghost and thus by the detonation depth. The notch and peak frequencies may shift byup to 620% due to source depth. For energy propagating at 208 forward of the vertical,the peak frequency will increase by 6%.

Geco-Prakla supplied a theoretically calculated source wavelet for the airgun array,including the surface ghost, at far-field (Fig. 4). The source from each individualairgun is known from tests in a Norwegian fjord, and the interference effect of eachairgun in the array is calculated numerically. Since the airgun source was fired at 10 mdepth, the first notch caused by the surface ghost lies at 75 Hz, a frequency muchhigher than those considered in this study. The effect of the directivity of the source is



Figure 3. (a) Theoretical explosive source wavelet including the surface ghost, for a 200 kgPowergel explosive at 125 m depth. At far-field, the initial burst and its ghost are much broaderthan shown. (b) As (a) after application of 2.5–12.0 Hz band-pass filter.

negligible at the low frequencies considered. In the worst case, with a maximumfrequency of 12 Hz, the wavelength in water is 125 m, and for a ray turning in the crustat a velocity of 4.5 km/s, the ray angle in the water column is less than 208 from thevertical. For a frequency of 8 Hz and a ray angle 208 from the vertical in the watercolumn, energy from the front and back of the array is out of phase by ,1/20th of aperiod. For arrivals from deeper in the crust, with ray angles closer to the normal, themaximum phase lag would be only ,1/50th of a period.

The spectra of the explosive and airgun sources are illustrated in Fig. 5. Thetheoretical source spectra from the airgun array and from a typical 200 kg explosivesource are shown in Fig. 5a. For comparison, examples of the spectra of seismic signalsfrom the airgun array and from an explosive source recorded at a seismometer stationin Iceland are shown in Fig. 5b. In Fig. 5b, the source signals have been attenuated bythe frequency response of the seismometer recording package and by absorption in the



Figure 4. (a) Geco-Echo airgun array source wavelet, including the surface ghost, at far-field.While the individual airguns have been calibrated by experiment, the effect of their interferencein the array is theoretically calculated. (b) As (a) after application of 2.5–12.0 Hz band-passfilter.

earth during their passage from source to receiver. The low-frequency end of thespectrum is attenuated primarily by the frequency response of the seismometer andrecording electronics, while the decrease of energy at the high-frequency end of thespectrum is due mainly to absorption of the seismic signal during transmission throughthe earth. The notches in the spectrum of the explosive charges due to the detonationdepth are particularly marked in both the theoretical and observed spectra, althoughthe precise frequencies at which the notches occur are slightly different due probably toa difference between the actual detonation depth of the observed explosive charge andthe assumed depth used to calculate the theoretical source spectrum. The relativeamplitude changes with frequency of the theoretical airgun source are reproduced well



Figure 5. (a) Calculated source frequency spectra. Solid line: Geco-Echo airgun array. Dashedline: theoretical marine explosive source (200 kg Powergel explosive at 125 m depth). (b)Observed source frequency spectra from station 010. Average spectra between 14.2 km and35.5 km offset over a 3 s window (see Fig. 8). Solid line: airgun array. Dashed line: marineexplosive source. The low frequencies of the observed spectra are attenuated by the frequencyresponse of the seismometer and recording package, while the high frequencies are absorbed bythe passage of the seismic signal through the earth.

in the observed spectrum. A study of noise and signal spectra indicates high noise levelsbelow 2.5 Hz and severe attenuation in the crust of frequencies above 12 Hz. A band-pass filter between 2.5 and 12.0 Hz is therefore applied to both source functions(Figs 3b and 4b). The results of these calculations are shown in Table 1.

Definitions

Measured amplitudes

For an acoustic wave, the instantaneous acoustic pressure P is related to theinstantaneous particle velocity v by the equation

v ¼Prc

; ð3Þ

where r is the density of the medium and c is the wave velocity. The signal produced bythe recording seismometers is proportional to the ground velocity and therefore to theacoustic pressure, and the constant of proportionality is independent of frequency.

RMS amplitude

In this study, the RMS amplitude of the seismic trace is measured for one second ofsignal (with noise) beginning at the first break. The apparent signal-to-noise ratiosquoted, S/N app

RMS, are given by the equation

S=NappRMS ¼

tarrþ1:0tarr RMS amplitude

RMS noise; ð4Þ

where tarr is the arrival time.The apparent signal strength is higher than the true signal strength, due to the

addition of noise. Assuming that the noise and signal are uncorrelated, the true signal-to-noise ratio, S/N true

RMS, is given by the equation

ðS=NappRMSÞ

2 ¼ ðS=N trueRMSÞ2 þ 1: ð5Þ



Table 1. Theoretical comparison of airgun and explosive seismicsources after application of a 2.5–12.0 Hz band-pass filter. Pressuremeasured in bar-metres. Impulse measured in bar-metre-seconds.

Airgun Explosive Ratiosource source (%)

RMS acoustic pressure* 3.2 41 8Initial burst acoustic pressure 11 65 17Maximum acoustic pressure 11 170 6.5Initial burst impulse 0.14 1.5 9

* RMS acoustic pressure calculated for a 1 second period beginning atthe initial burst.

322R

.K.S

taples,R.W

.Hobbs

andR

.S.W

hite

q1999

European

Association

ofG

eoscientists&

Engineers,G

eophysicalProspecting,47,313

–339

Figure 6. Common-receiver gathers from station 010 with a reduction velocity of 8 km/s and a 2.5–12.0 Hz band-pass filter (P-wave arrivalson the vertical channel). (a) Explosive shot record, trace-normalized and clipped. (b) Airgun shot record.



Fig

ure

6.

Con

tinue

d.

324R

.K.S

taples,R.W

.Hobbs

andR

.S.W

hite

q1999

European

Association

ofG

eoscientists&

Engineers,G

eophysicalProspecting,47,313

–339

Figure 7. Common-receiver gathers from station 410 with a reduction velocity of 8 km/s and a 2.5–12.0 Hz band-pass filter (P-wave arrivalson the vertical channel). (a) Explosive shot record, trace-normalized and clipped. (b) Airgun shot record.



Fig

ure

7.

Con

tinue

d.

Picking uncertainty

The traveltime pick is made by hand at the onset of the wavelet and no automatedpicking algorithms, such as wavelet correlation, are used. The traveltime pickinguncertainty is largely dependent on the amplitude of the first peak relative to the noiselevel:

S=Nappfirst peak ¼

max: amplitudefirst peak��2

pRMS noise

: ð6Þ

The√

2—

factor accounts for the fact that peak and RMS measurements are beingcompared.

If a signal is particularly weak then it may be difficult to establish when the true firstarrival occurs. In this case the uncertainty in picking an arrival would be much greaterthan the uncertainty in picking a specified wave, and may be of the order of a wholeperiod. The use of a combination of unfiltered, zero-phase and minimum-phasefiltered data helps to indicate the correct arrival.

Semblance

Semblance is a measure of the multiple trace-to-trace correlation of data given by

S ¼

�t2

t1

ðPN

i¼1 aiðtÞÞ2

NPN

i¼1 aiðtÞ2dt; ð7Þ

where t1 and t2 are limits in time, N is the number of traces considered and ai(t) is theinstantaneous amplitude of trace i. When considering a wide-angle arrival with a finitearrival velocity, the traces should be reduced at the required velocity before semblancecalculation.

For a system with perfectly correlated signal s, accompanied by random uncorrelatednoise n, an expansion of the above equation (Staples 1997) can be reduced to give

S ,s2

s2 þ n2 þn2

s2 þ n2

1N

: ð8Þ

Even without noise, the semblance will reduce as the number of traces is increased,due to variations in the signal coda with offset. When the signals across the offset rangeconsidered become uncorrelated, the semblance falls rapidly to zero. For a wide-angledata set such as the one used here, the effect of a combination of noise and finite signalcorrelation is seen.

Observed data

Wide-angle common-receiver record sections of both airgun and explosive sources areshown in Figs 6 and 7. The record sections from station 410 (Fig. 7) show arrivals to300 km offset on the airgun section and to 375 km offset on the explosive section, andsome evidence for arrivals to 480 km offset on both sections. Even on these record





Figure 8. Signals received on vertical seismometer at station 010 (Vattarnes lighthouse) in theoffset range 10–40 km, which represent rays turning within the shallow crust. (a) 200 kgexplosive data. (b) Airgun data. (c) and (d) Single airgun traces. Precursors due to the zero-phase filter are seen on the explosive section. A reduction velocity of 5.75 km/s is applied to thedata, and the absolute time scale is arbitrary. All traces are band-pass filtered 2–40 Hz.



Figure 9. Signals received on vertical seismometer at station 410 in the offset range 110–130 km,which represent rays turning within the mid-crust. (a) 200 kg explosive data. (b) Airgun data. (c)A single airgun trace. The precursors due to the zero-phase filter are seen on the explosivesection. Reduction velocities of 6.0 km/s and 7.5 km/s were applied to different parts of the dataso as to produce first arrivals at an approximately constant reduced time. The absolute time scaleis arbitrary. All traces are band-pass filtered 2–40 Hz.



Figure 10. Signals received on vertical seismometer at station 410 in the offset range 230–280 km, which represent rays turning within the deep crust. (a) 200 kg explosive data. (b) Airgundata after a 4-fold stack. (c) An airgun trace after a 4-fold stack. A reduction velocity of 8.0 km/swas applied to the data, and the absolute time scale is arbitrary. All traces are band-pass filtered2–40 Hz.

sections, it is clear that at short offsets the airgun section can be picked with a smalleruncertainty than the explosive sections, since the explosives produce a more emergentsignal.

Observed RMS amplitudes

The observed data are shown in detail in Figs 8–10. Table 2 shows the RMS amplitudeat various shot–receiver offsets both for a single explosive shot and for an average overfive airgun shots. The noise level varies more at station 010 than at station 410 (Fig.11), and this is most probably due to variations in sea-wave noise on the coastline. Boththe signal strength and the signal-to-noise ratios are greatest at short offsets.

The airguns show an RMS amplitude approximately 6–13% of the explosive’s RMSamplitude in the 10–40 km and 110–130 km offset ranges, and approximately 3–5%that of the explosive’s RMS amplitude in the 230–280 km offset range. The airgunsignal is attenuated below the noise level more rapidly than is the signal from theexplosives, since it carries less energy at low frequencies compared with the explosivesignal (Fig. 5).

Stacking

The effect on RMS amplitudes of stacking airgun data is demonstrated in Fig. 12 andTable 3. For perfectly correlated signals, it would be expected that the true signal levelwould increase as the number of traces, N. This is observed, within uncertainties, foraddition of traces up to the 8-fold stack, and then it gradually begins to break down.This indicates that nearby airgun arrivals (across a range of 600 m) have almostperfectly correlated seismic signals, but that the correlation breaks down over longerranges.



Table 2. Observed RMS signal amplitudes after application of a 2.5–12.0 Hz band-pass filter.

Airgun Explosive

Offset Noise SignalappRMS Noise Signalapp

RMS(km) (mV) (mV) S=Napp

RMS (mV) (mV) S=NappRMS sairgun=sexplosive (%)

Station 010 Station 01014.2 1.77 47.95 27.1 6.05 792.19 130.9 6.135.5 0.76 11.86 15.6 4.41 134.66 30.5 8.8

Station 410 Station 410129.9 0.31 2.32 7.5 0.28 24.53 87.6 9.4248.0 0.44 0.53 1.2 0.33 7.74 23.46 *

* Equation (5) indicates that the true airgun signal at 248.0 km offset is 0.30 mV, and thereforethat sairgun/sexplosive ¼ 3.8%.



Figure 11. Amplitude against offset for marine source arrivals at stations 010 (Vattarneslighthouse) and 410 (inland desert station) in the offset range 10–275 km. The apparent RMSsignal level is shown. The RMS noise levels at the coastal station 010 are generally higher andmore variable than at station 410 due to sea-waves on the coast.

It would be expected that the noise level would increase as√

N—

if it is random and isuncorrelated from shot to shot. If the noise level increases by more than a factor of

√N—

,this may be indicative of the occurrence of correlated wrap-around from the previousshots. No correlated wrap-around energy is observed in this data set and none would beexpected since the constant distance-interval firing method, with its attendant randomvariations in the time elapsed between successive shots, is an efficient method ofeliminating correlated wrap-around (McBride et al. 1994).

The signal-to-noise ratio of airgun data at 230–280 km offset is lower than that of theexplosive shot data, even after stacking.



Figure 12. The effect of stacking airgun arrivals on record sections. Arrivals in 247–262 kmoffset range received at station 410. (a) Unstacked airgun data. (b) 4-fold stacked airgun data. (c)16-fold stacked airgun data. (d) 64-fold stacked airgun data. (e) Unstacked 200 kg explosive datafor comparison.

Picking uncertainty

Table 4 shows the amplitudes of the assumed first signal peak, which are one of themost important controls on the uncertainty of picking the arrival times.X Shallow crustal diving rays (10–40 km offset) at station 010 (Fig. 8).The first break of the explosives arrival is picked easily, even on an isolated trace, butdue to the sparse data the picking uncertainty is as high as 50 ms. The first break of theairguns is picked easily, even on an isolated trace, and the first arrival can be picked withan uncertainty of 1–2 samples (10–20 ms) when neighbouring traces are visible. Theamplitude of the first airgun peak is 4–20% of the amplitude of the first explosive peak.However, the airguns can be picked more accurately, due to the correlation of denselyspaced data.X Mid-crustal diving rays (110–130 km offset) at station 410 (Fig. 9).



Table 4. First-peak amplitudes after application of a 2.5–12.0 Hz band-pass filter.

Airgun Explosive

Offset Noise Signalappfirst peak Noise Signalapp

first peak(km) (mV) (mV) S=Napp

first peak (mV) (mV) S=Nappfirst peak sairgun=sexplosive (%)

Station 010 Station 01014.2 1.77 105 41.9 6.05 820 95.8 12.835.5 0.76 8.4 7.8 4.41 150 24.1 5.6

Station 410 Station 410129.9 0.31 0.81 1.84 0.28 15.03 38.0 5.4

Table 3. The effect on RMS amplitudes of stacking airgun data at , 248 km offset afterapplication of a 2.5–12.0 Hz band-pass filter.

Airgun Explosive

Fold of Noise SignalappRMS Signaltrue

RMS SignaltrueRMS

stack (mV) (mV) (mV) S=N trueRMS (mV) S=N true

RMS

1 0.44 0.53 0.30 0.68 7.73 232 0.63 0.87 0.60 0.954 0.89 1.50 1.21 1.368 1.26 2.66 2.34 1.86

16 1.81 4.82 4.46 2.4632 2.66 8.47 8.04 3.0264 3.80 13.13 12.56 3.31

128 5.09 20.51 19.87 3.90256 6.81 27.90 27.05 3.97

The first break of the explosive arrival can be picked on an isolated trace, but again dueto sparse data the picking uncertainty is , 50 ms. The first break of the airguns is hardto pick on an isolated trace, but with two or three neighbouring traces visible, it can bepicked with an uncertainty of , 50 ms. Since later peaks in the airgun coda aresomewhat stronger than the first peak, arrivals can be followed easily even if the firstpeak is somewhat weak. A 4-fold stack improves the signal-to-noise ratio and allowspicking with an uncertainty of , 20 ms on an isolated trace. The amplitude of the firstairgun peak is only 2–6% of the amplitude of the first explosive peak. However, theairguns still produce data that are preferable for analysis in terms of the number ofrecorded traces and the ease of picking.X Deep crustal diving rays (230–280 km offset) at station 410 (Fig. 10).The first break of the explosive arrival can just be picked on an isolated trace. There issignificant doubt as to which peak represents the true first arrival even whenneighbouring traces are compared. The uncertainty of the chosen arrival on theunfiltered data is ,50 ms, but there is some uncertainty whether a much smaller arrival,some 300 ms earlier, should be identified as the first arrival. Energy from the airgunsource cannot be easily identified from an isolated trace. However, trace-to-tracecoherence allows us to identify airgun energy on unstacked data, even if the first arrivalcannot be identified. After a 4-fold stack it is possible to pick an airgun arrival, althoughthere is no reliable means of establishing whether this represents the first arrival,without reference to the explosive shot records. Increased fold of stacking fails toreduce the picking uncertainty. For these ranges explosives create a better source, sincethe first-break arrival time cannot be established from the airgun data alone.



Figure 13. Semblance of airgun arrivals for various numbers of traces N. The data are takenfrom station 410, in the offset range 230–280 km.

Semblance

At large offsets, where the signal-to-noise ratio of airgun arrivals is less than 1,semblance becomes an important factor in identifying arrivals. The semblance iscalculated for a range of numbers of traces, N (Fig. 13).

For the airgun source, the actual source signal is similar between shots. Thereforefor a small number of traces the reduction of semblance with N is caused mainly bythe effect of noise (equation 8). Figure 13 shows a near-linear section for N< 8(1/N > 0.125). This indicates near-perfect trace correlation over ranges less than600 m. The gradient tends towards 0.75, indicating a true signal-to-noise ratio of 0.58,which is close to, but less than, the figure calculated in Table 3 (S/N true

RMS < 0.68). As agreater number of airgun traces is considered, the effect of geological structure causesvariations in the coda due to scattering and multipathing, and the semblance fallsrapidly towards zero for N > 8 (1/N < 0.125).

The semblance is also calculated for a fixed number of traces (four) following nostack, 2-fold, 4-fold, 8-fold and 16-fold stacks (Fig. 14). The peak in semblance for a 4-fold stack indicates that this is the preferred stack to use before traveltime analysis.

Summary and conclusions

In this study we have examined arrivals in the frequency range of 2.5–12.0 Hz for a48-gun, 9324 cu. in. airgun source and a 200 kg explosive source.1 The theoretical modelling has correctly estimated, within uncertainties, the relativepower of the explosives and the airguns. The theoretical RMS pressure generated by



Figure 14. Semblance of airgun arrivals over four traces for various folds of stack. The data aretaken from station 410, in the offset range 230–280 km.

the airgun array in the relevant frequency range should be approximately 8% of thatgenerated by the explosive charge. The observed ratio of the airgun to explosive RMSpressure is 6–13% in the 10–40 km and 110-130 km offset ranges, and 3–5% in the230–280 km offset range. The theoretical initial burst pressure generated by airguns inthe relevant frequency range should be approximately 17% of that generated by theinitial burst of the explosive charge. The observed ratio of the airgun to explosive first-peak pressure is 4–20% in the 10–40 km offset range and 2–6% in the 110–130 kmoffset range.2 The variation of observed pressure ratios with offset is a result of attenuation of thehigher signal frequencies. Little of the airgun source energy lies below 6 Hz, whereas asignificant proportion of the explosive source energy lies below 6 Hz, due primarily tothe greater source depth.3 At short offsets, the airgun array provides a preferable seismic source to theexplosives, due to the excellent spatial coherence of densely spaced traces, although a4-fold trace stack may be necessary to increase the signal-to-noise ratio. At largeroffsets, explosives provide a preferable source due to their greater energy. During theproject discussed here, the wind never exceeded force 5 and the sea-swell was nevergreater than 2 m. Airguns provided a preferable source up to 80 km offset for a coastalstation (which suffers greater environmental noise) and up to 160 km offset for aninland station (which is quieter).

Airguns provide an environmentally more acceptable source than do large explosiveshots, and it is easier to maintain tight control on the repeatability of the sourcewaveform of airguns. Explosives are more prone to waveform variations due todiffering depths of detonation, are difficult to handle, and have a more uncertain shotinstant unless they are fired electrically. The logistics of firing explosives are such thatin practice they can only readily be fired at a much greater shot spacing than airguns.Only basic processing has been used in this study so as to enhance the signal-to-noiseratio of the data. Due to the shot-to-shot variability of the explosive sources, only band-pass frequency filtering is possible. However, for the data from the airgun source, somestacking of adjacent traces was found to be beneficial in addition to frequency filtering.Because the airgun source is so consistent from shot to shot, more powerful 2D filters,such as KL filtering, may be useful, and may extend the useful range of the airguns toover 200 km.

Historically, airguns have been used for normal-incidence profiling and explosiveshave been used for refraction seismics. However, this study shows that underfavourable environmental conditions and with suitable care in designing good low-frequency content, large airgun arrays are suitable for wide-angle profiles out to 160 kmoffset, and are preferable to explosives. This is sufficient offset to give turning rayswhich penetrate to the lower crust in most oceanic and sedimentary basin settings.Where control is required on Moho depths or the upper mantle structure in areas ofthick crust, explosives are still necessary in order to provide sufficient signal-to-noiseratio at the largest offsets.

As oil exploration moves into difficult frontier areas, such as those in deep water on



the Atlantic margin, or in areas where salt or basalt overlies prospective sediments,wide-angle data acquisition using either fixed ocean-bottom receivers or two-shipsynthetic aperture profiles (White et al. 1999) is increasingly being adopted. The abilityto use powerful airgun arrays for underway wide-angle data acquisition is an importantelement in the success of these methods.

Acknowledgements

We thank the following for assistance in the field: Tim Minshull, John McBride, PeterMaguire and John Smallwood, members of the universities of Cambridge, Leicesterand Iceland and of the Lamont-Doherty Earth Observatory, and the officers and crewof RRS Challenger and MV Geco Echo. Funding was provided by the UK NaturalEnvironment Research Council (NERC) through its BIRPS programme, the BritishInstitutions Reflection Profiling Syndicate Industrial Associates and the University ofCambridge. The receivers used in this study were loaned by the NERC GeophysicalEquipment Pool. Department of Earth Sciences, Cambridge, contribution no. ES5270.

References

Arons, A.B. 1948. Secondary pressure pulses due to gas globe oscillation in underwaterexplosions. II. Selection of adiabatic parameters in the theory of oscillation. Journal of theAcoustical Society of America 20, 277–282.

Arons, A.B. 1954. Explosive shock wave parameters at large distances from the charge. Journal ofthe Acoustical Society of America 26, 343–346.

Brandsdottir, B., Menke, W., Einarsson, P., White, R.S. and Staples, R.K. 1997. Faroe–IcelandRidge Experiment, 2, Crustal structure of the Krafla central volcano. Journal of GeophysicalResearch 102, 7867–7886.

Helmburger, D.V. 1967. Head waves from the oceanic Mohorovic discontinuity. PhD dissertation,University of California, San Diego.

McBride, J.H., Hobbs, R.W., Henstock, T.J. and White, R.S. 1994. Short note on the‘wraparound’ multiple problem of recording seismic reflections in deep water. Geophysics 59,1160–1165.

McBride, J.H., Lindsey, G.A., Hobbs, R.W., Snyder, D.B. and Totterdell, I.J. 1993. Someproblems in velocity analysis for marine deep seismic profiles. First Break 11, 345–356.

Richardson, K.R., Smallwood, J.R., White, R.S., Snyder, D. and Maguire, P.K.H. 1998. Crustalstructure beneath the Faroe Islands and the Faroe–Iceland Ridge. Tectonophysics 300, 159–180.

Staples, R.K. 1997. Crustal structure above the Iceland mantle plume. PhD dissertation, Universityof Cambridge.

Staples, R.K., White, R.S., Brandsdottir, B., Menke, W., Maguire, P.K.H. and McBride, J.H.1997. Faroe–Iceland Ridge Experiment, 1, Crustal structure of northeastern Iceland. Journalof Geophysical Research 102, 7849–7866.

Weston, D.E. 1960. Underwater explosions as acoustic sources. Proceedings of the Physical Society76, 233–249.

White, R.S., Fruehn, J., Richardson, K.R., Cullen, E., Kirk, W., Smallwood, J.R. and Latkiewicz,



C. 1999. Faeroes Large Aperture Research Experiment (FLARE): Imaging through basalts.In: Geology of the Northwest European Continental Margin, Geological Society of London (inpress).

White, R.S., McBride, J.H., Maguire, P.K.H., Brandsdottir, B., Menke, W.H., Minshull, T.A.,Richardson, K.R., Smallwood, J.R., Staples, R.K. and the FIRE Working Group. 1996.Seismic images of crust beneath Iceland contribute to long-standing debate. EOS 77, 197 &200–201.



A comparison between airguns and explosives as wide-angle seismic sources

Documents

Transcript of A comparison between airguns and explosives as wide-angle seismic sources