42. INFLUENCE OF POROSITY AND WATER SATURATION ON THECOMPRESSIONAL-WAVE VELOCITIES OF BASALTS FROM THE NORTH PHILIPPINE SEA

David M. Fountain, Department of Geology, University of Montana, Missoula, Montana.

INTRODUCTION

The effects of water saturation and open pore spaceon the seismic velocities of crystalline rocks are extreme-ly important when comparing laboratory data to in situgeophysical observations (e.g., Dortman and Magid,1969; Nur and Simmons, 1969; Christensen and Salis-bury, 1975). The existence of fractured rocks, flow brec-cias and drained pillows in oceanic crustal layer 2a, forinstance, may appreciably reduce seismic velocities inthat layer (Hyndman, 1976). Laboratory data assessingthe influence of porosity and water saturation on seis-mic velocities of oceanic crustal rocks would certainlyaid interpretation of marine geophysical data.

Igneous rocks recovered during Leg 58 of the DeepSea Drilling Project, in the Shikoku Basin and DaitoBasin in the North Philippine Sea, are extremely ves-icular, as evidenced by shipboard measurements of por-osities, which range from 0 to 30 per cent (see reports onSites 442, 443, 444, and 446, this volume). Samples withthis range of porosities afford an excellent opportunityto examine the influence of porosity and water satura-tion on seismic velocities of oceanic basalts. This paperpresents compressional-wave velocities to confiningpressures of 1.5 kbars for water-saturated and air-driedbasalt samples from the North Philippine Sea. Samplesused in this study are from sites 442, 443 and 444 in theShikoku Basin and Site 446 in the Daito Basin.

Excellent negative correlation between porosity andcompressional-wave velocity demonstrates that water-filled pore space can significantly reduce compressional-wave velocities in porous basalts. Velocities measured inair-dried samples indicate that the velocity differencebetween dry samples and saturated samples is small forporosities exceeding 10 per cent, and very large forlower porosities.

EXPERIMENTAL TECHNIQUES AND DATABasalt samples used in this study were placed in sea

water aboard ship upon recovery and kept saturated un-til arrival in the laboratory. Cylinders 1.25 or 2 cm indiameter and 3 to 5 cm in length were cut and weighedto determine wet-bulk densities. The water-saturatedcylinders were wrapped with 100-mesh copper screenand then jacketed with copper foil. The copper screenallowed maintenance of pore pressures at values lowerthan external pressures. Barium-titanate transducers(1 MHz frequency) were attached to the core ends. Theentire sample assembly was placed in a pressure vessel,where hydrostatic confining pressures as high as 1.5kbars were applied. Compressional-wave velocities were

then measured at 0.2-kbar increments to 1.0 kbar, andat 1.5 kbar. Table 1 presents compressional-wave veloc-ities for the water-saturated samples and also lists wet-bulk densities and the mean atomic weights for the sam-ples. Calculations of mean atomic weight were based ongeochemical analyses of Leg 58 igneous rocks (site re-ports, this volume).

Samples with least apparent alteration were selected,air-dried for periods greater than 48 hours, and re-weighed. Effective porosities and grain densities werecalculated (Table 2). Values for effective porosities andgrain densities are minimum values, because all porewater was probably not evacuated. The air-dried sam-ples were subjected to pressures as high as 1.5 kbars,and measurements of compressional-wave velocity wererepeated. Table 2 presents compressional wave veloci-ties for air-dried samples.

RESULTS

Figure 1 shows the relationship between compres-sional wave velocity at 0.5 kbar and wet bulk density forwater saturated Leg 58 basalts. Only samples from Sites442, 443, and 444 were included in the least-squares fit,because of the obvious deviation of Site 446 data fromthe trend. This distinctive deviation results from thehigher mean atomic weights of the Site 446 samples(Table 1). Shikoku Basin samples have mean atomicweights around 22, typical of oceanic basalts. Site 446samples, because of higher iron and titanium contents(Site 446 report, this volume), have mean atomicweights close to 23; therefore, they should not plot onthe velocity-density relationship for lower mean atomicweights (Birch, 1960). A better correlation between ve-locity and wet-bulk density is obtained for selected sam-ples from Sites 442, 443, and 444 which have no filledvesicles and minimal apparent alteration (Figure 2).

The range of wet-bulk densities in Figure 2 is primari-ly related to the porosity of the samples. Figure 3 showsa strong negative correlation between porosity and wet-bulk density for the selected Shikoku Basin samples.Also plotted on Figure 3 are theoretical lines relatingporosity to wet-bulk density for given grain densities.Most samples plot near the theoretical lines for graindensities of 2.8 and 2.9 g/cm3, in agreement with themeasured grain densities (Table 2). The influence ofgrain density on wet-bulk density, therefore, is small incomparison to the influence of porosity.

Figure 4 shows the relationship between compres-sional-wave velocity at 0.5 kbar and porosity for theselected Shikoku Basin samples. The strong negativecorrelation between porosity and velocity is close to the

935

D. M. FOUNTAIN

TABLE 1Wet-Bulk Densities, Mean Atomic Weights, and Compressional-Wave Velocities

for Water-Saturated Basalts of DSDP Leg 58

Sample(Interval in cm)

442A-31-1, 83-93442A-32-2, 54-57442A-33-1, 97-100442A-34-1, 82-85442B-5-2, 106-109

442B-7-1, 49-52442B-9-1, 110-114442B-11-2, 123-126442B-13-1, 38-42442B-15-1, 78-82

442B-16-1,121-12444349-3, 137-142443-50-1, 117-120443-53-1, 115-119443-54-1, 90-92

443-54-3, 7-10443-54-7, 64-67443-55-1, 78-81443-56-2,50-52443-57-1, 70-74

443-58-3, 137-141443-59-3, 98-102443-60-5,113-116443-61-4, 77-79443-62-2,43-45

443-63-3, 103-107443-64-1, 24-27444A-25-1, 8-12446A-5-1, 12-16446A-7-3, 80-83

446A-10-5, 113-117446A-14-1, 85-88446A-15-3, 64-70446A-20-1, 96-99

Wet-Bulk

Density(g/cm3)

2.5312.4612.6002.3952.440

2.5292.4792.4712.3502.312

2.6572.8712.8732.8362.872

2.8842.9072.5002.4842.628

2.4922.7822.8212.8682.690

2.7092.7712.8292.9342.760

2.5852.6612.7172.786

Mean AtomicWeight

22.0721.7421.9521.9121.85

21.9621.9022.3122.5322.19

21.9122.2922.3322.1922.27

22.3222.3622.3622.2222.26

21.7122.0622.0322.1921.88

22.0221.8522.3222.5222.73

22.6922.9222.8822.88

0.2

5.0554.6905.2504.0624.595

4.7454.4244.3774.1584.415

5.0806.2896.3305.6886.032

6.1476.1484.4504.2754.734

4.1415.7645.8705.9915.180

5.4465.6405.8206.0845.010

4.9804.2504.7684.948

Compressional-Wave Velocity (km/s) atVarious Pressures (kbar)

0.4

5.1084.7535.3254.0894.694

4.7974.5024.4304.2304.467

5.0986.3316.3905.7446.101

6.2146.1984.5224.3214.794

4.2315.8185.9306.0505.270

5.5015.7305.8986.1105.088

5.0084.3104.8374.993

0.6

5.1444.7955.3874.1174.755

4.8404.5524.4574.2974.498

5.1506.3606.4335.8006.140

6.2636.2404.5564.3504.858

4.2885.8515.9706.1035.337

5.5345.7945.9676.1395.150

5.0354.3454.8765.020

0.8

5.1794.8265.4284.1444.786

4.8714.5824.4854.3524.519

5.1826.3806.4635.8566.165

6.3006.2774.5844.3764.888

4.3195.8776.0306.1505.384

5.5605.8366.0246.1645.192

5.0604.3674.9025.042

1.0

5.2064.8505.4524.1664.805

4.8864.5994.5134.3904.540

5.2066.3966.4865.9106.185

6.3346.3044.5994.4004.918

4.3375.8946.0476.1875.412

5.5785.8786.0626.1855.217

5.0824.3834.9125.057

1.5

5.2484.8765.4744.2244.730

4.9024.6304.5664.4464.590

5.2556.4306.5175.9866.245

6.3736.3574.6354.4574.964

4.3715.9256.1406.2355.445

5.6055.9106.0986.2205.247

5.1274.4174.9505.077

theoretical values predicted for water-saturated basaltsby the time-average equation (Wyllie et al., 1956). Thefirst-order control of velocity by porosity is clearlydemonstrated by Figure 4. The second-order influenceof grain density is demonstrated by Figure 5. Althoughgrain density and velocity correlate well, the high valuefor the slope of the regression line indicates that the in-fluence of grain density is small when compared to theinfluence of porosity. The compressional-wave veloci-ties of the selected samples are primarily dependentupon the influence of water-filled pore space. Thestrong negative correlation between porosity and veloc-ity (Figure 4) was discovered using 1-MHz transducers.Similar strong negative correlations between porosityand velocity were observed for shipboard measurementsusing 0.4-MHz transducers. These results demonstratethat the compressional waves travel through vesicularbasalts at an aggregate velocity which depends onlyupon the velocity through the solid rock, the velocitythrough the fluid, and the proportion of the porousrock saturated with fluid. The time-average equation

presented by Wyllie et al. (1956) is the best descriptionof the relationships presented here. This behavior isdistinctly different than the data for the Lau Basinbasalts discussed by Christensen and Salisbury (1975).In that case, compressional waves travelled throughvesicular basalts at speeds too great to be aggregatevelocities. Christensen and Salisbury (1975) interpretedthe velocities to be framework velocities.

The importance of water saturation on compres-sional-wave velocities is demonstrated by Figure 6,which shows the difference between velocities throughwater-saturated and air-dried samples [ΔVp = Vp

(saturated) - Vp (air-dried)] versus the porosity of thesample. Site 446 samples are included in Figure 6,because only the difference in velocity is important, notthe absolute values. Velocities measured at 0.5 kbarwere used in Figure 6. Negative values of Δ Vp apparent-ly reflect errors in measurement which are less than 3per cent. Figure 6 demonstrates that the velocity dif-ference between the water-saturated and air-driedsamples is large for porosities greater than 2 and less

936

COMPRESSIONAL-WAVE VELOCITIES

TABLE 2Wet-Bulk Densities, Effective Porosities, Grain Densities, and Compressional-Wave

Velocities for Air-Dried Basalts of DSDP Leg 58

Sample(Interval in cm)

442A-34-1, 82-85442B-7-1, 49-52442B-11-2,123-126442B-13-1, 38-42442B-16-1, 121-12444349-3, 137-142443-50-1, 117-120443-53-1, 115-119443-54-1, 90-92443-54-3, 7-10443-54-7, 64-67443-58-3, 137-141443-60-5, 113-116443-61-4, 77-79443-62-2,43^5443-64-1, 24-27444A-25-1, 8-12446A-5-1, 12-16446A-7-3, 80-83446A-10-5,113-117446A-15-3, 64-70446A-20-1, 96-99446A-23-2, 46-49

Wet-BulkDensity(g/cm3)

2.3952.5292.4.712.3502.657

2.8712.8732.8362.8722.884

2.9072.4922.8212.8682.690

2.7712.8292.9342.7602.585

2.7172.7862.711

EffectivePorosity

(%)

20.7012.6117.7223.77

8.47

1.441.081.960.720.60

0.3714.650.951.024.40

4.542.510.657.421

19.62

5.785.964.89

GrainDensity(g/cm3)

2.7592.7502.7882.7712.810

2.8992.8932.8732.8852.895

2.9142.7482.8392.8872.769

2.7902.8172.9472.9012.972

2.8232.8992.799

Compressional-Wave Velocity (km/s) at

0.2

3.7104.3514.0254.0304.560

6.0286.3285.7156.1746.165

6.1333.7066.0185.9544.668

5.2355.6105.9484.4104.656

4.2254.0684.378

Various Pressures (kbar)0.4

3.9224.5554.1494.1544.643

6.1046.3855.7796.2146.2206.1743.8546.0856.0074.742

5.3035.6886.0054.5404.762

4.3194.1664.450

0.6

4.0264.6904.2424.2264.703

6.1636.4265.8406.2386.258

6.2003.9566.1316.0424.802

5.3555.7446.0314.6234.837

4.3954.2514.506

0.8

4.0964.7844.3134.2724.746

6.2066.4575.8856.2546.2906.2224.0246.1646.0684.8505.3945.7936.0654.6874.876

4.4554.3264.544

1.0

4.1524.8554.3694.3074.7756.2426.4835.9286.2676.3176.2444.0726.1946.0804.8845.4295.8356.0804.7354.9084.4974.3844.567

1.5

4.2354.9454.4584.3644.8436.3106.5425.9906.2886.3676.2964.1566.2436.1324.9535.4825.9226.1154.8274.9804.5634.4834.600

than 10 per cent. The difference is small to non-existentfor higher porosities. The large magnitude of the veloci-ty difference at low porosities is similar to the observa-tions made by Nur and Simmons (1969) for low-porosity crystalline rocks. The difference betweenvelocities for dry and saturated samples in the Nur andSimmons experiment was attributed to the influence ofpores in the form of cracks rather than in the form ofround holes. Leg 58 samples, however, are vesicularover the entire range of porosity. It is difficult,therefore, to separate the importance of crack porosityfrom vesicularity in this experiment. Gregory (1976)also found that the difference between velocities deter-mined for water-saturated and dry sedimentary-rocksamples was greatest for the low-porosity samples.

Figure 6 also shows that the nature of the fluid in thepore space has very little influence on the aggregate ve-locity. This is surprising, because Figure 4 demonstratesthat the time-average equation works well for the water-saturated basalts studied here. The time-average equa-tion, however, does not work for air-filled pore space,as demonstrated by Nur and Simmons (1969). Wyllie etal. (1958) noted similar behavior of the Berea sandstonewhen different saturating fluids were used. They notedlittle difference between the velocities for the water-saturated, porous sandstone and the dry sandstone,although the data for the water-saturated sample fol-lowed the predictions of the time-average equation.

CONCLUSION

Data presented in this study demonstrate that com-pressional-wave velocities of basalts recovered duringDSDP Leg 58 are primarily dependent upon the porosi-ty of the samples. The strong negative correlation be-tween velocity and porosity and the second-order influ-ence of grain density clearly document this dependence.Porosity is mainly in the form of vesicles, although theinfluence of crack porosity can not be neglected. The in-fluence of water-saturated pore space can be large. Ve-locities of samples with porosities around 20 per cent,for instance, can be reduced by as much as 30 per cent.The time-average equation (Wyllie et al., 1956) ade-quately describes the variation of compressional-wavevelocities of Leg 58 basalts as a function of porosity.

The influence of water saturation is most pronouncedfor samples with porosities less than 10 per cent. Atgreater porosities, the quantitative effect of water-filledpore space and air-filled pore space is about the same.This effect is similar to that observed for porous sedi-mentary rocks (Gregory, 1976) and low-porosity crystal-line rocks (Nur and Simmons, 1969).

The results discussed here support the idea that frac-tures in oceanic crustal layer 2a will considerably reducethe seismic-refraction velocities through the layer (e.g.,Hyndman, 1976). Although the scale of porosity andthe frequency of waves used in this study differ tremen-

937

D. M. FOUNTAIN

7.0

6.0

5.0

4.0

• SITES 442, 443, and 444

Δ SITE 446

2.2 2.6 3.0

WET-BULK DENSITY (g/cm3)

Figure 1. Compressional-wave velocity at 0.5 kbar ver-sus wet-bulk density for Leg 58 basalts. The linearrelationship for Sites 442, 443, and 444 is Vp = 3.74 ρ- 4.69, with a correlation coefficient of 0.905.

dously from the scale of fractures in the oceanic crustand frequencies of seismic-refraction experiments, thisstudy demonstrates that porosities between 0 and 25 percent can significantly reduce compressional-wave veloc-ities in oceanic crustal rocks. In particular, the time-average equation appears to adequately describe theporosity dependence of velocity in basalts saturatedwith sea water for frequencies which are sensitive to theopen pore space.

This study is only a preliminary examination of theinfluence of porosity and water saturation on seismicvelocities in deep-sea basalts. Further work on thisproblem should focus on the possible frequency de-pendence of the relationships discussed here, the relativeimportance of crack porosity versus general volumeporosity, the importance of the percentage of watersaturation, and the role of pore pressure.

ACKNOWLEDGMENTS

Laboratory work was supported by a small grant from theUniversity of Montana Research Grant and Fellowship Pro-gram. Dr. Nikolas I. Christensen kindly furnished laboratoryfacilities at the University of Washington. Mr. James Schultzgenerously provided his able technical assistance during thelaboratory experiments. M. H. Salisbury, R. D. Hyndman,and N. I. Christensen reviewed the manuscript.

2.2 2.6 3.0

WET-BULK DENSITY (g/cm3)

Figure 2. Compressional-wave velocity at 0.5 kbar ver-sus wet-bulk density for selected samples (no filledvesicles, minimum apparent alteration) from Sites442, 443, and 444. The linear relationship is Vp =4.17Q - 5.84, with a correlation coefficient of 0.982.

REFERENCES

Birch, F., 1960. The velocity of compressional waves in rocksto 10 kilobars, 1. /. Geophys. Res., 65, 1083-1102.

Christensen, N. I., and Salisbury, M. H., 1975. Structure andconstitution of the lower oceanic crust. Rev.Geophys., 13,57-86.

Dortman, N. B., and Magid, M. S., 1969. New data on veloc-ity of elastic waves in crystalline rocks as a function ofmoisture. Internat. Geol. Rev., 11, 517-524.

Gregory, A. R., 1976. Fluid saturation effects on dynamicelastic properties of sedimentary rocks. Geophysics, 41,895-921.

Hyndman, R. D., 1976. Seismic velocity measurements ofbasement rocks from DSDP Leg 37. In Aumento, F.,Melson, W. G., et al., Init. Repts. DSDP, 37: Washington(U. S. Govt. Printing Office), pp. 373-387.

Nur, A., and Simmons, G., 1969. The effect of saturation onvelocity in low porosity rocks. Earth Planet. Sci. Lett., 7,183-193.

Wyllie, M. R. J., Gregory, A. R., and Gardner, L. W., 1956.Elastic wave velocities in heterogeneous and porous media.Geophysics, 21, 41-70.

, 1958. An experimental investigation of factors af-fecting elastic wave velocities in porous media. Geophysics,23, 459-493.

938

COMPRESSIONAL-WAVE VELOCITIES

3.0

2.2

V P = 2.9

10.0 20.0POROSITY (%)

30.0

Figure 3. Wet-bulk density versus porosity for selectedsamples from Sites 442, 443, and 444. The linear rela-tionship is Q = -0.024 Φ + 2.88, with a correlationcoefficient of - 0.985. The dashed lines describe thetheoretical relationship between porosity and wet-bulk density for grain densities of 2.8 and 2.9 g/cm3.

o.o 10.0 20.0

POROSITY (%)

30.0

Figure 4. Compressional-wave velocity at 0.5 kbar ver-sus porosity for selected samples from Sites 442, 443,and 444. The linear relationship is Vv = -0.985 Φ +- 0.954. The dashed curves are the theoretical curves

predicted by the time-average equation for solid-rockseismic velocities of 6.15 and 6.5 km/s.

939

D. M. FOUNTAIN

7.0

2.6 3.0 "0.0

GRAIN DENSITY (g/cm3)

10.0 20.0

POROSITY (%)

30.0

Figure 5. Compressional-wave velocity versus graindensity for selected samples from Sites 442, 443, and444. The linear relationship is Vp = 11.87ρ - 28.09,with a correlation coefficient of 0.873.

Figure 6. Vp versus porosity of selected samples fromSites 442, 443, 444 and 446. ΔVp is the difference be-tween the seismic velocity at 0.5 kbar of the watersaturated sample and the velocity after the samplehas been air-dried.

940