Equipment

The seismograph used was a 24 trace Texas Instruments 7000 Portable Seismograph System with a basic frequency range from 5 to 500 c./sec. and with a selection of gain, filter, mixing, and automatic gain control settings providing a large number of operating characteristics. Automatic gain control was very rarely used, and mixing only occasionally with no appreciable improvement in the records. The party was also equipped with two Vector geophone cables, each having 12 take-outs at 30 m. intervals, and with a variety of geophones for measuring high and low frequency ground motion, both vertical and horizontal.

Gravity values were measured with a Frost gravimeter, which was generously loaned by the Lamont Geological Observatory of Columbia University. This instrument is free of drift, an invaluable asset when several months elapse between readings at the base station, and has a very wide range, enabling readings to be made on the Byrd traverses over 10 degrees of latitude and 2,500 m. of elevation without resetting. The authors are greatly indebted to the Lamont Observatory and especially to Professor Worzel for the use of this instrument.

For magnetic measurements an Arvela vertical component magnetometer, having a reading accuracy of about 10 gamma, was used. This instrument is very small, light, and easy to use, and with it an observation can be made in 10 min. or less. Altimetry, important both for reduction of the gravity data and for mapping, was conducted with three Wallace and Tiernan altimeters. Two of the instruments formed a matched pair which followed each other very well over a wide range of elevation.

Methods Employed

The following description pertains to the Sentinel and Horlick traverses; the Little America to Byrd traverse differed only in the distances between stops. The normal operating procedure was to travel every other day, making stops every 3 nautical miles (5.5 km.) en route to read altimeters and for gravimetric and magnetic measurements. Of the three traverse vehicles, one travelled 3 nautical miles (5.5 km.) ahead of the other two; altimeter readings at each vehicle were then compared to give the difference in elevation over the 3 nautical mile (5.5 km.) interval. At each stop the air temperature and wind velocity were measured to enable corrections for temperature and for horizontal variations in atmospheric pressure to be applied to the altimeter data. In a normal travelling day 30 nautical miles (55.5 km.) would be covered, the next day then being devoted to seismic and glaciological pit studies.

The primary seismic measurement at each station was the determination of the ice thickness by reflection shooting. The spread was laid out either in line or in the form of an “L”. One-fourth of the geophones were normally placed in the horizontal longitudinal position in order to detect possible shear of transformed compressional-shear reflections. Several shots were fired at the corner of the spread with different gain and filter settings. A charge of 1 lb. (0.45 kg.) at a depth of 4 m. was usual on the Sentinel traverse where reflection quality was very high; on the Horlick traverse best results were usually obtained with the same charge in a 10 m. hole or, occasionally, with a seven shot pattern 2 m. above the surface. Shooting in a “sprung” hole, i.e., one in which a charge had already been fired, was found to be of value since the use of such a hole not only increased by a factor of five the amount of seismic energy produced from a given charge size, but also appeared to improve the signal to noise ratio.Reference Bentley ¹

Reflection records varied greatly in quality, most of those on the Sentinel traverse being excellent, whereas those on the Horlick traverse were generally quite poor. The controlling factor in record quality was not the strength of the echo, although this varied considerably, but the amount of prolonged surface noise following the shot and interfering with the reflected arrival. Such noise has been reported by other observers and can be expected everywhere on sufficiently cold firn. A study of the cause of this noise is currently in progress but no conclusions have yet been reached.

As will be discussed further in another paper, a low velocity layer was discovered at the base of the ice sheet. Uncertainty as to the thickness of this layer and its wave propagation characteristics diminishes the accuracy with which echo times can be converted to ice thickness values. Using the average thickness estimated for the layer and assuming a velocity of 3,600 m./sec. within it, an average velocity of 3,820 m./sec. throughout the ice column was calculated and has been used for all thickness computations. The maximum error in total ice thickness which would result from the use of this value for the average velocity is estimated to be ±40 m. The error in relative thickness between two neighboring stations would be considerably less than this, but it is difficult to reckon owing to lack of information concerning the variability of the basal layer from place to place.

The gravity values measured between the seismic stations were employed to provide detail in the topographic profile. On a broad, flat plateau free air anomalies represent a first approximation to isostatic anomalies. In order, therefore, to minimize the effect of the compensating (negative) mass at depth corresponding to the load of the ice, free air anomalies were used to compute ice thickness. However, since the reflection shooting provided control every 30 nautical miles (55.5 km.), Bouguer anomalies would have produced negligibly different results. The gravity observations were used, then, to interpolate changes in sub-ice rock surface elevations between seismic soundings. This was accomplished by equating changes in free air anomaly relative to a seismic station to changes in ice thickness and adding or subtracting this difference to the seismically determined elevation of the rock surface relative to sea-level. For these calculations densities of 0.9 g./cm.³ for the ice and 2.67 g./cm.³ for bedrock were assumed. Thus 1 mgal anomaly represents 13.56 m. change in ice thickness.Reference Ostenso ²

Considerable difficulty was experienced in trying to correct traverse magnetic observations for diurnal changes in field strength. It was found that throughout most of a season the traverse party would be too far from any of the permanent base stations to permit effective control by means of the station magnetographs. For this reason the magnetic data have been used only to obtain the regional gradient and to give the general magnetic character of the basement rock.Reference Ostenso and Bentley ³

The altimetric survey was conducted using a modified leap-frog method with the normal corrections applied for air temperature and temporal changes in barometric pressure. In addition an adjustment was made for horizontal pressure differences by noting the surface wind speed and direction with each altimeter reading and relating these to the geostrophic gradient.Reference Ostenso and Bentley ⁴ Generally good results were obtained from the altimetry. The standard error of one measurement of elevation difference between two stops was about 1 m., and the estimated maximum error on a closed traverse loop was ±15 m. relative to Byrd Station. The latter error is naturally less the closer the field station is to Byrd Station.

The principal facts, i.e., latitude, longitude, observed gravity, free air anomaly, surface elevation, bedrock elevation and ice thickness, of all 719 observation sites from the three traverses are tabulated in the Appendix (p. 896). The values of observed gravity are on an absolute scale, being tied to the University of Wisconsin pendulum station at McMurdo Sound. The base value at Byrd Station (982.6009 gal) was established by multiple ties via air with the McMurdo Sound station using Worden, LaCost and Frost gravimeters.Reference Ostenso and Bentley ³ The program of establishing and strengthening a network of base stations in Antarctica is continuing and the value for the Byrd Station is subject to a slight adjustment (probably not in excess of ±2 mgal) as more data become available. Such a change in base value would result in an overall adjustment of the gravity surveys but would not change the free air anomalies.

Sample Records

Figure 2 (p. 911) shows three reflection records, the upper one from the Sentinel traverse, the middle from the Horlick Traverse. Both were produced by a 1 lb. (0.45 kg.) charge at a depth of 4 m., and the gain and filter settings were the same in each case. It can be seen from the upper record that the surface waves are rapidly attenuated with the result that after about 0.7 sec. all traces have essentially returned to background noise level. This permits the recording of an excellent echo from the ice—rock interface at the ice thicknesses normally encountered. In striking contrast to this record is the one from the Horlick traverse, exhibiting the prolonged surface noise which proved a major obstacle to successful reflection shooting. The noise was incoherent and looked much the same as this sample even when the filters were wide open. This interference was serious enough to prevent the recording of a satisfactory echo during the first 278 km. of the Horlick traverse, and to spoil throughout the summer any possibility of observing the weak reflections from the surface of the basal low velocity layer.

Evidence of sub-ice reflection energy can be seen shortly after the main echo on the first record. From some shots reflected energy continued to arrive for 0.5 sec., probably indicating a considerable amount of morainal material. As yet, the sub-ice reflection data have not been examined in detail.

The bottom record in Figure 2 shows reflections obtained with a 5 lb. (2.27 kg.) charge in a sprung hole on the Sentinel traverse. The relative gain level can be seen to be high by comparing reflection amplitudes with the upper record. Note, nevertheless, how rapidly the surface waves die out despite the absence, in this case, of any low cut filtering. This record also shows energy from the surface of the basal low velocity layer arriving 0.2 sec. before the main bottom echo.

Results

Seismic reflection and gravity results have been combined to draw the cross-sections of the ice sheet shown in Figures 3, 4 and 5. The dashed line in these figures shows the estimated position of sea-level relative to the rock surface (adjusted sea-level) after removal of the ice and allowance for isostatic rebound of the rock floor according to Archimedes’ Principle. From the Ross Ice Shelf—Rockefeller Plateau boundary to the end of leg 1 of the Sentinel traverse the picture is generally the same. The ice—rock interface is rough with the ice thickness varying between 600 and 2,700 m. With the exception of a few peaks, the entire rock floor is at present below sea-level, the major portion below adjusted sea-level.

Fig. 3. Profile from Little America V to Byrd Station. Dashed line represents adjusted sea-level

Fig. 4. Profile along the Sentinel Mountains traverse route. Dashed line represents adjusted sea-level

Fig. 5. Profile along the Horlick Mountains traverse route. Dashed line represents adjusted sea-level

Sentinel traverse leg 2 shows quite a different picture, with great ice thickness (a maximum of over 3,500 m.) and a smooth bottom below the estimated adjusted sea-level by as much as 1,200 m. Leg 3, paralleling the strike of the Sentinel Mountains, shows a basal profile of another character. The rock floor is very rough and almost entirely above sea-level, in several places breaking the surface to form nunataks. Leg 4 extends from the mountainous zone across a depression narrower but deeper than the one on leg 2 and back into the moderately rough area to the west. It was in this depression on leg 4 that a maximum ice thickness of 4,270 m. was discovered. This is believed to be the greatest thickness of ice discovered to date.

The ice thickness profile along the Horlick traverse is shown in Figure 5. Leg 1 is characterized by subglacial topography of moderate relief with the rock surface mostly below sea-level. It will be noted that the decrease in surface elevation from 1,500 to 800 m. is apparently not controlled by the level of the rock floor. Leg 2 runs along the Horlick Mountains and shows a rise in level of both the ice and the underlying rock surfaces. The first half of leg 3, which runs between the Horlick Mountains and another small group of mountains, exhibits a generally high rock level with rough topography. Data for the second half of this leg are sketchy as no gravity values were obtained after km. 1,380, but seismic reflections show the rock floor to be mostly below sea-level. At about km. 1,570 the extension of the deep depression found on leg 4 of the Sentinel traverse is seen. This feature is also shown near the west end of the Ellsworth—Byrd traverse profile,Reference Behrendt and Thiel ⁶ and a suggestion of its existence is found between km. 185 and 280 on Horlick traverse leg 1.

Figure 6 shows a comparison of the measured vertical component of the magnetic field, Z, with the rock surface topography on the Sentinel Traverse. Although, as mentioned previously, it has not been possible to correct these data for diurnal variations, a considerable change in their general character along the route is apparent. On leg 1, particularly on the first part of the leg, Z exhibits large variations which correlate fairly well qualitatively with the variations in rock level. This is best shown over the known volcano at the end of the leg. Along legs 2 and 3 the variations in Z are much less and show no relation to the sub-ice topography. The smoothest part of the magnetic curve is in fact found on leg 3 where the rock surface is the roughest. On leg 4 changes in Z are more noticeable but there is little apparent correlation with the bottom topography. The magnetic data, then, show that the traverse route passes from a region of rocks with high magnetic susceptibility (probably crystalline) eastward into a region where the underlying rocks have little or no magnetic susceptibility.

Fig. 6. Profile along the Sentinel Mountains traverse route showing the relationship between bottom topography and changes in the vertical component of the earth’s magnetic field (Z)

Although a detailed examination has not yet been carried out, rock samples collected from Mount Takahe on the Sentinel traverse and from the Executive Committee Range by the Byrd 1959–60 traverse group as well as the shapes of the peaks themselves, indicate a volcanic origin of these mountains. On the other hand, low grade metamorphic rocks were found in the foothills of the Sentinel Mountains (the main range was not reached) and other nunataks on that traverse;Reference Anderson ⁷ visits to the Horlick Mountains and neighboring peaks yielded samples of granite and sedimentary rocks.Reference Long ⁸ Thus the magnetic and geological evidence is in agreement in indicating a zone of volcanic material in the north-west part of Marie Byrd Land and of granite and rocks of sedimentary origin in the south and east. Granite and metasediments are known along the coast in the Edsel Ford Ranges, but nowhere else northwest of Byrd Station.

Long refraction shooting on the SentinelReference Bentley ¹ and Ross Ice Shelf Reference Crary ⁹ traverses has given the wave velocities in the rock beneath the ice at several places (Fig. 7). The two sections on the Ross Ice Shelf are taken from Crary’s preliminary results and are subject to minor revision. The velocities in the deepest layer shown in each column are in agreement within the limit of error of their determination, and their average, 6.1 km./sec., is representative of granitic continental crustal material. The only exception is the eastern Ross Ice Shelf where soundings were not sufficiently deep to reach this layer. Above the crustal material, however, there is an important difference between the average of 5.2 km./sec. from the Sentinel Mountains and km. 1,590 on the east side of the depression, and the value of 4.3 km./sec. found at Byrd Station and under the Ross Ice Shelf. The first two sections show overlying ocean sediments with a velocity of 2.2 km./sec. This provides further evidence that the marked physiographic depression defines the dividing line between two different geological provinces, Byrd and Little America Stations both being to the west of the dividing boundary.

Fig. 7. Seismic velocity sections determined by long refraction studies on the Ross Ice Shelf and in West Antarctica

Ice Surface Topography

The ice surface topography shows some interesting features as is shown in Figure 8. Two definite high areas are apparent, one in the east between the Sentinel and Horlick Mountains, and the other in the north-west in the vicinity of the Executive Committee Range. Between these areas is a saddle indicating flow of the ice from each high toward the other and off to the south-west and north-east. The large change in the ice level along Horlick traverse leg 1 (Fig. 5) where there is little change in the average elevation of the rock floor has already been pointed out; it is in general true that large variations in the elevation of the rock floor are not reflected in the ice surface elevation, which must therefore be determined by the pattern of ice flow.Reference Nye ^{10 ,} Reference Nye ¹¹ Thus the ice surface contours suggest that originally there were two separate ice sheets, one centered in the Executive Committee Range area and the other along the Sentinel—Horlick Mountains axis, and that these two converged across the low water-filled zone between them. Such a convergence undoubtedly represents a rather complex glacial morphology with intervening stages where the ice sheets were first joined by sea ice and later by an ice shelf before becoming completely ice-filled. It is interesting to note the decided asymmetry of the ice sheet about the Sentinel—Horlick Mountains axis. To the east, with no obstacles, the ice surface slopes down rapidly to the Filchner Ice Shelf; to the west the converging flow has produced a relatively flat surface over the broad reaches of the interior of West Antarctica.

Fig. 8. Elevation of the ice surface in West Antarctica

Although leg 2 of the Horlick traverse ran quite close to the mountains the surface contours can be seen to strike more or less perpendicular to the mountain front. This indicates that there is not much ice movement from the south and suggests that the Horlick Mountains chain, although not continuous above the ice surface, nevertheless forms an effective block against ice flow from the high South Polar Plateau.

Subglacial Topography

The contour map of rock surface elevation is shown in Figure 9. Because of the limitations imposed by the scale of this map topographic features less than 90 km. in extent were averaged out prior to contouring. From this figure it can be seen that, with the exception of the high spot centered around lat. 83° S., long. 105° W., the whole region between the Sentinel—Horlick Mountains axis and the Executive Committee Range area is below sea-level, the vast majority at least 500 m. below. This depression, although becoming gradually shallower, broadens to the north-east, suggesting a continuous channel extending from the Ross Sea to the Bellingshausen and Amundsen Seas. A conspicuous feature within this trough is a deep inner trough which reaches a maximum depth of more than 2,500 m. between Byrd Station and the Sentinel Mountains. From Figures 4 and 5 it is seen that if the ice sheet were removed the channel would be an open water passageway even after isostatic rebound. The expected isostatic rise of the land surface after removal of the ice and allowing for the weight of overlying water would bring the −2,500 m. contour up to −1,900 m. and the present —500 m. contour would be the approximate boundary of the water-filled channel. These calculations are based upon assumed densities of 3.3, 2.67, 1.03 and 0.9 g./cm.³ for mantle rock, crust, sea-water and ice respectively, and the apparently valid assumption of present isostatic equilibrium.Reference Ostenso ² No allowance was made for the rise in sea-level which would result from the melting of the Antarctic Ice Sheet. This rise (100 m.) would be small compared with the rise of the rock surface (although disastrously large for many portions of the world). If the entire Antarctic Ice Sheet were to melt, it would very likely do so at a rate greater than that of the isostatic rise of the land. For this reason a much larger area than indicated by the −500 m. contour would probably be submerged; however, it is felt that this contour presents a fair estimate of the extent of land and water before the growth of the ice. At any rate, it is clear that the Ross—Bellingshausen Sea connection is fundamental in nature and is not simply the result of a low plain depressed below sea-level by the overlying weight of ice, as, for example, is the case in Greenland. Such a conclusion is further substantiated by the gravity studies which support that there is a marked rise in the elevation of the Mohorovicic discontinuity (determined from Bouguer gravity anomaly values beneath the channel), indicating a thinning of the crust.Reference Ostenso ^{2 ,} Reference Bentley ¹²

Fig. 9. Rock surface elevations of West Antarctica relative to sea-level

CraryReference Crary ¹³ points out the existence of deep water under the southern boundary of the Ross Ice Shelf, which occupies the position of a foredeep in front of the Queen Maud Range. From Figure 9 it can be seen that this trough extends more or less as an unbroken feature past the Horlick Mountains and nunataks to the north, but is not apparent in front of the Sentinel Mountains. A connection between this deep and that found under the Filchner Ice ShelfReference Behrendt and Thiel ⁶ is possible through the gap between the Horlick traverse leg 3 and the Ellsworth—Byrd traverse, and in fact Thiel Reference Thiel ¹⁴ on the 1959–60 airborne traverse obtained a seismic sounding of 1,700 m. below sea-level at lat. 84° 45′ S., long. 88° W. If this represents a continuation of the deep it must be very narrow indeed for the sounding was made between two nunataks separated by only 150 km. in the north-south direction. It is clear that if such a connection exists, it (a) is a minor feature and (b) runs contrary to the topographic trend of the region as a whole. Certainly there is no evidence to support the existence of a major topographic connection between the Ross and Weddell Seas.

In Figure 10 is shown an isopach map of the region surveyed. Within the area delineated by the dotted line the average ice thickness was calculated to be 1,950 m. and the total ice volume 1,836,000 km.³. These computations were done by first plotting in detail the ice thicknesses at all the observation sites on a large-scale map. The area between successive contours was then carefully measured and multiplied by the thickness value intermediate between the bordering contour. Assuming an average ice thickness of 1,500 m. for the remainder of West Antarctica a total ice volume of 3,675,000 km.³ was obtained.

Fig. 10. Ice isopach map of West Antarctica

Conclusions

The following are the major conclusions reached from the topographic work done to date in Marie Byrd Land:

1. A major below sea-level connection probably exists between the Ross Sea and the Bellingshausen—Amundsen Seas beneath the ice of West Antarctica. This connection, defined by a 400 km. wide channel, is deep enough surely to have existed before the land surface was depressed by the weight of the overlying ice sheet. Within this channel, between Byrd Station and the Sentinel Mountains, there is a deeper inner trough where a depth of more than 2,500 m. below sea-level was found.

2. The combination of seismic, magnetic and geologic evidence leads to the conclusion that the channel separates West Antarctica into two provinces, with granite and rocks of sedimentary origin found to the east and south, and a volcanic region to the north-west.

3. There is no major channel or depression connecting the Ross and Weddell Seas. Although the possibility of a narrow deep channel joining the two areas has not been eliminated, all available evidence points to essential continuity of the mountainous chain from the southern boundary of the Ross Ice Shelf through the Horlick and Sentinel Mountains to the Palmer Peninsula.

4. Ice surface topography indicates that the present ice flow is outward from two centers, one in the vicinity of the Executive Committee Range, and the other between the Horlick and Sentinel Mountains. The surface slope is relatively steep toward the Ross and Filchner Ice Shelves, but between the two centers the converging flow has produced a fairly broad, flat saddle.

5. From the configuration of the ice and rock surfaces it is concluded that the West Antarctic Ice Sheet originated as two separate ice sheets in two separate mountainous areas. As these sheets expanded they converged over the open water between, were probably joined at first by a floating ice shelf, which then grew thick enough to fill the trough completely and produce the present single grounded ice sheet.

Acknowledgements

In drawing the various maps data from the 1958–59 airborne traverse, the Ellsworth—Byrd traverse and a short traverse from Byrd Station to the Executive Committee Range have been used. For these data the authors wish to express their appreciation to Edward Thiel, Rev. Edward Bradley, John Pirrit, Leonard LeShack and William Chapman. Results from the last part of the Horlick traverse have very kindly been provided by Frank Chang and William Chapman. Our gratitude also goes to the various members of the Byrd traverse parties whose help made the collection of the data possible, and to the U.S. Navy and in particular Air Development Squadron 6 without whose unfailing support the traverse program would never have been possible.