Introduction

One of the objectives of the Ross Ice Shelf Geophysical and Glaciological Survey (RIGGS) was to determine density profiles from the surface down to the firn–ice boundary (upper 70–100 m) by seismic measurements of the variation of the compressional-wave (P-wave) velocity with depth and then by relating P-wave velocity to density using an empirical formula presented by Reference KohnenKohnen (1972). In this paper we present the results of seismic refraction experiments performed at station J-9, the site of the Ross Ice Shelf drilling project. The results are interesting because they indicate the existence of a large lateral density change between two sites only 2 km apart in a region far removed from any apparent large stress gradients.

Station J-9 is located at 82° 22′ 30" S., and 168° 37′ 33" W. (Fig. 1). One of the refraction experiments (Profile 1) was conducted during the 1974–75 Antarctic field season, the other (Profile 2) in the 1976–77 season. Both profiles were oriented north-south (magnetic). The receiver spacing varied from 2 to 10 m, depending on the shot-point distance. The energy source was either a hammer blow on a wooden stake or the detonation of an explosive charge. The 1974–75 measurements were made with a Texas Instruments 7000B 24-channel seismograph, those in 1976–77 were made with an SIE 24-channel seismograph. Differences due to instrumental effects were found to be negligible.

Fig. 1. Map of the Ross Ice Shelf showing the location of station J-9 where the measurements of this paper were made. Circled letters denote ice streams.

The 1976–77 refraction experiments included a seismic recording along two other lines at angles of 60° on either side of Profile 2, for the purpose of investigating the dependence of velocity on direction. A comparison of the time–distance curves for the three lines showed only negligible differences, indicating that the constant-velocity horizons beneath Profile 2 are essentially horizontal.

Data Analysis

Travel times were read from the seismograms with an estimated error of 0.3 ms using a magnification of seven times. Time–distance curves for Profiles 1 and 2 arc displayed in Figure 2; a clear divergence of as much as 3 ms is evident.

Fig. 2. Compressional-wave travel-time curves for the two profiles at J-9. Solid circles denote the 1974–75 measurements; open circlse indicate the measurements made about 2 km away in 1976–77.

Both data sets were analyzed by fitting a non-linear expression to the time-distance curves (Reference KirchnerKirchner, unpublished),

(1)

where x is the shot-detector distance, t is the predicted travel time, and the a_i are the constants to be determined by the best fit. This method has the advantage of being free from observer bias and it ensures complete reproducibility. Also, the slopes can easily be obtained by differentiating Equation (1). Equation (1) is evaluated using non-linear regression analysis (see Reference KirchnerKirchner, unpublished, for a more detailed discussion).

After obtaining the velocities as a function of distance, velocities as a function of depth were calculated using the standard Wiechert–Herglotz–Bateman integral. Densities were then derived using Kohnen′s empirical equation (Reference KohnenKohnen, 1972),

where ρ(z) is the density (in Mg m⁻³) at the depth z at which the P-wave velocity equals v P(z) (in m s⁻¹), and, v P_i ═ 3860 m s⁻¹, the P-wave velocity in ice. The errors in the densities are estimated to be ±0.018 Mg m⁻³ at the surface, decreasing to ±0.006 Mg m⁻³ at 40 m depth and to about ±0.002 Mg m⁻³ at the final depth. (This analysis does not take into account the effect of temperature, the correction for which would increase the density by a maximum of 0.002 Mg m⁻³.)

Reference RobertsonRobertson (unpublished), who originally analyzed Profiel 1, found good agreement between densities derived seismically and those measured on cores from the 100-m hole nearby (personal communication from C. C. Langway, Jr, 1975 (Fig. 3). On the other hand, densities derived from Profile 2, also shown in Figure 3, are significantly greater. The indication is that there is a significant horizontal variation in density between the two sites. Two possible causes can be suggested for this: The first is that some physical change occurred in the J-9 area during the two-year time period. However, this seems highly unlikely since the differences are marked at depths greater than 20 m but negligible near the surface.

Fig. 3. Density–depth curves. Densities measured on cores are shown by crosses; those determined from the seismic velocities are shown by solid circles for the 1974–75 profile and by open circles for the 1976–77 profile.

A more plausible possibility is that the effect is associated with the interaction between ice stream B and the ice shelf. The occurrence of large stresses is indicated by the heavily-crevassed boundary area that terminates about 100 km up-stream from station J-9 (fig. 10 in Reference Jezek, Jezek, Bentley and CloughJezek and others, 1979). For an ice-shelf velocity of about 350 m a⁻¹ (Reference ThomasThomas, 1976[a]), that distance corresponds to an estimated age of about 300 years. The disturbed surface from the crevassed zone would thus be buried at J-9 at a depth corresponding to 300 years of undisturbed snow accumulation, i.e. about 30 m (Reference ThomasThomas, 1976[b]). This agrees roughly with the depth to the top of the density anomaly. In support of the idea of up-stream deformation is the earlier discovery by Reference RobertsonRobertson (unpublished) of significantly different values for the maximum P-wave velocity along reversed profiles, indicating a dipping structure (not necessarily dipping smoothly), at RIGGS 1973–74 base camp (station B.C.) about 30 km up-stream.

No more detailed interpretation can be made without making further velocity–depth maps. It is worth emphasizing, however, that the assumption of lateral homogeneity of density in low-stress regions may be invalid in regions down-stream of high-stress zones.