DEEP SOUNDING : ECHO RESULTS 115

SWITHINBANK, C. W. M. 1968. Radio echo sounding of Antarctic glaciers fromlight aircraft. IUGGjIASH General Assembly of Bern, 1967, Commission ofSnow and Ice: Reports and Discussions, p. 405-14.

SWITHINBANK, C. W. M. 1970. Jce movement in the McMurdo Sound area ofAntarctica. In: International Symposium on Antarctic Glaciological Explora-tion (ISAGE), Hanover, New Hampshire, U.S.A., 3-7 September 1968. Cam-bridge (Pub. No. 86 of IASH), p. 472-87.

WAITE, A. H. 1966. International experiments in glacier sounding, 1963 and1964. Can. J. Earth Sei., Vol. 3, No. 6, p. 887-92.

WALFORD, M. E. R. 1964. Radio sounding through an ice shelf. Nature, Vol. 204,No. 4956, p. 317-19.

Measurements of electromagnetic wave velocity in the EastAntarctic ice sheet

BY

JOHN W. CLOUGH and CHARLES R. BENTLEYUniversity of Wisconsin, Department of Geology and Geophysics,

Geophysical and Polar Research Center,6118 University Avenue, Middleton, Wisconsin, 53562

Contribution 223

ABSTRACT

Several measurements of electromagnetic wave velocity in the ice ofDronning Maud Land have been carried out by means of wide-anglereflection profiling and comparison of electromagnetic and seismic echotimes. The mean velocity is found to be 171 ±2 m//nsec, corresponding,for a temperature of — 10°C, to a dielectric constant of 3-12 ±-05. Thisvalue agrees with some field measurements made elsewhere on polar ice,but is significantly different from some others. Echo time comparisonsprobably indicate that a major portion of the ice sheet is strongly aniso-tropic. The existence, and horizontal continuity over at least a few hun-dred metres, of at least 15 internal reflectors at depths between 250 and1250 m has been confirmed. Higher resolution measurements are neededto define velocity variations in the upper 250 m of the ice sheet.

IntroductionElectromagnetic sounding (EMS) of polar ice, first accomplished in 1958

at Wilkes Station, Antarctica, by A. H. Waite (Waite and Schmidt, 1961),is now an established technique. There have been, however, very fewmeasurements of the electromagnetic wave velocity ve, or, equivalently, therelative dielectric constant e', either in polar ice or near the 30Hz operatingfrequency of sounding equipment. Laboratory measurements suggestthat e' lies between 3-1 and 3-2 for pure ice, although there are divergentresults (see review by Evans, 1965).

Two field studies with EMS equipment have been made. Jiracek (1967),using the wide-angle reflection technique at several different Antarctic

116 ISAGE

localities, none of them on the main ice sheet, found e' = 3-2 (ve = 168 m//usec) for pure ice. Pearce and Walker (1967), comparing soundings withthe ice thickness known from the drill hole at Camp Century, Greenland,found a higher value: e' = 3-31 (ve = 165 m//xsec). Apparently, manymore field measurements are desirable, not only for more accurate icethickness determination, but also because velocity variations if they reallyexist would yield valuable information about the physical nature of theice deep within polar ice sheets.

Velocity variations in the upper parts of an ice sheet are also of interestbecause of their dependence on density and on the geometry of the ice-airmixture. Since the velocity decreases with depth, it obviously cannot bedetermined by refraction techniques. Instead, one can in principle takeadvantage of the succession of internal reflectors which have been found toexist down to a depth of 1000 m or more in the ice, and employ the wide-angle reflection method. At the same time, the attitude and horizontalcontinuity of the reflecting surfaces can be investigated.

In this paper we report on several measurements of electromagneticwave velocity in the East Antarctic ice sheet by wide-angle reflections and bythe comparison of electromagnetic and seismic reflection times. Observa-tions were carried out during the 1965-66 and 1967-68 oversnow traversesin Dronning Maud Land.

Wide-angle basal reflections

The wide-angle reflection technique has the dual advantage of being insitu, and of being independent of depth determination by other methods.The use of a common reflecting point eliminates (to the first order) changesin travel time due to bottom topography and bottom slope. At traversestation 840 (79CS, 07°W) two wide-angle profiles, one with a commonreflecting point and one with a fixed end, were completed along the sameline. Standard plots of reflection time squared versus distance squared(T2 — X2 plots) are presented in Figs. 1 and 2. The indicated values forve are 172-8 ± 0-7 m//xsec for the common-reflecting-point profile, and171-1 ± 1-2 m//xsec for the fixed-end profile, the latter having beencorrected for the effect of the subglacial topography. Correction for thenear-surface layers yields (rounded off) ±1 vc = 172 ± 1 m//xsec and170 ± m/usée, respectively, for the ice, corresponding to e = 3-07 +•05. This is smaller than previous field determinations, although it over-laps the range of laboratory measurements (e.g., Auty and Cole, 1952).

Laboratory experiments on Arctic glacier ice have indicated a slightvariation of e' with temperature (W. B. Westphal, reported in Jiracek,1967), although independence of temperature is observed in pure ice pro-duced artificially (Auty and Cole, 1952). Using Westphal's results andcorrecting to — 10°C, assuming a mean temperature in the Antarctic ice of—50°C, we obtain e' = 3-12 ± -05 (ve = 170 ± 1 m//xsec), still slightlyless than previous measurements.

DEEP SOUNDING : ECHO RESULTS 117

24 -

22

EMS T 2 - X 2

MILE 840COMMON REFLECTING POINT

I I I I I I Z400 800 1200 1400 1600 1800 2000

DISTANCE (METERS)

2200 2400 2600

FIG. 1. T2-X2 for station 840, common-reflecting-point profile. X's denote pointsnot used in the least-square analysis.

For comparison, we present a T2 — X2 plot (Fig. 3) for a commonreflecting-point wide-angle profile carried out on the Barnes Ice Cap onBaffin Island under the auspices of the Geographical Branch of theCanadian Department of Energy, Mines and Resources, during May 1967.The mean ice cap temperature is about — 12°C. The indicated velocity,171-4 ± 0-8 m/^sec (e' = 3-07 ± 0-03), which requires no correction fornear surface variations, is essentially the same as that for Dronning MaudLand.

Travel time comparisons

The travel times for vertical electromagnetic and seismic reflections at13 stations, covering a range of ice thickness from about 2200 m to 3400 mshow a reasonably consistent relationship (Fig. 4). The travel times havebeen adjusted by 0-12 /xsec and —03 sec, respectively, to correct for theeffect of the near-surface velocity structure. The least-square regressionline equation is

te = (22-8 ± 1-0) tp + 2-0 ±1-5 , (1)where te and tp are the electromagnetic and compressional wave traveltimes in /xseconds and seconds, respectively. For a normal seismic velocityvp =• 3900 m/sec, Equation (1) yields ve = 170 ± 7 m//xsec. This agrees

118 ISAGE

24 -

22

EMS T2-X2

MILE 840

- t i l lBOTTOM

I I IPROFILE

1500 M-1600 M

1700 M

I I400 800 1200 1400 1600 1800 2000

DISTANCE (METERS)

2200 2400 2600

FIG. 2. T2-X2 plot for station 840, fixed-end profile. Note that the bottom profilecovers the same distance as the T2-X2 plot, but with a linear scale. The data pointsand the fitted velocity have not been corrected for the bottom topography.

with the independent determination of ve discussed above, but clearlythe precision is not sufficient to gain much new velocity information.Furthermore, a problem remains concerning the positive te intercept inEquation (1) which may affect the velocity ratio, as we shall discuss next.

For a simple model of the ice sheet as a homogeneous, isotropic slab ofice lying on a discrete reflecting base, te should clearly be directly propor-tional to tp. The te intercept of + 2 /xsec seen in Fig. 4, however, appearsto be real. Furthermore, similar time differences between the two tech-niques have been reported by Jiracek (1967), Robin et al. (in press) andD. Carter (personal communication). Several possible explanationsshould be considered :

(1) Systematic errors in time measurement are responsible. This maybe true in part, but it is unlikely that these errors could be as large as 2/xsec.The error in seismic travel time may be taken as about -005 sec, whichcorresponds approximately to a 0-1 /^sec change in intercept. The resolu-tion of the electromagnetic travel times is estimated to be 0-5 /xsec orbetter. (It is assumed that the time required for the signal to pass throughthe receiver and display equipment is the same for the echo and the initial

DEEP SOUNDING : ECHO RESULTS 119

7 —

EMS T - XBARNES ICE CAPBAFFIN ISLAND

I A & 8 10X HUNDREDS OF METERS

FIG. 3. T2-X2 plot for the Barnes Ice Cap, Baffin Island.

pulse.) It is difficult, therefore, to see how systematic errors could causemore than 1 /4 of the observed intercept.

A good check on the index error for vertical electromagnetic reflectionsis provided by T2 — X2 plots from wide-angle profiles. Because the trans-mitter and receiver are separated by hundreds of metres along these pro-files, the direct pulse no longer saturates the receiver, so that the intercepttime may be a more accurate measure of the true vertical reflection timethan a vertical sounding. This intercept is, in fact, often observed to beless than the vertical time, the difference reaching 0-75 /xsec in one case(Fig. 1) but more often being 0-2 /̂ isec or less (Fig. 2 and other profiles notpresented here). Note that the discrepancy appears too abruptly at dis-tances less than about one tenth of the ice thickness to be attributable toanisotropy.

120 1SAGE

40EMS TRAVEL TIME VS.SEISMIC TRAVEL TIME

EAST ANTARCTICA

.2 4 .6 .8 1.0 1.2 1.4 1.6

SEISMIC TRAVEL TIME (SEC.)

1.8 2.0

FIG. 4. Comparative electromagnetic and seismic comprcssional wave echo times,Dronning Maud Land. Solid line indicates least-square regression fit; dashed line isexplained in text.

(2) There is a real difference in reflecting boundary for the two wavetypes, the electromagnetic reflector lying deeper. This possibility is dis-cussed by Jiracek (1967) in the case of Roosevelt Island, where there is anadditional complication in an apparent requirement for the wave velocityto increase beneath the ice. He found a satisfactory model difficult todevise. Even without the velocity restriction in the present case, it is stillnecessary to postulate a layer more than 100 m thick, which must present alow electromagnetic but high acoustic impedance contrast with the ice.(Geometrical properties of the boundaries cannot be called upon becauseof the near equality of seismic and electromagnetic wave lengths—about15 m and 5 m, respectively.) Suitable earth material does exist, e.g., high-aluminium-clay permafrost (Cook, 1960), but there is no geological supportfor the existence of such a thick, extensive and uniform layer of specialisedcomposition. It is significant in this regard that no EMS echoes fall nearly2 /xsec below the regression line in Fig. 4. The hypothesis of different,widely separated reflecting boundaries does not appear tenable for

DEEP SOUNDING : ECHO RESULTS 121

Dronning Maud Land. (We do not deny the possible validity of thishypothesis elsewhere.)

(3) Either ve or vp or both differ considerably from the expected value.To get an idea of the velocities required, let us force the regression linethrough the origin, obtaining a slope of 24-3 /xsec/sec. Then for ve = 170m//xsec we would have vp = 4130 v/sec, whereas vp = 3900 would implyve = 160m/jiisec(e' = 3-5). The only conceivable cause of such discrepantvelocities would be anisotropy.

Measurements of e' on single crystals at frequencies near 30 MHz havenot been reported. Experiments now in progress indicate, however, thatthe difference between e' parallel to, and normal to the c-axis is less than1 per cent in the microwave region (W. B. Westphal, personal communica-tion). Since there are no critical frequencies in the intervening frequencyband, it is probable that the anisotropic variation of electromagnetic wavevelocity at 30 MHz is less than 1 m//xsec, and is thus negligible in the prob-lem at hand.

Compressional wave velocities are much more strongly affected by aniso-tropy, the maximum value along the c-axis exceeding the polycrystallineaverage by more than 200 m/sec. Strong anisotropy has been observed inthe lower part of the ice column in the deep drill hole at Byrd Station(Gow et al, 1968), and has been shown by seismic measurements to bewidespread in west Antarctica (Bentley, 1969, in press). In the Byrdcore a strong preferred alignment of c-axes near the vertical was observed.

Let us then suppose the East Antarctic ice sheet to be made up partly ofisotropic, and partly of anisotropic ice. If, in the investigated region ofDronning Maud Land, ice thickness changes from place to place reflectchanges in the thickness of the isotropic fraction, the regression slope of te

on tp would equal vp/ve in isotropic ice, and the regression line would havea non-zero intercept, as in Fig. 4.

To attempt a quantitative fit to our data, we may postulate a morespecific model. We will assume an extreme case: a layer with a constantthickness of 2500 m in which there is perfect vertical alignment of c-axes,and a layer of isotropic ice of variable thickness, in which vp = 4100 m/secand 3900 m/sec, respectively, for vertically propagating waves. Thecorresponding te vs tp relation would then follow the solid line in Fig. 4 fortp > 1-2 sec, and the dashed line for tp < 1*2 sec, the latter segmentreferring, of course, to a total ice thickness of less than 2500 m, so that theanisotropic ice thickness would perforce be changing. With this model apositive intercept of 0-5 /xsec still remains—this much, perhaps, could beattributed to systematic errors.

The model described is too extreme to seem very satisfactory. It isworth noticing, however, that, in many instances, velocity variations inWest Antarctica can be explained only by extreme models. Further-more, if the bottom several hundred metres of ice are isotropic, as atByrd Station, major changes in total ice thickness over rugged subglacialterrain might indeed occur primarily in the isotropic layer. The model

122 ISAGE

thus at least has the merit of qualitative support from independentglaciological data. We conclude that seismic anisotropy, together withsome systematic errors in time measurement, is the most likely explana-tion for the observed intercept.

Wide-angle internal reflectionsDetailed measurements of wide-angle reflections internal to the ice were

made at traverse station 500 (76°S, 09cE). Fig. 5 shows eight oscilloscope

?käk j

4 4 0

430

4 2 0

410

16

OSCILLOSCOPE DISPLAYS

FIG. 5. Oscilloscope displays of a section of a detailed wide-angle profile, showinginternal reflections. Numbers along side the photographs give the distance betweensource and receiver. Travel times are indicated relative to the arrival of the directpulse through the air.

displays from a portion of this profile. The traces are stacked to aid in thecorrelation of events from one record to another, which is the major diffi-culty in the analysis of these data. The first pulse on the left, extending toabout 2 jasec, is the direct pulse through the air. The depth correspondingto 2 /xsec is approximately 200 m ; consequently, any reflections within thefirst 200 m are masked by the direct pulse and results cannot be obtainedfor this zone. The echo time for the ice-rock interface is off-scale; all the

DEEP SOUNDING : ECHO RESULTS 123

reflections shown are from within the ice. A quick inspection of the photo-graphs shows that some events are easy to correlate from one record to thenext, whereas others are more variable in nature and maybe present only ina portion of the records.

200 400 600 800X-METERS

1000 1200 1400 1600

1 6 - •

200 400 600 800 1000 1200

X - M E T E R S1400 1600

FIG, 6. T2-X2 plot for detailed wide-angle profile of internal reflections, station 500.

All distinguishable reflections were included on a T2 — X2 plot (Fig. 6).Connecting lines were drawn, however, only where the corresponding cor-relation could be made on the oscillograms; the partial lines thus representarrivals which were consistent for a range of distances, but were missing orextremely variable at other distances.

Mean velocities from all T2 — X2 lines extending from zero distance, andfrom the line which would have an intercept of 9-5 /xsec, were plotted as afunction of depth (Fig. 7). Error bars were determined by estimatinggraphically the maximum and minimum slopes which appeared subjectivelyreasonable.

Superimposed on the plot is a computed mean velocity-vs-depth curvedetermined from the empirical velocity-density relationship ve = c/(l +O-85/a) where c is the electromagnetic wave velocity in free space, and p isthe density in gm/cm3 (Robin et al, in press), and the density-depth data

124 ISAGE

0

200

4 0 0

igkl 600ki

800

1000

1200

i—•-

i—•

VELOCITY-DEPTHCURVE

MILE 5 0 0WIDE-ANGLE

160 170 180 190 200VELOCITY (METERS//^ SEC)

210

FIG. 7. Electromagnetic wave velocity vs depth for station 500, computed from linesof Fig. 6. Solid curve is a computed model of velocity vs depth. Error bars wereestimated subjectively.

from the Byrd Station drill hole (Gow, personal communication). It isclear that the resolution is insufficient to detect real deviations in velocityfrom the mean curve. The good general agreement does, however, provethe reality of some fifteen internal reflectors between depths of 250 m and1250 m and it demonstrates the horizontal continuity of some of thereflectors over distances of hundreds of metres. (Note that the reflectionlines are limited in length by ray path geometry, not by the horizontalextent of the reflectors.) Since a dip of 1° on a reflecting surface would

DEEP SOUNDING : ECHO RESULTS 125

produce a velocity error of only 1 or 2 m//xsec, we can conclude only thatthe reflectors do not have dips greater than a few degrees.

Profiles with larger station spacings than those at station 500 were com-pleted at station 840. The character of the return echoes varies signifi-cantly over the distance from one station to the next, making it difficult orimpossible to correlate individual arrivals. For this reason the reflectionlines in the T2 — X2 plots (Figs. 8 and 9) and the resulting velocity-depthprofile (Fig. 10) are less well determined than those already discussed.Nevertheless, the velocity-depth profile agrees satisfactorily with the pre-dicted result.

14'

12

10

8

6

4

I I I 1

EMS T-X2

MILE 840 NEAR~ COMMON REFLECTING

^ ^ \ ^ ^ ^ \ ^ ^

P:'^''î;:i755 3 3-^5^:^76^2,3.0': S = ' l 1 1 1

1

S U R F A C E

POINT•

^ 169.4 , 5.5

172.6,4.5,3.7

1

1

•

•

^ ^-"^^-167.5

58.2 ,6.3

1

1 1

•—^^-^170.9 ,9.3

^^169 .7 ,9.0-171.2,8.73.2,8.1 —,7.4

—

—

• * -

1 1400 600 800 1000 1200 1400 1600

D I S T A N C E (METERS)

1800

FIG. 8. T2-X2 plot for detailed wide-angle profile of internal reflections, station 840,common-reflecting-point profile. Indicated velocities and intercepts are given along-side each line.

Conclusions

(1) The electromagnetic wave velocity in the glacier ice of central Dron-ning Maud Land is 171 ± 2 m//Asec, as determined in situ by wide-anglereflection profiling. This is equivalent, after a small temperature correc-tion to e' = 3-12 ± 0-05 at a temperature of — 10°C. This is significantlysmaller than values measured for the ice of Skelton Inlet and extrapolatedfor other Ross Ice Shelf stations (Jiracek, 1967), and for Camp Century onthe Greenland ice sheet (Pearce and Walker, 1967), but is in close agreementwith measurements on Roosevelt Island (extrapolated from Jiracek, 1967)and on the Barnes Ice Cap. The explanation for these differences is notyet known.

126 ISAGE

14 —

Uj10

—1 1——i r

EMS T-X2

MILE

i: V i

: 840

1 1

"^X/166.1+10.2

• ,-X/l69.4+7.94

^•"T X/I7I.5+6.85 —

-^</169.l+5.35

180+4.12

—

400 600 800 1000 1200 1400 1600

D I S T A N C E (METERS)1800

FIG. 9. T*-X2 plot for detailed wide-angle profile of internal reflections, station 840,fixed-end profile. Travel-time equations are given alongside each line.

(2) Comparative electromagnetic and compressional-wave reflectiontimes are consistent with ve = 171 m//nsec, but the precision is not high.The relative delay of about 2 /usec observed for all electromagnetic echoesalmost certainly reflects a real glaciological phenomenon, although afraction of the delay may result from unidentified electronic causes. Ahypothesis of different reflecting boundaries for the two wave types isrejected as geologically unreasonable and completely ad hoc. It isprobable that it is actually the compressional waves which are speeded upby propagation through a thick zone of anisotropic ice in which thecrystal c-axes are largely vertical; it is further likely that the zone ofanisotropic ice stays relatively constant in thickness throughout the areaof Dronning Maud Land investigated.

(3) Correlative wide-angle reflections have confirmed the existence of atleast 15 internal reflectors at depths between 250 m and 1250 m in the iceat one location in Dronning Maud Land. These reflectors are continuous,or nearly continuous, over horizontal distances of at least a few hundredmetres. Mean velocities in the upper 1200 m determined from these pro-files are in good agreement with estimates based on an approximate density-depth profile and an empirical velocity-density relation. The precision

DEEP SOUNDING : ECHO RESULTS 127

200

400

kj600

^ 8 0 0

1000

I I I I I I I I I I I I I I I I I

I I I I

VELOCITY

DEPTH CURVE —

I I 1 I I I I I 1 M I I I I I I I160 170 180 190 200

VELOCITY (METERS//* SEC)

210

FIG. 10. Velocity vs depth for station 840, computed from the lines in Fig. 8 (solidcircles) and Fig. 9 (open circles). Solid curve is a computed model of velocity vs depth.

of these measurements is not yet high enough to ascertain the reality ofsome possible anomalies. High-resolution, short-pulse reflection studiesof the upper 250 m would be very valuable.

Acknowledgements

We wish to thank John Beitzel, and Carl K. Poster for their invaluableassistance with the field programme, and we gratefully acknowledge thesupport of the National Science Foundation under grant GA552.

REFERENCES

AUTY, R. P. and COLE, R. H. 1952. Dielectric properties of ice and solid D2O,J. Chem. Phys. Vol. 20, p. 1309-14.

BENTLEY, C. R., (in press). Seismic anisotropy in the West Antarctic ice sheet, Am.Geophys. Union Antarctic Monograph Series, Claciology II.

BENTLEY, C. R. 1969. Seismic anisotropy in the West Antarctic ice sheet. In:Gow, A. J. et al., Eds., International Symposium on Antarctic GlaciologicalExploration (ISAGE), Hanover, New Hampshire, U.S.A., 3-7 September 1968.Cambridge (Pub. No. 86 of IASH), p. 128-130.

COOK, J. C. 1960. RF electrical properties of salty ice and frozen earth. J.Geophys. Res., Vol. 65, p. 1767-71.

EVANS, S. 1965. Dielectric properties of ice and snow—a review. / . Glac, Vol. 5,p. 773-92.

Gow, A. G., UEDA, H. T. and GARFIELD, D. E. 1968. Antarctic Ice Sheet:preliminary results of first core hole to bedrock. Science, Vol. 161, p. 1011-13.

JIRACF.K, G. R. 1967. Radio sounding of Antarctic ice. University of WisconsinGeophysical and Polar Research Center Research Report Series, 67—1.

128 1SAGE

PEARCH, P. C. and WALKER, J. W. 1967. An empirical determination of the relativedielectric constant of the Greenland Ice Cap. J. Geophys. Res., Vol. 72, No. 22,p. 5743-47.

WAITE, A. H.anrfScHMiDT, S. J. 1961. Gross errors in height indication from pulsedradar altimeters operating over thick ice or snow. IRE Convention Record,Part 5, p. 38-54.

Seismic anisotropy in the west Antarctic ice sheet

BY

CHARLES R. BENTLEY

University of Wisconsin, Department of Geology and Geophysics,Geophysical and Polar Research Center,

6118 University Avenue, Middlcton, Wisconsin 53562

ABSTRACT

Seismic reflection and refraction measurements made during severaltraverses in west Antarctica have provided clear indication of anisotropicwave propagation in the ice sheet. These indications include marked in-creases in the refracted compressional and shear wave velocities at depths ofhundreds of metres in the ice, dependence of the propagation velocities ofwaves reflected from the bottom of the ice on the angle of incidence andalso on geographic azimuth, and a variation of shear wave propagationvelocity with direction of polarization.

All of these characteristics are found for propagation through a singlecrystal of ice. An analysis of the velocity variations has therefore beenmade on the assumption that the ice at depth behaves approximately as asingle crystal, i.e., that there is perfect alignment of the c-axes. This is not,of course, a physically realistic assumption, but will give nearly the sameelastic wave velocity structure as a random, or axially symmetric, distribu-tion of c-axes within a cone having an apex angle of, say, 30° or less.

To find the best model fitting the observations at each seismic profile, aset of tables and curves relating propagation velocity for P and S waves toangle of incidence at the base of the ice, with the inclination and azimuth ofthe "single crystal" axis as parameters, was prepared. The procedure usedwas then to seek the crystal orientation corresponding to the set of curveswhich best fit all the seismic observations at each station. Data employedincluded reflected P waves, converted P to S reflections, and P and S refrac-ted arrivals. Reflected S-waves, although often present, were found to benot very definitive owing to a four-fold multiplicity of possible wave types.They, therefore, were not used in model selection.

A best-fitting model having been selected, the corresponding crystalattitude in space was then compared at each station with the expecteddirection of ice flow. At those stations where the most confident selec-tions could be made, the azimuth of the c-axis orientation was downslope.