Moore_McAdoo_NF Sedi Geol

Intended for publication as a research article in Sedimentary Geology
Landward fining from multiple sources in a sand sheet deposited by the 1929
Grand Banks tsunami, Newfoundland
Andrew L. Moorea*, Brian G. McAdoob, and Alan Ruffmanc
a
Department of Geology, Kent State University, Kent, OH 44242
Department of Geology and Geography, Vassar College, Poughkeepsie, NY
c
Geomarine Associates, Ltd., Halifax, NS
b
*Corresponding author: Tel: +1-330-672-9465, Fax: +1-330-672-7949, Email:
[email protected]
1
ABSTRACT (500 word maximum):
2
A sandy deposit from the 1929 Grand Banks tsunami in Newfoundland contains sediment
3
from two distinct sources, one from a gravel shoreline close to the deposit, and one from a sandy
4
bay-mouth bar some 200 m seaward of the deposit. We took approximately 100 core samples of
5
this deposit in an attempt to characterize lateral grain size trends within the sand. Although the
6
coarse fraction does fine with distance inland, the fine fraction does not change size over the
7
study area, and the aggregate grain size changes in no systematic way.
8
9
We interpret this deposit to represent the mixture of material picked up at the bar with
material picked up at the gravel shoreline. The bar material does not fine in part because it is
10
already fairly well sorted, but also because it is far from its source. The shoreline material, on the
11
other hand, is poorly sorted so that the tsunami took only those grains it was capable of moving,
12
and deposited them closer to their source.
13
We estimated the size of the tsunami by determining the flow depth-flow velocity
14
combinations required to advect sand from the bar to the back of the deposit, and by estimating
15
the shear velocity required for motion of the largest grain we found during our survey. This
16
modeling indicates an average flow depth of about 2.5-2.8 m over the area, at a flow velocity of
17
1.9-2.2 m/s. This estimate compares well with eyewitness accounts of a maximum flow depth of
18
7 m at the shoreline if our estimate represents an average over the whole study area. (263 words)
19
20
KEYWORDS: Tsunami, Newfoundland, sedimentology
21
22
23
1. Introduction
Since the onset of modern tsunami sedimentology with Konno’s groundbreaking study of
24
deposition following the 1960 Chilean tsunami (Konno, 1961), much research has been focused
25
not only on describing and interpreting the deposits of modern tsunamis (e.g., Dawson et al.,
26
1996; Gelfenbaum and Jaffe, 2003; Minoura et al., 1997; Nanayama et al., 2000; Nishimura and
27
Miyaji, 1995; Sato et al., 1995; Shi et al., 1995) but also on the deposits of prehistoric tsunamis
28
(e.g. Atwater and Moore, 1992; Benson et al., 1997; Bourgeois and Johnson, 2001; Dawson et
29
al., 1988; Goff et al., 2000; Hemphill-Haley, 1995; Minoura et al., 1994; Nanayama et al., 2003;
30
Pinegina and Bourgeois, 2001). The deposits of historic tsunamis play an important transitional
31
role between these two—generally, eyewitness accounts can help constrain both the tsunami
32
flow parameters and the original extent of deposition, but at least some “taphonomic” processes
33
have occurred, so that the deposit will begin to approximate a paleotsunami deposit. Historic
34
tsunamis, then play an important role in interpreting not only what changes may take place in a
35
tsunami deposit, but also in interpreting the hydraulics of ancient tsunamis.
36
The 1929 Grand Banks tsunami plays an especially important role in this regard.
37
Although its effects were relatively local, both the tsunami and its source have been well studied
38
(e.g., Doxsee, 1948; Hasegawa and Kanamori, 1987; Piper et al., 1988). Most important, Heezen
39
and Ewing (1952) presented evidence that the tsunami was generated by an offshore landslide in
40
a landmark paper that detailed the behavior of the resulting turbidity current using the location
41
and timing of trans-Atlantic telegraph cables. Later studies (Piper et al., 1999; Piper et al., 1985)
42
showed that the evolution of the landslide was more complex than first envisioned, including a
43
series of delayed, retrogressive slumps that added to the ultimate magnitude of the failure.
44
Regardless, the tsunami was recorded not only in Maritime Canada, but also as far away as
45
Charleston, South Carolina (USA), the Azores, and Portugal (Fine et al., 2005; Ruffman, 1997).
46
The 1929 Grand Banks tsunami, then, stands as one of the few historically documented
47
landslide-generated waves; investigating its deposits should allow a better understanding not
48
only of the distribution and organization of landslide generated tsunami deposits, but also leaves
49
open the possibility of interpreting those deposits to infer something of the hydraulics that
50
produced them.
51
52
The 1929 Grand Banks earthquake and tsunami
53
The M=7.2 1929 Grand Banks earthquake and subsequent tsunami remain Canada’s
54
worst earthquake-related disaster, killing 27 people on Newfoundland’s Burin Peninsula.
55
Eyewitnesses reported damage along nearly 50 km of deeply embayed coastline facing the
56
epicenter (Ruffman, 1991; Ruffman, 1996). Tuttle and others (2004) searched this coastline for
57
deposits from the 1929 event, but found deposits only at Taylor’s Bay, near the western end of
58
the damage area (Fig. 1). Witnesses at Taylor’s Bay described three waves coming ashore
59
during the evening of November 18, 1929, reaching a maximum flow depth of perhaps 7 m, and
60
a maximum runup of 8.5 m. Survivors also described the deposition of a layer of pebbles and
61
sand in the wake of the tsunami (Tuttle et al., 2004). Taylor’s Bay provides an almost ideal
62
setting to create, trap, and preserve tsunami deposits, which is perhaps why it alone either had
63
originally deposited or preserved a sand sheet from the 1929 event.
64
Tuttle and others (2004) document the deposit characteristics and geometry at Taylor’s
65
Bay. They show one to three units of very coarse- to fine-grained sand present in the marsh
66
peats that consistently fine upward and landward, which is characteristic of tsunami deposits. In
67
addition to the sand fraction, the deposit consists of saltwater tidal marsh grasses and littoral,
68
benthic and planktonic marine diatoms, suggesting that the high energy, turbulent wave gathered
69
sediment from the seafloor as it approached the beach.
70
In this study, we revisit Taylor’s Bay with the aim of continuing the work done by Tuttle
71
and others (2004) by learning more about the wave characteristics by studying the sediment. We
72
are able to distinguish sediment sources with different grain size by very careful analysis of
73
homogenized sediment cores. By isolating sediment sources, we are able to better model
74
sediment transport, using the characteristic landward-fining to estimate wave velocity, hence
75
amplitude. Taylor’s Bay provided an excellent locale to test this model as eyewitness accounts
76
constrain our findings.
77
78
2. Field Survey
79
80
Geologic setting
Taylor’s Bay is a glacially carved bay along the southern coast of the Burin Peninsula
81
(Fig. 1). Soils here are shallow (typically less than 50 cm), and rest on an eroded surface of the
82
Proterozoic Marystown Group, a sequence of rhyolitic volcaniclastics (O'Brien et al., 1995) that
83
also outcrop along the sides of the bay, forming low cliffs. A bay-mouth bar has formed along
84
the back of the bay (Fig. 1) and encloses a shallow “pond” that has a shingle shoreline fronted by
85
pioneering marsh. The bar is composed of well sorted medium sand, and is dominantly quartz, in
86
contrast to the other sediments in the area, which are generally lithic fragments. In 1929, the
87
coast road ran along the top of this bar, and the tsunami destroyed it. In 1977 the road was
88
replaced, and now runs along the former shoreline of the back bay pond.
89
The tsunami deposit at Taylor’s Bay is found in a freshwater marsh immediately
90
landward of the pond. The marsh is dominantly heather and other woody plants. In November,
91
1929, winter would have been well underway in Newfoundland, but the woody structure of the
92
heather would have remained, and acted as a trap for incoming sediment low in the water
93
column, thus acting as a trap.
94
95
96
Field observations of the tsunami deposit
Field observations of the sediment deposits at Taylor’s Bay suggest a very heterogeneous
97
deposit, with few obvious patterns with the exception of the expected landward thinning with
98
increasing elevation. The layer of sand forms a sharp contact with the underlying peat, which is
99
~40 cm thick throughout the area. The basal peat often contains angular sands and gravels eroded
100
from the underlying bedrock. The tsunami sand fines upward from very coarse sand with pebble-
101
sized material in places, to very fine sand. There is little evidence of in situ vegetation at the
102
boundary of the deposit with the underlying peat, and in the thickest sections, the deposit
103
consists of three subunits separated by peaty laminations (Tuttle et al., 2004), suggesting that the
104
tsunami flow initially eroded some peat before depositing the sand. The contact between the
105
deposit and the overlying peat is less distinct, due to bioturbation (Ruffman and Tuttle, 1994).
106
The thickness of the deposit varies by up to 25 cm within several meters. On average, the
107
deposit is around 4 cm thick, filling in depressions and thinning over paleosurface highs, such as
108
boulders and grass root balls. The deposit is discontinuous over the sampled area, and thins
109
noticeable inland.
110
The tsunami deposit consists of an olive (5Y 5/3), strongly coarse skewed, moderately
111
sorted medium sand. The sand often displays some bimodality (Fig. 2), but this is not seen
112
throughout the deposit. The grains themselves are subrounded to subangular, and primarily
113
composed of lithic fragments and quartz. Most of the lithics are cherty, and indistinguishable
114
from the local bedrock. Carbonate grains are rare or absent. The sand is slightly denser than
115
quartz (2.77 g/cm3), consistent with the density of the local bedrock. The sand commonly
116
appears massive in exposure (Fig. 3); no sedimentary structures were visible in the study area.
117
118
119
Deposit architecture: methods
To determine areal trends within the deposit, we collected sediment data on a 10 m x 10
120
m grid using a 2 cm diameter friction corer. Each hole was driven to refusal, and each line of
121
cores was extended landward until two successive holes showed no sand. At each location, we
122
recorded the stratigraphy of the core and collected the sand sheet (if present) for later analysis.
123
A secondary goal of this study was to demonstrate a rapid method of characterizing a
124
tsunami deposit in the field. In the rapid-assessment phase of modern tsunamis (such as the
125
recent 26 December 2004 Sumatran event), it is critical to be able to move into a location and
126
obtain sediment information before it is disturbed. By taking a number of rapid borings over a
127
large area, we use the lateral variations in grain size without attempting to preserve vertical
128
stratigraphy. During the reconnaissance phase of tsunami surveys, geologist can aid in
129
observations while covering larger areas in a timely manner, rather than focusing on only a few
130
locations along a linear transect. During this first trial of the method (where we could not
131
actually see the deposit) we were able to cover the entire 6.6 hectare deposit, collecting data and
132
samples at over 100 locations, all mapped using a laser theodolite in just over 10 hours. The
133
results were consolidated into a GIS database.
134
In the lab, all samples were heat-treated in a drying oven set to 125° C for 48 hours to
135
ensure that the samples were inert. After heating, samples were immersed in 3% hydrogen
136
peroxide for 48 hours to remove most organic material, then decanted to remove minor amounts
137
of silt, clay, and residual organic matter. The resulting sand and gravel was then dried in a drying
138
oven set to 80° C for 24 hours or until dried.
139
To determine the grain size within the deposit, we dry sieved each sample at ¼-Φ
140
intervals from –2Φ to 4.5Φ, and weighed the resulting sample splits (Figure 2). Samples lost less
141
than 5% of their mass during sieving. Sediment density was determined by taking a 10 g aliquot
142
of sediment from a sample chosen at random, and determining sediment density of this aliquot in
143
a Micromeritics Accupyc 1330.
144
145
Deposit architecture: results
146
By collecting the data over a semi-regular grid, we were able to plot the data in map view
147
in an attempt to recognize spatial trends. This has an advantage over transects in that we can see
148
how sediment characteristics might vary over two dimensions. For example, the elevation to the
149
base of the sand layer gives us an idea of the paleosurface (Fig. 4). The base of the sand layer is
150
deeper near the ocean, and comes closer to the surface landward as peat production is likely to be
151
higher in the marshy area near the pond. There is a zone of thick sediment near the northern end
152
of the sampled area that may represent a pre-existing low such as a drainage channel. The
153
isopach map of the area (Fig. 5) shows that the tsunami deposit thickness largely tracks with the
154
base of the sand, suggesting that the sand simply fills in topographic lows.
155
Although we expected median grain size to fine landward, with little longshore variation,
156
we were puzzled by the lack of pattern we determined by sieving (Fig. 6). Further, the most
157
likely source for the tsunami sediment, the bay-mouth bar, does not contain sand coarser than ~1
158
mm, although sediment up to 8 mm is common in the tsunami deposit. We considered that a
159
second source of sediment must have mixed with the bay-mouth bar sand, forming the bimodal
160
tsunami deposit; in order to separate these two populations, we tried to increase the resolution of
161
our grain size curves.
162
163
3. Peak deconvolution
164
Methods
165
To increase the resolution of our grain size curves, we used a Retsch Camsizer, an
166
optically based instrument capable of determining grain size to within ±1% over the range 30 µm
167
to 30,000 µm (~5Φ to -5Φ). The instrument images a falling curtain of sediment at 25 Hz, then
168
determines the grain size of each particle in the image, in our case by determining the cross-
169
sectional area of the particle and then reporting the diameter of a circle of equivalent area. This
170
tends to increase the grain size relative to sieving, but probably yields a result more compatible
171
with settling tube. Because the instrument made between 2 million and 7 million individual
172
measurements on our samples (depending on sample size), the resulting dataset can easily
173
support 1/16 Φ to 1/32 Φ resolutions (Figure 2).
174
Camsizer analysis of grain size shows that the sand sheet is a mixture of at least two
175
distinct sediment populations—a coarser peak centered near ½ Φ, and a finer peak centered near
176
2 Φ. These populations are present in differing admixtures (Figure 7), resulting in median grain
177
size dominated by the peak ratio rather than by shifts in either peak.
178
To determine if shifts in the size of either peak might be areally correlated, we used peak
179
deconvolution software (Seasolve PeakFit v. 4.12) to separate the two peaks. We used a quasi
180
log-normal distribution to model each population, and were thus able to adjust not only the mean
181
and standard deviation of each peak, but also the skewness. For each sample, we used Camsizer
182
data at 1/16 Φ resolution, yielding nearly 100 data points to define each grain size curve.
183
Figure 8 is typical of these results. Sediment in the deposit can be separated into four
184
distinct peaks—a low-amplitude coarse-skewed peak centered around -1 Φ, a poorly sorted
185
coarse-skewed peak centered near ½ Φ, a moderately sorted coarse-skewed peak centered near 2
186
Φ, and a low-amplitude fine-skewed peak near 3 Φ. The -1 Φ peak is defined mostly by data
187
points containing fewer than 30 particles each, and probably represents rare occurrences where
188
two particles touched during imaging, and were recorded as a single larger grain. Similarly, the 3
189
Φ peak probably consists mostly of organic residue from the digestion of rootlets and peat within
190
the sample. The 2 Φ peak is remarkably similar in shape and location with the grain size
191
signature of the modern beach ridge, and is probably derived from the 1929 beach ridge in
192
approximately the same location as the modern. The ½ Φ peak is attended by two smaller peaks
193
that commonly form in optical imaging when non-spheric grains are sampled. These peaks
194
represent grains that fell with their ab face showing (for the coarser sub-peak) and with their cb
195
face showing (for the finer sub-peak), and thus properly belong with the coarse peak The
196
presence of these peaks with the coarse peak, and their absence from the fine peak, suggests that
197
the fine peak sands are somewhat more spheric than the coarse peak sands.
198
The source of the coarse peak is somewhat harder to determine than the source of the fine
199
peak. The coarse peak sands are subangular to subrounded and composed of lithics
200
indistinguishable from the underlying Marystown Group. No modern source for these grains is
201
plainly visible, but this is, in part, because a modern road, laid in 1977, covers what would have
202
been the 1929 shoreline. We suspect that the coarse peak came from this shoreline, but cannot
203
collect samples for comparison.
204
205
Results
206
The results of the grain size peak deconvolution are presented in Figure 9. The D50 data
207
show an overall coarser mixture of sediment with no clear trends in grain size. Similarly, the
208
narrow range of the fine fraction of the sediment shows very little spatial variation, and if
209
anything, a slight coarsening landward. The coarse fraction, however, shows a clear and distinct
210
landward fining trend, one that is not clear in the homogenized D50, nor in the fine fraction where
211
there is not a similar range of grain size.
212
Further evidence that the coarse peak had its source in a shoreline running under the
213
modern road comes from analysis of the proportion of each population present over the area. We
214
determined the relative amount of “coarse” and “fine” material for each sample by determining
215
the ratio of the height of the fine peak to the coarse peak. A map of this ratio over the study area
216
demonstrates that seaward of approximately the position of the modern road, the peak ratio is
217
quite high (7-10), then becomes abruptly low (~1) near the road, and becomes high again
218
landward (Figure 9). This pattern is consistent with dune sand being swept in by the tsunami, and
219
mixing with a coarser sediment population near the road. Before the road, no coarse particles
220
were available to the tsunami, so none are recorded in its deposits. At the pond edge (now buried
221
under the road), the tsunami was able to pick up coarse material and add it to the deposit.
222
However, this material was heavier, and so deposited more quickly than did the dune sand,
223
resulting in a sudden “pulse” of coarse material that decayed with distance from its source.
224
225
4. Hydraulic estimation
226
227
By using high resolution grain size data and deconvolving the resulting data, we can see
228
that the Taylor’s Bay deposit results from the mixture of at least two separate grain size
229
populations. The finer peak matches well the grain size signature of the modern dune (Fig. 8),
230
suggesting that the dunes are the source of fine sand in the tsunami deposit. The coarser material
231
is more difficult to locate, but in 1929 a shingle beach ran along shore where the modern coast
232
road now lies. This beach is the only available source for coarse sands and gravels in the area—
233
bedrock, although near the surface here, cannot be a source for the particles in the deposit
234
because the particles show signs of rounding by transport. Equally, the peat underlying the
235
tsunami deposit tends to be free of coarse material, suggesting that the gravel fraction of the
236
deposit could not have been winnowed from the peat during passage of the wave.
237
To estimate the size of the tsunami at Taylor’s Bay, we started with the assumption that,
238
because the source for the coarse fraction was heterogeneous with respect to grain size, at least
239
some of the grains were too large for the tsunami to move. This contrasts with some recent
240
tsunamis (e.g. the 2004 Indian Ocean tsunami in Sumatra), where the tsunami’s source material
241
was fine enough that all grains could have been moved by the tsunami. If this assumption holds,
242
however, than we can estimate the shear velocity of the flow by determining the critical shear
243
stress required to move the largest grains found in the tsunami deposit. The largest grain we
244
found in 96 cores was a subangular lithic fragment 2.4 cm by 1.5 cm by 1.3 cm. For fine gravel,
245
dimensionless critical shear stress, τ*cr, is often considered to be constant at about 0.045-0.06
246
(e.g., Pitlick, 1992; Wilcock and Southard, 1989). Using these estimates, and assuming the
247
particle density to be equal to the bulk density of the tsunami deposit (2.77 g/cm3), we
248
determined the shear velocity to be between 10.8 and 12.5 cm/s. Such shear velocities yield
249
Rouse numbers from 0.55 to 0.63 for the fine peak mean, and 1.8 to 2.1 for the coarse peak
250
mean, suggesting that the grains may have traveled to the deposit in suspension and incipient
251
suspension, respectively.
252
Because the flow can be treated, at least on shore, as a short-duration unidirectional flow,
253
the law of the wall should hold within the flow. The depth-averaged law of the wall can be
254
written:
255
256
U=
u*   h  z 0   
 ln − 1 −   
h   
κ   z 0 
(1)
257
258
where z0 is commonly taken to be D84, which we know, κ is von Karman’s constant, which is
259
also known, and u* is the shear velocity just estimated. This yields one equation, and two
260
unknowns (Fig. 10).
261
Similarly, if the finer grains were to have traveled in suspension from their source (at the
262
beach ridge) to the farthest part of the flow, the maximum time available for travel is the time it
263
would take for these grains to settle from the top of the flow, a time given by:
264
h
l
=t =
ws
U
265
(2)
266
where ws is the known settling velocity, and ℓ the distance from the beach ridge to the location of
267
the last grains. This also yields one equation and two unknowns (Fig. 10). Solving this two
268
equation system (i.e. the intersection of the two lines in Figure 10) yields an average flow depth
269
of 2.5-2.8 m and an average flow velocity of 1.9-2.2 m/s, with larger flow depths associated with
270
slower flow velocities. These flow estimates yield Froude numbers of 0.37-0.44 and Reynolds
271
numbers of about 5.4 x 106, indicating that the flow was fully turbulent and subcritical during
272
deposition of the sand. Although this flow depth is considerably less than the 7 m maximum flow
273
depth reported by Tuttle and others (2004), it is not inconsistent with it. The maximum flow
274
depth necessarily occurred near the shoreline—just as necessarily the flow depth reached zero at
275
the maximum runup. Our estimate represents an average flow depth over the area rather than an
276
estimate of either extreme.
277
278
Error
279
280
Each of the two lines in Figure 10 is based on the parameters we measured at Taylor’s
281
Bay. Here we describe the effect on our estimate if the measured parameters are incorrect. First,
282
the effect of increasing the size of z0 (i.e. roughening the ground surface) is to shift “LOTW”
283
down and to the right. Effectively, this increases the flow depth of the wave at the expense of
284
flow velocity. Increasing the size of u* (e.g. discovering larger particles in the tsunami deposit)
285
shifts “LOTW” up and to the left. This decreases flow depth, but increases flow velocity. On the
286
other hand, increasing either the advection length of the fine particles or increasing their grain
287
size shifts “Advection Length” up and to the right, increasing both flow depth and velocity.
288
These changes cannot continue unconstrained, however. Flow behind a bore (as behind a
289
tsunami) must be subcritical, so depth-velocity combinations that yield a Froude number much in
290
excess of 1 are unlikely. Equally, flow velocities equal to the shear velocity are unrealistic. Last,
291
the flow depth cannot exceed the runup height. These constraints form a region of permissible
292
flow depth-velocity combinations on Figure 10.
293
As stated previously, because we are unlikely to have found the largest particle to have
294
been moved by the tsunami, our estimate of u* probably represents a minimum. Our estimate of
295
D84 (and therefore z0), however, is based on nearly 100 measurements of grain size on the bed,
296
and is probably relatively precise. Similarly, our estimate of the grain size (and therefore the
297
settling velocity) of the fine peak is relatively precise, but our estimate of the distance traveled
298
by those grains is necessarily an underestimate. Because “LOTW” has a relatively flat slope in
299
FIG, the velocity solution is relatively insensitive to changes in the advection length, but the flow
300
depth is relatively insensitive. Doubling the advection length, for example, effectively doubles
301
the flow depth, but only increases flow velocity by about 5%. We conclude, then, that our flow
302
depth estimate is necessarily minimum estimate because of the possibility that the original
303
deposit extended farther inland, but that distal portions have been eroded post-deposition. Our
304
flow velocity estimate, however, is relatively insensitive to these changes, and is probably a
305
reasonable estimate.
306
307
308
5. Summary
309
Grain size trends in the Taylor’s Bay tsunami deposit are complicated by the occurrence
310
of two sand sources. These sources are relatively distinct, however, and can be distinguished
311
from each other in grain size analyses of sufficient resolution (here 1/16 Φ). Once separated, the
312
individual sediment populations do appear to fine away from their source, although the dune
313
sand population is so far from its source that little fining should be apparent.
314
We estimated the flow depth of the tsunami to have been 2.5 to 2.8 meters, with a flow
315
velocity of 1.9 to 2.2 m/s by estimating the time dune sand would take to travel in suspension
316
from its source at the bay mouth bar to the marsh where the deposit formed. This, coupled with
317
an estimate of the shear velocity from the largest grain to have been moved, yields a single
318
permissible flow depth-flow velocity combination. This estimate is relatively sensitive to the
319
distance traveled by the dune sand, however, and therefore probably represents a minimum
320
estimate. Our estimate is not inconsistent with eyewitness accounts of the wave, and should be
321
applicable wherever the source material contains grains larger than can be moved by the tsunami.
322
323
324
Acknowledgements:
We thank Justin Minder for his help with fieldwork at Taylor’s Bay. This research was
325
supported by the Vassar College Environmental Science Research Fund and the Kent State
326
University Research Council.
327
References:
328
Atwater, B.F. and Moore, A.L., 1992. A tsunami about 1000 years ago in Puget Sound,
329
330
331
332
Washington. Science, 258: 1614-1617.
Benson, B.E., Grimm, K.A. and Clague, J.J., 1997. Tsunami deposits beneath tidal marshes on
northwestern Vancouver Island, British Columbia. Quaternary Research, 48: 192-204.
Bourgeois, J. and Johnson, S.Y., 2001. Geologic evidence of earthquakes at the Snohomish delta,
333
Washington, in the past 1200 yr. Geological Society of America Bulletin, 113(4): 482-
334
494.
335
336
337
Dawson, A.G., Long, D. and Smith, D.E., 1988. The Storegga Slides: evidence from eastern
Scotland for a possible tsunami. Marine Geology, 82: 271-276.
Dawson, A.G., Shi, S., Dawson, S., Takahashi, T. and Shuto, N., 1996. Coastal sedimentation
338
associated with the June 2nd and 3rd, 1994 tsunami in Rajegwesi, Java. Quaternary
339
Science Reviews, 15: 901-912.
340
341
342
Doxsee, W.W., 1948. The Grand Banks earthquake of November 18, 1929. Publications of the
Dominion Observatory, 7(7): 323-335.
Fine, I.V., Rabinovich, A.B., Bornhold, B.D., Thomson, R.E. and Kulikov, E.A., 2005. The
343
Grand Banks landslide-generated tsunami of November 18, 1929: preliminary analysis
344
and numerical modeling. Marine Geology, 215: 45-57.
345
346
347
Gelfenbaum, G. and Jaffe, B., 2003. Erosion and Sedimentation from the 17 July, 1998 Papua
New Guinea Tsunami. Pure and Applied Geophysics, 160(10-11).
Goff, J.R. et al., 2000. Evidence for an earthquake and tsunami about 3100-3400 years ago, and
348
other catastrophic saltwater inundations recorded in a coastal lagoon, New Zealand.
349
Marine Geology, 170: 231-249.
350
Hasegawa, H.S. and Kanamori, H., 1987. Source mechanism of the magnitude 7.2 Grand Banks
351
earthquake of November 1929: double couple or submarine landslide. Bulletin of the
352
Seismological Society of America, 77: 1984-2004.
353
354
355
Heezen, B.C. and Ewing, M., 1952. Turbidity currents and submarine slumps and the 1929
Grand Banks earthquake. American Journal of Science(250): 849-873.
Hemphill-Haley, E., 1995. Diatom evidence for earthquake-induced subsidence and tsunami 300
356
years ago in southern coastal Washington. Geological Society of America Bulletin, 107:
357
367-378.
358
Konno, E., ed., 1961. Geological observations of the Sanriku coastal region damaged by tsunami
359
due to the Chile earthquake in 1960. Contributions to the Institute of Geology and
360
Paleontology, Tohoku University, 52: 1-40.
361
362
363
364
365
366
367
Minoura, K., Imamura, F., Takahashi, T. and Shuto, N., 1997. Sequence of sedimentation
processes caused by the 1992 Flores tsunami. Geology, 25: 523-526.
Minoura, K., Nakaya, S. and Uchida, M., 1994. Tsunami deposits in a lacustrine sequence of the
Sanriku coast, northeast Japan. Sedimentary Geology, 89: 25-31.
Nanayama, F. et al., 2003. Unusually large earthquakes inferred from tsunami deposits along the
Kuril trench. Nature, 424: 660-663.
Nanayama, F. et al., 2000. Sedimentary differences between the 1993 Hokkaido-nansei-oki
368
tsunami and the 1959 Miyakojima typhoon at Taisei, southwestern Hokkaido, northern
369
Japan. Sedimentary Geology, 135: 255-264.
370
Nishimura, Y. and Miyaji, N., 1995. Tsunami deposits from the 1993 Southwest Hokkaido
371
earthquake and the 1640 Hokkaido Komagatake eruption, Northern Japan. In: K. Satake
372
and F. Imamura (Editors), Tsunamis: 1992-1994, their generation, dynamics, and hazard.
373
Birkhäuser, Basel, pp. 719-734.
374
O'Brien, S.J., O'Driscoll, C.F., Green, B.A. and Tucker, R.D., 1995. Pre-Carboniferous geology
375
of the Connaigre Peninsula and the adjacent coast of Fortune Bay, southern
376
Newfoundland. Current Research, Newfoundland Department of Mines and Energy
377
Geological Survey, Report 1995-1: 267-296.
378
Pinegina, T.K. and Bourgeois, J., 2001. Historical and paleo-tsunami deposits on Kamchatka,
379
Russia: long-term chronologies and long-distance correlations. Natural Hazards and Earth
380
Systems Sciences, 1: 177-185.
381
Piper, D.J.W., Cochonat, P. and Morrison, M.L., 1999. The sequence of events around the
382
epicentre of the 1929 Grand Banks earthquake: initiation of debris flows and turbidity
383
current inferred from sidescan sonar. Sedimentology, 46(1): 79-97.
384
Piper, D.J.W., Shor, A.N., Farre, J.A., O'Connell, S. and Jacobi, R., 1985. Sediment slides and
385
turbidity currents on the Laurentian Fan: sidescan sonar investigations near the epicenter
386
of the 1929 Grand Banks earthquake. Geology, 13: 538-541.
387
Piper, D.J.W., Shor, A.N. and Hughes Clarke, J.E., 1988. The 1929 Grand Banks earthquake,
388
slump and turbidite current. Geological Society of America Special Paper 229: 77-92.
389
390
391
Pitlick, J., 1992. Flow resistance under conditions of intense gravel transport. Water Resources
Research, 28: 891-203.
Ruffman, A., 1991. Notes on the recurrence rate of a November 18, 1929-like event in the
392
Laurentian Slope (LSP) seismic source zone or of similar shelf-edge/slope events off
393
eastern Canda. In: J. Adams (Editor), Proceedings of the Workshop on Eastern Seismicity
394
Source Zones for the 1995 Seismic Hazard Maps. Geological Survey of Canada Open
395
File Report 243, pp. 371-396.
396
Ruffman, A., 1996. Tsunami runup mapping as an emergency preparedness planning tool: the
397
1929 tsunami in St. Lawrence, Newfoundland. Emergency Preparedness Canada, Report
398
for Contract 94-D025.
399
Ruffman, A., 1997. Tsuami runup mapping as an emergency preparedness planning tool: the
400
1929 tsunami in St. Lawrence, Newfoundland. Geomarine Associates, contract report for
401
Emergency Preparedness Canada (EPC), Ottawa, Ontario, 1, 107 pp.
402
Ruffman, A. and Tuttle, M.P., 1994. In search of an on-land record of the 1929 Grand Banks
403
tsunami, Geological Association of Canada; Mineralogical Association of Canada;
404
Canadian Geophysical Union, Joint Annual Meeting, pp. 96.
405
Sato, H., Shimamoto, T., Tsutsumi, A. and Kawamoto, E., 1995. Onshore tsunami deposits
406
caused by the 1993 Southwest Hokkaido and 1983 Japan Sea earthquakes. In: K. Satake
407
and F. Imamura (Editors), Tsunamis: 1992-1994, their generation, dynamics, and hazard.
408
Birkhauser, Basel, pp. 693-717.
409
Shi, S., Dawson, A.G. and Smith, D.E., 1995. Coastal sedimentation associated with the
410
December 12th, 1992 tsunami in Flores, Indonesia. In: K. Satake and F. Imamura
411
(Editors), Tsunamis: 1992-1994, their generation, dynamics, and hazard. Birkhauser,
412
Basel, pp. 525-536.
413
Tuttle, M.P., Ruffman, A., Anderson, T. and Jeter, H., 2004. Distinguishing tsunami from storm
414
deposits in eastern North America: the 1929 Grand Banks tsunami versus the 1991
415
Halloween storm. Seismological Research Letters, 75: 117-131.
416
Wilcock, P.R. and Southard, J.B., 1989. Bed load transport of mixed size sediment: fractional
417
transport rates, bed forms, and the development of a coarse surface layer. Water
418
Resources Research, 25: 1629-1641.
419
420
421
422
Figure Captions
423
Figure 1. Location map of the study area at Taylor’s Bay, Newfoundland.
424
425
Figure 2 . A. Bimodal grain size distribution of sand within the deposit, determined by sieving.
426
The horizontal line shows ± 1σ from the mean, shown by the vertical crossbar. The inverted
427
triangle shows the median grain size. Bins for which fewer than 30 particles were retained are
428
lightly shaded. B. Bimodal grain size distribution of sand within the deposit, determined
429
optically. Error at the peaks is approximately ±1% of the total weight.
430
431
Figure 3. Cross section through the marsh at Taylor’s Bay.
432
433
Figure 4. Elevation of the base of the sand over the study area. Light colors represent low
434
elevations, and dark colors high elevations. Sample locations are marked with open
435
circles.
436
437
Figure 5. Thickness of the tsunami deposit over the study area. Light colors represent thicker
438
deposits, and dark colors represent thinner deposits. Sample locations are marked with
439
open circles.
440
441
Figure 6. Median grain size of the tsunami deposit over the study area. Light colors represent
442
coarser sand, and dark colors represent finer sand. Sample locations are marked with
443
open circles.
444
445
Figure 7. Grain size plots of Taylor’s Bay deposit showing different mixtures of two main peaks.
446
447
Figure 8. Peak deconvolution of a sample typical of the Taylor’s Bay deposit. Black dots
448
represent Camsizer data; gray peaks are individual sediment populations within the total.
449
The black line represents the sum of the gray peaks. The gray shaded area is a Camsizer
450
analysis of dune sand from the beach ridge at the mouth of Taylor’s Bay.
451
452
Figure 9. Grain size trends and ratio of fine to coarse material along a flow-parallel transect.
453
Dashed vertical line marks approximate location of the 1929 pond shoreline.
454
455
Figure 10. Graphical solution of Equations 1 and 2—our flow depth-flow velocity estimate for
456
the 1929 Grand Banks tsunami is the intersection of these two lines. Gray areas are
457
hydraulically improbable combinations of flow depth and velocity.
458
458
459
460
Figure 1.
460
461
462
Figure 2.
462
463
464
465
Figure 3.
465
466
467
Figure 4.
467
468
469
Figure 5.
469
470
Figure 6.
471
472
473
474
Figure 7.
474
475
476
Figure 8.
476
477
Figure 9.
478
479
480
Figure 10.