Page 7 - gt1606

Figure 3. (A) Long-profile of the Virgin River (V.R.) from The Narrows to the
town of Springdale, with extrapolated pre-slide profile (black dashed line).
Reconstructed top of rock avalanche deposits is compared to modeled profile.
Inset shows simplified schematic of the UGS core. Clay lake bed deposits are
projected onto the long-profile. 36× vertical exaggeration. (B) Cross section
through the Sentinel rock avalanche deposit showing reconstructed base of
slide, top of slide, and original valley profile, compared to model results and turbulence parameter � = 200; (2) unit weight � = 20 kN/m3; and
modern topography. Qts—sandy talus cone at base of source; Jts—Temple Cap
Formation; Jnw—Navajo Sandstone, white; Jnp—Navajo Sandstone, pink;
Jnb—Navajo Sandstone, brown; Jkt—Kayenta Formation, Tenney Canyon
Tongue; Jnl—Navajo Sandstone, Lamb Point Tongue; Jk—Kayenta Formation;
Jms—Moenave Formation, Springdale Sandstone; no vertical exaggeration.
measurement uncertainty due to high boron content; Z4, with
likely inheritance from pre-slide exposure; and Z8, which was
highly weathered with possible inheritance. The remaining nine
ages are consistent, indicating a mean age of 4.8 ± 0.4 ka (error:
±1�). No evidence for a concurrent paleo-seismic event in the
area has been identified, which might have triggered the rock
avalanche (Stenner et al., 1999).
Runout Modeling
We performed numerical runout simulation of the Sentinel
rock avalanche using the equivalent-fluid code DAN3D
(McDougall and Hungr, 2004). Our goal was to reproduce the
observed deposit extents and thickness, and in doing so validate
the hypothesized single-event failure scenario. DAN3D allows
modeling of rock avalanche motion over 3D terrain based on
input of slide path topography, source thickness, and material GSA TODAY | www.geosociety.org/gsatoday
properties. We used a modified version of DAN3D that allows
initial rigid-body motion (Aaron and Hungr, 2014). We selected a
Voellmy rheology for basal shear resistance and assigned uniform
material properties across the runout path. The best-fitting input
parameters were (1) Voellmy friction coefficient f = 0.27 and
(3) an internal friction angle �i = 35°. The relatively high Voellmy
friction coefficient for the Sentinel rock avalanche (cf. Hungr and
Evans, 1996) indicates comparably low mobility, which is due in
part to the cross-valley flow orientation and likely dry rock
avalanche and substrate materials.
Runout modeling results for the Sentinel rock avalanche are
shown in Figure 4 (see also animation in the Supplemental Data
Repository [see footnote 1]). After 4.5 s of initial rigid-body
displacement (~20 m southeast), the mass collapses, fluidizes, and
rapidly accelerates. The ensuing rock avalanche travels predomi-
nantly southeast, crossing Zion Canyon in less than 20 s with
maximum velocity of 90 m/s and energetic runup on the opposite
valley wall. Debris then spreads laterally up and down canyon.
After ~50 s, the distal ends of the deposit have been reached and
most of the material has come to rest, traveling a maximum
source-toe path distance of 2.4 km at an average velocity of ~50
m/s. The modeled deposit boundaries match well with those
mapped in the field (Figs. 4E and 4F), although our simulated
deposit exceeds the mapped boundaries in areas to the east and
southwest. Excessive material spreading in the initial part of our
simulation may account for some of this discrepancy. The thick-
ness and overall distribution of mass within the simulated deposit
also match well with our topographic reconstruction.
Independent velocity constraint can be obtained from estimated
runup (h) on the opposing valley wall as: V = (2ghmax)0.5 (Jibson
et al., 2006), where g is gravity and hmax = 200 m, indicating a
minimum velocity needed to achieve this runup of 62 m/s (value
does not account for basal shear resistance).
SENTINEL LAKE
The Sentinel rock avalanche blocked Zion Canyon over a
distance of more than 3 km, damming the Virgin River and
creating a lake. We estimate an initial breach elevation of the land-
slide dam of 1420 m, which is lower than the maximum height of
~1445 m and corresponds to a topographic low in the northern-
central region of the deposit. At this initial level, Sentinel Lake
would have covered an area of ~6.2 km2 (Fig. 2C). Assuming that
discharge of the Virgin River was similar to modern mean values
(USGS, 2014), we estimate that ~5–10 yr were required to fill the
lake to this high-stand.
Preserved clay beds and mollusks indicate that a calm, deep-
water environment was sustained by Sentinel Lake for several
centuries (Hamilton, 1976, 2014). Lake-bed deposits are found as
low as 1285 m (UGS core log data) and as high as 1315 m (Fig. 3),
representing at least 30 m maximum thickness. Core and surface
observations show that these clay beds are gray and yellow in color
and layered at the centimeter scale. We estimate an approximate
stable water elevation for Sentinel Lake of 1345 m based on the
maximum elevation of identified clay deposits and morphology of
the landslide dam. Field evidence of debris slumps and clay-filled
channels indicates that this elevation varied slightly over time.
7