Tsunami-generated sediment wave channels at Lake Tahoe, California-Nevada, USA

Lake Tahoe straddles the Nevada-California (USA) state line, and the major obtuse angle in that boundary occurs within the lake at 39°N, 120°W. This 34-km-long lake at the boundary between the Basin and Range province and the Sierra Nevada occupies a fault-generated basin. The lake surface is 1900 m above sea level, and the maximum depth is 500 m.

In the late Quaternary, a gigantic ∼12 km³ landslide detached from the west wall of Lake Tahoe and slid 15 km east across the lake. It left hundreds of angular fragments on the lake floor, 3 of which are >1 km (Gardner et al., 2000) (Fig. 1). The landslide, named the McKinney Bay landslide, is the second largest Quaternary landslide on the continent, exceeded only by the landslide on the northwest side of Mount Shasta, California. Rapid movement of the slide across the lake would have splashed water high above the shoreline, producing giant tsunamis (Schweickert et al., 2000, 2004; Moore et al., 2006; Ward, 2013).

New data permit a revised assessment of this megasplash. Observations with a remotely operated vehicle have revealed some of the effects of wave activity associated with the landslide. New analyses of submerged sediment wave channels and delta terraces (defined herein) have shown their similarity to analogous features in the marine environment. Sediments recovered from shallow piston cores indicate that the sediment wave channels are not currently forming, but are of about the same age as the principal landslide. Subaerial field observations around the lake have begun to focus on the nature of backwash and/or tsunami effects and deposits.

The Lake Tahoe region has been studied extensively since the early geologic mapping of the region by Lindgren (1896). In this summary, we concentrate on the giant landslide and the consequences of that event. The first recognition of landsliding in the lake resulted from single-beam low-resolution echo sounding coupled with a seismic reflection survey that identified mounds in the center of the lake presumably deposited by landslides (Hyne et al., 1973).

A major advance in our understanding of lake history was the multibeam sonar survey that produced a high-precision bathymetric map (gridded at 10 m) of the lake (Gardner et al., 1999, 2000). A U.S. Army Corps of Engineers airborne lidar survey mapped nearshore bathymetry unattainable by boat-mounted sonar (U.S. Geological Survey, 2014). This work revealed the McKinney Bay landslide amphitheater on the west side of the lake beneath McKinney Bay, and the giant scattered landslide blocks in its debris field extending 15 km east across the lake (Fig. 1).

The survey also mapped 15 sediment wave channels around the lake that extend from near the shoreline to the flat floor nearly at 500 m depth. It is important to note that sediment wave channels like those at Lake Tahoe are not common in other large deep lakes. The survey also depicted subaqueous terraces that ring the lake (Fig. 1). Commonly two terraces are present that are ∼10–15 m apart in depth; the deepest is at 25–30 m.

Dredging of the landslide blocks revealed that they are composed of fine-grained lithified sediment derived from an upfaulted section of lake beds bordering the west side of the lake (Moore et al., 1999, 2006). Core from this sediment in two drill holes flanking the dam at the outlet of the lake contain volcanic ash layers, dated from 2.1 to 0.75 Ma (Verosub et al., 2004); therefore, the landslide is derived from an upfaulted section of poorly lithified lake beds deposited in an ancient Lake Tahoe at least as long ago as 2.1 Ma.

Based on the dimensions and location of landslide blocks, Gardner et al. (2000) proposed that when the east-moving landslide blocks reached the east wall of the lake basin, they bounced off the wall and traveled north and south as well as back west several kilometers. Such a ricochet seems unlikely because upon striking the wall so forcefully, one would expect that the blocks composed of poorly lithified lake beds would disintegrate. However, the prodigious energy of this mammoth landslide cannot be denied.

Organic material in 21 3-m piston gravity cores taken from lake-bottom sediments was dated by radiocarbon analysis (Smith et al., 2013). The cores contain thin turbidites that alternate with normal sediments. Nearly all of the cores penetrated the 7700–8000 yr old Mount Mazama ash (Bacon, 1983) and bottomed near a 12 ka horizon with radiocarbon-dated organic material. None of the cores penetrated the landslide, indicating that it is older than 12 ka. The cores taken as a whole indicate that ∼3 m of sediment was deposited in the past 12 k.y., an average sedimentation rate of ∼0.25 mm/yr.

Bathymetric mapping and subbottom seismic profiling by Hyne et al. (1973) did not identify the McKinney Bay reentrant as the source of the landslide, but did determine, based on the area and thickness of debris spread out on the lake floor, that the volume of landslide material was ∼10 km³. Ward (2013), in his modeling of the landslide, reported a volume of 5 km³; however, no documentation was provided on how this volume was calculated.

The estimated size of the present-day west wall reentrant from which the landslide moved is ∼5 × 6 × 0.33 km, indicating a volume of 10 km³ for the material removed from the wall (Moore et al., 2006). In addition, a volume of ∼2.5 km³ of glacial moraines was apparently removed by landslide movement and backwash. The resulting total volume of material that has moved out of the amphitheater is ∼12.5 km³.

The landslide was from the center of an upfaulted section of poorly consolidated Pleistocene Tahoe lake beds, with some included volcanic rock (Figs. 1 and 2). This section originally underlay an area ∼12–13 km long paralleling the central west side of the lake. Two remnants remain on each side of the landslide amphitheater, the Tahoe City shelf remnant on the north and the Sugar Pine Point remnant on the south. Both remnants average ∼4 km in length (north-south) and 4.5 km in width (east-west). The volume of each, assuming a 0.33 km thickness, is ∼6 km³. Both of these masses of rock are capable of breaking free in the future, in which case either one would produce a devastating landslide of about half the volume of the McKinney Bay slide. The Sugar Pine Point remnant is perhaps the more likely candidate because its east and north sides are precipitous, whereas the Tahoe City remnant is more buttressed on the south and east by steps 1–2 km wide and 250–350 m high. However, failure of the Tahoe City remnant would lower the sill impounding the lake at its outlet, thereby causing a catastrophic spill of water down the Truckee River Canyon.

The age of the landslide is controversial. It was first estimated as 300 ka (Gardner et al., 2000) by using a sedimentation rate of 0.15 mm/yr based on radiocarbon ages in short sediment cores (reported by Hyne et al., 1972).

This age was contested in Schweickert et al. (2000) because of field evidence that showed that Tioga-age glacial moraines at the mouths of several canyons, including McKinney, General, and Meeks (Fig. 2), were eroded by tsunamis apparently generated by the landslide; it was proposed that the landslide was postglacial and perhaps Holocene in age.

Kent et al. (2005) adopted an age of 60 ka for the landslide based on an extrapolation of an average sedimentation rate of 0.4 mm/yr determined by radiocarbon dating in a 3-m-deep core east of the Stateline North Tahoe fault; they arrived at the age by extrapolating to the bottom of a 24-m-thick sediment cover on the landslide measured on seismic records. Kent et al. (2005) suggested, however, that an earlier sedimentation rate may have been higher and that the age may be younger.

Dingler et al. (2009) and Smith et al. (2013) both adopted the 60 ka age for the landslide, but Dingler et al. did adjust the age to “ca. 40–60 ka” in their Discussion section (p. 1102).

Extrapolating the low present-day sedimentation rate back to the time of the landslide is unrealistic, as argued in Schweickert and Lahren (2006). The age reported by Kent et al. (2005) is too old because several important processes produced a sedimentation storm following the landslide (Schweickert et al., 2000; Schweickert and Lahren, 2006). Eight of these processes are as follows. (1) As the landslide moved across the lake, it plowed through the sediment-covered lake floor and stirred up ancient lake beds as well as recently deposited sediments, much of which would have drifted back on top of the landslide debris. (2) The 223 poorly lithified lake sediment blocks mapped within the landslide (36 longer than 500 m and 27 protruding more than 40 m above the sediment cover; Gardner et al., 2000) would have impinged on one another during landslide movement and thereby partly disaggregated into clouds of sediment. (3) Landslide movement created an energetic train of giant waves, which repeatedly swept across the lake (Ward, 2013), reaching the sides and bottom and stirring up sediment probably for weeks to years. (4) Several smaller landslides occurred after the main event both within and outside the McKinney Bay reentrant (Gardner et al., 2000; Schweickert et al., 2000; Smith et al., 2013), all of which would have stirred up sediment. (5) Large sediment wave channel systems (n = 15) and associated delta terraces (described in the following) were formed where splash-induced tsunamis reentered the lake burdened with abundant clastic material eroded from onshore. The volume of the tsunamis would be enhanced by snowmelt if they occurred in the winter. (6) Land erosion and the resulting lake sedimentation rate would have been augmented in the decades following the landslide because runoff would be accelerated due to the denudation of land vegetation by tsunamis. (7) Tioga maximum moraines were trimmed back and removed by the landslide-induced tsunamis; subsequent increased runoff during glacial retreat would have increased the sedimentation rate. (8) Three families of earthquakes would have dislodged sediment from the walls of the lake to be deposited on top of the landslide on the deep lake floor. First, the landslide was likely triggered by a major tectonic earthquake and its aftershocks. Second, the rapid 15 km movement of 12 km³ of blocky debris across the lake floor would have imparted vibrations into both the substrate and the lake water. Third, the rapid movement of several gigatons of landslide debris from the upthrown to the downthrown side of the major West Tahoe–Dollar Point fault (Fig. 2; Schweickert et al., 2004) would likely have induced subsequent displacement and quakes along the fault.

Geologic mapping on the west side of the lake indicates that Tioga glacial moraines at the mouths of General, McKinney, and Meeks Creeks are trimmed back and that moraines at the mouths of Blackwood and Ward Creeks have been removed (Schweickert et al., 2000; Howell et al., 2013). The moraines were apparently eroded by landslide-related tsunamis, indicating that the landslide was postglacial (Fig. 2; Schweickert et al., 2000). Cosmogenic dating of remaining Tioga moraines nearby has provided an age of 21 ka (Howle et al., 2012).

The bottom sediment taken in a series of 21 lake-bottom cores yielded radiocarbon dates as old as 12 ka (Smith et al., 2013), yet did not penetrate landslide debris. On the basis of this and the moraine evidence, the age of the landslide is constrained between 21 and 12 ka.

Sediments on terraces near Cave Rock on the east side of the lake (Fig. 1) were cored and dated by radiocarbon and optically stimulated luminescence methods (Kent el al., 2005). The estimated age of the terrace abrasion surface at 21 m depth at the base of overlying sediment is 19.2–17.3 ka. This age was later reported as 19.2 ± 1.8 ka (Dingler et al., 2009). However, the interpretation of the data is controversial (Schweickert and Lahren, 2006). Even so, this age for the terrace is in accordance with our concept that the Cave Rock delta terrace (described later) is essentially synchronous with the McKinney Bay landslide.

Numerous large boulders, some of which are found 55 km downstream at Verdi near the California-Nevada state border, provide possible evidence for large rushes of landslide-splashed water out the lake outlet and down the Truckee River Canyon. The largest boulder above the ground surface measures 12 × 6 × 3 m (Birkeland, 1968).

East-west boulder ridges as much as 2 km long occur on the submerged Tahoe City shelf that flanks McKinney Bay on the north. They have been attributed to giant ripple marks generated at the head of the landslide flank (Moore et al., 2006). Similar boulder ridges not previously reported are present on the Tahoma Shelf south of McKinney Bay (Schweickert, 2013, personal observation).

On the west side of the lake, a gently sloping surface extends from Tahoe City south past Ward and Blackwood Creeks to Sugar Pine Point (Fig. 2). This probable tsunami erosion surface extends to 1 km inland at General Creek and attains an elevation of ∼30 m above lake level; it cuts across Tahoe and Tioga moraine deposits. Upon this surface is a deposit ∼1 m thick of sandy pebble gravel with beach-rounded granitic pebbles.

Another probable tsunami erosion surface has been identified at Emerald Point, where Tioga-age recessional moraines have been flattened and are overlain by sandy pebble gravel similar to that in the north. On the east side of the lake, the soil has been stripped to 30 m above lake level on the point south of Skunk Harbor.

The large area of low terrain bordering the lake in the south and southeast sectors accounts for ∼50% (75 km²) of the total land surface area between the lake and 100-m elevation above the lake. This area would have sequestered much of the splash and runup water. A sheet of loose sand <2 m thick covers parts of this large low-lying area. Apparently the entire area was washed by large waves. If the landslide and splash occurred in the winter, the volume of the resulting backwash would have been increased by the melting of snow on the ground.

Runback into the lake would have been slowest at the south coast because the water would temporarily have ponded in an area extending 10 km from the shoreline. This southern zone would have sequestered a large volume of water before it washed back into the lake. This may account for the higher terraces, and lack of terraces, in the southern part of the lake. The processes of sediment wave channel and delta terrace formation both continued longer in the south after drainback was completed in other areas and the lake level had risen.

Ward (2013), in modeling a ∼5 km³ landslide, reported a wave height >50 m with 5 sizable waves occurring over a period of 19 min. Gardner et al. (2000) cited modeling that suggested a wave height of 101 m.

If 100 m waves had been uniformly distributed around the lake, then ∼150 km² of adjacent land, or an area equal to 30% of the lake area, would have been inundated.

Aside from the initial splash, model simulations indicate that large-scale water oscillations would follow the landslide, five or more large splashes would occur over a period of less than an hour, and disturbed water would continue for a long period (Ward, 2013). Moreover, the remains of several smaller landslides within the landslide scar (Gardner et al., 2000; Smith et al., 2013) indicate that additional wave activity would have followed the major landslide.

There are 15 systems of subaqueous sediment wave channels around the margin of Lake Tahoe (Fig. 1). The channel systems range from 1.8 to 10.4 km long and are commonly associated with one of two subaqueous terraces at their heads, which apparently formed at the post-slide lake shoreline. In general, the systems are erosional in their upper parts with steep-walled, nearly flat-bottomed channels carved into the substrate, but are constructional in their lower parts, where the channels merge into depositional aprons or fans. The sediment wave channels and their branches are steepest at the head where the slope commonly attains 15°. Most of the channel systems are between 5° and 1° with lower distal values (Fig. 3A).

The general term “sediment wave channel” is used for the entire feature because sediment waves are ubiquitous in them, even though the lowermost part is a depositional apron with no channel or with indistinct channels. The overall character of these systems is similar, all appearing to be of the same age and to have formed in a similar fashion.

The distinctly erosional channels with convex-upstream bedforms change to depositional aprons in their lower part with convex downstream bedforms. Generally the transformation from erosional to depositional occurs about three-quarters of the way down the overall length of the system. The occurrence of bedforms that are either convex upstream or downstream seems to depend simply on the nature of the substrate elevation contours. Bedforms that fill the bottom of channels are convex upstream, whereas those that cover the cone- or ridge-shaped aprons are convex downstream.

Two cores 3 m long (cores 2 and 3 of Smith et al., 2013) are near the edge of the lower part of one of the largest sediment wave channel aprons (Tahoe Keys; Fig. 1). The cores did not penetrate material of the apron, indicating that the sediment wave channel is older than the age of the sediment at the bottom of the cores. Both cores contain the 8 ka Mazama ash. Detailed features of the sediment wave bedforms, including the giant ripple marks (Figs. 1, 3, and 4), appear in the sonar images, indicating that the several meters of overlying lake sediment is insufficient to obscure the bedforms. This evidence suggests that the landslide and probably all the sediment wave channel systems are of about the same age and occurred before 8 ka.

The upper or proximal ends of most sediment wave channels are generally made up of several adjacent tributaries or finger-like branches (Figs. 5–8). In most systems, the branches are side by side and generally originate along a line parallel to, but a few hundred meters offshore, the present shoreline. Most originate and erode into the delta terraces. The finger tributaries contain small sediment waves with discernable wavelengths as small as 40–50 m (Fig. 4). Others show no such bedforms, but the 10 m gridded data provide poor resolution of such small features.

The nearshore ends of finger tributaries of the channels commonly have blunt cirque-like heads 50–250 m across at the head of a trench as deep as 75 m; these cirque-like features were carved into the subaqueous terraces.

Multiple tributaries feed into the 10.4-km-long Incline sediment wave channel system at the northeast corner of the lake (Figs. 1 and 5). The northern end has more than 10 variably distinct branches that head between 10 and 15 m depth, except for one that reaches 5 m depth. These branches drain a broad coastal zone, indicating that they were fed by a 2-km-wide sheet of water entering the lake. Three more channel systems enter from the east, contribute to the main system, and were fed by an additional 3-km-wide sheet flow.

The Glenbrook sediment wave channel system on the east-central coast has six branches (Figs. 1 and 6). The 3-km-long system is fed from a 2-km-wide coastal zone.

The western branch of the Stateline channel system at the southeastern corner of the lake (Fig. 1) is fed by 19 small side by side channel branches each 150–250 m wide that drain a 4-km-wide-span of coastline. The water that fed this system clearly was not channelized, but entered the lake as a nearly continuous sheet flow along a 4-km-wide front. Downstream, the branches merge into an incised 600-m-wide channel that contains bedforms with wavelengths of as much as 300 m (Fig. 4).

The Tahoe Keys system at the south end of the lake contains 20 individual side by side finger branches that merge downstream into much larger erosional channels (Figs. 1 and 8). The individual branches are 50–100 m wide and drain a coastal zone where apparently a uniform sheet of water 3.4 km wide entered the lake. The bedforms in these fingers (generally convex upslope) have wavelengths of 40–50 m.

The surface flow that fed the combined Stateline and Tahoe Keys systems (Fig. 1) entered the lake in a nearly continuous sheet of water 8 km wide at the south and southeast sector of the lake. We can only speculate on the depth of the water, but it must have been more than several meters. This region is now the site of the city of South Lake Tahoe, the most densely populated region in the Lake Tahoe Basin. If such a sheet of water flowed into the lake now, it would cause unimaginable devastation.

The middle parts of the larger sediment wave courses contain the largest distinct channels with steep walls as high as 150 m. They contain crescent-shaped sediment waves that are convex upstream and include the largest sediment waves observed. In the Incline system (Fig. 5), the main channels are 250–420 m wide, with bedforms that have wavelengths of 250–300 m and amplitudes of 3–4 m (Gardner et al., 2000). In the Stateline system, the western main channel is 600 m wide with bedforms to 300 m in wavelength, whereas the eastern channel is 300 m wide with bedforms also to 300 m in wavelength (Figs. 1, 3B, and 7). In the Tahoe Keys system, the main channels are commonly 600–800 m wide and contain sediment waves with maximum wavelengths of 300–450 m (Figs. 1, 3, and 8). These ripple mark–like sediment waves are asymmetrical, with their lee sides shorter and steeper than their stoss sides. Sediment wave wavelengths are about one-half the width of the channel that they occupy (Fig. 3B).

In the Stateline channel system, the sediment wave wavelength in the finger tributary region is 50–100 m, the wavelength in the main incised channels is 150–300 m, and the wavelength in the lower apron is 100–200 m (Figs. 4 and 7). The longest sediment waves have developed just below the point where the wide bundle of upper finger tributaries with small sediment waves merges to form the main incised channel. Below the main incised channel, where the system widens to develop the apron or fan, the bedforms become shorter in wavelength (Figs. 4 and 7).

Drainage in the central incised channel part of the system is narrowest, producing an hourglass-like plan for the entire system. The largest sediment waves are apparently favored where the narrowed course leads to the greatest current velocity.

The lower parts of the sediment wave channel systems are broad, and resemble alluvial fans on land (Fig. 1). Most of the material carried downslope and eroded from the incised channels was apparently deposited in the aprons at the lower termini of the systems. The aprons from several channels commonly coalesce, producing compound aprons. Coalescence of the Baldwin Beach and Tahoe Keys systems formed an apron 4 km wide (Fig. 1), and the joining of several channels in the Glenbrook system produced an apron 2 km wide (Fig. 6).

The deepest discernible termini of individual fans commonly extend to depths within 20–40 m of the flat lake floor. No doubt the distal parts of the fans extend farther and are masked by the younger sediment, which is thicker on the deep flat floor of the lake. No evidence indicates that any one fan is younger than any other, which suggests that the fans formed simultaneously.

The volume of the larger aprons, based on their bulge above the regional slope, is perhaps 0.1–0.2 km³. The aprons are mantled with convex-downslope sediment waves, with maximum wavelengths of 100–300 m (Fig. 3B). These sediment waves also are asymmetric with the lee sides shorter and steeper than the stoss sides, but the asymmetry is not as pronounced as the asymmetry of the sediment waves upstream in the incised channels.

The apron in the lower part of the Glenbrook sediment wave channel on the east side of the lake has grown over, and partly buried the edges of, two isolated blocks of the McKinney Bay landslide (Fig. 6). Blocks on the north side and on the southwest side of the apron are covered by sediment waves of the apron. Therefore, this sediment wave system is at least partly younger than the giant landslide.

Recent multibeam sonar surveys in several marine settings (Cartigny et al., 2011) have provided insight into the development of sediment waves similar to those in Lake Tahoe. Within the axial channel of Monterey Canyon, central California, crescent-shaped sediment waves occur that are convex upcanyon with wavelengths of 20–80 m and amplitudes of as much as 2.5 m (Paull et al., 2010). The dune-like features are asymmetric, with short, steep lee limbs on their downstream side and long, flatter stoss limbs on their upstream side. These bedforms have features similar to those within the Tahoe channels and occur in similar water depths. Instruments installed on the bottom of Monterey Canyon demonstrate that the bedforms periodically become activated and that boulder-sized concrete instrument mounts have been carried downstream. These gravity-flow movements occur several times a year and are in part triggered by significant ocean wave disturbances (Paull et al., 2010); they suggest that localized faulting and en masse failures within the axial channel may combine with turbidity currents in forming the sediment waves.

A remarkable field of sediment waves occurs just outside the Golden Gate entrance to San Francisco Bay (Barnard et al., 2006), where strong tide-driven currents flow through the narrow rocky strait and commonly exceed 2.5 m/s. The field of giant bedforms covers ∼4 km² and includes sediment waves with wavelengths to 220 m and heights of 10 m.

Very large sediment waves have been mapped offshore of the South Sandwich volcanic arc in the South Atlantic Ocean (Leat et al., 2013). Sediment waves with wavelengths of 1.4–3.7 km and amplitudes of 85–200 m occur at depths of 700–2800 m on the flanks of Montagu volcano.

Sonar multibeam bathymetry reveals that sediment wave channels are currently forming off the entry of the Squamish River into the Pacific at Howe Sound, British Columbia (Hughes Clarke et al., 2012). The 100-km-long river enters the sea at an active 350-m-wide delta. The river discharges 400–800 m³/s of sediment-laden water into the sea and has developed submarine sediment wave channels remarkably similar to those of Lake Tahoe (Fig. 7).

The Squamish system has multiple small nearly parallel finger tributaries in its uppermost reach where fed at the river delta; 8 individual side by side branches 20–35 m wide drain a large part of the river flow over a span of 350 m. The tributaries carry sediment waves of 12–15 m wavelength. Downstream, the branches merge into a 100-m-wide channel, which contains sediment waves with wavelengths of 30–40 m that are convex upstream. The channels, somewhat smaller and steeper than most in Lake Tahoe (Figs. 3B and 7), are as much as 1.7 km in mapped length, and 100–200 m wide.

Turbidity currents formed by the deposition and collapse of delta sediments that were deposited where the river enters the sea surge downslope and carve the finger tributaries. These join together to feed the incised channel systems. During the Squamish sonar study, five mass-wasting events were triggered by major (>20,000 m³) collapses of the lip of the river delta (Hughes Clarke et al., 2012).

Repeat mapping indicates that the waveforms migrate upstream during periods of increased river flow (Hughes Clarke et al., 2012). An explanation for this migration is that the suspended sediment in the turbidity flow is supercritical and erosional on the steep lee faces of the waves and subcritical and depositional on the gently sloping stoss faces (Cartigny et al., 2011). This difference causes lee erosion of the bedform crests and stoss-side deposition, thereby causing the wave crests to migrate upstream even though sediment is moving downstream.

The Squamish example indicates that vigorous sediment-laden currents are capable of establishing sediment wave channel systems very similar to those of Lake Tahoe, but they require the flow of a major river. Although the data on deep lakes is sparse, it is clear that sediment wave channels like those at Lake Tahoe are not common. Modern multibeam imagery of Crater Lake, Oregon (Bacon et al., 2002), Lake Lucerne, Switzerland (Hilbe et al., 2011), or Lake Ohrid, Macedonia-Albania (Sebastian Krastel, 2013, oral commun.), has not displayed sediment wave channels.

Paull et al. (2010) proposed four possible causes for the formation of crescent-shaped bedforms in Monterey Canyon: (1) tidal currents, (2) turbidity currents, (3) slumping of canyon floor fill, and (4) remobilization of canyon fill. Paull et al. (2010, p. 755) noted that “whether the bedforms are generated by erosion associated with cyclic steps in turbidity flows or internal deformation associated with slumping during gravity-flow events remains unclear.”

The sediment waves at Lake Tahoe are clearly not of tidal origin. Analogy with the Squamish River system supports the notion that the Tahoe systems formed from sediment buildup in delta terraces deposited from giant tsunami sheet flows that returned sediment-laden splash water to the lake. The prime agents responsible for the development of the sediment wave channels are apparently the turbidity currents generated from the collapse and disaggregation of these deltas. In addition, perhaps large masses of the deltas became detached, slumped downslope, and played a role in shaping the sediment wave channels. The large volume and high velocity of the density flows eroded the steep upper and middle parts of the sediment wave systems and deposited sediment in the low gradient lower aprons.

The sonar bathymetric map of Lake Tahoe, augmented by the airborne lidar bathymetric mapping system (capable of penetrating of 25–30 m of water), shows two shallow subaqueous terraces. Terraces on the east side of the lake are more continuous and deeper than those on the west side (Fig. 1). All terraces have a slight (0.1–0.3 m/km) tilt down to the north (Fig. 9). The terraces are interpreted to have formed by simultaneous depositional and erosional processes and are referred to as delta terraces.

On the east shore, the depths of both the shallow and deep terraces (measured at the midpoint, the flattest part of each terrace) increase on average 0.24 m/km in the 36 km from south to north (Fig. 9). The middle of the shallow terrace ranges from 5 to 13 m south to north, and the middle of the deep terrace ranges from 18 to 23 m south to north. The terraces are remarkably parallel, and on average the deeper terrace is 13 m below the shallower terrace (Fig. 9).

On the west side of the lake, the delta terraces and sediment wave systems are best developed south of the main landslide reentrant (Fig. 1). As measured in the middle, the upper terrace descends south to north from 5 to 9 m in depth. The poorly developed lower terrace on the west side descends south to north from ∼12–13 m depth.

In general, the upper terrace on the west side ranges to 4 m above that on the east, and the lower terrace on the west side ranges to ∼9 m above that on the east.

Two distinctly erosional terraces flank the landslide reentrant on the west side of the lake: the 5-km-wide Tahoe City shelf on the north and the 2-km-wide shelf on the south side, by Sugar Pine Point. Remotely operated vehicle (ROV) dives show that these platforms are carved into a part of the uplifted paleo-Tahoe sedimentary prism, the part that escaped landsliding. These platforms are covered with cobble-boulder ridges or boulder megaripples that apparently represent material deposited by giant waves flanking the head of the landslide (Moore et al., 2006). These shelves, marginal to the landslide reentrant, remain high-standing parts of the sediment mass. They now pose the risk of future large-scale landsliding capable of producing destructive tsunamis.

ROV dives with a video camera on the east side of the lake explored the terraces at Zephyr Cove and found that both upper and lower terraces are covered with unconsolidated sand with embedded, scattered, somewhat rounded granitic boulders as large as meter size.

The northern of two cirques at the southern side of the Zephyr Cove sediment wave channel was investigated where the branch head is cut into the lower edge of the lower terrace. ROV observations indicate that there, at a water depth of 75 m, the cirque headwall cuts through 1–2 m of unlithified sands into a sequence of lithified, probably pre-landslide, lake sediments.

It is surprising that the prominent reentrant on the northwest Tahoe shoreline, Agate Bay, contains no delta terraces or well-developed sediment wave channels. The bay has a few narrow channels on the west side that begin halfway down the slope and extend to the deep lake floor. The channels contain no sediment waves, and have not developed the depositional aprons common at the lower end of other sediment wave channels. A possible explanation for this anomaly is that the subaqueous slope below this coast is unique in its particularly gentle slope down to a depth of 200 m. This underwater slope may inhibit the development of high splashes on land and may also lessen the energy of backwash turbidity currents flowing down the subaqueous slope.

The sediment wave channels are common in the regions where the subaqueous sediment terraces are well developed (Figs. 1 and 9). The two northern channels, the seven eastern channels, and the five western channels originated at a terraced shoreline. The two channels along the gently shelving south coast (Baldwin Beach and Tahoe Keys) are not associated with deeper terraces and occur where the shallow terrace is shallowest (∼5 m).

Sediment wave channels are not present in the region of the McKinney Bay landslide reentrant. Perhaps if they developed in this region, they were removed by the extreme oscillation of bottom currents associated with the main or subsequent smaller slides.

The cirques at the upper ends of finger tributaries have eroded into the outer margin of one or both of the terraces (Figs. 1 and 5–9). The extent of erosion into the terraces is as follows, proceeding clockwise from the north. The head of the Crystal Bay channel system seems to have removed most of the lower terrace and part of the upper. The north tributaries of the Incline system drain a 1.5 km span of the terrace and removed most of the lower terrace and part of the upper (Fig. 5). The middle and southern tributaries have eroded mainly into the lower terrace but partly into the upper terrace (Figs. 5 and 9).

The Secret Harbor channel preserves only part of the lower terrace, but most of the upper. The Skunk Harbor channel head has apparently obliterated the lower terrace and eroded into the upper terrace. The branches of the Glenbrook channel system have eroded through most of the lower terrace, leaving only small remnants; some branches have reached deep into the upper terrace (Figs. 6 and 9). The Logan Shoals channel has eroded deeply into the lower terrace and only slightly into the upper (Fig. 6). The Cave Rock channel has eroded only slightly the outer part of the lower terrace. The Zephyr Cove channel has eroded slightly into the lower terrace and has not reached the upper.

In general, the Stateline channel system has eroded most of the lower terrace, leaving only small remnants; in its eastern part, the branches drain a 1.2 km coastal span; in the middle part, a 2.3 km span; and in its southwestern part, a 1.9 km span. This system alone, therefore, seems to have resulted from a sheet of water 5.4 km wide crossing the southeastern shoreline of the lake (Fig. 1).

The Tahoe Keys system extends to 3–8 m depth, leaving only a shallow gentle slope above with no apparent terrace. Its 20 finger tributaries span a 3.4 km stretch of coastline, indicating that it was fed by a subaerial sheet flow of that width (Fig. 8). A shallow shelf and ridge occurs offshore in the divide between the Tahoe Keys and the Baldwin Beach channel systems. Steep slopes in this passive erosional remnant explored by ROV dives down to 100 m show well-bedded ancient lake beds. Finger tributaries of the Baldwin Beach system reach upward to 5–10 m depth with only a gentle slope above.

The western sediment wave channel systems are smaller than those in the south, and most were fed by drainage through major glaciated canyons. The Meeks Bay system is the largest and was fed from three glacial canyons, including General Creek and Meeks Creek that carried sediment down to merge in a single apron on the lake floor. The Meeks Bay cirques cut into the upper terrace and head at ∼5 m water depth. The Meeks Creek canyon feeds 20 finger branches, while the other 2 have only 2 branches each. The Lonely Gulch channel system heads in 3 branches, and the Rubicon Point and Emerald Point channels each have 3–4 branches that begin at 10 m depth. A remnant of the upper terrace is preserved along the divide between the Lonely Gulch and Rubicon Point channels. The small Emerald Point system heads in a narrow reentrant north of Emerald Point.

The Emerald Bay system, which extends out from Emerald Bay, is distinctly different than the other systems and may have a special origin perhaps related to the fact that it heads in a young glacial canyon and fjord. The system is ill defined and is studded with elongate piles of giant boulders as much as 40 m in size that are clustered more than halfway down the apron at 300 m depth.

Sediment wave channels as well as systematic delta terraces are absent within the McKinney Bay amphitheater. If such features ever existed, they no doubt were obliterated by recent small slides from the walls of the main landslide amphitheater.

Most of the sediment wave channels, or at least their uppermost parts, are younger than one or both terraces into which they have eroded (Fig. 9). Several of the branch channels only show erosion on the lower terrace, leaving the upper terrace unmarked. In places, large parts of both terraces are continuous and uneroded. That the sediment wave channel systems were created by copious amounts of sediment-bearing flowing water that crossed the shoreline, incised the main channels, and built the depositional aprons demonstrates that such large-scale flows passed by the present site of the existing delta terraces. Therefore, the sediment wave channels are partly older than the remaining delta terraces. Parts of the terraces must be younger than the channels, because otherwise the sediment-laden water flowing into the lake would have passed over them and eroded them. This contradictory relationship suggests that the channels and the delta terraces are of about the same age, and that the processes that formed them overlapped in time.

The sediment wave channels appear to be the same age as the landslide and the delta terraces. This suggests that the channel systems were produced by voluminous backwash from enormous waves generated by the movement of the giant landslide. As noted herein, such large waves eroded glacial moraines at the mouths of General, McKinney, and Meeks Creeks and removed glacial moraines from the mouths of Blackwood and Ward Creeks, all of which were glaciated canyons (Schweickert et al., 2000). Similarly, giant waves produced 2-km-long boulder ridges on the Tahoe City Shelf (Moore et al., 2006). We estimate that the megasplash onto the surrounding countryside temporarily lowered lake level to that of the lower terrace, which is now 20–30 m below the present eastern shoreline (Fig. 9), but was then considerably less because of subsequent tectonic lowering of the east side of the lake. The surface area of the lake is 500 km² and its volume 150 km³. For every 10 m lowering of lake level, 5 km³ of water need to be splashed from the lake, part of which drenched the lake margins and part of which was lost down the outlet.

The enormous splashes inundated large areas around the lake and the resulting extensive backwash of sediment-laden water poured back across the shoreline in wide swaths or sheet flows. As the backwash crossed the shoreline and entered the lacustrine environment, its velocity was dramatically reduced, rendering it incapable of carrying its sediment load. As a result, much of the load was deposited just below the lowered shoreline as deltas, which rapidly merged with one another, shaped by seiche activity. This concurrent deposition broadened the deltas producing the lower delta terrace. Rapid growth of these oversteepened delta terraces caused them to collapse, locally forming the channel heads or cirques. Collapsed sediment mixed with drainback water formed turbidity currents. These hyperpycnal flows moved downslope and carved the finger tributary channels, which merged downstream into the major sediment wave channels on the steep middle slopes of the lake basin. These channelized currents incised their channels into the soft lake sediments that carpet the lake bottom, and produced the bedforms in the channels. Further downslope where the slope decreased (Fig. 3), the suspended sediment in the turbidity flows was eventually dumped, producing depositional aprons in the deep lake.

The sediment wave channels and associated finger tributaries at Lake Tahoe closely resemble those currently forming where the Squamish River is flowing into Howe Sound north of Vancouver (Fig. 7) (Hughes Clarke et al., 2012). This similarity suggests that the volume of the Tahoe backwash was briefly equivalent to more than 15 Squamish Rivers in full flood.

It is uncertain why two sets of terraces formed. Perhaps the lower terrace is related to the greatest initial lowering of the lake due to splashing over the countryside and overflow down the outlet. It would have been formed as nearby splash water rapidly returned. Later the slow backwash from the broad lowlands south of the lake where the splash had moved farthest inland may have been responsible for depositing the upper terrace.

When all backwash ceased flowing, the lake still probably remained below its sill because of lake water lost to overflow down the outlet as well as to infiltration into the ground. However, the generally high precipitation within the basin would have restored the lake to its outlet level within a few seasons.

Later, continued basin-range faulting occurred on major north-south faults within the lake basin; all of these faults are downthrown to the east (Schweickert et al., 2000; Kent et al., 2005). This faulting depressed the eastern terraces a few meters relative to those on the west. Such tectonic activity may also have been responsible for the gentle northward downtilting of the terraces (Fig. 9).

Only recently have we gained some insight into the consequences of the giant McKinney Bay landslide in Lake Tahoe. The immense water disturbance created by the collapse of ∼12 km³ of poorly lithified sediments in the lake and its rapid movement 15 km east across the lake bottom splashed lake water onto the surrounding terrain to a possible height of 100 m and possibly lowered the lake level 10 m. The backwash from this splash formed enormous tsunami-like sheets of water many kilometers wide that exceeded the volume of 15 Squamish Rivers in full flood. These tsunamis must have decimated life in the Lake Tahoe Basin within the backwash zone. Entry of the sediment-laden splash water into the lake formed extensive deltas at the lowered shoreline. These deltas merged into terraces, which collapsed and triggered mass-wasting events and turbidity flows that fed many sediment wave channel systems that deposited fans on the 500-m-deep lake floor.

Two remnants of landslide-susceptible Pleistocene lake-bed sediments remain on either side of the reentrant of the McKinney Bay landslide, the Tahoe City remnant on the north and the Sugar Pine Point remnant on the south. Both of these remnants are capable of failing in the future and producing huge landslides about half the size of the McKinney Bay slide. The Sugar Pine Point area is believed to be more unstable because its east side is precipitous. The Tahoe City remnant is probably more stable because of buttressing by a broad step, but its proximity to the lake outlet raises the possibility of rapid lake drainage occurring if it were to fail.

Systematic examination and dating of subaerial sediments surrounding the lake may provide additional clues to the nature and distribution of the splash-produced tsunamis. How many waves were there? What heights did they attain? What size material did they carry? Where are the deposits best preserved? What are the ages of such deposits?

When longer sediment cores (e.g., 10–30 m length) are taken from the deep lake, the landslide deposit and sediment wave channel systems could be sampled and dated. Sediments immediately overlying the landslide, and the sediment wave channel debris, should contain abundant wood fragments carried into the lake from the surrounding forests by the splash backwash. Useful constraints on the age of the landslide could be acquired by dating this material.

More extensive shallow-water drilling of the delta terrace system may provide new information on the nature of sediments that compose it. The ages of the terraces may provide the most reliable age of the landslide.

On-land drilling in areas where the ancient lithified lake beds occur could provide important constraints on these landslide-prone materials and on the early history of the lake. Such areas would include Sugar Pine Point and the broad flat region along the south lakeshore. Extensive lake beds are exposed in the ridges between the sediment wave channels offshore of the south coast. These probably extend south to underlie the south coastal plain.

Further study of the cirque-like heads of the tributaries of the sediment wave channel systems may help elaborate the processes that created these giant systems. The tributary heads can be examined by ROV video systems and perhaps by ROVs capable of collecting samples; some are within scuba-diving range.

The high-energy, erosive-depositional processes unleashed by the McKinney Bay landslide are complex and interwoven. However, this one-time event in a closed lake basin provides a benchmark that can be compared to the processes that produced similar features in the ocean.

We thank Katherine Coble for information on sediment wave studies, Peter Dartnell for help in interpreting the Tahoe multibeam survey, Barry Moring for valuable assistance with the illustrations, Joel Robinson for aiding in lidar plots, and Steven N. Ward for consultations concerning his computer models of landslide-induced wave activity. Winifred Kortemeier aided in field and boat studies, and Thomas Adamek led the ROV-mounted camera work. Critical reviews by Ginger Barth, Stephanie Ross, and two anonymous reviewers greatly improved the manuscript.