Engineers, geologists, and seismologists have spent recent years evaluating and verifying the earth conditions under and around Martis Creek dam near Truckee, California, US, to better understand its seismic and seepage challenges and ensure that the dam continues to operate effectively and safely.

In the most recent effort, the Sacramento District of the US Army Corps of Engineers (USACE) and its consultants sought to better assess the causes of seepage and hydrologic deficiencies and further assess subsurface anomalies found during previous subsurface investigations. The team combined advanced remote sensing technology, namely light detection and ranging (LiDAR), with more traditional geomorphic and paleoseismic trenching techniques to gain a clearer definition of these anomalies. The team knew LiDAR could help strip away the vegetation to produce a so-called bare earth model in hopes this would allow them to visualise these features more effectively. In doing so, they discovered something unexpected: a previously unrecognised fault extending across the North Tahoe Basin, now named the Polaris Fault.

Dam design and construction

USACE named Martis Creek Dam one of its highest risk dams in its inventory due to its history of excessive seepage and the consequences of flooding to downstream communities if it failed. Owned and operated by the USACE Sacramento District, the dam and reservoir are located approximately 4.8km east of Truckee on Martis Creek, a tributary to the Truckee river.

Constructed in 1972, the 34m high zoned, rolled earthfill dam has a maximum storage capacity of 25,163m3x106 at the spillway crest and 42,678m3x106 at the maximum spillway design flood pool elevation. The dam’s function is to provide flood control and future water supply, but heavy seepage coincident with higher pool elevations during test reservoir fillings have prevented USACE from allowing the dam to achieve its full design potential.

The dam consists of three zones:

• 1) An upstream impervious zone.

• 2) A downstream zone (Zone 2) identified as random fill.

• 3) A vertical and horizontal drain.

The original dam design did not include a seepage cutoff barrier but instead designed and constructed an upstream clay blanket within the reservoir and upstream toe of the embankment. The embankment foundation was stripped to a nominal depth of about 30cm on the abutments, and the soft alluvium in the valley floor was removed to a maximum depth of about 1.8m and an average depth of about 0.9m (USACE, 1972), to reach a low permeability foundation layer referred to as the Blue Silt zone.

Between the upstream impervious zone and clay blanket, the potential for excessive seepage was thought to be resolved. These seepage control measures were apparently inadequate as uncontrolled seepage was judged excessive during test reservoir fillings at various times from construction through to 1995. Modifications to seepage control measures were made from 1972 to 1995 but were not completely successful.

Previous investigations

Geologic and foundation investigations were performed during the dam design and again in the 1980s to understand the cause of seepage deficiencies. During the most recent round of studies, detailed mapping and geomorphic reconnaissance coupled with hollow-stem auger and sonic boreholes at various locations around and on the dam were performed. The US Geological Survey (USGS) and subcontractor have also conducted geophysical investigations using seismic, resistivity, self-potential and time-domain electromagnetic methods. Boreholes were drilled to confirm and improve the existing stratigraphic model of a relatively continuous and simple sequence with a Blue Silt sandwiched between more granular, sandy layers above and below. But several boreholes in the left abutment of the dam either found the Blue Silt to be missing or at a distinctly different elevation than surrounding layers. When USACE first mapped a fault trace near the dam using LiDAR imagery, changes to the drilling programme and the USGS geophysical survey helped confirm the fault.

LiDAR studies

While the scientists and engineers focused on characterising the dam’s foundation, it became evident that better quality base mapping was required. This terrain model would also be useful for the hydrologists to evaluate the true capacity of the watershed and Martis Creek dam’s capacity to store and discharge this volume of water. They were also studying the consequences of a failure of the dam known as a breach condition to see how this might affect a large population in Reno and Sparks downstream along this narrow Truckee river drainage.

The hydrologists needed a good quality terrain model to reflect the ground surface digitally for use in hydrology and hydraulic computer models. To do this, they obtained LiDAR imagery from a local water purveyor and began studies based on this imagery.

LiDAR is a technology that uses focused beams of light (laser pulses) and measures the time it takes the light pulse to go from the sensor to an object where it reflects back to the sensor. Using the speed of light, the travel time is converted to a distance. With airborne systems whose flying height is known, the elevation of the ground can be calculated. Modern LiDAR scanning systems are capable of emitting more than 150,000 pulses per second. When combined with aerial photography, LiDAR is a powerful tool for mapping and studying the earth’s surface.

As the team interacted to collectively resolve the dam’s issues, a very important discovery was made by USACE. Its geologists used the bare-earth data to produce a three-dimensional terrain model using specialised LiDAR software. This data represents those data points that were the latest returns recorded, meaning they had travelled the farthest and are the ones most likely to have been reflected off of the ground surface. By focusing on these data points, vegetation is effectively removed and the modeled surface is that closest to representing the ground. The result was amazing. When the intensity data (effectively a measure of the strength of the return pulse) was looked at, painted stripes and numbers on the runway at the airport near the dam could even be seen.

In the terrain model produced they observed several linear features that are indicative of faults in the vicinity of the East Martis Creek Dam. Once focused in on that area, a sharp vegetation lineament was also observed in the aerial photography. This find was significant because the recent seismic hazard assessment for the dam had not indicated any Quaternary-age faults in the immediate area, and the lineaments identified in the LiDAR imagery projected towards the dam.

This result led the team to acquire the remaining LiDAR data from the utility district and analyse them for more evidence of faulting. With these new data the broader context of what the team was looking at became apparent.

A linear feature that cuts across a broad alluvial terrace below the dam was identified. This feature continued to the north where it crossed the Truckee river and offset at least two generations of terrace risers. Projecting south towards the dam, a subtle feature was identified cutting across the terrace riser below the spillway and projecting between the spillway and left abutment of the dam. Features indicative of the fault included aligned hills, ridgelines, depressions and mole-tracks (characteristic sequences of small ridges and depressions). Some of these features have only 0.3-0.6m of relief, but the LiDAR was able to depict them.

Each feature was carefully digitised in GIS and set the stage for extensive field geomorphic studies to see and understand these features on the ground surface. By comparing aerial photos with the LiDAR data, the team was able to fill in many of the gaps in the LiDAR lineament with features that are not reflected in the topography but result in changes in vegetation cover.

Once the team was able to ground map and examine these numerous features along the lineament, they selected a few candidate sites for trenching. These paleoseismic trenches are not the usual trenching and logging. They are logged in tremendous detail searching for rotated cobbles and smeared clay threads that may represent faulting. Among these candidate sites, the team selected the best sites based on right-of-entry and access along with the locations most likely to reveal strong evidence of the ground disturbance along this lineament. One site was where a strong vegetation contrast was visible in both aerial imagery and on the ground. The second site was higher near Martis Creek and was coincident with a well-defined scarp strongly evident in the LiDAR data.

Exploring lineament into the ground

In the fall of 2008, the team began trenching at the two locations simultaneously. A team of geologists excavated and logged the subsurface condition in detail at both paleoseismic trench sites. Early in the logging at both sites, the evidence was clear.

In both cases, well-defined shears in the ground offset otherwise continuous layers of the fan deposits, and both coincided precisely with the surface feature that had been mapped at that location. The lineament was clearly a fault, or perhaps two different strands of the same fault that are parallel.

Not all faults are considered a problem. Historically, geologists have based the interpretation of the hazard of fault rupture to be consistent with their age or activity. Younger faults are thought to be active, and older faults are thought to be dormant or inactive. In the case of USACE and at the time these studies were performed, their regulations define this activity as 35,000 years. If the fault appears to have moved recently the fault needs to be considered as a location where ground rupture may occur and is also recognised as a potential source of an earthquake.

The fan deposits were sampled and examined to attempt to determine an age, but no good quality results could be obtained. Samples were taken and tested using mass radiocarbon dating techniques and yielded less than 1000-year ages but could not be relied on because of the averaging that occurs within the samples contaminated by possibly younger roots and soils. Earth deposits are also qualitatively dated-based on specialised methods of assessing their colour, climate, and presence of clay and mineralisation that are strong indicators of age. In both trenches, the fault was interpreted to be possibly as young as 10,000 years, or Holocene.

Two distinct earthquake events can be identified in Trench 1, with the most recent event rupturing the basal contact of the uppermost soil unit. Apparent vertical offset during this event is about 12.5cm, whereas the largest earthquake event produced an estimated 0.3m of apparent vertical offset. The horizontal offset could not be determined for either event, but is likely to be significantly larger than the vertical offset. The fault exposure in Trench 2 offsets volcanic bedrock and the overlying fan deposits also in an apparent down-to-the-west sense of displacement.

The fault that had been previously mapped across the North Tahoe Basin is now known to be an active fault capable of ground rupture and needs to be considered as another seismic source for future earthquakes. The fault was named the Polaris fault after a nearby historic locality along the Truckee river, north of Martis Creek dam.

Studies continue to further assess the Polaris fault regarding its true age, recurrence interval (frequency of earthquakes), and relationship to other nearby fault systems. This will require identification and logging of several additional paleoseismic trench sites so that multiple earthquake ruptures can be directly observed and dated. These studies are currently progressing.

Bruce Hilton is a principal engineering geologist for Kleinfelder.

Ronn Rose is an engineering geologist with the Sacramento District of the US Army Corps of Engineers. His current position is that of the district dam safety programme manager.

Lewis Hunter is the senior geologist for the Sacramento District, US Army Corps of Engineers.