Earth Imaging Journal: Remote Sensing, Satellite Images, Satellite Imagery
Breaking News
2018 URISA Exemplary Systems in Government Award Recipients Announced
URISA Is pleased to announce the recipients of 2018...
Create a Functional Map Application in a Day with Jibestream’s Indoor Mapping Platform
TORONTO - Jibestream, the premier indoor mapping platform provider,...
Ubisense Announces myWorld for Salesforce on the Salesforce AppExchange, the World’s Leading Enterprise Apps Marketplace
Ubisense today announced it has launched myWorld for Salesforce...
Maxar’s SSL Selected to Define Next-Generation Secure Satellite Communications for the U.S. Air Force
HERNDON, VA - SSL, a Maxar Technologies company (formerly MacDonald,...
Airgon Technical Seminar at Commercial UAV Expo: Meet Our Guest Speakers
Please join us on Monday, October 1st for our...

LandslidePhoto_USGS_Godt_ By Matteo Luccio

On March 22, 2014, four miles east of Oso, Wash., a portion of an unstable hill collapsed, sending mud and debris across the North Fork of the Stillaguamish River, covering an area of approximately one square mile and devastating a rural neighborhood, killing 43 people. It also dammed the river, causing extensive flooding upstream and blocking State Route 530. Excluding landslides caused by volcanic eruptions, earthquakes or dam collapses, the Oso slide was the deadliest single landslide event in U.S. history.

LiDAR data—if collected, analyzed and disseminated to local residents and policy makers—can be invaluable in mapping landslide hazards. In turn, such mapping can greatly help mitigate the consequences of landslides. After the Oso disaster, interest in landslide hazards grew significantly in Washington as well as Oregon, California and other mountainous states.

A map describes slope inclination in the area around the Oso landslide (outlined in red).

A map describes slope inclination in the area around the Oso landslide (outlined in red).

Data and Accuracy Requirements

The type, amount and accuracy of data required to assess landslide hazards depends on the intended use. A hazard assessment over a wide area for general planning purposes could be based on coarse data, such as the locations of mountains and steepness of terrain. Conversely, a landslide hazard assessment for a particular potential subdivision in a mountainous area would require much more detailed information and higher spatial resolution, according to Dave Montgomery, a professor of geomorphology in the University of Washington’s Department of Earth and Space Sciences and a member of the Oso team of Geotechnical Extreme Events Reconnaissance (GEER), a group sponsored by the National Science Foundation.

Typically, the first step in mapping landslide hazards is to inventory previous landslides—their location, size, direction, type, depth, slope and geology—because the key to understanding the future is understanding the past, points out Bill Burns, an engineering geologist with the Oregon Department of Geology and Mineral Industries (DOGAMI).

The Role of LiDAR

LiDAR data collected from an aerial platform give the precise position of the ground at each laser point. Five to 10 years ago, researchers didn’t anticipate the resolution and accuracy that now can be achieved with LiDAR.

“It has been a breakthrough in revealing the really small morphological features,” says Joseph Wartman, an associate professor in the Department of Civil and Environmental Engineering of the University of Washington and co-leader of GEER’s Oso team.

A key feature of LiDAR is its ability to penetrate vegetation, as long as it’s not too dense. By processing LiDAR data via algorithms, researchers can mathematically remove tree-canopy vegetation, revealing the geomorphology of the “bare earth” beneath it. LiDAR enables geologists to recognize patterns of previous landslides by providing very-high-resolution topographic data of the actual ground surface. It gives geologists “a brand new pair of glasses for looking at terrain,” says Montgomery, because it represents the topography with much higher spatial resolution and greater clarity than previous technologies.

“A typical landslide will have a steep head and a flat upper part, and then a concave part in the head and a convex part at the toe,” explains Jeff Keaton, principal engineering geologist with AMEC Americas and co-leader of GEER’s Oso team.

In 2009, a study of how many landslides had been mapped in the state of Oregon before DOGAMI started using LiDAR found 12,000 mapped landslides. Using LiDAR, the agency now has 41,000 landslides in that same database, and it has only covered a small portion of the state.

“People mapped landslides in Oregon for 60 years using aerial photography and walking around on the ground, and they couldn’t see them,” notes Burns.

LiDAR data chronicle the Oso landslide area, from 2003 (top) to 2013 before the landslide (middle) and 2014 after the landslide (bottom).

JdLC-D_Oso Landslide Area 2003 LiDAR MapJdLC-E_Oso Landslide Area 2013 LiDAR Map

JdLC-G_Oso Landslide Area 2014 Lidar MapAssessing Ground Composition

Landslide hazard assessment includes trying to assess the underground composition of a hillside to see if it’s made of hard rock or sand. Geologists look for exposures or use geologic maps to see what’s supposed to be there. Shallow geophysics (e.g., ground-penetrating radar and shallow seismic sensors) can help reveal the internal structure and composition of a slope.

Another method is exploratory borings. “I’ve gone down into borings several feet in diameter,” says Montgomery. “You get lowered down into them to log a profile through suspected or potential landslides.”

Types of Landslides

According to Montgomery, there are basically two types of landslides: large, deep-seated ones and shallow ones. Typically, the former involve a whole piece of a mountain, while the latter involve just the surficial soil and perhaps the weathered rock on top of the fresh rock. LiDAR data can be used to identify most of the deep-seated landslides that have already happened, and sufficiently good LiDAR data can reveal recent shallow landslides. Alternatively, inputting high-fidelity LiDAR terrain data into models can help predict where shallow landslides may take place.

“Mapping the risk requires figuring out where we’ve had landslides in the past, what the geology there was, and where else we have those same conditions,” says Burns. It’s also common for portions of landslides, or entire landslides, to re-activate. “The inventory really helps us with the susceptibility maps,” he adds.

Keaton counts four different categories of landslides:

  1. Translational landslides, which are slopes that move by sliding.
  2. Earth movements that gradually lean forward until they no longer have sufficient support to keep them on the slope and topple over.
  3. Spreads, which are translational slides riding on a surface that tends to be destabilized by earthquake shaking.
  4. Flows, in which earth material becomes fluidized and moves as a slurry.

Sometimes a landslide moves from an unstable condition to a stabilized condition, thereby reducing the potential for future movements. Furthermore, the transition of earth material from one place on a slope to another can reduce the surface angle and make it more stable, provided that groundwater conditions and other external factors don’t exceed some threshold.

A hillshade map shows ground conditions based on 2003 LiDAR data.

A hillshade map shows ground conditions based on 2003 LiDAR data.

Pattern Recognition and Analysis

Collecting LiDAR data is just the first step in mapping landslide hazards and, ultimately, reducing risks due to geologic hazards. Next, the data must be analyzed and interpreted. Currently, this often is done by experts who look at the data and interpret geologic forms as landslides or features of past landslides.

“The human eye and brain are pretty powerful integrators of information,” says Montgomery. “So a well-trained geologist can map many landslides, or potential landslide source areas, simply from having good topographic data from LiDAR.”

Landslides alter the morphology of the ground surface in specific ways. Researchers have developed visualizations that highlight those specific features and make them easier to see. Translating LiDAR data into evidence about past landslides requires high-quality data (i.e., greater than eight pulses per square meter, multiple returns, classified point clouds, many ground-control points, etc.) as well as someone extremely familiar with what landslides look like, such as an engineering geologist. Such assessment needs to be done in a GIS on a powerful computer that can take the data, process them, and construct the polygons surrounding the landslides.

This process can be automated further by generating algorithms to identify deep-seated slides or areas potentially prone to shallow landsliding. Due to the work that has already been done to automate that process, Wartman believes automation will be “the next frontier” in the field. “An enormous amount of LiDAR data will become available to us,” he says, “and I don’t think it is practical to have individuals look at them and visually interpret them.”

An additional and critically important step is to make such information usable by members of the public or public officials. “That involves consideration of risk,” says Wartman, “which is not just thinking what the likelihood is of a geologic hazard occurring in the next 10 or 20 years, but, instead, thinking about what the consequences would be on human populations, the systems we use to support modern life, and public and private property.”

Conveying Risk

There’s a big difference between mapping all the areas where landslides might occur and predicting which ones will occur. Notes Montgomery, “I can hit one button in a GIS and generate a landslide-hazard map that says, ‘Everything here is steeper than 25 degrees. You might have a landslide.’ It just wouldn’t be a terribly good map.”

In any steep terrain, he points out, there are many places that are potential landslide initiation sites, but only a few of them have failed historically.

“If you want to generate a prediction of which hillside will fail when, good luck,” adds Montgomery. “If you want to identify areas that have a higher potential for generating landslides and maybe even getting into what the actual risk of failure is, those are much more achievable goals.”

What’s commonly called a hazard map, based on where past landslides have occurred, isn’t really a hazard map, says Keaton. “It’s a map that shows where slope movements have occurred, but if we don’t connect to it how often the ground moves at every location and how much the ground moves, then we are missing that element we need to [determine] what kind of damage can occur.”

In addition, such a map misses the potential for landslide run-out. In Oso, for example, all the damage was where the landslide ran out, not on the slope where it originated.

A photo shows the Oso landslide as seen from the south.

A photo shows the Oso landslide as seen from the south.

The Puget Sound LiDAR Consortium

The Puget Sound LiDAR Consortium (PSLC) is an informal group of local agency staff and federal research scientists devoted to developing public-domain high-resolution LiDAR topography and derivative products for the Puget Sound region. It came together in fall 1999 around the issue of earthquake hazards, with initial participants from Kitsap County, the Kitsap Public Utility District, the city of Seattle, the Puget Sound Regional Council, NASA and the U.S. Geological Survey (USGS). It has since been joined by Clallam and Island counties.

Although it has no rules or formal membership, these governments and agencies work in concert. From a legal perspective, PSLC is no more than a set of interagency agreements that allow purchases of LiDAR surveying via a contract negotiated by Kitsap County.

At the moment, PSLC is an unfunded, volunteer effort maintained by people distributed at the agencies generating the data and then providing them to someone at the University of Washington, who has been putting the data together and organizing them on a volunteer basis. Consortium members are committed to bringing all the data together into a single Web site, where they can be served and made available at no charge; because they were collected with public money, they are public data.

The Oso Disaster

LiDAR data for Oso had been collected and processed in 2003 and 2013, and was available to PSLC members at no cost, so why wasn’t the disaster avoided? The subdivision that became Steelhead Drive was built in the early 1960s, therefore much of the community was already there by the time the first LiDAR data were collected.

“The two LiDAR collections done prior to the catastrophic 2014 failure were not collected as part of a landslide hazard-assessment program,” notes Montgomery. Before the Oso landslide, the state of Washington had only one staff person working half-time in the Department of Natural Resources to serve the statewide landslide hazard assessment.

“The LiDAR documented everything but did not detect anything, because detection requires an interpretation,” adds Keaton.

Ultimately, Wartman argues, the failure of the LiDAR data to avert the disaster is due to a lack of systematic programs for assessing LiDAR data. “LiDAR data is not going to do anything until it is interpreted, and then presented and made accessible to members of the public. Those last two critical steps were never taken.”


Prevention and the Future

Montgomery served on the SR 530 Landslide Commission appointed by Washington Governor Jay Inslee to recommend what the state should do to try to prevent future disasters such as the one in Oso. One of the commission’s top two recommendations was to expand the landslide hazard-mapping program; get statewide LiDAR data; and hire people to start analyzing the data, making the maps and making them publicly available.

“I would love to see the type of statewide mapping where anybody would be able to go online, call up an area that they are interested in, look at the LiDAR data for themselves, superimpose on them a geologic map, superimpose on it a hazard map that has been conducted and vetted by the state’s geologists, and be able to access and make maps via the Web,” adds Montgomery. “All the technology exists already; what doesn’t exist yet is the support structure.”

Another important development is USGS’ recent implementation of the 3D Elevation Plan (3DEP) to cover the entire United States (except Alaska) with LiDAR.

“For the most part,” Wartman predicts, “we are going to have continuous coverage of the United States within 10 years. The next big challenges are in assessing those data, finding the means to address this enormous amount of LiDAR data that soon will become available and translating it to members of the public.”

Editor’s Note: For more information, visit the related GEER report.


Other Ways to Detect Landslides

A variety of technologies besides LiDAR can help detect landslide hazards, including ground-penetrating radar (GPR), which uses the dialectric properties of the earth-water system. It requires an antenna that generates electromagnetic frequency energy and another nearby antenna that receives energy reflected by the ground. Researchers then interpret the reflected data.

“We often do this visually,” explains Jeff Keaton, principal engineering geologist with AMEC Americas and co-leader of GEER’s Oso team, “but we can do it with a method that allows us to make horizontal or vertical slices through the earth. The antennas have to be in contact with the ground or very close to it.”

For example, if a landslide aligned clay minerals in the sliding process, GPR can be used to map its base by revealing the contrast between those minerals and the adjoining material. Beyond that, however, GPR isn’t useful for mapping landslide hazards, because of the following:

  1. Researchers need data collected across wide expanses, but GPR is a tool carried by someone walking. By contrast, an airplane flying LiDAR can collect large areas quickly.
  2. GPR doesn’t illustrate the ground surface, only what’s underneath the ground.

Slopes also can be evaluated by ascertaining their physical properties, usually by drilling. Acoustic seismic methods, for example, measure the speed of sound. This is typically done by striking a metal plate on the ground with a sledgehammer, which generates an energy pulse that moves through the subsurface, while a linear array of geophones measure the pulse’s arrival time.

Another method uses a seismograph and a trigger on the hammer—when it strikes the plate, it turns on the timer. Measuring when energy vibrations are received at each geophone’s location, and knowing the distance from the plate, allows researchers to calculate the seismic velocity, which they then translate into material shear strength.

Matteo Luccio is a freelance writer with Pale Blue Dot, specializing in geospatial technologies; e-mail:

Comments are closed.