USGS Science⌕ Search

SEARCH · USGS Science

Results for “Landslides”

Search indexed USGS publications on groundwater, aquifers, geologic maps, mineral resources and earthquakes. Explore source records by subject and place.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 487 records · Page 27Linked to original sources
USGS2025

Harnessing geospatial artificial intelligence and deep learning for landslide inventory mapping: Advances, challenges, and emerging directions

Recent advancements in artificial intelligence (AI) and deep learning enable more accurate, scalable, and automated mapping. This paper provides a comprehensive review of the applications of AI, particularly deep learning, in landslide inventory mapping. In addition to examining commonly used data sources and model architectures, we explore innovative strategies such as feature enhancement and fusion, attention-boosted techniques, and advanced learning approaches, including active learning and transfer learning, to enhance model adaptability and predictability. We also highlight the remaining challenges and potential research directions, including the estimation of more diverse variables in landslide mapping, multimodal data alignment, modeling regional variability and replicability, as well as issues related to data misinterpretation and model explainability. This review aims to serve as a useful resource for researchers and practitioners, promoting the integration of deep learning into landslide research and disaster management.

Remote Sensing↗
USGS2015

Discussion and Review of “Seismology of the Oso-Steelhead landslide” by Hibert, Stark, and Ekstrom

In this study, Hibert et al. use the seismic waves generated by the disastrous 22 March 2014 landslide near Oso, Washington to investigate the dynamics of the event by combining long-period source inversions and short period analysis. Overall the paper is concise and well written and takes an innovative approach by integrating both high- and low-frequency seismic data together in the interpretation. It also demonstrates the useful prospect of resolving a trajectory of the center of mass of the landslide from a few traces of noisy seismic data. However, the main point of discussion that I would like to address stems from differences between the findings by Hibert et al. and the findings of a recently published multidisciplinary paper on the Oso landslide (Iverson et al., 2015). I am a co-author on the paper and performed the seismic analysis, which was similar to that of Hibert et al., yet the interpretation of how the event unfolded differs significantly between the two papers. In my opinion, this stems largely from a difference in information, some of which Hibert et al. unfortunately did not have access to prior to submission because it was just published in Iverson et al. (2015). However, differences also seem to stem from technical details of the methods and data used. Typically, such details are only of interest to seismologists, but in this case they make a substantial difference in understanding how the event unfolded and how it became so disastrous. For that reason, they are worthy of discussion here. The crux of the difference lies in two points 1) Iverson et al. (2015) found a multi-stage initiation of the main event, while Hibert et al. did not and 2) Iverson et al. (2015) found the second high-frequency event to be from a significantly smaller debris fall whereas Hibert et al. estimate it to be a substantial mass. I detail these two points below, then follow with an additional comment regarding the trajectory estimation made by Hibert et al. and some minor comments.

Natural Hazards and Earth System Sciences↗
USGS2006

Submarine landslides in the Santa Barbara Channel as potential tsunami sources

Recent investigations using the Monterey Bay Aquarium Research Institutes (MBARI) Remotely Operated Vehicles (ROVs) "Ventana" and "Tiburon" and interpretation of MBARI's EM 300 30 kHz multibeam bathymetric data show that the northern flank of the Santa Barbara Basin has experienced massive slope failures. Of particular concern is the large (130 km2) Goleta landslide complex located off Coal Oil Point near the town of Goleta, that measures 14.6-km long extending from a depth of 90 m to nearly 574 m deep and is 10.5 km wide. We estimate that approximately 1.75 km3 has been displaced by this slide during the Holocene. This feature is a complex compound submarine landslide that contains both surfical slump blocks and mud flows in three distinct segments. Each segment is composed of a distinct head scarp, down-dropped head block and a slide debris lobe. The debris lobes exhibit hummocky topography in the central areas that appear to result from compression during down slope movement. The toes of the western and eastern lobes are well defined in the multibeam image, whereas the toe of the central lobe is less distinct. Continuous seismic reflection profiles show that many buried slide debris lobes exist and comparison of the deformed reflectors with ODP Drill Site 149, Hole 893 suggest that at least 200 000 years of failure have occurred in the area (Fisher et al., 2005a). Based on our interpretation of the multibeam bathymetry and seismic reflection profiles we modeled the potential tsunami that may have been produced from one of the three surfical lobes of the Goleta slide. This model shows that a 10 m high wave could have run ashore along the cliffs of the Goleta shoreline. Several other smaller (2 km2 and 4 km2) slides are located on the northern flank of the Santa Barbara Basin, both to the west and east of Goleta slide and on the Concepcion fan along the western flank of the basin. One slide, named the Gaviota slide, is 3.8 km2, 2.6 km long and 1.7 km wide. A distinct narrow scar extends from near the eastern head wall of this slide for over 2 km eastward toward the Goleta slide and may represent either an incipient failure or a remnant of a previous failure. Push cores collected within the main head scar of this slide consisted of hydrogen sulfide bearing mud, possibly suggesting active fluid seepage and a vibra-core penetrated ???50 cm of recent sediment overlying colluvium or landslide debris confirming the age of ???300 years as proposed by Lee et al. (2004). However, no seeps or indications of recent movement were observed during our ROV investigation within this narrow head scar indicating that seafloor in the scar is draped with mud. ?? 2006 Author(s). This work is licensed under a Creative Commons License.

Natural Hazards and Earth System Sciences↗
USGS1993

The Golden bypass landslide, Golden, Colorado

Slope instability along a new highway bypass in Golden, Colorado, became a major concern in 1993. Rains and snowmelt accelerated movement of a landslide that had begun to develop before the bypass was opened to traffic in July of 1991. The downslope movement of earth materials increased significantly in 1993. During the first few months of the year, the landslide pushed onto the west shoulder of the road and crumpled the pavement beneath the south-bound lane. As we prepare this article (September, 1993), the slide continues to encroach onto the highway, posing a persistent problem despite repeated efforts to slow or stop its movement. As this article will show, permanent solutions to landslide problems of this kind are difficult to obtain.

Colorado↗
USGS1993

Landslides caused by the Klamath Falls, Oregon, earthquakes of September 20, 1993

The Klamath Falls earthquakes caused landslides throughout an area of about 420 sq km and as far as about 29 km from the epicenter, a distribution that is typical for magnitude 6 earthquakes (see graphs on following pages). Most of the landslides were rock falls or shallow, highly disrupted rock slides from roadcuts, quarries, and steep bluff faces. Such landslides are also among the most common types caused by other earthquakes of comparable magnitude. In the Klamath Falls area, the most numerous earthquake-induced rock falls were along the east-to southeast-facing flank of a ridge immediately south and west of Howard Bay (locality 1 on the accompanying map), 18 km east-southeast of the epicenter of the magntiude 6.0 shock at 10:45 p.m. This ridge is more than 240 m high and has slopes steeper than 45 ° in places. The upper part of the ridge is composed of material from basaltic lava flows, an the lower slopes are covered with colluvium and talus deposits containing abundant boulders.

Oregon↗
USGS2001

The rainfall-triggered landslide and flash-flood disaster in northern Venezuela, December 1999

Rainstorms in December 1999 induced thousands of landslides along the northern slopes of the Cordillera de la Costa mountain range principally in the state of Vargas, Venezuela . Rainfall accumulation of 293 millimeters during the first 2 weeks of December was followed by an additional 911 millimeters of rainfall on December 14 through 16. The landslides and floods inundated coastal communities resulting in a catastrophic death toll estimated at between 15,000 and 30,000 people. Debris flow damage to houses, buildings, and infrastructure in the narrow coastal zone was severe. Flash floods on alluvial fans at the mouths of rivers draining the coastal mountain range also contributed to the general destruction. In time scales spanning decades to centuries, the alluvial fans along this Caribbean coastline are areas of high geomorphic activity. Because most of the coastal zone in Vargas consists of steep mountain fronts that rise directly from the Caribbean Sea, the alluvial fans provide the only relatively flat areas upon which to build. Rebuilding and reoccupation of these areas requires careful determination of hazard zones to avoid future loss of life and property. A limited assessment of the distribution and character of landslides is currently in progress by the U.S. Geological Survey in cooperation with the Venezuelan Ministry of Environment and Natural Resources.

Conference Paper↗
USGS1999

Evaluation of landslide hazards with ground-penetrating radar, Lake Michigan coast

Ground - penetrating radar (GPR) and boreholes were used to investigate a landslide -prone bluff at Sleeping Bear Dunes National Lakeshore on the northeastern coast of Lake Michigan . Based on borehole observations, sediment underlying the area is homogeneous, consisting of well-sorted, medium to coarse sand. GPR penetrated up to 20 m deep in these sediments , revealing the late Quaternary stratigraphy in great detail. We define four units, or radar facies, based on criteria similar to those used in seismic stratigraphy. Directly beneath a landslide at Sleeping Bear Point (and nowhere else in this survey) is a deeply incised, channel-fill deposit that intersects the shoreline at a high angle. The buried channel is at least 10 m deep and 400 m wide, and it might be a subglacially carved feature of Pleistocene age. A prominent, planar unconformity marks the upper surface of the channel deposit, which is overlain by stratified beach and dune material. Several crosshole GPR surveys were performed in the vicinity of the landslide: 1) a constant offset profile (COP), 2) a multiple offset gather (MOG), and 3) a vertical radar profile (VRP). Tomographic analysis of these data determined the velocity structure of sandy sediment that underlie the failed bluff. Because GPR velocity is dependent on electrical properties, we use it as a proxy for geotechnical properties of the soils. Our working hypothesis is that the hidden channel may act as a conduit for pore water flow between upland regions and Lake Michigan, and thereby locally reduce soil strength and promote slope failure.

Conference Paper↗
USGS2017

Integrating landslide and liquefaction hazard and loss estimates with existing USGS real-time earthquake information products

The U.S. Geological Survey (USGS) has made significant progress toward the rapid estimation of shaking and shakingrelated losses through their Did You Feel It? (DYFI), ShakeMap, ShakeCast, and PAGER products. However, quantitative estimates of the extent and severity of secondary hazards (e.g., landsliding, liquefaction) are not currently included in scenarios and real-time post-earthquake products despite their significant contributions to hazard and losses for many events worldwide. We are currently running parallel global statistical models for landslides and liquefaction developed with our collaborators in testing mode, but much work remains in order to operationalize these systems. We are expanding our efforts in this area by not only improving the existing statistical models, but also by (1) exploring more sophisticated, physics-based models where feasible; (2) incorporating uncertainties; and (3) identifying and undertaking research and product development to provide useful landslide and liquefaction estimates and their uncertainties. Although our existing models use standard predictor variables that are accessible globally or regionally, including peak ground motions, topographic slope, and distance to water bodies, we continue to explore readily available proxies for rock and soil strength as well as other susceptibility terms. This work is based on the foundation of an expanding, openly available, case-history database we are compiling along with historical ShakeMaps for each event. The expected outcome of our efforts is a robust set of real-time secondary hazards products that meet the needs of a wide variety of earthquake information users. We describe the available datasets and models, developments currently underway, and anticipated products.

Conference Paper↗
USGS1983

Seedling establishment on a landslide site

Two landslide scars (slide tracks) were plot sampled one and two years after landsliding to determine plant species involved in ecesis. The study site is located in a blockfield cove on Massanutten Mountain, northern Virginia. Old growth forest, adjacent to the slide tracks, is composed primarily of hemlock (Tsuga canadensis) and sweet birch (Betula lenta). These species are the first and foremost colonizers on the slide tracks. Dendrochronological analysis shows that landsliding is frequent at the study site and this disturbance may impart non-equilibrium conditions on the forest. Seed bed characteristics, light availability, and mode of dispersal may be important factors in plant ecesis on slide tracks

Virginia↗
USGS2002

Landslide triggers and types

The term “landslide trigger” refers to an external stimulus, such as intense rainfall, rapid snowmelt, seismic shaking, volcanic eruption, stream/coastal erosion, or natural dam failure, which causes an immediate or near-immediate response in the form of landslide activity. In this paper, we discuss the types of landslides caused by these triggering processes and present a limited number of worldwide examples of their occurrence.

Conference Paper↗
USGS2011

Effects of soil-engineering properties on the failure mode of shallow landslides

Some landslides mobilize into flows, while others slide and deposit material immediately down slope. An index based on initial dry density and fine-grained content of soil predicted failure mode of 96 landslide initiation sites in Oregon and Colorado with 79% accuracy. These material properties can be used to identify potential sources for debris flows and for slides. Field data suggest that loose soils can evolve from dense soils that dilate upon shearing. The method presented herein to predict failure mode is most applicable for shallow (depth <5 m), well-graded soils (coefficient of uniformity >8), with few to moderate fines (fine-grained content <18%), and with liquid limits <40.

Landslides↗
USGS2020

Observations on the May 2019 Joffre Peak landslides, British Columbia

Two catastrophic landslides occurred in quick succession on 13 and 16 May 2019, from the north face of Joffre Peak, Cerise Creek, southern Coast Mountains, British Columbia. With headscarps at 2560 m and 2690 m elevation, both began as rock avalanches, rapidly transforming into debris flows along middle Cerise Creek, and finally into debris floods affecting the fan. Beyond the fan margin, a flood surge on Cayoosh Creek reached bankfull and attenuated rapidly downstream; only fine sediment reached Duffey Lake. The toe of the main debris flow deposit reached 4 km from the headscarp, with a travel angle of 0.28; while the debris flood phase reached the fan margin 5.9 km downstream, with a travel angle of 0.22. Photogrammetry indicates the source volume of each event is 2-3 Mm3, with combined volume of 5 Mm3. Lidar differencing, used to assess deposit volume, yielded a similar total result; although error in the depth estimate introduced large error and masks expected increase due to dilation and entrainment. The average velocity of the rock avalanche-debris flow phases, from seismic analysis, was ~25-30 m/s, and the velocity of the 16 May debris flood on the upper fan, from super-elevation and boulder sizes, was 5-10 m/s. The volume of debris deposited on the fan was ~104 m3, 2-orders of magnitude less than the avalanche/debris flow phases. The 13 May landslide was apparently triggered by rapid snowmelt; with debuttressing triggering the 16 May event. While spring 2019 was warm, it wasn’t unusual. It is likely that progressive glacier retreat and permafrost degradation were the conditioning factors; precursor activity was noted at least 1 yr previous; thus, the mountain was primed to fail and average seasonal snowmelt tipped the balance.

British Columbia↗
USGS2023

The 2022 Chaos Canyon landslide in Colorado: Insights revealed by seismic analysis, field investigations, and remote sensing

An unusual, high-alpine, rapid debris slide originating in ice-rich debris occurred on June 28, 2022, at 16:33:16 MDT at the head of Chaos Canyon, a formerly glacier-covered valley in Rocky Mountain National Park, CO, USA. In this study, we integrate eyewitness videos and seismic records of the event with meteorological data, field observations, pre- and post-event satellite imagery, and uncrewed aircraft vehicle imagery to characterize the event and future hazards it may pose. Deformation of the eventual slide mass preceded rapid failure by decades, starting in the early to mid-2000s, accelerating in 2018 (the warmest year on record), and reaching ~ 20 m/year in 2021. The main event, which was preceded by smaller sliding episodes earlier that day, had a volume of ~ 2.1 million m 3 , reached peak velocities of about 5 m/s, slid on a surface up to 80 m deep, and moved up to ~ 245 m downslope in < 2 min. We observed blocks of frozen debris (permafrost) in the landslide deposits. Within ~ 2 weeks, these blocks had melted and became dry, conical debris mounds (molards). We hypothesize that the rapid slide was induced by gradually increasing long-term air temperatures that thawed ice and increased pore pressures. The presence and suspected influence of permafrost on the occurrence of this landslide indicate other slopes in the park, and other moderate-to-low latitude alpine regions may experience similar slope stability issues as temperatures continue to warm.

Colorado↗
USGS2025

Utility of a swath laser rangefinder for characterizing mass movement flow depth and landslide initiation

Mass movements such as debris flows and landslides are some of the deadliest and most destructive natural hazards occurring mostly in alpine and volcanic settings. With ever-growing populations located downslope from known debris flow channels, early warning systems can help prevent loss of life. Geophysical and technological advances have improved monitoring and detection capabilities in recent years; however, they can often be cost prohibitive and resource intensive, making them less accessible to disadvantaged populations. We tested and validated a readily available and cost-effective two-dimensional swath laser rangefinder in a controlled experimental setting against two independent flow-depth lasers. The swath laser successfully recorded cross-sectional changes in flow depth from four debris flows and a water-only flood, in addition to geomorphic changes associated with landslide initiation. The results suggest that a swath laser could be integrated into systems for debris flow detection and characterization of mass movements in natural settings, thus improving the ability to monitor these hazards.

Landslides↗
USGS2016

Elucidating the role of vegetation in the initiation of rainfall-induced shallow landslides: Insights from an extreme rainfall event in the Colorado Front Range

More than 1100 debris flows were mobilized from shallow landslides during a rainstorm from 9 to 13 September 2013 in the Colorado Front Range, with the vast majority initiating on sparsely vegetated, south facing terrain. To investigate the physical processes responsible for the observed aspect control, we made measurements of soil properties on a densely forested north facing hillslope and a grassland-dominated south facing hillslope in the Colorado Front Range and performed numerical modeling of transient changes in soil pore water pressure throughout the rainstorm. Using the numerical model, we quantitatively assessed interactions among vegetation, rainfall interception, subsurface hydrology, and slope stability. Results suggest that apparent cohesion supplied by roots was responsible for the observed connection between debris flow initiation and slope aspect. Results suggest that future climate-driven modifications to forest structure could substantially influence landslide hazards throughout the Front Range and similar water-limited environments where vegetation communities may be more susceptible to small variations in climate.

Geophysical Research Letters↗
USGS2009

Deposition of Mount Mazama tephra in a landslide-dammed lake on the upper Skagit River, state of Washington

The cataclysmic eruption of Mount Mazama, Oregon, at c. 6730 (super 14) C yr BP, deposited tephra over 1.0 X 10 (super 6) km (super 2) of north-western North America. Primary tephra fall accumulated to a thickness of 2 cm in the upper Skagit River watershed, Washington. Mazama tephra eroded from this watershed was deposited in Lake Ksnea, of 14 km length and 40 m depth. This lake was created when a landslide blocked the Skagit River at 7040 (super 14) C yr BP. Horizontally bedded, dark grey silt and clay were deposited slowly by suspension settling in Lake Ksnea before the eruption of Mount Mazama. The 2-cm-thick primary Mazama tephra layer abruptly caps 7 m of pre-eruption sediments, and is overlain by as much as 17 m of Mazama tephra deposited relatively rapidly on a delta at the mouth of Damnation Creek. Most of a 13-m-thick section is composed of lacustrine tephra containing rhythmic stratified beds deposited by suspension settling. Turbidity currents deposited centimetre-scale, cross-bedded silt and tephra at the top of some rhythmite beds. Lower in this section, tephra containing abundant fine-grained terrestrial sediments and other sedimentary structures interrupts the rhythmite beds. These structures include faulted and warped beds, flame structures and pendants created by soft-sediment deformation. Tephra deposits are overlain conformably with cross-bedded sands throughout most of a 200-m-long section. Coarse alluvial gravels and landslide deposits unconformably overlie the tephra and sand at several locations. The deposits described are interpreted as an inversely graded, prograding delta sequence composed almost entirely of Mount Mazama tephra. Despite a lack of age control on the rate of tephra deposition, the sedimentology of this section indicates that the tephra delta was deposited within 1 yr or less

Oregon↗
USGS2010

The Block composite submarine landslide, southern New England slope, U.S.A.: A morphological analysis

Recent multibeam surveys along the continental slope and rise off southeast New England has enabled a detailed morphological analysis of the Block composite landslide. This landslide consists of at least three large debris lobes resting on a gradient less than 0.5 °. The slide took place on gradients of between 1 ° and 5 ° in Quaternary sediments likely deposited at the time of low sea level and high sedimentation rates associated with glaciations. The slide debris lobes are very close to each other and cover an area of about 1.125 km 2 of the sea floor. With an average thickness of 50 m, the total volume of the deposit is estimated at 36 km 3 . In some cases, the departure zone appears to be near the crest of the continental slope, at a water depth between 500 and 2,000 m with debris spreading over about 20 km at a depth ranging from 2,500 to 2,600 m. From preliminary analysis, at least one lobe of the Block Composite slide (lobe 2) would require further study to evaluate its tsunamigenic potential.

Book chapter↗
USGS1983

Landslide risk assessment

Landslide risk can be assessed by evaluating geological conditions associated with past events. A sample of 2,4 16 slides from urban areas in West Virginia, each with 12 associated geological factors, has been analyzed using SAS computer methods. In addition, selected data have been normalized to account for areal distribution of rock formations, soil series, and slope percents. Final calculations yield landslide risk assessments of <0.50=low risk, 0.50 to 1.50=moderate risk, and >1.50=high risk. The simplicity of the method provides for a rapid, initial assessment prior to financial investment. However, it does not replace on-site investigations, nor excuse poor construction. ?? 1983 Springer-Verlag New York Inc.

Environmental Geology↗