This data release contains displacement and pore-water pressure data from a field experiment performed November 14, 2003, at Mount Kaba-san, Japan. This experiment generated a shallow landslide, induced by water infiltration from overhead sprinkling, that mobilized into a debris flow. More information about this experiment can be found in Ochiai and others (2004). Extensometer data recorded the ground-surface locations (displacement) and pressure transducers recorded dynamic pore-water pressures within the hillslope leading up to and through rapid failure. Data were recorded at a 100-Hz sampling frequency on a National Instruments data-acquisition system. The accompanying cross-sectional diagram (Japan_exp_cross-section.png) illustrates the general instrument configuration at the start of the experiment. Extensometers (linear position transducers) were attached to the overhead sprinkling structure and their wire cables (that extend with displacement) were attached to ground anchors downslope of the instruments. These instruments with retractable wire rope cables were manufactured by UniMeasure. Extensometer 4 had a longer wire length to better record the transition from slow sliding motion to rapid debris flow. Extensometer Model Approx. wire length range (m) 3 JX-PA-80-N11-11S-111 2 4 HX-PA-400 8 5 JX-PA-80-N11-11S-111 2 Piezometers were directly buried pressure transducers encased in custom-built cylindrical housings (43 mm in diameter, 93 mm long) with sintered filters at their bases (refer to Japan_exp_piezometer_images.jpg). The transducers within the enclosures were manufactured by Druk, model PDCR800, with a range of 70 kiloPascals (kPa) gauge pressure. Each enclosure had two sealable tubes that extended to the ground surface after burial. These tubes allowed water to be injected into the enclosure through one tube and air to escape out the other tube. This configuration enabled full water saturation of the enclosed transducer sensors, rapid pressure transmission to the sensor diaphragms, and better recording of dynamic responses during rapid failure. This data release contains the following files: (1) Metadata for this data release. (2) Data (in csv format) from the entire experiment (until about six minutes after rapid failure) downsampled to 1-Hz for ease of viewing and plotting. Time in seconds denotes time after overhead sprinkling commenced. (3) Data (in csv format) from the rapid-failure period (24620-24640 seconds) at 100-Hz to portray dynamic responses during rapid failure. (4) Image of experiment cross section showing instrument locations and landslide failure mass. (5) Images of piezometers consisting of pressure transducers and enclosures. Extensometer data is presented as raw recorded position (m) and corrected position (m) to reflect downslope, rather than oblique, displacement. Refer to Processing Steps in the metadata for more information on extensometer data corrections. Pore-pressure data is presented in recorded centimeters of head and converted to kPa. Reference cited Ochiai, H., Okada, Y., Furuya, G., Okura, Y., Matsui, T., Sammori, T., Terajima, T. and Sassa, K., 2004, A fluidized landslide on a natural slope by artificial rainfall: Landslides, v. 1, p. 211-219.

This inventory was originally created by Yagi and others (2009) describing the landslides triggered by the M6.9 Eastern Honshu, Japan earthquake that occurred on 2008-06-13 at 23:43:45 UTC. Care should be taken when comparing with other inventories because different authors use different mapping techniques. This inventory also could be associated with other earthquakes such as aftershocks or triggered events. Please check the author methods summary and the original data source for more information on these details and to confirm the viability of this inventory for your specific use. With the exception of the data from USGS sources, the inventory data and associated metadata were not acquired by the U.S. Geological Survey (USGS) and thus have not been reviewed for accuracy and completeness by the USGS. They are presented as part of this data series for convenience of the user only, as part of an effort to make published ground-failure inventories more accessible from a single aggregated site. No warranty, expressed or implied, is made regarding the display or utility of the data on any other system or for general or scientific purposes, nor shall the act of distribution constitute any such warranty.

This inventory was originally created by Keefer and Manson (1998) describing the landslides triggered by the M 6.9 Loma Prieta, California earthquake that occurred on 18 October 1989 at 00:04:15 UTC. Care should be taken when comparing with other inventories because different authors use different mapping techniques. This inventory also could be associated with other earthquakes such as aftershocks or triggered events. Please check the author methods summary and the original data source for more information on these details and to confirm the viability of this inventory for your specific use. With the exception of the data from USGS sources, the inventory data and associated metadata were not acquired by the U.S. Geological Survey (USGS) and thus have not been reviewed for accuracy and completeness by the USGS. They are presented as part of this data series for convenience of the user only, as part of an effort to make published ground-failure inventories more accessible from a single aggregated site. No warranty, expressed or implied, is made regarding the display or utility of the data on any other system or for general or scientific purposes, nor shall the act of distribution constitute any such warranty.

Landslide susceptibility models show the potential of landslide occurrence at a location. These models are pivotal for reducing losses associated with landslides (Godt et al., 2022). In this data release, we include susceptibility results from the associated manuscript by Woodard and Mirus (2025). This manuscript shows how a morphometric model can create consistent and effective susceptibility models over large regions (> 100 km2) by analyzing the terrain’s topography. The model assumes that areas with high relative slope and hillslope area in comparison to the rest of the terrain are more susceptible to landsliding. As the model’s only input is elevation data, it mitigates the data biases common in the data-driven statistical methods (e.g., machine learning) generally used over these scales. We compare the morphometric model outputs to a parsimonious national susceptibility map and logistic regression machine learning models. The national susceptibility map is available in Belair et al., (2024). The two logistic regression models are trained on the landslide data available in the Willamette Valley Hydrologic Unit Code (HUC) 4 watershed (DOGAMI, 2024). To account for the effects of the sampling ratio of event to non-event data points, we create two logistic regression models. The first uses a 1:1 sampling ratio of landslide to non-landslide points and the second uses all the data within the training data which results in a 1:33 sampling ratio. Environmental datasets requisite for the logistic regression models are all derived from the three-dimensional elevation program (3DEP) (U.S. Geological Survey, 2019a) preprocessed within the National Hydrography Dataset (U.S. Geological Survey, 2019b). The morphometric model was derived using only the 3DEP dataset without any input of where landslides have occurred. All model outputs are shown with slope units. This data release includes the following files: 1) logistic regression results with 1:1 sampling ratio over Willamette Valley HUC4 watershed (1709) (Logistic_1709_1.zip); 2) logistic regression results with 1:33 sampling ratio over Willamette Valley HUC4 watershed (1709) (Logistic_1709_All.zip); 3) morphometric results with uniform weights over the Willamette Valley HUC4 watershed (1709) (Morph_Uniform_1709.zip); 4) morphometric results with area weights over the 1701 HUC 4 watershed (Morph_Area_1701.zip); 5) morphometric results with area weights over the 1702 HUC 4 watershed (Morph_Area_1702.zip); 6) morphometric results with area weights over the 1703 HUC 4 watershed (Morph_Area_1703.zip); 7) morphometric results with area weights over the 1704 HUC 4 watershed (Morph_Area_1704.zip); 8) morphometric results with area weights over the 1705 HUC 4 watershed (Morph_Area_1705.zip); 9) morphometric results with area weights over the 1706 HUC 4 watershed (Morph_Area_1706.zip); 10) morphometric results with area weights over the 1707 HUC 4 watershed (Morph_Area_1707.zip); 11) morphometric results with area weights over the 1708 HUC 4 watershed (Morph_Area_1708.zip); 12) morphometric results with area weights over the 1709 HUC 4 watershed (Morph_Area_1709.zip); 13) morphometric results with area weights over the 1710 HUC 4 watershed (Morph_Area_1710.zip); 14) morphometric results with area weights over the 1711 HUC 4 watershed (Morph_Area_1711.zip); 15) morphometric results with area weights over the 1712 HUC 4 watershed (Morph_Area_1712.zip). 16) shape file field descriptors (Field_Descriptors.txt) Each zip-file contains the vector shapefiles of interest which can be extracted using most archiver software. References Cited DOGAMI. (2024). SLIDO (Version 4.5) [Data set]. https://pubs.oregon.gov/dogami/SLIDO/4.5/SLIDO_Release_4p5_wMetadata.gdb.zip. Gina M Belair, Jeanne M Jones, Sabrina N Martinez, Benjamin B Mirus, & Nathan J Wood. (2024). Slope-Relief Threshold Landslide Susceptibility Models for the United States and Puerto Rico [Data Release]. U.S. Geological Survey.

Revised dataset of landslides triggered by the 2011 M9.1 Tohoku earthquake. Point data are presented as a csv with new failure types, source type, and impacts to the build environment classification for mapped landslides. This dataset is an extension of the Wartman et al. (2013) landslide inventory following the Tohoku earthquake, and includes the original landslide IDs and type classification of the original authors. Associated landslide polygons and original mapping details are provided at https://doi.org/10.5066/F7H70DB4

Two active landslides at and near the retreating front of Barry Glacier at the head of Barry Arm Fjord in southern Alaska could generate tsunamis if they failed rapidly and entered the water of the fjord. Landslide A, at the front of the glacier, is the largest, with a total volume estimated at 455 M m3. Historical photographs from Barry Arm indicate that Landslide A initiated in the mid twentieth century, but there was a large pulse of movement between 2010 and 2017 when Barry Glacier thinned and retreated from about 1/2 of the toe of Landslide A. Interferometric synthetic aperture radar (InSAR) investigations of the area between May and November, 2020, revealed a second, smaller landslide (referred to as Landslide B) on the south-facing slope about 2 km up the glacier from Landslide A. Landslide-generated tsunami modeling in 2020 used a worst-case scenario where the entire mass of Landslide A (about 455 M m3) would rapidly enter the water. The use of multiple landslide volume scenarios in future tsunami modeling efforts would be beneficial in evaluating tsunami risk to communities in the Prince William Sound region. Herein, we present a map of landslide structures and kinematic elements within, and adjacent to, Landslides A and B. This map could form at least a partial basis for discriminating multiple volume scenarios (for example, a separate scenario for each kinematic element). We mapped landslide structures and kinematic elements at scale of 1:1000 using high-resolution lidar data acquired by the Alaska Division of Geological and Geophysical Surveys (DGGS) on June 26, 2020 and high resolution bathymetric data acquired by the National Oceanic and Atmospheric Administration (NOAA) in August, 2020. The predominate structures in both landslides are uphill- and downhill-facing normal fault scarps. Uphill-facing scarps dominate in areas where downslope extension from sliding has been relatively low. Downhill-facing scarps dominate in areas where downlslope extension from sliding has been relatively high. Strike-slip and oblique-slip faults form the boundaries of major kinematic elements. Four major kinematic elements, herein named the Kite, the Prow, the Core, and the Tail, are within, or adjacent to Landslide A. One major kinematic element, herein named the Wedge, forms Landslide B. Kinematic element boundaries are a result of cumulative, differential patterns and amounts of movement that began at inception of the landslides. Elements and/or their boundaries may change location as the landslides continue to evolve. Kinematic elements mapped in 2020 may or may not reflect patterns of historical short-term, episodic movement, or patterns of movement in the future. We were not able to field check our mapping in 2020 because of travel restrictions due to the COVID-19 pandemic. We hope to field check the mapping in the summer of 2021. In this data release, we include GIS files for the structural and kinematic map; metadata files for mapped structural features; and portable document files (PDFs) of a location map, and the structural and kinematic map at a scale of 1:5000. Lidar and bathymetric data used to map landslide structures will be released by DGGS and NOAA in 2021.

This data release contains (1) geotechnical reports describing colluvium strength and grain size distribution, (2) hydrological monitoring data (rainfall and soil volumetric water content), and (3) shapefiles of mapped landslides from 1996 and 2021 that occurred in the Columbia River Gorge, Oregon. The geotechnical reports describe test results from a sieve and hydrometer analysis (ASTM D422) to characterize the grain size distribution and from consolidated drained direct shear tests (ASTM D3080) to characterize soil shear strength. Hydrological data includes a time history of rainfall and volumetric water content from a monitoring station in the Columbia River Gorge, spanning 10/28/2022 to 2/13/2023. The mapped landslide shapefiles represent shallow landslide source areas, assumed to have failed during storms in 1996 and 2021.

We performed hourly monitoring of conditions at the Two Towers landslide located in northern California near the town of Zenia. Monitored conditions included rainfall, groundwater head, horizontal total stress, horizontal effective stress, vertical soil deformation, and subsurface displacement. Data were acquired November 11, 2014–July 22, 2017, except for times during which power failure occurred; data for these times are given as “NAN” (not a number). Rainfall data are provided in millimeters during the past hour (mm/hr). Groundwater heads are provided in meters (m) above the landslide base. Horizontal stresses are provided in kilopascals (kPa). Vertical soil deformation data are provided in terms of length (centimeters, cm) of the sensor. Cumulative landslide displacement is provided in millimeters (mm). Rainfall was measured at the landslide middle monitoring location using a tipping-bucket rain gauge with resolution of 0.254 mm and accuracy of ±2% to 250 mm/hr (resolutions and accuracies stated herein are as specified by sensor manufacturers and accounting for datalogger resolution). A vibrating-wire total-stress plate sensor was installed with near-vertical orientation in the floor of an excavated pit at the middle monitoring location. This sensor measured total horizontal stress applied to its 230-mm-diameter surface with resolution of 0.014 kPa and accuracy of ±0.069 kPa. The sensor was installed within a slot slightly wider than the plate itself with its center at a depth of 1.83 m, and a vibrating-wire fluid pressure transducer with the same resolution and accuracy as the total stress sensor was installed adjacent to the cell to measure fluid pressure and therefore provide a means of calculating horizontal effective stress. The pit was backfilled after sensor installation with material removed during its excavation. The remaining sensors were installed within 6.35-cm-diameter holes bored using hand equipment. These included electronic, vibrating-wire fluid pressure transducers (piezometers) with resolutions of 0.014 kPa and 0.086 kPa, and respective accuracies of ±0.069 kPa and ±0.344 kPa. Boreholes were backfilled above transducers first with ~0.3 m of material obtained during boring followed by bentonite granules to the ground surface. Piezometers were installed at depths of 3.66 m and 6.07 m at the upper monitoring location, 3.95 m and 5.69 m at the middle monitoring location, and 2.62 and 3.66 m at the lower monitoring location. Landslide basal depths were identified at approximately 6.3 m, 7.9 m, and 3.6 m at the upper, middle, and lower monitoring locations, respectively. A 30.48-cm-long biaxial tilt sensor installed within PVC casing (slope inclinometer) was used to monitor landslide displacement at the lower monitoring location. The slope inclinometer has 0.003 mm displacement resolution and long-term displacement accuracy of ±0.23 mm. A vibrating-wire length sensor was installed in a borehole to measure near-surface vertical deformation at the middle monitoring location. This sensor measured length with 0.0375 mm resolution and ±0.15 mm accuracy. The sensor’s upper and lower ends were anchored within cement grout such that its length was measured over a depth range (at installation) of 0.20-1.72 m. All sensors contain thermistors and readings are temperature compensated, with the exception of the rain gauge. These data support a study described in Schulz, W.H., Smith, J.B., Wang, G., Jiang, Y., and Roering, J.J., 2018, Clayey landslide initiation and acceleration strongly modulated by soil swelling: Geophysical Research Letters, DOI:10.1002/2017GL076807.

We performed laboratory time-dependent creep and stress-oscillation earthquake simulations in uniaxial loading in compression and tension on four blocks of porphyritic dacite collected from pyroclastic deposits of Mt. Fugen, of the Unzen-dake volcanic complex in Japan (referred to as ‘Unzen volcano’). Experiments were carried out in the Experimental Volcanology and Geothermal Research Laboratory at the University of Liverpool. Loading experiments were carried out in a 100 kN Instron 8800 uniaxial press at room temperature using both compression tests, and Brazilian disk tests to provide the indirect tensile strength (hereafter referred to as tensile strength/tension tests). Load was recorded by an Instron Dynacell 2527 load cell at 100 Hz, which has an accuracy of ±0.1% of the full load capacity (100 kN). Strain is recorded using an Instron LVDT (Linear Variable Differential Transformer) Deflection Sensor, which has an accuracy of ±0.00001 mm or ±0.05% of the measured displacement, whichever is the largest. For the compression tests, cylindrical specimens with a 20-mm diameter were cored and then cut to a length of 40 mm to form cylinders with a 2:1 length:diameter ratio. The cylinders were axially compressed between two parallel plates in the loading frame during testing. For the Brazilian tests, cylindrical specimens with a 40-mm diameter were cored and then cut to a length of 20 mm to form disks with a 2:1 diameter:length ratio. The disks were diametrically compressed between two parallel plates in the loading frame to induce tensional stresses in the orthogonal direction. The specimen dimensions and loading frame setup follow American Society for Testing and Material (ASTM) standards for unconfined (uniaxial) compression and Brazil tensile strength testing (ASTM Standard D7012 (ASTM, 2014) and ASTM Standard D3967 (ASTM, 2016), respectively). Following preparation, specimens were oven dried at 60°C for 24 hours and then placed into a vacuum chamber for one week prior to mechanical experiments.