Geology Reports⌕ Search

SEARCH · Geology Reports

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 199 records · Page 11Linked to original sources

Landslides caused by the M 7.6 Tecomán, Mexico earthquake of January 21, 2003

The Tecomán, Mexico earthquake (also called the “Colima earthquake”) of January 21, 2003 (M 7.6) triggered several hundreds of landslides in the coastal cordilleras of Colima State, near the earthquake source, and several thousands in the volcanic highlands north and northwest of Colima City. These landslides, mostly shallow and disrupted failures, caused minor damage to roads, to a railroad, and to irrigation systems. In one area, extensive, post-earthquake rock-fall activity indicates a possible long-term instability that could threaten dwellings and other infrastructure located nearby. In the coastal cordilleras, most of the landslides were generated by failures of artificially cut slopes, especially along roads. The rocks of the coastal cordilleras are generally well indurated, and landslides occurred only where the rocks were made locally susceptible by weathering, the presence of prominent discontinuities with unfavorable orientations, or intense fracturing or shearing. In contrast to the coastal cordilleras, the volcanic rocks to the north were more susceptible to the occurrence of seismically triggered landslides. The greatest number and concentrations of landslides occurred there, and the landslides were larger than those in the coastal cordilleras, even though this volcanic terrain was farther from the earthquake source. Here, stretches of river bluffs several hundred meters long had been stripped of vegetation and surficial material by coalescing landslides, and several days after the main shock, thousands of small rock falls were still occurring each day, indicating an ongoing hazard. The high susceptibility of volcanic materials to earthquake-generated landslides conforms to findings in other recent earthquakes.

Engineering Geology↗

Study of a prehistoric landslide using seismic reflection methods integrated with geological data in the Wasatch Mountains, Utah, USA

An integration of geological and geophysical techniques characterizes the internal and basal structure of a landslide along the western margin of the Wasatch Mountains in northern Utah, USA. The study area is within a region of planned and continuing residential development. The Little Valley Landslide is a prehistoric landslide as old as 13??ka B.P. Drilling and trenching at the site indicate that the landslide consists of chaotic and disturbed weathered volcanic material derived from Tertiary age volcanic rocks that comprise a great portion of the Wasatch Range. Five short high-resolution common mid-point seismic reflection profiles over selected portions of the site examine the feasibility of using seismic reflection to study prehistoric landslides in the Wasatch Mountain region. Due to the expected complexity of the near-surface geology, we have pursued an experimental approach in the data processing, examining the effects of muting first arrivals, frequency filtering, model-based static corrections, and seismic migration. The results provide a framework for understanding the overall configuration of the landslide, its basal (failure) surface, and the structure immediately underlying this surface. A glide surface or de??collement is interpreted to underlie the landslide suggesting a large mass movement. The interpretation of a glide surface is based on the onset of coherent reflectivity, calibrated by information from a borehole located along one of the seismic profiles. The glide surface is deepest in the center portion of the landslide and shallows up slope, suggesting a trough-like feature. This study shows that seismic reflection techniques can be successfully used in complex alpine landslide regions to (1) provide a framework in which to link geological data and (2) reduce the need for an extensive trenching and drilling program. ?? 2007 Elsevier B.V. All rights reserved.

Engineering Geology↗

Geology and tsunamigenic potential of submarine landslides in Santa Barbara Channel, Southern California

A large submarine landslide complex and four small landslides developed under the Santa Barbara Channel, suggesting a potential hazard from landslide-generated tsunamis. We integrate offshore stratigraphy and geologic structure, multibeam bathymetric information, and several kinds of seismic-reflection data to understand how and when the submarine landslides formed. Seismic-reflection data show that mass failure along the slope began at least 200 ka ago. Landslides appear as zones of poor reflectivity having an irregular upper surface, and these zones alternate vertically with strong parallel reflections. The emplacement ages of two of the three main landslide lobes are well established at 8 and 10 ka. The source material for the youngest part of the landslide complex was sediment of probable late Pleistocene and Holocene age that accumulated in a shelf-edge delta. Directly under this delta, growth of faults and anticlines was particularly intense and tended to oversteepen the deltaic deposits. These active structures also formed migration pathways and reservoirs for aqueous and hydrocarbon fluids from the deep basin. Tsunami deposits have been described from a low-lying area near Santa Barbara, and numerical modeling of tsunamis generated by hypothetical landslides in Santa Barbara Channel indicates a moderate to severe threat [Borrero, J.C., Dolan, J.F. and Synolakis, C.E., 2001. Tsunamis within the eastern Santa Barbara Channel. Geophys. Res. Lett., 28(4): 643-646.], involving wave runups of 2-20 m, for a range of assumed landslide volumes. Inundation from these waves, however, is expected to be highly focused so that only narrow (???10-km) sections of the shoreline would be affected. Crown Copyright ?? 2005 Published by Elsevier B.V. All rights reserved.

Marine Geology↗

The frequency and distribution of recent landslides in three montane tropical regions of Puerto Rico

Landslides are common in sttep mountainous areas of Puerto Rico where mean annual rainfall and the frequency of intense storms are high. Each year, landslides cause extensive damage to property and coccasionally result in loss of life. Average population density is high, 422 people/km2, and is increasing. This increase in population density is accompanied by growing stress on the natural environment and physical infrastructure. As a result, human populations are more vulnerable to landslide hazards. The Blanco, Cibuco, and Coamo study areas range in surface area from 276 to 350 km2 and represent the climatologic, geographic, and geologic conditions that typify Puerto Rico. Maps of recent landslides developed from 1:20 000-scale aerial photographs, in combination with a computerized geographic information system, were used to evaluate the frequency and distribution of shallow landslides in these areas. Several types of landslides were documented-rainfall-triggered debris flows, shallow soil slips, and slumps were most abundant. Hillslopes in the study area that have been anthropogenically modified, exceed 12?? in gradient, and greater than 300 m in elevation, and face the east-northeast, are most prone to landsliding. A set of simplified matrices representing geographic conditions in the three study areas was developed and provides a basis for the estimation of the spatial controls on the frequency of landslides in Puerto Rico. this approach is an example of an analysis of the frequency of landslides that is computationally simple,. and therefore, may be easily transferable to other settings.

Geomorphology↗

SlideDetect: Spatio-temporal landslide detection using a three-dimensional convolutional neural network

Landslides pose a serious and ongoing threat to both human lives and infrastructure worldwide; therefore, it is of interest to predict where and when landslides are likely to occur. Advances in machine learning techniques have spurred numerous studies aimed at estimating relative landslide propensity, but are limited to spatial (as opposed to temporal) prediction due to the sparsity of landslide timing data. We address this data gap by training SlideDetect, a 3-dimensional convolutional neural network (3D CNN), to identify landslides based on their spatial and temporal occurrence within multitemporal image stacks. We use an inventory of landsides triggered by the 2018 Hokkaido earthquake and two years of monthly composite optical imagery spanning this event. The model can identify not only landslide location but also landslide date with an area under the precision-recall curve (PR-AUC) of 0.84. We further present a new standard for presenting PR curve results that explicitly compares model performance at different confidence thresholds, allowing for clearer model evaluation and comparison. Our new approach to constraining landslide timing paired with this more consistent and objective method for evaluating model performance shows considerable promise, and with further application and testing, SlideDetect could enhance the data availability and tools needed to advance landslide hazard and risk assessments.

JGR Machine Learning and Computation↗

Unsteady, nonuniform landslide motion: 1. Theoretical dynamics and the steady datum state

Unsteady, nonuniform motion of persistently active landslides is a process of widespread importance. A general, three-dimensional theory aimed at elucidating this process is developed from physical principles and field measurements of landslide behavior. The theory employs a versatile constitutive model that represents landslides as deformable bodies composed of frictional, nonlinear viscoplastic material. The three-dimensional theory is reduced to a mathematically tracTable form by defining an ideal landslide datum state that consists of steady, unidirectional shear flow driven by ground-water seepage and gravitational forces. Solution of the datum-state equation of motion yields vertical landslide velocity profiles that can represent deformation styles ranging from shear-thickening viscoplastic flow to perfectly plastic frictional slip. This range of theoretical profiles encompasses the range of profiles measured in four persistently active northern California landslides. Also obtained from the datum-state equation of motion is an analytical solution for datum-state landslide sediment fluxes. An important feature of this solution is that it represents theoretical landslide sediment fluxes as a family of continuous-functions. The solution thus provides a mathematical basis for a general perturbation analysis of the kinematics of unsteady, nonuniform landslide motion, which will be presented in a companion paper in a subsequent issue of the Journal.

Journal of Geology↗

Searching for seismic precursors - The Barry Landslide hazard clean up

The Barry Landslide, located in Barry Arm of Prince William Sound, Alaska, poses a major hazard due to its steep, unstable slopes and the potential for a massive landslide-generated tsunami. With an estimated volume of 500–700 million cubic meters, the Barry Landslide could trigger highly destructive waves. In this study, we focus on seismic signals from the Barry Landslide, which are critical for providing timely tsunami warnings. Since the summer of 2020, the region has been instrumented to monitor the landslide, but the seismic record is complicated by the presence of nearby glaciers and frequent regional earthquakes. Among these signals, we analyze a specific class of short-duration, high-frequency seismic events that exhibit strong seasonal variability, increasing in rate from late summer to mid-winter before ceasing abruptly in late winter or early spring. Our analysis suggests that the source of these signals is likely near or beneath Cascade Glacier, adjacent to the landslide, rather than within the landslide mass itself. We apply detection algorithms to construct a time history for this signal type, which we then compare with environmental factors like precipitation, temperature, and slope displacement data from ground-based radar and remote sensing. Correlations indicate that these seismic events may be driven by seasonal hydrological changes, particularly the freeze-up of subglacial water pathways. While these events are not directly linked to landslide motion, they serve as indirect markers of subsurface hydrological conditions that influence slope stability. Our findings highlight the complex interplay between glaciers, groundwater, and landslide dynamics, emphasizing the need for multi-parameter monitoring to assess evolving geohazards in the region.

Alaska↗

The Springdale, Utah, landslide: An extraordinary event

The most dramatic geologic effect of the M-5.7 St. George, Utah earthquake of 2 September 1992 was the triggering of the 14,000,000-m3 Springdale, Utah landslide. The roughly 10 m of landslide movement destroyed three houses, threatened several condominiums, disrupted utility lines, and temporarily closed the southwest entrance to Zion National Park. The seismic triggering of this landslide is puzzling because its distance from the earthquake epicenter, 44 km, is much greater than the farthest distance (18 km) at which similar landslides have been triggered in worldwide earthquakes of the same magnitude. Other Colorado Plateau earthquakes also have produced landslides far beyond worldwide distance limits, which suggests that regional variations in ground-shaking attenuation may require different landslide-triggering distance limits for different seismotectonic regions. Slope stability analysis and historical records of landslide movement suggest that the Springdale landslide was only slightly above limit-equilibrium conditions at the time of the earthquake. Dynamic stability analysis using Newmark's permanent-displacement method indicates coseismic landslide displacement of only 1-8 cm; this rather modest displacement probably induced enough deformation in the montmorillonitic clays along the failure surface to reduce shear strength and destabilize the slide, which continued to move for several hours after the earthquake.

Utah↗

The 3D Elevation Program—Landslide recognition, hazard assessment, and mitigation support

The U.S. Geological Survey (USGS) Landslide Hazards Program conducts landslide hazard assessments, pursues landslide investigations and forecasts, provides technical assistance to respond to landslide emergencies, and engages in outreach. All of these activities benefit from the availability of high-resolution, three-dimensional (3D) elevation information in the form of light detection and ranging (lidar) data and interferometric synthetic aperture radar (IfSAR) data. Research on landslide processes addresses critical questions of where and when landslides are likely to occur as well as their size, speed, and effects. This understanding informs the development of methods and tools for hazard assessment and situational awareness used to guide efforts to avoid or mitigate landslide impacts. Such research is essential for the USGS to provide improved information on landslide potential associated with severe storms, earthquakes, volcanic activity, coastal wave erosion, and wildfire burn areas. Decisionmakers in government and the private sector increasingly depend on information the USGS provides before, during, and following disasters so that communities can live, work, travel, and build safely. The USGS 3D Elevation Program (3DEP) provides the programmatic infrastructure to generate and supply lidar-derived superior terrain data to address landslide applications and a wide range of other urgent needs nationwide. By providing data to users, 3DEP reduces users’ costs and risks and allows them to concentrate on their mission objectives. 3DEP includes (1) data acquisition partnerships that leverage funding, (2) contracts with experienced private mapping firms, (3) technical expertise, lidar data standards, and specifications, and (4) most important, public access to high-quality 3D elevation data.

Fact Sheet↗

Landslide monitoring in the Atlantic Highlands area, New Jersey

Shallow and deep-seated landslides have occurred episodically on the steep coastal bluffs of the Atlantic Highlands area (Boroughs of Atlantic Highlands and Highlands) in New Jersey. The oldest documented deep-seated landslide occurred in April 1782 and significantly changed the morphology of the bluff. However, recent landslides have been mostly shallow in nature and have occurred during large storms with exceptionally heavy rainfall. These shallow landslides have resulted in considerable damage to residential property and local infrastructure and threatened human safety. The recent shallow landslides in the area (locations modified from New Jersey Department of Environmental Protection) consist primarily of slumps and flows of earth and debris within areas of historical landslides or on slopes modified by human activities. Such landslides are typically triggered by increases in shallow soil moisture and pore-water pressure caused by sustained and intense rainfall associated with spring nor’easters and late summer–fall tropical cyclones. However, the critical relation between rainfall, soil-moisture conditions, and landslide movement has not been fully defined. The U.S. Geological Survey is currently monitoring hillslopes within the Atlantic Highlands area to better understand the hydrologic and meteorological conditions associated with shallow landslide initiation.

New Jersey↗

Precipitation thresholds for landslide occurrence near Seattle, Mukilteo, and Everett, Washington

Shallow landslides along coastal bluffs frequently occur in the railway corridor between Seattle and Everett, Washington. These slides disrupt passenger rail service, both because of required track maintenance and because the railroad owner, Burlington Northern Santa Fe Railway, does not allow passenger travel for 48 hours after a disruptive landslide. Sound Transit, which operates commuter trains in the corridor, is interested in a decision-making tool to help preemptively cancel passenger railway service in dangerous conditions and reallocate resources to alternative transportation. Statistical analysis showed that a majority of landslides along the Seattle-Everett Corridor are strongly correlated with antecedent rainfall, but that 21-37 percent of recorded landslide dates experienced less than 1 inch of precipitation in the 3 days preceding the landslide and less than 4 inches of rain in the 15 days prior to the preceding 3 days. We developed two empirical thresholds to identify precipitation conditions correlated with landslide occurrence. The two thresholds are defined as P 3 = 2.16-0.44P 15 and P 3 = 2.16-0.22P 32 , where P 3 is the cumulative precipitation in the 3 days prior to the considered date and P 15 or P 32 is the cumulative precipitation in the 15 days or 32 days prior to P 3 (all measurements given in inches). The two thresholds, when compared to a previously developed threshold, quantitatively improve the prediction rate. We also investigated rainfall intensity-duration (ID) thresholds to determine whether revision would improve identification of moderate-intensity, landslide-producing storms. New, optimized ID thresholds evaluate rainstorms lasting at least 12 hours and identify landslide-inducing storms that were typically missed by previously published ID thresholds. The main advantage of the ID thresholds appears when they are combined with recent-antecedent thresholds because rainfall conditions that exceed both threshold types are more likely to induce two or more landslides than conditions that exceed only one threshold type.

Washington↗

Landslides triggered by the August 14, 2021, magnitude 7.2 Nippes, Haiti, earthquake

The August 14, 2021, magnitude 7.2 Nippes, Haiti, earthquake triggered thousands of landslides on the Tiburon Peninsula. The landslides directly caused fatalities and damage and impeded response efforts by blocking roads and causing other infrastructure damage. Adverse effects of the landslides likely will continue for months to years. This report presents an assessment of potential postearthquake landslide-related geologic hazards for the Tiburon Peninsula and a preliminary map of the landslides triggered by the earthquake. This hazard assessment is based on an emergency analysis of the currently available, postearthquake satellite imagery. In this report, we highlight specific areas of concern that may benefit from more detailed assessment and longer-term monitoring. Our mapping efforts revealed that at least 4,893 landslides were triggered across the Tiburon Peninsula by the earthquake and subsequent rainfall from Tropical Cyclone Grace. We also observed hundreds of landslide deposits potentially restricting flow in rivers and streams. In addition, we observed landslides that likely affected roads by rendering them impassable or susceptible to subsequent damage from existing landslides. Because of the preliminary nature of this report and the limits of remote analyses, additional investigation and monitoring would be beneficial to accurately determine the threat posed by these hazards to people and infrastructure.

Nippes↗

Satellite interferometry landslide detection and preliminary tsunamigenic plausibility assessment in Prince William Sound, southcentral Alaska

Regional mapping of actively deforming landslides, including measurements of landslide velocity, is integral for hazard assessments in paraglacial environments. These inventories are also critical for describing the potential impacts that the warming effects of climate change have on slope instability in mountainous and cryospheric terrain. The objective of this study is to identify slow-moving landslides in the Prince William Sound region, southcentral Alaska, United States, which has had rapid deglaciation since the mid-1800s, and assess their tsunamigenic plausibility. We use an automated time series persistent scatterer interferometric synthetic aperture radar processing method with 7 years of Sentinel-1 data (2016–22) to identify 43 slow-moving slopes with average velocities ranging from approximately 0.2 to 21 millimeters per year. Landslide presence is confirmed using aerial imagery and previous landslide inventory records. We assess the tsunamigenic plausibility of the landslides using empirically derived estimates of landslide mobility based on modeled landslide volumes. Of the identified landslides, our preliminary analysis suggests that 11 have tsunamigenic potential if they were to fail rapidly and catastrophically. Although our estimate of tsunamigenic plausibility is preliminary and can be refined with additional observations and analyses, it can be used to prioritize ongoing and future hazard assessment, surveillance, and research efforts.

Alaska↗

The Loma Prieta, California, earthquake of October 17, 1989: Landslides

Central California, in the vicinity of San Francisco and Monterey Bays, has a history of fatal and damaging landslides, triggered by heavy rainfall, coastal and stream erosion, construction activity, and earthquakes. The great 1906 San Francisco earthquake ( M S =8.2-8.3) generated more than 10,000 landslides throughout an area of 32,000 km 2 ; these landslides killed at least 11 people and caused substantial damage to buildings, roads, railroads, and other civil works. Smaller numbers of landslides, which caused more localized damage, have also been reported from at least 20 other earthquakes that have occurred in the San Francisco Bay-Monterey Bay region since 1838. Conditions that make this region particularly susceptible to landslides include steep and rugged topography, weak rock and soil materials, seasonally heavy rainfall, and active seismicity. Given these conditions and history, it was no surprise that the 1989 Loma Prieta earthquake generated thousands of landslides throughout the region. Landslides caused one fatality and damaged at least 200 residences, numerous roads, and many other structures. Direct damage from landslides probably exceeded $30 million; additional, indirect economic losses were caused by long-term landslide blockage of two major highways and by delays in rebuilding brought about by concern over the potential long-term instability of some earthquake-damaged slopes.

California↗

Map of landslides triggered by the January 12, 2010, Haiti earthquake

The magnitude (M) 7.0 Haiti earthquake of January 12, 2010, triggered landslides throughout much of Haiti on the island of Hispaniola in the Caribbean Sea. The epicenter of the quake was located at 18.44°N., 72.57°W. at a depth of 13 kilometers (km) approximately 25 km southwest of the capital, Port-au-Prince. Although estimates vary widely, the most reliable surveys of casualties indicate that the earthquake caused 158,679 fatalities and more than 300,000 injuries. The U.S. Geological Survey compared publicly available satellite imagery acquired both before and after the earthquake and mapped 23,567 landslides that were triggered by the strong shaking. Our mapping from aerial photography and satellite imagery was augmented by field observations. Most of the landslides triggered by the earthquake were south of the Léogâne fault on the footwall and were fairly shallow falls and slides in weathered limestone (2–5 meters [m] thick) and volcanic rock and soil (generally <1 m thick). Landslides extended from the north to the south coasts of the southwestern peninsula (southwest of Port-au-Prince) and almost 60 km to the east and west of the epicenter. The highest concentration of landslides was on the steep limestone slopes of incised river valleys, but large numbers of landslides also occurred on gentler slopes in weathered volcanic rocks. Although some high landslide concentrations did occur near areas of maximum fault slip, the overall distribution of landslides appears to involve complex interactions between geology, topography, and strong shaking with limited spatial correlation between fault slip and landslides.

Scientific Investigations Map↗

Dendrogeomorphic Assessment of the Rattlesnake Gulf Landslide in the Tully Valley, Onondaga County, New York

Dendrogeomorphic techniques were used to assess soil movement within the Rattlesnake Gulf landslide in the Tully Valley of central New York during the last century. This landslide is a postglacial, slow-moving earth slide that covers 23 acres and consists primarily of rotated, laminated, glaciolacustrine silt and clay. Sixty-two increment cores were obtained from 30 hemlock (Tsuga canadensis) trees across the active part of the landslide and from 3 control sites to interpret the soil-displacement history. Annual growth rings were measured and reaction wood was identified to indicate years in which ring growth changed from concentric to eccentric, on the premise that soil movement triggered compensatory growth in displaced trees. These data provided a basis for an 'event index' to identify years of landslide activity over the 108 years of record represented by the oldest trees. Event-index values and total annual precipitation increased during this time, but years with sudden event-index increases did not necessarily correspond to years with above-average precipitation. Multiple-regression and residual-values analyses indicated a possible correlation between precipitation and movement within the landslide and a possible cyclic (decades-long) tree-ring response to displacement within the landslide area from the toe upward to, and possibly beyond, previously formed landslide features. The soil movement is triggered by a sequence of factors that include (1) periods of several months with below-average precipitation followed by persistent above-average precipitation, (2) the attendant increase in streamflow, which erodes the landslide toe and results in an upslope propagation of slumping, and (3) the harvesting of mature trees within this landslide during the last century and continuing to the present.

Scientific Investigations Report↗

Preliminary assessment of aggradation potential in the North Fork Stillaguamish River downstream of the State Route 530 landslide near Oso, Washington

On March 22, 2014, the State Route 530 Landslide near Oso, Washington, traveled almost 2 kilometers (km), destroyed more than 40 structures, and impounded the North Fork Stillaguamish River to a depth of 8 meters (m) and volume of 3.3&times;10 6 cubic meters (m 3 ). The landslide killed 43 people. After overtopping and establishing a new channel through the landslide, the river incised into the landslide deposit over the course of 10 weeks draining the impoundment lake and mobilizing an estimated 280,000&plusmn;56,000 m 3 of predominantly sand-sized and finer sediment. During the first 4 weeks after the landslide, this eroded sediment caused downstream riverbed aggradation of 1&ndash;2 m within 1 km of the landslide and 0.4 m aggradation at Whitman Road Bridge, 3.5 km downstream. Winter high flows in 2014&ndash;15 were anticipated to mobilize an additional 220,000&plusmn;44,000 m 3 of sediment, potentially causing additional aggradation and exacerbating flood risk downstream of the landslide. Analysis of unit stream power and bed-material transport capacity along 35 km of the river corridor indicated that most fine-grained sediment will transport out of the North Fork Stillaguamish River, although some localized additional aggradation was possible. This new aggradation was not likely to exceed 0.1 m except in reaches within a few kilometers downstream of the landslide, where additional aggradation of up to 0.5 m is possible. Alternative river response scenarios, including continued mass wasting from the landslide scarp, major channel migration or avulsion, or the formation of large downstream wood jams, although unlikely, could result in reaches of significant local aggradation or channel change.

Washington↗

Assessing locations susceptible to shallow landslide initiation during prolonged intense rainfall in the Lares, Utuado, and Naranjito municipalities of Puerto Rico

Hurricane Maria induced about 70 000 landslides throughout Puerto Rico, USA, including thousands each in three municipalities situated in Puerto Rico's rugged Cordillera Central range. By combining a nonlinear soil-depth model, presumed wettest-case pore pressures, and quasi-three-dimensional (3D) slope-stability analysis, we developed a landslide susceptibility map that has very good performance and continuous susceptibility zones having smooth, buffered boundaries. Our landslide susceptibility map enables assessment of potential ground-failure locations and their use as landslide sources in a companion assessment of inundation and debris-flow runout. The quasi-3D factor of safety, F 3 , showed strong inverse correlation to landslide density (high density at low F 3 ). Area under the curve (AUC) of true positive rate (TPR) versus false positive rate (FPR) indicated success of F 3 in identifying head-scarp points (AUC = 0.84) and source-area polygons (0.85 ≤ AUC ≤ 0.88). The susceptibility zones enclose specific percentages of observed landslides. Thus, zone boundaries use successive F 3 levels for increasing TPR of landslide head-scarp points, with zones bounded by F 3 at TPR = 0.75, very high; F 3 at TPR = 0.90, high; and the remainder moderate to low. The very high susceptibility zone, with 118 landslides km −2 , covered 23 % of the three municipalities. The high zone (51 landslides km −2 ) covered another 10 %.

Puerto Rice↗