Inventory of Large Slope Instabilities, Central and Eastern Prince William Sound, Alaska

Steep landscapes shaped by glaciers, particularly areas where ice has recently retreated, often experience gravitationally driven slope deformation. We refer to hillslopes that show evidence of such deformation as bedrock instabilities. These deep-seated instabilities may or may not develop into coherent landslides with through-going basal sliding surfaces. In coastal environments, failure of these bedrock instabilities as rapid landslides can generate tsunami waves that pose a hazard for communities, infrastructure, mariners, and important natural and cultural resources. In our study area, previous work used persistent scatterer interferometric synthetic aperture radar (PSInSAR) to identify undocumented instabilities moving at rates of approximately 0.2 to 21 millimeters per year between 2016 and 2022 (Schaefer and others, 2024). Satellite optical imagery and digital elevation models can be used to identify instabilities that have moved in the past but may not be actively deforming or may have been missed by the PSInSAR analysis due to suboptimal conditions (for example, poor lighting or satellite imaging geometry) or movement rates outside of the PSInSAR detection range. Here, we present an inventory of manually mapped bedrock instabilities in central and eastern Prince William Sound (cePWS), which expands the footprint of an existing inventory completed in 2023 (Higman and others, 2023). Slope instabilities included in this inventory are defined as large areas (> 0.01 km2) that exhibit evidence of slope deformation, including scarps, tension cracks, and signs of recent smaller-scale landslides. Deposits from past catastrophic rock avalanche failures are also included. All instabilities in this inventory were identified from a combination of field observations, satellite and airborne optical imagery, interferometric synthetic aperture radar (InSAR)-derived elevation models, and limited local light detection and ranging (lidar) elevation models. We leveraged multiple imagery sources rather than relying on a single source to overcome the inherent challenges in identifying and classifying topographic features in the rugged and complex coastal Alaskan landscape and to overcome limitations in a single data set, such as cloud cover in optical imagery or artifacts in a digital elevation model. We applied the method systematically to the entire study area, but due to limitations of the data and our identification methods (discussed in Methods Section), we do not consider this inventory to be exhaustive or complete, nor should it be considered a statistically representative sample of instabilities in this region. The data presented here represent bedrock instabilities in eastern Prince William Sound identified prior to May 2025.
 
This data release includes: (1) an inventory of instabilities in two forms (cepws_instabilities.shp, cepws_instabilities.csv), (2) a study area shapefile (cepws_studyarea.shp), (3) a document summarizing the data and methods (cepws_inventory_summary.pdf), and a (4) readme (readme.txt) with information on coordinate systems and folder contents.
 
Disclaimer: Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
Data Fields:

OBJECTID (fid): Software assigned unique identifier of each polygon
unique_id: Unique identifier of each polygon
Shape: Geometry data type
Name: Instability name
morph_type: Instability description based on morphological attributes
ls_features: Morphological attributes used to classify instabilities
ext_reason: Reasoning behind each extent delineation
conf_delin: Confidence in polygon delineation
conf_deform: Confidence deformation has occurred
hist_deform: Evidence of historical deformation
description: General description of the instability
notes: Unformatted notes field, includes additional information
image_link: Link to image source
prior_work: Reference to previous work
noted_by: Name of person(s) who identified the site
utm_y: Latitude of instability centroid (NAD 1983 UTM Zone 6N)
utm_x: Longitude of instability centroid (NAD 1983 UTM Zone 6N)
Shape_Length: Total length of polygon’s perimeter (m)
Shape_Area: Total area of polygon (m2)

Data Field Attributes:
morph_type


Inconclusive topographic signs of instability: Zone of potential deformation encompassing topographic evidence suggestive of gravitational deformation.
Instability with ambiguous boundaries: Zone of deformation encompassing topographic evidence strongly suggestive of gravitational deformation, with diffuse boundaries and no obvious headscarp or lateral margins.
Instability with potential headscarp: Zone of deformation with a downhill facing escarpment along parts or all of the upslope boundary that may be discontinuous or topographically diffuse.
Instability with developed headscarp: Zone of deformation features bounded on the upslope boundary by a sharp, continuous, downhill facing escarpment.
Repeat rock avalanche source: Slope area that appears to be the upslope initiation site of numerous rock avalanches.
Deposit: Landform composed of a mass of bedrock that has moved downslope from an original source area via gravity, often highly fractured or disintegrated. These include catastrophic rock avalanche deposits, and may represent the final evolution of some bedrock instabilities that progress to rapid failure.

hist_deform


No identified evidence of active deformation: Deformation features present, but topographically diffuse and/or overprinted with vegetation, and lacking any cross-cutting relationships between landforms and deformation features indicative of active deformation (such as scarps displacing lateral moraine deposits near present-day glacial termini).
Uncertain evidence of active deformation: Deformation features present, but relative activity cannot be determined based on appearance of these features alone, and no obvious cross cutting relationships are present between landforms and deformation features.
Suggestive evidence of active deformation: Deformation features present that appear topographically sharp, potentially including cross cutting relationships between landforms and deformation features indicative of active deformation.
Clear evidence of active deformation: Movement has been measured in recent history using methods such as InSAR, radar, repeat lidar, repeat imagery, etc. 

ls_features


Antiscarp: Uphill-facing scarp
Bedrock (downdropped) block: Intact block within deformation zone with diameter of a few meters or more, notably coarser than talus
Bench (bulging): Relatively narrow, lower gradient swath of terrain bounded by steeper slopes above and below
Headscarp: Well-defined downhill-facing scarp that bounds most of the evidence of deformation at its uphill extent
Hummocky terrain: Irregular, rough topography signifying past deformation or landslide runout
Lineation: Undifferentiated linear feature thought to be related to deformation, may include scarps, tension cracks, or shear zones
Lobate toe: Curved or rounded plan-view form at base of instability or landslide deposit
Normal scarp: Downhill-facing scarp within the mapped boundary of the instability, but not bounding the upslope extent like a headscarp
PSInSAR movement: Movement detected by persistent scatterer interferometric synthetic aperture radar (PSInSAR). Detection rates range from approximately 0.2 to 21 millimeters per year.
Rock avalanche: Type of landslide characterized by catastrophic collapse and rapid displacement of bedrock, may include interstitial ice or permafrost
Sackung: Linear graben-like trough or sequence of troughs formed near and parallel to ridgeline through gravitational deformation
Scarps: Steep surface produced by differential movements within displaced material, unknown or ambiguous scarp orientation
Shear zone: Zone of strongly deformed material produced by differential movement of material, potentially including oblique slip scarps with en echelon fractures or jagged, roughened ground at lateral margins of deformation zone
Sinkhole: Depression in the ground that has no natural external surface drainage
Source area: Area at or above head of instability exposed by downslope movement of material
Springs: Groundwater emerging onto the surface
Talus source area: Portion of slope which produce small failures of bedrock that may be deposited to form steep fans or cones
Tension cracks: Open subvertical bedrock crack

conf_deform


High: There is direct and compelling evidence of the instability (e.g., field observations, presence of headscarp, definitive deformation features observed in high-quality imagery like lidar)
Medium: There is some evidence to suggest the instability exists
Low: There is some ambiguous evidence to suggest bedrock deformation, but deformation features are relatively small, few in number, or visually ambiguous (potentially due to data quality and completeness) compared to medium and high confidence classes, which introduces uncertainty as to whether the instability exists

conf_delin


High: Can draw exact outline of instability because boundaries of instability are clearly visible by way of lateral shear zones, topographic inflections, and/or sharp contacts with undeformed ground
Medium: Can draw general shape of instability within tens of meters
Low: Can define a broad boundary that encloses deformation features but extent of deformation is ambiguous 

Methods:
In most cases the identification of bedrock instabilities was dependent on interpreting topographic features characteristic of bedrock landslides (ls_features) visible in imagery, remote sensing data, or digital elevation datasets. A list of landslide features and their definitions used to identify potential instabilities is described above. Lineations associated with slope deformation include antiscarps, normal scarps, tension cracks, and sackungen. Importantly, these deformation-related lineations need to be distinguished from other linear features, such as glacially plucked bedrock lineations, bedding, metamorphic foliations, intrusive dikes and sills, glacial moraines, glacial striations, and tectonic fault scarps. Other features that could indicate deformation include downdropped bedrock blocks, benches, and shear zones. As with lineations, expert judgment is needed to determine whether these features are due to deep-seated deformation or caused by other processes. For example, downdropped blocks could be related to shallow slope failures, benches could be caused by differential erosion of bedding, and shear zones could be due to tectonic movement. Some of the landslide features we identify may not indicate the presence of a slope instability on their own, but in the context of other features may add confidence to our interpretation of gravitational deformation. Ultimately, linking these features to slope deformation is a qualitative process of data interpretation.

We manually delineated a perimeter polygon for each identified instability. In most cases, the polygon represents the “plausible extent” of the site boundary. Where possible, upper edges were defined along the uppermost normal scarps (headscarp), lateral edges were defined along shear zones, and lower edges were assumed to be within some area of steep slope. In areas without clear boundary indicators, polygons were drawn to contain all the identified related features. Where the instabilities intersect glacial ice or waterbodies, we delineated along this contact, such that we only map the subaerial portions visible in the imagery. We recognize the true geometry of these instabilities may extend beyond our mapped polygon in the downslope direction. Moreover, in some regions with broadly distributed deformation features and no clear lateral margins, it was difficult to determine whether to map multiple separate instability polygons or a single instability that encompassed all identified features. Similarly, it was not always possible to determine which features we observe at the surface coalesce in the subsurface. These challenges mean that the mapped geometry of features assigned low  ‘conf_delin’ should be interpreted cautiously, and a high degree of uncertainty should be assumed.

This inventory consists of instabilities that are visible in remotely sensed data sources and available digital elevation data. Poor lighting conditions, less than ideal satellite geometry, snow cover, coarse data resolution, and other data limitations preclude the exhaustive compilation of all instabilities in the region. This inventory is particularly robust on alpine slopes covered with smooth unconsolidated sediment or tundra where surface deformation is especially clear in remote-sensed imagery. Instabilities in areas that are snow covered, very steep and rugged, or often in shadow may be under-represented because deformation features are obscured in the optical imagery.

This inventory was compiled independently of that of Schaefer and others (2024). Similar efforts in other locations (for example, Glacier Bay, Alaska, after Avdievitch and others (2020) and Kim and others (2022)) highlight the fact that the PSInSAR methods applied in Schaefer and others (2024) provide complementary, but often dissimilar, estimates of landslide location and extent. We maintained independence from this approach to ensure that the mapped extents of optically identified landslides were unbiased by these satellite-derived measurements of slope deformation.  Despite the independent approaches, our inventory shared common extents of the 4 slow-moving landslides identified in Schaefer and others (2024) within our study area.

Data types used to identify instabilities included remote sensing data and existing digital elevation datasets. The primary imagery source was ESRI World Imagery (ESRI, 2025b), a global compilation of satellite and aerial imagery at variable resolution. Most images were analyzed using ESRI Imagery Wayback (ESRI, 2025c), Google Earth (Google, 2025), QGIS (QGIS Development Team, 2025), and ArcGIS Online (ESRI, 2025a). Digital elevation data included the Alaska statewide 5-m InSAR-derived digital elevation models (U.S. Geological Survey, 2019) and the 2024 0.5-m lidar-derived digital terrain model in Cordova (Zechmann and others, 2024).

References:
Avdievitch, N.N., Schmitt, R.G., and Coe, J.A., 2020, Inventory map of submarine and subaerial-to-submarine landslides in Glacier Bay, Glacier Bay National Park and Preserve, Alaska: U.S. Geological Survey data release, https://doi.org/10.5066/P9GCDYT2.

ESRI, 2025a, ArcGIS Online, accessed on May 6, 2025 at https://www.arcgis.com/index.html.

ESRI, 2025b, World Imagery, accessed on May 6, 2025 at https://www.arcgis.com/home/item.html?id=10df2279f9684e4a9f6a7f08febac2….

ESRI, 2025c, World Imagery Wayback, accessed on May 6, 2025 at https://livingatlas.arcgis.com/wayback/.

Google, 2025, Google Earth, accessed on May 6, 2025 at https://www.google.com/earth/.

Higman, B., Lahusen, S.R., Belair, G.M., and Staley, D.M., 2023, Inventory of Large Slope Instabilities, Prince William Sound, Alaska: U.S. Geological Survey data release, https://doi.org/10.5066/P9XGMHHP.

Kim, J., Coe, J.A., Lu, Z., Avdievitch, N.N., and Hults, C.P., 2022, Spaceborne InSAR mapping of landslides and subsidence in rapidly deglaciating terrain, Glacier Bay National Park and Preserve and vicinity, Alaska and British Columbia, Remote Sensing of Environment 281: 113231, https://doi.org/10.1016/j.rse.2022.113231.

QGIS Development Team, 2025, QGIS Geographic Information System (v. 3.34.14), Open Source Geospatial Foundation Project, http://qgis.osgeo.org/.

Schaefer, L.N., Kim, J., Staley, D.M., Lu, Z., and Barnhart, K.R., 2024, Satellite interferometry landslide detection and preliminary tsunamigenic plausibility assessment in Prince William Sound, southcentral Alaska: U.S. Geological Survey Open-File Report 2023–1099, p. 22, doi:10.3133/ofr20231099.

U.S. Geological Survey, 2019, USGS Original Project Resolution AK_GlacierBay_2019_B19 collection: U.S. Geological Survey, https://www.sciencebase.gov/catalog/item/530f4226e4b0e7e46bd2c315, accessed January 10, 2025, at https://prd-tnm.s3.amazonaws.com/index.html?prefix=StagedProducts/Eleva….

Zechmann, J.M., Wikstrom Jones, K.M., and Wolken, G.J., 2024, Lidar-derived elevation data for Cordova, southcentral Alaska, collected August 18-19, 2023, and September 19 and 22, 2023: Alaska Division of Geological & Geophysical Surveys Raw Data File 2024-6, p. 11.