skip to main content


Search for: All records

Award ID contains: 1945431

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Abstract

    Landslides are notoriously difficult to predict because numerous spatially and temporally varying factors contribute to slope stability. Artificial neural networks (ANN) have been shown to improve prediction accuracy but are largely uninterpretable. Here we introduce an additive ANN optimization framework to assess landslide susceptibility, as well as dataset division and outcome interpretation techniques. We refer to our approach, which features full interpretability, high accuracy, high generalizability and low model complexity, as superposable neural network (SNN) optimization. We validate our approach by training models on landslide inventories from three different easternmost Himalaya regions. Our SNN outperformed physically-based and statistical models and achieved similar performance to state-of-the-art deep neural networks. The SNN models found the product of slope and precipitation and hillslope aspect to be important primary contributors to high landslide susceptibility, which highlights the importance of strong slope-climate couplings, along with microclimates, on landslide occurrences.

     
    more » « less
    Free, publicly-accessible full text available December 1, 2024
  2. Abstract

    To investigate how bedrock transforms to soil, we mapped the topography of the interface demarcating onset of weathering under an east‐west trending shale watershed in the Valley and Ridge province in the USA Using wave equation travel‐time tomography from a seismic array of >4,000 geophones, we obtained a 3D P‐wave velocity (Vp) model that resolves structures ∼20 m below land surface (mbls). The depth of mobile soil and the onset of dissolution of chlorite roughly match Vp = 600 m/s and Vp = 2,700 m/s, respectively. Chlorite dissolution initiates porosity growth in the shale matrix. Depth to the 2,700 m/s contour is greater under the N‐ as compared to S‐facing hillslopes and under sub‐planar as compared to concave‐up land surfaces. Broadly, the geometries of the ‘soil’ and ‘chlorite’ Vp contours are consistent with the calculated potential for shear fracture opening under weak regional compression. However, this calculated fracture potential does not consistently explain observations related to N‐ versus S‐facing aspect nor fracture density observed by borehole televiewer. Apparently, regional compression is only a secondary influence on Vp: the primary driver of P‐wave slowing in the upper layers of this catchment is topographic control of reactive water flowpaths and their integrated effects on weathering. The Vp result is best explained as the long‐term integrated effect of groundwater flow‐induced geochemical weathering of shale in response to climate‐driven patterns of micro‐ and macro‐topography.

     
    more » « less
  3. Abstract

    Weathering processes weaken and break apart rock, freeing nutrients and enhancing permeability through the subsurface. To better understand these processes, it is useful to constrain physical properties of materials derived from weathering within the critical zone. Foliated rocks exhibit permeability, strength and seismic anisotropy–the former two bear hydrological and geomorphological consequences while the latter is geophysically quantifiable. Each of these types of anisotropy are related to rock fabric (fractures and foliation); thus, characterizing weathering‐dependent changes in rock fabric with depth may have a range of implications (e.g., landslide susceptibility, groundwater modeling, and landscape evolution). To better understand how weathering effects rock fabric, we quantify seismic anisotropy in saprolite and weathered bedrock within two catchments underlain by the Precambrian Loch Raven schist, located in Oregon Ridge Park, MD. Using circular geophone arrays and perpendicular seismic refraction profiles, anisotropy versus depth functions are created for material 0–25 m below ground surface (bgs). We find that anisotropy is relatively low (0%–15%) in the deepest material sampled (12–25 m bgs) but becomes more pronounced (29%–33%) at depths corresponding with saprolite and highly weathered bedrock (5–12 m bgs). At shallow soil depths (0–5 m bgs), material is seismically isotropic, indicating that mixing processes have destroyed parent fabric. Therefore, in situ weathering and anisotropy appear to be correlated, suggesting that in‐place weathering amplifies the intrinsic anisotropy of bedrock.

     
    more » « less
  4. Abstract

    We used seismic refraction to image the P‐wave velocity structure of a shale watershed experiencing regional compression in the Valley and Ridge Province (USA). From estimates showing strong compressional stress, we expected the depth to unweathered bedrock to mirror the hill‐valley‐hill topography (“bowtie pattern”) by analogy to seismic velocity patterns in crystalline bedrock in the North American Piedmont that also experience compression. Previous researchers used failure potentials calculated for strong compression in the Piedmont to suggest fractures are open deeper under hills than valleys to explain the “bowtie” pattern. Seismic images of the shale watershed, however, show little evidence of such a “bowtie.” Instead, they are consistent with weak (not strong) compression. This contradiction could be explained by the greater importance of infiltration‐driven weathering than fracturing in determining seismic velocities in shale compared to crystalline bedrock, or to local perturbations of the regional stress field due to lithology or structures.

     
    more » « less
  5. Temporal and spatial variations of tectonic rock uplift are generally thought to be the main controls on long-term erosion rates in various landscapes. However, rivers continuously lengthen and capture drainages in strike-slip fault systems due to ongoing motion across the fault, which can induce changes in landscape forms, drainage networks, and local erosion rates. Located along the restraining bend of the San Andreas Fault, the San Bernardino Mountains provide a suitable location for assessing the influence of topographic disequilibrium from perturbations by tectonic forcing and channel reorganization on measured erosion rates. In this study, we measured 17 new basin-averaged erosion rates using cosmogenic 10Be in river sands (hereafter, 10Be-derived erosion rates) and compiled 31 10Be-derived erosion rates from previous work. We quantify the degree of topographic disequilibrium using topographic analysis by examining hillslope and channel decoupling, the areal extent of pre-uplift surface, and drainage divide asymmetry across various landscapes. Similar to previous work, we find that erosion rates generally increase from north to south across the San Bernardino Mountains, reflecting a southward increase in tectonic activity. However, a comparison between 10Be-derived erosion rates and various topographic metrics in the southern San Bernardino Mountains suggests that the presence of transient landscape features such as relict topography and drainage-divide migration may explain local variations in 10Be-derived erosion rates. Our work shows that coupled analysis of erosion rates and topographic metrics provides tools for assessing the influence of tectonic uplift and channel reorganization on landscape evolution and 10Be-derived erosion rates in an evolving strike-slip restraining bend.

     
    more » « less
    Free, publicly-accessible full text available August 17, 2024
  6. Landscapes are frequently delineated by nested watersheds and river networks ranked via stream orders. Landscapes have only recently been delineated by their interfluves and ridge networks, and ordered based on their ridge connectivity. There are, however, few studies that have quantitatively investigated the connections between interfluve networks and landscape morphology and environmental processes. Here, we ordered hillsheds using methods complementary to traditional watersheds, via a hierarchical ordering of interfluves, and we defined hillsheds to be landscape surfaces from which soil is shed by soil creep or any type of hillslope transport. With this approach, we demonstrated that hillsheds are most useful for analyses of landscape structure and processes. We ordered interfluve networks at the Calhoun Critical Zone Observatory (CZO), a North American Piedmont landscape, and demonstrated how interfluve networks and associated hillsheds are related to landscape geomorphology and processes of land management and land-use history, accelerated agricultural gully erosion, and bedrock weathering depth (i.e., regolith depth). Interfluve networks were ordered with an approach directly analogous to that first proposed for ordering streams and rivers by Robert Horton in the GSA Bulletin in 1945. At the Calhoun CZO, low-order hillsheds are numerous and dominate most of the observatory’s ∼190 km2 area. Low-order hillsheds are relatively narrow with small individual areas, they have relatively steep slopes with high curvature, and they are relatively low in elevation. In contrast, high-order hillsheds are few, large in individual area, and relatively level at high elevation. Cultivation was historically abandoned by farmers on severely eroding low-order hillsheds, and in fact agriculture continues today only on high-order hillsheds. Low-order hillsheds have an order of magnitude greater intensity of gullying across the Calhoun CZO landscape than high-order hillsheds. In addition, although modeled regolith depth appears to be similar across hillshed orders on average, both maximum modeled regolith depth and spatial depth variability decrease as hillshed order increases. Land management, geomorphology, pedology, and studies of land-use change can benefit from this new approach pairing landscape structure and analyses. 
    more » « less
  7. null (Ed.)
    Tectonic deformation can influence spatiotemporal patterns of erosion by changing both base level and the mechanical state of bedrock. Although base-level change and the resulting erosion are well understood, the impact of tectonic damage on bedrock erodibility has rarely been quantified. Eastern Tibet, a tectonically active region with diverse lithologies and multiple active fault zones, provides a suitable field site to understand how tectonic deformation controls erosion and topography. In this study, we quantified erosion coefficients using the relationship between millennial erosion rates and the corresponding channel steepness. Our work shows a twofold increase in erosion coefficients between basins within 15 km of major faults compared to those beyond 15 km, suggesting that tectonic deformation through seismic shaking and rock damage significantly affects eastern Tibet erosion and topography. This work demonstrates a field-based, quantitative relationship between rock erodibility and fault damage, which has important implications for improving landscape evolution models. 
    more » « less