The Alpine cryosphere changes at unprecedented speed, affecting the thermal, hydrological, and mechanical state and behaviour of rock slopes. While numerous studies investigated singular drivers for progressive rock slope failures, the knowledge of hydro-thermo-mechanically coupled processes remains scarce. In this paper, we investigate the 2024 permafrost rock slope failure at Platteikogel with a volume of 50 000 m3
Fluvial valley networks in mountain ranges record the interactions between climate, tectonics, and lithology. While drainage network analysis has transformed our understanding of these interactions in subaerial settings, the landscape evolution of ice-covered orogens is poorly known. The Gamburtsev Subglacial Mountains are a ∼ 600 km-long mountain range situated beneath the East Antarctic Ice Sheet. These mountains were an important nucleation site for the ice sheet approximately 34 million years ago and are now buried beneath ∼ 2 kmχ-mapping and stream power incision modelling to show that valley network morphology is consistent with a fluvial landscape that was locally overprinted by early mountain glaciation. The spatial distribution of channel steepness indices allows us to determine the position of major geological boundaries at the edges of the mountains. We also use independent estimates of denudation rates to evaluate competing scenarios for the timing of mountain uplift and valley incision, finding that uplift of the modern Gamburtsevs most likely commenced in the Mesozoic or early Cenozoic. Regional geomorphic analysis suggests that base level for some Gamburtsev fluvial catchments was set by enclosed interior basins associated with extensional faulting. These depocentres may retain detrital sedimentary material eroded from the Gamburtsevs prior to Antarctic glaciation and are potential targets for future sub-ice drilling campaigns.
Bedrock river bathymetry is dynamic, with incision rates dependent on sediment cover, supply, and mobility in the channel. However, the scale and fluctuation of this dynamic sediment storage is not well understood, particularly in large bedrock rivers where the bed is not visible at low flows. We used repeat, high resolution, multibeam bathymetric surveys from 2021–2023 to characterize bed and bank topography in nine bedrock canyons that are representative of a wide range of width, depth, slope, and velocity observed through the 375 km long Fraser Canyon in British Columbia. Change in elevation as high as 15 m is identified between surveys. We characterize patches of contiguous change to measure changes in sediment storage volume. Our observations reveal that channel morphology determines where storage occurs. We find that sediment is “staged” through canyons, initially being deposited in a canyon near a sediment supply site, then moving downstream as the initial deposit declines. Substantial changes in storage volume occur without substantial changes in patch footprint. These findings provide key context for interpreting the reach-scale structure of bedrock erosion, the long-term evolution of mountain river networks, and the moderation of sediment delivery to lowland environments.
Recent advances in deep-learning-based image segmentation have enabled the development of automated approaches to detect individual grains and measure them for geoscientific applications. These methods facilitate the creation of much larger and more precise datasets than traditional manual grain measurements. However, they typically perform best as specialized models trained on homogeneous, task-specific datasets, and often show reduced accuracy when used on different data types.Here, we present an updated framework, ImageGrains 2.0 that leverages Cellpose-SAM, a recently published next-generation deep-learning model originally developed for cell segmentation in biomedical research. It currently represents the state-of-the-art for dense segmentation in 2D and 3D biomedical datasets. It yields robust results and is capable to generalize across distinctly different image datasets. These properties allow us to re-train the model with geoscientific datasets comprising annotated images of fluvial gravel, coarse pro-glacial deposits, and X-ray computer tomography scans of glacial till and marine sand. We benchmark the segmentation performance of our method against ground-truth annotations, compare it to the performance of other segmentation methods, and we evaluate its measurement accuracy. Our results indicate that this approach outperforms existing methods and confirm that the outstanding performance of Cellpose-SAM is transferable to segment sediment grains. We analyze the size and shape of these segmented grains and find that an increase in grain segmentation accuracy leads to more precise and realistic morphometric results, e.g., more accurate grain size distributions. Additionally, we introduce an interactive graphical user interface for image annotation and correction of model predictions, facilitating the use of the framework for a broad range of image settings. Furthermore, this study underscores the importance of curating more publicly available datasets, which could pave the way towards the generation of a foundation model for segmenting granular particles in geoscientific imagery.
During rain, water infiltrates the ground, where it flows as groundwater toward nearby rivers. There, its emergence can entrain sediments, triggering seepage erosion and thereby influencing the development and expansion of river networks. To investigate this process, we construct an experimental aquifer, made of erodible plastic sediments. A reservoir beneath the aquifer supplies water at a controlled recharge rate. We find that seepage erosion, driven by the resulting groundwater flow, is sufficient to initiate the formation and growth of a drainage network. For a given recharge rate, network growth eventually ceases as the drainage system reaches a steady-state morphology, in which sediments are everywhere at the threshold of motion. This observation indicates that the recharge rate of the aquifer selects the size of the network. In our experiment, the depth of the aquifer is small compared to its lateral extent, so that the flow of groundwater obeys the Dupuit-Boussinesq equation. As in natural systems, the water table in our experiment intersects the drainage network at the elevation of the streams. This condition provides the necessary boundary conditions to solve for the Dupuit-Boussinesq equation and reconstruct the shape of the water table around the river network. The resulting numerical solution agrees well with piezometric measurements carried out in the experimental aquifer and reveals that groundwater flow converges toward channel tips, where its flux is maximal.
Geomorphology as a discipline is defined by the use of topographic form to understand surface processes on Earth and other planets. In practice this requires drawing connections between quantitative metrics of surface geometry and rates of erosion and deformation, to understand the spatial partitioning of different erosion processes and the feedback between them. Curvature, perhaps the most fundamental way to measure and categorize surfaces of any kind, also appears explicitly in many erosion models and is therefore of significance to geomorphology. However, there is ambiguity in how curvature of discretely sampled topographic surfaces such as digital elevation models is defined and calculated. In this study we use a formal surface theory approach to compute intrinsic and extrinsic curvature metrics, and associated shape-class distributions, of approximate steady-state fluvial topography of the Oregon Coast Range, USA. We develop a workflow, including careful spectral filtering to isolate wavelengths of interest, that provides a nuanced view of landscape geometry that is consistent and accurate across steep landscape regions. Two invariants of the curvature tensor – the mean and Gaussian curvatures – reveal systematic structure of topographic geometry in channel and ridge networks that captures transitions between hillslope, debris flow, and fluvial process regimes. Mean curvature and associated shape classes are equipartitioned between concave-down and concave-up elements, forming complementary branching structures that span the landscape. These results suggest that formal surface theory approaches could prove valuable in improving process regime identification from digital elevation data in fluvial landscapes.
Supercritical flows in mountain rivers create complex flow-obstacle interactions that govern infrastructure vulnerability and channel morphodynamics, yet current understanding remains focused mostly on smooth-bed assumptions that poorly represent natural gravel-bed channels, where grain-scale roughness fundamentally alters flow physics near the bed and around obstacles such as bridge piers and in-stream vegetation. This study quantifies how bed surface characteristics control velocity fields, turbulent structures, and bed stress patterns around obstacles in supercritical flow through high-resolution detached eddy simulations coupled with volume-of-fluid free surface tracking. We examined three morphodynamic states representative of natural channel evolution: smooth beds analogous to bedrock channels, rough flat beds representing post-flood recovery conditions where sediment has been deposited as relatively uniform gravel sheets, and equilibrium scoured beds representing quasi-steady morphodynamic states. Digital representation of detailed bed surface elevation, including individual sediment grains, was considered using Structure-from-Motion photogrammetry. Numerical simulations reproduced characteristic supercritical flow structures including wall-jet formations, horseshoe vortex systems, and reverse spillage phenomena across all bed configurations. We observed that grain-scale roughness completely transforms flow organization from coherent, predictable vortical structures to chaotic flow fields dominated by grain-roughness effects. While smooth beds exhibit symmetric stress distributions with organized patterns, rough beds generate highly skewed distributions with extreme spatial variability, where coefficient of variation increases from 37 % to 115 %. Individual grains work as micro-obstacles, creating localized stress concentrations exceeding smooth-bed conditions by factors of 2–3, which can fundamentally alter sediment transport mechanisms. An equilibrium scour hole creates hierarchical flow disturbances where large-scale topographic modifications interact with grain-scale disruptions to produce the most complex stress fields observed. These findings demonstrate that engineering design standards based on smooth-bed assumptions can significantly underestimate the spatial heterogeneity and peak stress magnitudes characteristic of natural rough-bed conditions. The transition from organized stress patterns in smooth beds to grain-scale dominated physics in rough beds necessitates fundamentally different approaches to flow prediction, infrastructure design, and morphodynamic modelling in steep channel environments.
Azerbaijan hosts the world's highest concentration of mud volcanoes (MVs) on Earth, including some of the largest edifices that produce kilometre-scale mud breccia flows. Traditionally, such flows have been attributed to major eruptive events; however, recent study suggests that many of these flows form through prolonged creeping processes, analogous to warm-based glacier motion. To assess the prevalence and characteristics of this phenomenon, we analysed historical satellite imagery from Google Earth for several dozen mud volcanoes across the Caspian Basin, complemented by field observations at selected sites. Our analysis reveals that 19 mud volcanoes exhibit measurable creeping surface displacement of preexisting mud flows, with rates ranging from a few metres to tens of metres per decade. While some volcanoes, such as Goturdag, display continuous deformation across entire flow lengths, most of them exhibit only episodic motion in connection to mud volcano eruptions. While most creeping flows occur outside inhabited areas, landforms such as Bozdaq Gobu represent a potential hazard to settlements and infrastructure. These findings highlight the importance of post-eruptive creep in mud volcano evolution and emphasise the need for ongoing monitoring and hazard awareness.
Understanding the spatio-temporal dynamics of suspended sediment source activation is essential for effective ecological management, risk assessment, and infrastructure planning. Provenance analysis, which traces sediment origins, plays a crucial role in these applications, but is often based on costly fingerprinting methods. In this study, we validate a time- and cost-effective fingerprinting approach based on X-ray diffraction (XRD) data. We implement and compare two non-linear inversion schemes (steepest descent and Quasi-Newtonian) applied to binned XRD data and spatial information on potential source areas, in order to invert detrital mineralogical data into erosion rate maps while quantifying posterior uncertainty and error propagation. Forward-inverse tests with synthetic data demonstrate consistent convergence of the posterior solution and reveal the influence of geological complexity, tracer selection, and signal blending on inversion performance. The application to real-world datasets from the Gornergletscher catchment further validates the practical utility and robustness of the model.
Understanding river incision model is crucial for predicting long-term landscape evolution. For the bedrock channel incision model (detachment-limited (DL) model: erosion rate E=KAmSnASK, m, and n) can be estimated via slope-area analysis if E10Be denudation rate, previous studies globally compiled the parameter values for variable lithology. However, limited data availability for soft sedimentary rock restricts the applicability of global compilation. In addition, measuring the 10Be concentration in sedimentary rock is challenging in humid and tectonically active regions. To address this, slope-area analysis was conducted in the Kamikita Coastal Plain, Japan, where lithology (Miocene to Pleistocene sedimentary rocks) and uplift rate (∼ 0.2 mm yr−1
Sediment characteristics and grain-size distribution are crucial for understanding natural hazards, hydrologic conditions, and ecosystems. However, traditional methods for collecting this information are costly, labor-intensive, and time-consuming. To address this, we present OrthoSAM, a workflow leveraging the Segment Anything Model (SAM) for automated delineation of densely packed pebbles in high-resolution orthomosaics. Our framework consists of a tiling scheme, improved seed (input) point generation, and a multi-scale resampling scheme. Validation using synthetic images shows high precision close to 1, a recall above 0.9, with a mean IoU above 0.9. Using a large synthetic dataset, the two-sample Kolmogorov-Smirnov test confirms that there is no significant difference between the predicted and the ground-truth grain size distributions. We identified a size detection limit of 30 pixels; pebbles with a diameter below this limit are not reliably detected. Applying OrthoSAM to orthomosaics from the Ravi River in India, we delineated 6087 pebbles with high precision (0.93) and recall (0.94), based on manual verification of each predicted mask. The resulting grain statistics include area, axis lengths, perimeter, RGB statistics, and smoothness measurements, providing valuable insights for further analysis in geomorphology and ecosystem studies.
Four-dimensional (4D) topographic datasets are increasingly available at high spatial and temporal resolution, particularly from permanent terrestrial laser scanning (PLS) time series. These data offer unprecedented opportunities to analyse rapid and complex morphological processes occurring in sandy coastal environments, such as sandbar welding or bulldozer activity, as well as their longer-term impacts on sandy beaches. However, studying these processes requires the extraction and recognition of recurrent topographical surface dynamics across time, which in turn demands novel, automated methods. This study presents a novel workflow that combines 4D objects-by-change (4D-OBCs) with unsupervised classification using Self-Organizing Maps (SOMs) and hierarchical clustering. Applied to a three-year PLS time series comprising 21 194 hourly point clouds, the method identifies 4412 instances of short-term surface dynamics. These are organized into two SOMs (64 nodes each) and further grouped into 31 clusters representing distinct dynamic types, such as berm deposition, large-scale backshore erosion, and human interventions (e.g., bulldozer activity). The classification results enable detailed spatiotemporal analyses of coastal morphodynamics. The SOM topology reveals seasonal patterns in surface activity, where, for example, winter is dominated by erosional activity over the whole beach but depositional activity mainly occurs in the intertidal area. The broader clusters facilitate interpretation of environmental responses and identification of changes in cross-shore zonation of types of dynamics, like berm formation. This approach demonstrates the potential of integrating PLS and unsupervised learning to characterize complex surface dynamics, through a fully automated extraction and classification workflow. While the interpretation of clusters and their relation to environmental variables in this study is performed through expert-based analysis, the methods provide a framework for targeted, data-driven investigation and prediction of morphodynamic processes in high-resolution 4D remote sensing datasets.
Reconstructing the transport histories and provenances of glacial sediments and ice-contact deposits (e.g. tills, moraines) in formerly glaciated regions remains a major challenge, particularly at icefield- to ice-sheet scales and over multi-millennial timescales. Yet such reconstructions are central to key questions in Quaternary science, including estimates of past glacial erosion rates and sediment fluxes, the role of subglacial sediment storage in erosion reduction, or the reconstruction of past ice-flow dynamics, ice divides, and transfluences. While numerical modelling can enable one to reproduce past glacial sediment transport via coupling glacier models with particle tracking, this becomes computationally unfeasible over large spatial domains and paleo timescales using traditional computing. As a result, no study to date has simulated glacial sediment transport using large particle numbers (tens of millions) across continental-scale icefields such as the one occupying the European Alps during the Last Glacial Maximum (LGM). Here, we overcome this limitation using the Instructed Glacier Model (IGM), which allows the coupling of 3D Lagrangian particle tracking with high-resolution glacier simulations, both accelerated on Graphics Processing Units (GPU). This unlocks the modelling of ice advection of millions of particles at minimal additional computational cost, allowing simulations of glacial sediment transport across the European Alps over multi-millennial timescales (40–18 ka) and at the unprecedented spatial resolution of 300 m. We achieve ∼ 50×
Physics-based morphodynamic modeling is essential for advancing river management science and understanding Earth's geomorphological evolution processes. However, their computational demands and long processing times hinder long-term applications. This paper introduces and tests a robust Deep Learning (DL) framework that opens the door to overcoming these challenges through integrating convolutional neural networks (CNNs) with long short-term memory (LSTM) architectures, trained using outputs from the physics-based HEC-RAS model. This framework facilitates rapid and continuous spatiotemporal predictions of hydrodynamic parameters and morphodynamic responses of flood events. Hydrodynamic predictions showed strong performance across the testing dataset, with mean RMSEs of 0.15 m and 0.04 m s−1R2
Wave-abrasion terraces are geomorphic marker horizons that provide information of past water levels, in marine and lacustrine environments. By integrating elevation measurements and age constraints, they serve as strain markers to assess vertical deformation rates associated with tectonic and/or climatic processes. As most geomorphic markers, wave-abrasion terraces are ephemeral features, and their topographic signature has variable levels of noise. Therefore, accurate and precise estimates of marine terrace morphology are essential to obtain significant uplift/subsidence rates. The open source TerraceM-3 enables operators to reduce non-systematic and systematic errors in terrace mapping by integrating machine learning techniques to replicate human mapping criteria, and standardized and reproducible workflows to handle systematic errors. In many regions, the availability of high-resolution topographic data remains relatively scarce limiting precision in geomorphic marker mapping. TerraceM-3 introduces a new module for downloading, filtering, and processing centimeter-resolution topographic data from the ICESat-2 satellite at global scale. The TerraceM-ICESat module produces vegetation-free profiles ready for assisted machine-learning mapping into a graphical user interface. Shallow bathymetry may be also extracted to extend the mapping of drowned terraces offshore. The new functionalities of TerraceM-3 were tested along tectonically active coasts in Peru and Algeria, revealing detailed deformation histories controlled by subducted seamounts and crustal faults. TerraceM-3 is designed to support research in tectonic geomorphology and paleoclimate studies by enhancing the precision and accuracy of wave-abrasion terrace mapping with applications in the assessment of coastal hazards.
Asymmetric tidal dunes with intermediate (10–17°) to low-angle slopes (< 10°), usually with an irregularly-shaped lee side, are often found in natural, constrained tidal environments such as tidal rivers, estuaries and tidal channels. However, previous studies on bedform flow dynamics have largely focused on high-angle dunes with a simple (straight) lee side, generally found in flume studies or small rivers. This study provides a detailed characterisation of the flow and turbulence over asymmetric tidal dunes under an idealised tidal flow condition based on laboratory measurements. Specifically, we aim to address how tidal dune shape, especially the lee side geometry, controls the properties of flow separation and resulting turbulence structures. Furthermore, we address how flow bidirectionality changes flow and turbulence over the same tidal dune geometry. To achieve this, we conducted large-scale, high-resolution flume experiments over two idealised dune morphologies which represent natural asymmetric tidal dunes with intermediate- to low-angle slopes. The flow condition was an idealised representation of tidal flow for which the same unidirectional steady currents were imposed first in one direction, then in the opposite direction. Our results show that for the case of an intermediate-angle tidal dune and when the flow was directed from the gentle stoss to the steep lee slope, a downward expanding turbulent wake and a small, near-bed permanent flow separation were detected. A small flow separation was also detected for the case of low-angle tidal dune. When the flow was reversed and directed from the steep stoss to the gentle lee slope, flow direction significantly altered the flow dynamics for both dunes as no permanent flow separation was observed and turbulence structure was similar to that over a flat bed. Interestingly, we demonstrated that a small intermittent flow separation can still form even for tidal dunes with very gentle slope (4°) provided that a short steep portion is present. This implies that low-angle dunes can generate flow resistance and can potentially contribute to sediment mobilisation above low-angle dunes. Overall, our study highlights the significant impact of dune morphology, particularly the lee side slopes, and flow direction on the flow and turbulence dynamics above asymmetric tidal dunes. Our findings can have further implications on the parameterisation of hydraulic roughness, estimation of sediment transport and the resulting morphodynamics in natural shallow water environments.
Bedrock river incision is a fundamental process driving the evolution of mountainous landscapes. Bedrock strength is often considered a primary control on incision rates and river profile morphology, with laboratory experiments showing a strong correlation between erosion rate and tensile strength. However, in natural settings, lithological boundaries frequently do not correspond to changes in the channel gradient. This study addresses this apparent paradox by integrating field observations with numerical experiments in the tributaries of the Abukuma River basin, northeastern Japan. Field surveys were conducted to measure bedrock tensile strength, riverbed gravel grain size, and the spatial distribution of lithologies. Despite more than an order-of-magnitude variation in bedrock tensile strength across the study area, the channel slopes remained nearly uniform. Numerical experiments were performed using three models of bedrock river erosion to investigate the underlying mechanisms. Among them, the sediment-flux-dependent model, which explicitly incorporates sediment cover and tool effects, most accurately reproduced the observed longitudinal profiles. The results reveal that the local lithology does not directly influence channel slope due to a negative feedback between sediment cover and river gradient. Higher bedrock erodibility reduces channel slope and sediment transport capacity, promoting sediment cover. The resulting sediment cover suppresses further erosion and offsets the effect of bedrock strength. These findings highlight the limited role of bedrock strength in controlling channel gradients and underscore the importance of sediment dynamics, particularly sediment supply and grain size, in shaping fluvial topography. Future research should explore how lithology-dependent variations in sediment characteristics influence river profile development.
The Normalized Difference Vegetation Index (NDVI) can be effectively used for monitoring the spatial and temporal dynamics of riparian vegetation. However, quantitative and efficient evaluations of the links between NDVI and bio-hydromorphological processes remain limited, particularly in river management contexts where dense in-channel vegetation can obstruct flow and reduce conveyance capacity. Using 200 cloud-free Sentinel-2 images (2015–2024) covering a 20-km reach of the Chikuma River (Japan), we evaluated the utility of high temporal resolution NDVI and greenness index (defined as NDVI > 0.2) as quantitative indicators of bio-hydromorphological dynamics and its implications for riverine management. The analysis focused on the relationships between NDVI dynamics, flood magnitude, relative elevation along lateral channel morphology, and seasonal vegetation variability within a frequently disturbed channel. The results show that NDVI fluctuations strongly correspond to flood disturbances at lower relative elevations, whereas vegetation at higher elevations remains relatively stable. The annual maximum greenness ratio was well described by a logistic model along the cross-sectional transects. Annual greenness ratio exhibited clear seasonal patterns, showing a late-summer (August–September) greenness peak. These spatiotemporal and seasonal NDVI characteristics demonstrate the potential of Sentinel-2 imagery to operationalize both the “when” (timing) of vegetation management and the “where” (priority zones defined by relative elevation), providing a transferable, remotely sensed basis for flood-risk mitigation in frequently disturbed riverine environments.
Recent recognition of a new type of fluvial system – fluvial fans – introduces a fan-shaped channel network that appears similar to that of river-dominated deltas. Deltas form where rivers enter lakes and oceans, while fluvial fans are terrestrial landforms. However, fluvial fans can reach the shorelines of oceans or lakes, and in such cases the distinction between fluvial fan and river-dominated delta channel networks becomes ambiguous. We currently lack fundamental understanding of these two landforms' morphometric differences, despite their high socioeconomic significance, vulnerability to natural hazards, and key differences in how these landforms respond to global climate change and urbanization. Here we review the relevant conceptual differences in delta and fluvial fan network morphodynamics, propose a set of quantitative morphometric criteria to distinguish fluvial fan and delta channel networks, and test these criteria on 40 deltas and 40 fluvial fans from across the world. This initial attempt to contrast and distinguish deltas and fluvial fans based on their channel network morphometrics demonstrates that quantifying channel network angles (mean of 74.0° for deltas and 55.0° for fluvial fans) and trends in normalized channel widths and lengths provide efficient criteria, but some ambiguities remain that need to be resolved in future work. This research advances our mechanistic understanding of fluvial fan and delta channel networks and the recognition of modern and ancient landforms on Earth and other planetary bodies, such as Mars and Saturn's moon Titan.
River erosion is a fundamental process that impacts, among others, mountain landscape evolution. Mountain rock lithologies often exhibit bedding, joints, and fractures that are thought to alter the incision efficiency of rivers compared to intact, massive rocks. The presence of close enough planar mechanical discontinuities allows the creation and entrainment of large blocks through plucking, a process that adds to abrasion, and potentially macroabrasion, by the transported sediment. Despite preliminary attempts to include shallow fracturing in theoretical models of bedrock incision and a couple of studies that quantified the relative importance of abrasion and plucking processes in situ, we are still lacking ways to systematically probe the role of fractures on bedrock erosion rates and processes. Due to the complex interactions at play, here we investigate this question via an experimental approach, using a new erosion mill designed to erode a fractured concrete disk with a diameter of 17 cm. We simulate vertical or dipping fractures by embedding a 3D-printed plastic mesh in the concrete, using BVOH – a plastic that softens in cold water – creating mechanical weaknesses with a controlled pattern. We explore 10 different geometries and run 4 additional experiments without fractures for control. We record the topographic evolution every 2 min by photogrammetry and derive erosion maps by measuring elevation changes between successive scans. Our results show that fractures influence the morphodynamical evolution of the disks and the relative contributions of abrasion and plucking. However, abrasion systematically remains the dominant erosion mechanism, with plucking contributing at most to 29 % of the total erosion for vertical fractures spaced by 20×20 mm2, 40 % for one specific dip angle (67°), and less than 10 % for most experiments. Average erosion rates show a modest (20 %) increase with the fraction of plucking, but do not show a clear relationship with fracture density and the presence of fractures. We suggest that the rate of erosion by plucking is limited by the depth and slow rate of horizontal fracture propagation between pre-existing vertical fractures, such that in our experimental setup, abrasion is systematically a dominant component. These findings emphasize the critical role of block preparation and loosening for plucking to be an effective process compared to abrasion. This new setup allows abrasion, macroabrasion, and plucking driven by bedload impacts to be studied in controlled situations, albeit with the well-known limits of abrasion mills and without the variety of natural processes that can drive fracture propagation. Further experiments should expand the parameter space of the erosion efficiency problem (i.e., sediment mass, grain size, flow velocity, intact rock mass strength, 3D fracture patterns) to help in developing mechanistic models applicable in natural environments.