The research, published in the journal Earth Surface Dynamics, tackles a surprisingly stubborn problem in geomorphology. Grain-size analysis is fundamental to understanding natural hazards, hydrologic conditions, and ecosystems, but traditional field methods are costly, labor-intensive, and slow. A trained observer recording sizes along a stream can typically measure only a few hundred pebbles, yet robust statistical analysis of grain-size distributions often requires far more measurements than that. Photo-based alternatives, from image texture analysis to machine-learning classifiers, have tried to close the gap, but each comes with compromises that limit their usefulness in the messy, shadow-strewn environment of a real mountain river.

OrthoSAM builds on the Segment Anything Model, or SAM, a foundation model released by Meta AI researchers in 2023 that was trained on more than one billion masks across eleven million images. SAM’s remarkable capability is zero-shot inference: it can segment objects in entirely new image domains, from cell microscopy to pebble photographs, without any additional training. Its architecture consists of three decoupled components, an image encoder, a prompt encoder, and a mask decoder, which together allow the model to generate precise segmentation masks from simple input prompts such as points or polygons. For scientific imagery, this flexibility is a genuine advantage over custom-trained neural networks, which tend to perform well only within the narrow conditions they were trained on.

But SAM was never designed for what river scientists need. The model rescales every input image to 1024 by 1024 pixels to fit its transformer architecture, which means a 24-megapixel photograph is effectively compressed down to roughly 0.7 megapixels. For a sprawling orthomosaic containing thousands of densely packed pebbles, that compression is catastrophic: small stones shrink into invisibility and larger objects lose mask quality. The standard automated scheme also applies a grid of 32 by 32 equally spaced input points, optimized for images containing a limited number of objects, not the hundreds to thousands of grains found on a single riverbed image. Hardware constraints compound the difficulty, since each input point generates three candidate masks that must all be held in GPU memory.

The Potsdam team, led by Vito Chan together with Aljoscha Rheinwalt and Bodo Bookhagen, engineered a workflow of three interlocking components to overcome these limits. First, a tiling scheme divides large orthomosaics into 1024 by 1024 pixel patches with a definable overlap, so that SAM processes each patch at full effective resolution. Masks that touch a tile border are discarded to avoid artificially over-segmented pebbles along the seams, and a filtering box ensures only one mask is kept per object. Second, an improved input point generator determines how densely the prompt grid must be spaced so that every object, even the smallest pebble of interest, receives at least one input point, while a centroid-based refinement step reduces duplicate and overlapping masks. Third, a multi-scale resampling scheme runs the segmentation at several resolutions and merges the results, allowing boulders too large to fit within a single tile to be captured in a coarser pass and stitched back into the final labeled map.

Validating such a system posed its own challenge, because reliable ground-truth datasets with thousands of individually delineated pebbles simply do not exist. The researchers therefore built a synthetic pebble generator, producing images of 10,000 by 10,000 pixels filled with up to 5,000 non-overlapping solid circles of random sizes, rendered in black and white, in color, with Gaussian noise, and with simulated shadows cast by hemispherical domes under a directional light source. Because the true positions and sizes of every circle are known, the team could rigorously quantify detection quality, mask accuracy, and the fidelity of the resulting size distributions.

The results were striking. On 1,872 synthetic black-and-white pebbles, OrthoSAM achieved a precision of 1.0, a recall of 0.87, and a mean Intersection over Union, the standard measure of mask accuracy, of 0.98. On 27,528 colored pebbles with shadows, the workflow reached a precision of 0.98, a recall of 0.94, and a mean IoU of 0.91. A two-sample Kolmogorov-Smirnov test confirmed that the predicted grain-size distributions were statistically indistinguishable from the ground truth, with p-values exceeding 0.25 in every synthetic image. The team also identified a crucial detection limit: pebbles with a diameter below 30 pixels are not reliably detected, a threshold that directly informs how field crews should plan camera distance and image resolution.

To prove the concept on real terrain, the researchers applied OrthoSAM to three orthomosaics from the Ravi River in the western Himalaya, images assembled from hundreds of photographs taken with a Sony camera and processed to a spatial resolution of 0.2 millimeters per pixel. The workflow delineated 6,087 pebbles across the three scenes, with manual verification of every predicted mask yielding a precision of 0.93 and a recall of 0.94. For each pebble, the software computes a rich table of measurements, including projected area, the lengths of the a- and b-axes, perimeter, color statistics, and a normalized isoperimetric ratio that serves as a proxy for roundness, all exported in a format ready for further geomorphic analysis.

The study is candid about the method’s remaining weaknesses. Shadows proved to be a subtle adversary: pebbles completely covered by shadow were segmented as accurately as those in uniform light, but partially shadowed pebbles, where stark contrast divides a single object into lit and dark halves, were responsible for most of the observed drop in precision and mask quality. Strong image noise also degrades accuracy, prompting the authors to recommend keeping camera ISO settings moderate during data collection. Because SAM performs only instance segmentation without any classification component, non-pebble objects such as marker panels, wooden branches, and structure-from-motion artifacts can also be delineated and must be filtered out before statistics are computed. The team calls for a community-based reference dataset for granular material segmentation, analogous to the benchmark datasets that transformed lidar research.

The implications reach well beyond one Himalayan river. Grain-size distributions govern how sediment moves through drainage basins, how flood hazards develop, and how aquatic ecosystems function, and the ability to sample more than ten thousand objects from a single virtual outcrop promises a new scale of observation for Earth scientists. OrthoSAM’s Python-based pipeline is openly available on GitHub and archived on Zenodo, and the Ravi River orthomosaics are published under an open license, meaning any research group with a capable GPU can begin counting stones. A 10,000 by 10,000 pixel synthetic image currently takes about four hours to process on a 16-gigabyte NVIDIA Quadro RTX 5000, so wider adoption will depend on computing resources, but the direction is clear. The humble pebble count, a field technique essentially unchanged since 1954, is being reborn as an automated, reproducible, and scalable digital science.

Subject of Research: Automated delineation of river pebbles and grain-size analysis from high-resolution orthophotos using an extended Segment Anything Model

Article Title: OrthoSAM: multi-scale extension of the Segment Anything Model for river pebble delineation from large orthophotos

Article References: Chan, V., Rheinwalt, A., & Bookhagen, B. (2026). OrthoSAM: multi-scale extension of the Segment Anything Model for river pebble delineation from large orthophotos. Earth Surface Dynamics, 14(3), 391-416. https://doi.org/10.5194/esurf-14-391-2026

Image Credits: AI Generated

DOI: 10.5194/esurf-14-391-2026

Keywords: OrthoSAM, Segment Anything Model, grain-size analysis, geomorphology, river pebbles, orthomosaics, deep learning, image segmentation, remote sensing, Himalaya, Ravi River, sediment transport

The team collected surface sediment samples from the inner shelf before and after the monsoon and analyzed their sand, silt, and clay fractions. Inverse Distance Weighting interpolation in ArcGIS, run with a power parameter of two and a twelve-point fixed-radius neighborhood, transformed these point measurements into continuous maps of seabed texture. The results are striking: sand concentrations reach 99 percent in high-energy offshore zones during the pre-monsoon period and 98 percent after the monsoon, while silt climbs to 69 percent pre-monsoon and 67 percent post-monsoon in sheltered nearshore and estuarine settings. Clay, the finest fraction, peaks at 11 percent before the monsoon and 8 percent after it. These patterns trace the boundary between energetic open waters that winnow away fines and calm, protected embayments where fine particles settle out.

Grain-size statistics, computed with the Folk and Ward logarithmic method using GRADISTAT v9.1, add texture to the picture. Mean grain sizes ranged from 1.533 to 3.767 phi before the monsoon and from 1.267 to 4.233 phi afterward, spanning medium to very fine sands. Most samples were poorly sorted, about 70 percent pre-monsoon and 76.6 percent post-monsoon, a signature of deposition under fluctuating currents and river influence. Skewness values, ranging from minus 0.620 to 0.614 phi pre-monsoon and minus 0.425 to 0.496 phi post-monsoon, were predominantly positive, indicating the accumulation of fine sediments under low-energy conditions. Localized negative skewness at a handful of nearshore and offshore stations betrayed pockets of active winnowing, where waves strip away fines and leave coarser grains behind.

Correlation analysis quantified the hydraulic sorting that governs the system. Sand and silt displayed a strong inverse relationship, with coefficients near minus 0.97 after reprocessing, and sand and clay correlated negatively at roughly minus 0.84 to minus 0.88. Silt and clay, by contrast, co-occurred strongly, with positive correlations of about 0.79 to 0.85, reflecting their shared fate in low-energy depositional environments such as estuarine mouths and tidal flats. Notably, the authors report that an initially tabulated perfect inverse correlation of minus 1.00 between sand and silt turned out to be an artifact of output rounding, and the corrected values did not change the study’s central interpretation. Most telling of all, inter-seasonal correlations were exceptionally high, with pre- and post-monsoon sand fractions matching at roughly 0.95 and silt fractions at 0.96, evidence that the monsoon redistributes sediment without dismantling the underlying spatial order.

To disentangle the sediment populations driving these patterns, the team applied End-Member Modelling Analysis using the EMMAgeo package in R. Three statistically robust end members emerged. The first, peaking near 1.0 phi, represents coarse to medium sand winnowed and transported by high-energy offshore wave action; its contribution fell from 43 percent pre-monsoon to 31 percent post-monsoon as hydrodynamic energy waned. The second, peaking near 2.5 phi, is fine sand redistributed by littoral drift and mid-shelf reworking, and it held nearly steady at 38 to 34 percent across seasons, marking it as a year-round player. The third end member, peaking near 5.0 phi, captures silt-clay mixtures deposited in quiet estuarine and nearshore waters, and its share nearly doubled, rising from 19 percent to 35 percent after the monsoon, consistent with monsoon-fed river discharge delivering suspended fines while wave reworking diminished.

Canonical Correlation Analysis sharpened the seasonal contrast. Pre-monsoon samples clustered tightly within narrow canonical score ranges, indicating a uniform sand-silt-clay distribution under steady wave energy and limited river input. Post-monsoon samples, however, spread widely across the score plot, a statistical fingerprint of the turbulence unleashed by monsoon waves, increased river discharge, and runoff that resuspend seabed sediments and sweep finer particles shoreward. Complementary Empirical Orthogonal Function analysis of the interpolated sand grids identified two dominant modes explaining about 65 percent of total variance: an offshore-onshore gradient accounting for roughly 45 percent, and a nearshore fine-sediment accumulation zone adjacent to the estuarine mouth contributing about 20 percent, with stronger expression after the monsoon.

Principal Component Analysis and Hierarchical Cluster Analysis then distilled the dataset into its essential geography. The first two principal components explained 48.0 and 26.6 percent of variance respectively, a cumulative 74.6 percent, and the analysis passed standard suitability checks with a Kaiser-Meyer-Olkin value of 0.81 and a significant Bartlett’s test of sphericity. High positive scores on the first component marked sand-dominated, high-energy shelf environments, while the second component separated sites by finer distinctions between clay and silt. Cluster analysis, using Ward’s linkage and Euclidean distance, grouped samples into three recurring settings: sand-rich high-energy zones exposed to wave action, transitional sand-silt environments shaped by littoral drift and estuarine mixing, and fine-rich, low-energy mud deposits near estuarine mouths. With a cophenetic correlation coefficient of 0.89 and 86.7 percent of samples retaining consistent cluster membership across seasons, these three depositional provinces proved to be the enduring architecture of the Nagapattinam inner shelf.

The study also sounds a note of caution about human fingerprints on the seabed. Stations near jetties, seawalls, and harbor structures consistently fell within transitional clusters, showing elevated silt and clay contents and poorer sorting, hallmarks of sediment trapped on the updrift side of structures and starved supply downdrift. Dredging near fishing harbors resuspends fine sediments that later settle in sheltered nearshore and estuarine zones, contributing to the fine-fraction anomalies visible in the interpolation maps. The region’s new jetty for transporting offshore oil and gas adds another engineered element to a coastline already juggling natural and anthropogenic pressures. Model performance metrics confirmed that predictions were most reliable for sand, with an R-squared of 0.91 pre-monsoon, while clay showed the largest relative errors, at 48.3 percent pre-monsoon and 55.8 percent post-monsoon, reflecting the challenge of tracking suspended fines.

The authors acknowledge that a two-season sampling design cannot capture disturbances from cyclones, storm surges, or extreme river discharge events, and they call for continued monitoring, sediment-transport modeling, and wave-climate analysis using wave-rose diagrams. Still, the takeaway is compelling for coastal managers and ecologists alike: the Nagapattinam inner shelf is not a chaotic system at the mercy of the monsoon but a structured one, in which seasonal pulses of energy and river discharge redistribute sediment within a stable, three-part granulometric framework. That stability, maintained by persistent wave regimes, littoral drift, and the steady delivery of Cauvery sediments, provides a baseline against which future erosion, habitat change, and infrastructure impacts can be measured, in a monsoon-dominated coastal segment that has, until now, lacked such a detailed assessment.

Subject of Research: Seasonal sediment dynamics and grain-size distribution on the Nagapattinam inner shelf of the Cauvery basin, southern India

Article Title: Seasonal variations of inner shelf sediment dynamics and granulometric characteristics along the Nagapattinam coast of the Cauvery basin, Southern India

Article References: Pitchaimani, V. S., Abishek, S. R., Joe, R. J. J., & Promilton, A. A. A. (2026). Seasonal variations of inner shelf sediment dynamics and granulometric characteristics along the Nagapattinam coast of the Cauvery basin, Southern India. Discover Oceans, 3(1), Article 4. https://doi.org/10.1007/s44289-025-00109-x

Image Credits: AI Generated

DOI: 10.1007/s44289-025-00109-x

Keywords: sediment dynamics, granulometric analysis, Nagapattinam coast, Cauvery basin, monsoon, inner shelf, grain size analysis, IDW interpolation, end-member modelling, PCA, coastal geomorphology, Bay of Bengal