The effort behind the data was anything but routine. High Mountain Asia contains the largest glacierized area outside the polar regions, yet field research there is notoriously difficult. Low barometric pressure, extreme weather and steep terrain challenge both personnel and instruments, and the region’s stations are frequently disabled by riming, heavy snowfall, battery depletion and communication failures. The Everest stations were installed as part of the National Geographic and Rolex Perpetual Planet Everest Expeditions, with Sherpa climbers bolting custom aluminium tripods designed by Campbell Scientific directly to rock and guying them to additional anchor points. The Balcony station at 8,430 metres was toppled by extreme winds and anchor failure in January 2020, prompting a return expedition in May 2022 that installed the new Bishop Rock station at 8,810 metres and upgraded the South Col site.

Each station carries a carefully engineered suite of sensors. Air temperature and relative humidity are measured with Vaisala HMP155A-L5-PT probes housed in naturally ventilated 14-plate solar radiation shields, while barometric pressure is recorded with Vaisala PTB110 and PTB210 sensors inside the datalogger enclosures. Wind speed and direction initially came from R. M. Young 05108-45 anemometers, but all four units installed at the two highest sites failed in the extreme conditions, and the team replaced them with a polycarbonate version of the sensor plus a custom Richards C5C anemometer and pitot tube made by the Mount Washington Observatory. Radiation components are captured with Hukseflux pyranometers and pyrgeometers, and precipitation is measured at the two lowest stations with OTT Pluvio2 weighing gauges fitted with double-Alter wind shields to reduce the wind-induced undercatch that plagues snowfall measurement in exposed mountain terrain.

The engineering details reveal how altitude dictated design choices. At the lower stations, sensors sit 2 metres above the ground and batteries live inside the datalogger box; at the highest sites, sensors were mounted lower, at about 1.5 metres, to reduce both wind loading and the leverage, or torque, that a taller pole would exert on the tripod base during severe gusts. Batteries at the upper stations were placed in separate insulated boxes to protect them from extreme cold, and both boxes were bolted to rock to cut wind drag. The lower stations weigh roughly 60 kilograms each, the upper ones about 52 kilograms including the experimental pitot tube. Data are logged on Campbell Scientific CR1000X dataloggers at intervals from 10 minutes to daily, and the published archive provides quality-controlled hourly values in Coordinated Universal Time.

Because raw readings from such a hostile environment are riddled with artefacts, the team applied a multistage quality control procedure. Relative humidity values exceeding 100 percent were capped at that physical limit, and artificially low readings caused by the logger computing humidity with respect to water rather than ice at sub-zero temperatures were corrected using Buck’s equations for saturation vapour pressure. Periods of zero wind speed combined with zero directional variability were flagged as sensor freezing and removed rather than mistaken for calm conditions. Night-time incoming shortwave radiation below 7 watts per square metre was set to zero, and an albedo-based correction recalculated compromised radiation values whenever fresh snow or rime on the upward-facing sensor pushed the apparent surface albedo above the realistic threshold of 0.95.

The resulting climatology, though based on a short record, paints a vivid seasonal picture. Mean annual temperatures were 4.1 degrees Celsius at Phortse, minus 3.1 degrees at Base Camp and minus 10.2 degrees at Camp II. At South Col, July was the warmest month with a mean of minus 12.2 degrees and a striking diurnal range of 10.2 degrees, while February averaged minus 29.7 degrees. Precipitation clearly delineates the seasons: winters are predominantly dry, amounts build through the pre-monsoon, and the June-to-September monsoon delivers 72 percent of annual precipitation at Phortse and 77 percent at Base Camp. The mean annual precipitation gradient between the two sites, separated by roughly 1,500 metres, was minus 107 millimetres per kilometre, but it weakens by almost half during the monsoon, indicating that relative precipitation drops with elevation are far steeper outside the monsoon season.

Temperature gradients proved equally revealing. Between Phortse and Base Camp the mean temperature gradient was minus 4.8 degrees Celsius per kilometre, least negative in winter and most negative in the pre-monsoon. When the four stations with minimal data gaps were combined, the gradients became more strongly negative, at minus 6.5 degrees per kilometre in the pre-monsoon, minus 5.7 in the monsoon and minus 6.0 annually. The authors caution that the temperature-altitude relationship in the Khumbu region is fundamentally non-linear, which limits direct comparison with lapse rates calculated across different elevational ranges, and they therefore provide a non-linear equation for the gradient between Phortse and South Col. Radiation measurements add further nuance: maximum daily incoming longwave radiation during the monsoon exceeds 350 watts per square metre at Phortse but stays below about 250 at South Col, likely reflecting the colder, drier, less cloudy atmosphere aloft.

A central contribution of the study is a rigorous comparison with ERA5, the fifth-generation global reanalysis produced by the European Centre for Medium-Range Weather Forecasts that many researchers use as a stand-in for sparse mountain observations. Extracted from the nearest grid point at the 350 hectopascal pressure level, ERA5 temperatures tracked observed variability at South Col reasonably well, with coefficients of determination above 0.6 in both 2019 and 2022, and even higher agreement of 0.85 with the short Balcony record in 2019. But the reanalysis systematically underestimated air temperature, showed far less diurnal variability because the pressure level represents the free atmosphere rather than the surface, and consistently overestimated mean wind speeds, with mean absolute errors of 4.8 and 5.3 metres per second in 2019 and 2022. Relative humidity comparisons carried mean absolute errors above 20 percent, though 10-day running means from both datasets clearly captured the monsoon onset at South Col on 1 July 2019 and 14 June 2022.

The practical implications reach well beyond atmospheric science. Observations from South Col reveal mean winds that can exceed 30 metres per second and gusts above 60 metres per second, conditions capable of blowing mountaineers off their feet and inducing cold injuries. Pressure data from the network have already shown that oxygen availability at the summit varies on synoptic timescales, meaning the apparent elevation of Everest, how high the mountain would feel without supplemental oxygen, can shift by almost 750 metres, and a winter ascent without bottled oxygen may at times be impossible. The team anticipates that ERA5, once its time-varying biases are corrected with empirical-statistical or machine-learning approaches, could be used to gap-fill and extend the intermittent records from Everest’s upper slopes, enabling hyper-local forecasts that help expeditions identify optimal climbing windows.

The archive also underpins research on the region’s fragile cryosphere and its role as a freshwater source for downstream communities. It has already been used to estimate surface energy balances at the summit and at South Col Glacier, revealing a high-altitude ice system acutely sensitive to changes in effective precipitation because of extremely high insolation and its responsiveness to albedo variations. Combined with the longer-running EvK2CNR and GLACIOCLIM networks at lower elevations, the new data allow quantification of elevational gradients in key meteorological variables across roughly five vertical kilometres, covering the entire glacierized range of the Khumbu region, information essential for distributed glacier and hydrological modelling. And because meteorological measurements from the highest reaches only began in 2022, the authors note that scientists are still at the very beginnings of exploring the weather of this extreme environment, with questions the data might answer likely to grow rapidly across disciplines far beyond the climate sciences.

Subject of Research: High-altitude weather station observations across the Mount Everest region of Nepal

Article Title: Weather station data from the Mount Everest region, Nepal: 3810–8810 m above sea level

Article References: Khadka, A., Perry, L. B., Matthews, T., Sherpa, T. G., Shrestha, C. B., Shrestha, D., Aryal, D., Tuladhar, S., Pradhananga, N., Kayastha, D., Raichle, B., Athans, P., Sherpa, D. Y., Garrett, K., Wheeler, G., Young, T., & Elmore, A. (2026). Weather station data from the Mount Everest region, Nepal: 3810–8810 m above sea level. Earth System Science Data, 18(10), 7253-7267. https://doi.org/10.5194/essd-18-7253-2026

Image Credits: AI Generated

DOI: 10.5194/essd-18-7253-2026

Keywords: Mount Everest, automatic weather stations, Himalaya, climate data, ERA5 reanalysis, temperature lapse rate, monsoon, glaciology, high-altitude meteorology, Khumbu, quality control, open data

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 insulating power of a rocky blanket has been recognized since the classic experiments of Østrem in 1959. A continuous debris layer just a few centimeters thick can dramatically slow the melting of ice beneath it, which is why many Himalayan and Alpine glaciers today sport long, sluggish tongues of debris-mantled ice that seem stubbornly resistant to warming. But the Zurich team argues that treating this debris layer as a static feature fundamentally misleads our understanding. Debris does not simply sit on a glacier; it is delivered by rockfalls and avalanches, entrained into the ice in the accumulation zone, carried along inside the glacier for centuries, and only much later melts out onto the surface in the ablation zone. Every stage of that journey couples back to the dynamics of the ice itself.

To capture this complexity, the researchers built a numerical flowline model based on the Shallow Ice Approximation that couples ice flow to a depth-resolved, two-dimensional advection scheme for englacial debris. Unlike earlier models that assumed a uniform debris concentration throughout the ice, their approach tracks debris concentration on a grid of 40 vertical layers and 50-meter horizontal spacing, allowing debris to be supplied at a prescribed rate and location entirely independently of the climate forcing. A novel entrainment scheme controls exactly how much debris enters the ice in the accumulation area, while a terminal ice-cliff boundary condition, in which surface debris slides off when ice thins below a critical height, prevents the unrealistic piling up of debris at the snout. The team validated the model against benchmark experiments from previous studies and tested its sensitivity to the ice rheology, grid resolution, and the so-called characteristic debris thickness that governs how strongly a given layer suppresses melt.

One of the study’s most striking quantitative results concerns how much warming debris cover can offset. The researchers introduced a metric they call dELA: the difference between the equilibrium line altitude of a debris-free glacier and that of a debris-covered glacier of equivalent size. They found that steady-state glacier size scales linearly with the rate of debris input, meaning that each increment of rocky supply buys a proportional amount of protection against a rising snowline. Intriguingly, this compensation is roughly independent of elevation, and because debris-covered tongues are long and thin, the effect on glacier length is about twice as strong as the effect on ice volume. More debris does not merely thicken the blanket; it stretches the glacier, and it does so in a predictable, linear fashion.

The real surprises emerged when the team subjected their model glaciers to abrupt climate change. When the equilibrium line altitude was raised by 100 meters, equivalent to roughly 1 to 1.25 degrees Celsius of warming, the debris-covered glacier entered a first phase lasting several decades in which its terminus refused to retreat at all, even as the ice beneath the debris thinned rapidly. Debris melted out of the stagnant ice and accumulated on the surface, expanding and thickening the insulating layer. Then came phase two: the weakened tongue suddenly collapsed, retreating abruptly within a few decades and shedding much of its debris cover in the process. Finally, in a third phase lasting centuries, the glacier slowly re-advanced as a thick new debris layer built up, eventually stabilizing at a length slightly beyond its post-collapse minimum. A debris-free glacier subjected to the same warming retreats smoothly and almost immediately by comparison.

This non-monotonic response, thinning, collapse, and then re-advance, is rooted in the slow journey of debris through the ice. When the climate warms, the equilibrium line climbs toward the debris source, shortening the englacial trajectory that debris must travel before melting out. Meanwhile, the thinner, slower glacier transports ice less efficiently, so the same absolute debris supply is spread over less ice flux, concentrating debris and thickening the surface layer. All three of these mechanisms push in the same direction: warming, counter-intuitively, increases both the extent and the thickness of debris cover. But because the enhanced debris concentrations must first be advected through the glacier interior, their full effect on the surface emerges only after a lag of centuries, producing the delayed re-advance that the authors identify as a previously unstudied feature of the debris-glacier system.

The model also demonstrates that debris-covered glaciers are extraordinarily good at filtering out short-term noise. When the researchers forced their glacier with sine-wave oscillations in climate or debris supply, fluctuations on decadal timescales had essentially no impact on glacier geometry. Even century-scale variations in debris input left the terminus almost perfectly stable, and only oscillations with wavelengths of 500 years produced a measurable response, one so phase-shifted that glacier volume appeared inversely related to debris supply. In practical terms, this means that the state of a debris-covered glacier today reflects long-term trends rather than recent disturbances, even more strongly than is the case for clean-ice glaciers. Short-term variability in rockfall activity or decade-scale climate wiggles simply never make it through the system’s long memory.

Large debris supply events are a different story. In a spike experiment, the team deposited a full meter of debris across the input zone in a single year, mimicking a catastrophic rock avalanche, and then watched the consequences unfold. The pulse of debris traveled englacially for centuries before surfacing, temporarily increasing both debris thickness and glacier extent for more than a hundred years. Similarly, step changes in debris input produced no reaction at all for roughly 300 years, followed by very slow adjustment. The authors caution that this debris advection response time depends critically on where debris enters the glacier relative to the equilibrium line: material deposited high in the accumulation zone travels a long, slow path through the ice, delaying its surface appearance by many centuries, whereas debris dropped directly onto the ablation zone takes effect almost immediately.

When the researchers moved from idealized forcing to something closer to reality, feeding their model a reconstructed 2,000-year equilibrium line history for the Alps followed by future scenarios from the IPCC, the filtering effect became even more apparent. Decadal fluctuations were smoothed out of the length response almost entirely, and the thick debris layers and re-advances seen in steady conditions only appeared during periods when the climate remained roughly constant for several centuries. Long-term climate variability, the study concludes, actively suppresses the formation of thick debris covers. Adding a realistic, undulated bed topography based on radar surveys of Switzerland’s Zmuttgletscher introduced yet another wrinkle: steep bedrock steps can cause sections of the thinning tongue to detach entirely, leaving pockets of stagnant dead ice and producing a stepped, stuttering retreat. Under the most extreme high-emission scenario, once the snowline climbed above the debris deposition zone into the headwall, the delayed re-advance vanished altogether and the glacier settled into centuries of extremely slow, relentless shrinkage.

The implications reach well beyond glacier modelling. Because englacial transport lags and dynamic feedbacks entangle the debris record with past climate and ice flow, reconstructing historical debris supply rates or climate conditions from the debris covers we observe today is inherently ambiguous, and may be impossible when both variables have changed. The authors also note that their results are sensitive to the characteristic debris thickness parameter, which varies widely between glaciers and studies, underscoring the need for glacier-specific field constraints. What the study offers instead is a clearer physical intuition: debris-covered glaciers are not simply slower versions of clean-ice glaciers but systems governed by their own distinct clock, one set by the centuries-long passage of rock through ice. Their apparent stability can be the most dangerous signal of all, masking thinning ice that may be poised for abrupt collapse, while their long memory means that today’s expanding debris covers are, in part, echoes of supply decisions made by mountains hundreds of years ago.

Subject of Research: Transient response and debris-transport feedbacks of debris-covered glaciers under changing climate and debris supply

Article Title: Feedbacks and timescales in the modelled transient response of debris-covered glaciers

Article References: Hardmeier, F., Ferguson, J. C., & Vieli, A. (2026). Feedbacks and timescales in the modelled transient response of debris-covered glaciers. The Cryosphere, 20(9), 5365-5392. https://doi.org/10.5194/tc-20-5365-2026

Image Credits: AI Generated

DOI: 10.5194/tc-20-5365-2026

Keywords: debris-covered glaciers, glacier dynamics, englacial debris transport, ice flow modelling, climate change, response times, equilibrium line altitude, glacier retreat, The Cryosphere, numerical simulation, Himalaya, Alps

The challenge the researchers faced is one that defines hydrogeology across the entire Himalayan arc. Unlike the vast alluvial aquifers of the plains, mountain groundwater is not stored in generous underground sponges. Instead, it occupies a patchwork of fractures, joints, weathered zones and thin valley-fill deposits, all controlled by an extraordinarily complex geological history. The Soar Valley sits within terrain dominated by Precambrian and Paleozoic rocks, bounded to the south by the Main Boundary Thrust and to the north by the Main Central Thrust, with the North Almora Thrust, Berinag Thrust and Munsiari Thrust slicing the landscape into juxtaposed units of contrasting permeability. Limestone and dolomite of the Deoban and Gangolihat formations can store and transmit water through solution-widened fractures, while the slates and phyllites of the Mandhali Formation, locally known as Soar Slates, resist flow except along cracks and weathered zones. Whether a given hillside yields water or sheds it can change within a few hundred metres.

To bring order to this complexity, the team adopted a multi-criteria decision analysis framework built on the Analytic Hierarchy Process, or AHP, a weighting method developed by mathematician Thomas Saaty that has become a workhorse of environmental mapping. Eight thematic layers were assembled: rainfall, geology, lineament density, land use and land cover, geomorphology, drainage density, slope and elevation. Each layer was derived from a different source, including India Meteorological Department rainfall data averaged over 2021 to 2024, Valdiya’s classic 1980 geological map of the Kumaon Himalaya, lineaments extracted from ASTER elevation data and cross-checked against the Geological Survey of India’s Bhukosh database, a 2025 Sentinel-2 land cover product from Esri, and geomorphological units from the same GSI archive. Every dataset was resampled to a common 30-metre grid and projected into a single coordinate system so that the layers could be stacked and compared pixel by pixel.

The AHP procedure required the researchers to judge, pairwise, how much each factor matters for groundwater occurrence, using Saaty’s one-to-nine importance scale. When the resulting comparison matrix was solved, rainfall emerged as the heaviest influence at roughly 25.7 percent, followed closely by geology at 24.1 percent and lineament density at 20.2 percent. Land use and land cover received a moderate 11 percent, geomorphology 7.8 percent, drainage density 5.2 percent, slope 3.6 percent and elevation just 2.4 percent. The logic is straightforward: rain is the ultimate source of recharge, rock type governs how much water the subsurface can hold, and fractures act as the highways along which water infiltrates and moves. Slope and elevation, by contrast, exert only indirect control by shaping runoff and topographic position. Crucially, the team checked the internal consistency of their judgments using Saaty’s consistency ratio, obtaining a value of 4.4 percent, comfortably below the 10 percent threshold above which expert judgments are considered unreliable.

With weights in hand, the researchers ran a weighted overlay in ArcGIS, multiplying each layer’s weight by the favourability rating of its subclasses and summing the results into a Groundwater Potential Index. The index was then classified into five zones, from very low to very high. The headline result is striking: only 0.01 percent of the study area, a fraction of a square kilometre, qualified as very high potential, while 92.76 percent fell into the moderate and high categories and 7.23 percent into the low and very low classes. The moderate zone alone covered 426.73 square kilometres, or 68.11 percent of the area, with the high zone adding another 154.42 square kilometres. In other words, despite receiving between roughly 1,858 and 2,442 millimetres of monsoon rainfall a year, the landscape offers almost no places where every favourable condition coincides. The best prospects cluster on the valley floor and in low-relief pockets where permeable lithology, gentle slopes and structural discontinuities overlap.

The spatial pattern tells a coherent geological story. High lineament densities, exceeding about 0.084 kilometres per square kilometre, mark zones of secondary porosity where tectonic deformation has cracked otherwise impermeable rock, and these patches align with some of the model’s more promising areas. Permeable units such as the Thalkedar Formation, the Damtha Group and the Gangolihat Formation scored well, whereas the compact quartzites of the Berinag Formation were rated low. Geomorphology reinforced the divide: active floodplains, piedmont alluvial plains, valley fills and water bodies earned high recharge ratings, while the highly dissected hills and valleys that blanket 87.52 percent of the study area were rated unfavourably. Land cover added a further layer of nuance, with tree cover and rangeland dominating the landscape and built-up areas, at 7.79 percent, concentrating in the urbanising valley where sealed surfaces suppress infiltration. The authors caution that satellite land-cover classification in fragmented Himalayan terrain carries real uncertainty, with small terraced fields easily confused with rangeland.

What separates this study from many earlier AHP-based groundwater maps is its quantitative validation. The team compiled 91 spring locations from systematic fieldwork across the valley and supplemented them with spring records from the CHIRAG Spring Atlas of Uttarakhand. Using a receiver operating characteristic analysis, they tested how well the model’s continuous potential index discriminated between locations where springs actually occur and the full range of index values. The result was an area under the curve of 0.795, a figure conventionally read as good discriminatory performance. Importantly, the spring data played no role in setting the AHP weights, so the validation tested an independent prediction rather than a model tuned to its own answers. The authors are careful to note the caveat that no independently confirmed absence dataset was available, meaning the statistic measures discrimination with respect to springs rather than absolute predictive accuracy, and that clustered spring distributions may influence the result.

The practical implications reach well beyond academic mapping. Because the zones represent relative suitability under the chosen criteria and thresholds rather than measured aquifer storage or sustainable yield, the authors position the map as a screening tool for prioritising follow-up work. Areas flagged as high or very high potential warrant detailed hydrogeological investigation, careful well-site assessment and controlled development, while the low and very low zones call for recharge enhancement, spring-shed protection, rainwater harvesting and runoff management. The team also emphasises protecting traditional mountain water infrastructure, including community ponds, naula and dhara spring structures and natural drainage channels, from encroachment and filling, and urges that urban planning in the valley incorporate permeable surfaces, green infrastructure and community-based water management so that growth does not sever the recharge pathways the aquifers depend on.

For a state where more than half of rural households depend on springs that many studies suggest are drying or becoming seasonal, the Soar Valley map offers something rare: a spatially explicit, quantitatively tested starting point for decisions about where to drill, where to recharge and where to build. The framework itself, integrating climatic, geological, structural, topographic and land-surface controls in a single validated model, is designed to be transferable to other fast-urbanising Himalayan valleys facing the same squeeze of rising demand and naturally constrained supply. The authors recommend that future work add seasonal groundwater-level monitoring, spring-discharge measurement and aquifer characterisation to sharpen the picture further. In a mountain system where a single misplaced borewell or paved-over recharge zone can undo decades of natural storage, knowing where the water is likely to be, and where it is not, may prove as valuable as the water itself.

Subject of Research: GIS-based groundwater potential zone mapping in the Kumaon Lesser Himalaya

Article Title: Hydro-geospatial modelling of groundwater potential zones in the soar valley and adjacent terrains of the Kumaon Lesser Himalaya using an integrated AHP–GIS framework

Article References: Pant, D., Gururani, K., Upadhyay, R., Singh, R. A., & Singh, A. K. (2026). Hydro-geospatial modelling of groundwater potential zones in the soar valley and adjacent terrains of the Kumaon Lesser Himalaya using an integrated AHP–GIS framework. Discover Geoscience, 4(1), Article 395. https://doi.org/10.1007/s44288-026-00766-1

Image Credits: AI Generated

DOI: 10.1007/s44288-026-00766-1

Keywords: groundwater, Himalaya, AHP, GIS, remote sensing, springs, hydrogeology, lineaments, Uttarakhand, recharge, multi-criteria decision analysis, urbanization

The stakes could hardly be higher. Landslides rank among the world’s most destructive natural hazards, and a global analysis of fatal non-seismic landslide events recorded 4,862 such disasters between January 2004 and December 2016, killing 55,997 people, with Asia the hardest-hit continent. India carries a disproportionate share of the burden: roughly 0.42 million square kilometres, or 12.6 percent of the country’s land surface excluding snow-covered terrain, is susceptible to landslide hazard. In the Himalayan region alone, landslides claimed more than 5,000 lives between 2007 and 2015, and the June 2013 extreme rainfall disaster in Uttarakhand destroyed over 250 villages and killed approximately 6,000 people. The 2021 Chamoli disaster further devastated hydropower infrastructure, bridges, and settlements in the Rishiganga and Dhauliganga valleys.

The research team, led by Sidra of Jamia Millia Islamia in New Delhi, built their analysis on an unusually detailed foundation: a landslide inventory of 852 manually digitised landslide locations, compiled from Google Earth satellite imagery spanning 2012 to 2022. Each landslide was mapped as a polygon and then converted to presence points, which were split into a training set of 627 points and an independent testing set of 225. To give the machine learning classifiers the negative examples they need, the team generated 852 non-landslide points in stable areas, carefully avoiding a 150-metre buffer around known landslides to reduce spatial autocorrelation and enforcing a minimum spacing of 50 metres to prevent clustering.

Fourteen conditioning factors fed the models, each chosen for its documented influence on slope instability and its relevance to the geology, hydrology, and human geography of the corridor. Topographic variables—elevation, slope, aspect, and curvature—were derived from the ALOS PALSAR digital elevation model at 30-metre resolution. Hydrological factors included drainage density and distance to streams, while geological inputs comprised lithology, geology, distance to faults, and distance to lineaments drawn from Geological Survey of India data at 1:50,000 scale. Environmental and anthropogenic layers included land use and land cover, the Normalized Difference Vegetation Index derived from Sentinel-2 imagery, distance to roads from OpenStreetMap, and rainfall data from the India Meteorological Department.

The Frequency Ratio model, a classical bivariate statistical technique, quantified the association between each factor class and landslide occurrence by dividing the percentage of landslide pixels in a class by the percentage of the study area that class occupies. Values above one signal elevated susceptibility. The results were telling: the steepest slopes, between 45 and 78.18 degrees, showed the highest frequency ratio of 1.36, while barren land registered a striking 3.76 and built-up areas 2.64. Bare soil and rock with sparse vegetation scored 3.62, and areas within 486 metres of streams scored 1.56. Intriguingly, susceptibility was highest at the lowest elevations, between 815 and 1,130 metres, likely reflecting the concentration of road construction, settlements, and land-use modification at these altitudes.

Where the statistical model treats each factor in isolation, the machine learning approaches capture the nonlinear interplay that governs real slope failures. Random Forest, an ensemble method that grows hundreds of decision trees on bootstrap samples and aggregates their votes, was trained with 500 trees using the scikit-learn library. XGBoost, a gradient boosting algorithm that sequentially adds trees to correct the residuals of previous ones while penalising model complexity through a regularised objective function, was trained with 400 trees at a maximum depth of five and a learning rate of 0.05. Both models received the same balanced presence-absence dataset.

Performance was evaluated using the area under the receiver operating characteristic curve, a standard measure of a model’s ability to discriminate between landslide and non-landslide locations, alongside success rate curves and a rigorous spatial cross-validation scheme. The GroupKFold algorithm partitioned the data into spatial blocks so that all samples within a block went entirely to either training or testing, minimising spatial leakage. Random Forest achieved the highest ROC-AUC of 0.880, with a mean spatial cross-validation score of 0.888 plus or minus 0.021. XGBoost followed closely at 0.867, with a spatial mean of 0.876 plus or minus 0.026. The Frequency Ratio model trailed at 0.761, with a spatial mean of 0.805 plus or minus 0.025.

The spatial concentration of predicted risk was equally revealing. In the Random Forest map, the very high susceptibility class covered just 10.4 percent of the corridor yet contained 66.8 percent of all mapped landslides, while the very low class spanned 26.6 percent of the area with a negligible 0.3 percent of events. XGBoost performed similarly, concentrating 65.3 percent of landslides within 13.4 percent of the area. The success rate curves for both machine learning models rose steeply at the outset, indicating sharp discrimination between stable and unstable ground, whereas the Frequency Ratio curve climbed gradually, reflecting its weaker ability to delineate high-risk zones.

Feature importance analysis from both machine learning models converged on a consistent story. Land use and land cover emerged as the single most influential factor in Random Forest, contributing 0.1608 of the total importance, followed by elevation, aspect, NDVI, and slope—together accounting for roughly 52 percent of model importance. XGBoost likewise ranked land use and land cover first, with elevation and distance to streams next. The prominence of road proximity, which ranked sixth in Random Forest and carried weight comparable to slope itself, underscores how slope cutting, excavation, and drainage alteration during highway construction destabilise the very corridors that connect Himalayan communities.

The authors acknowledge limitations, including a temporal mismatch between the landslide inventory compiled from 2022 imagery and the 2024 land cover and vegetation datasets, and the absence of field verification of the susceptibility zones. Even so, the maps produced offer highway planners, engineers, and disaster managers a practical tool for prioritising slope stabilisation, guiding development, and targeting monitoring along one of India’s most critical and most fragile mountain routes. As climate extremes intensify and infrastructure expands across the Himalaya, the study makes a compelling case that data-driven machine learning, validated in space as well as in random splits, should become the default instrument for anticipating where the mountains will give way.

Subject of Research: Comparative landslide susceptibility mapping using Frequency Ratio, Random Forest, and XGBoost models along the NH-7 corridor in Chamoli district, Uttarakhand, India

Article Title: Landslide susceptibility mapping using Frequency Ratio, Random Forest, and XGBoost Models along NH-7 between Nandprayag and Vishnuprayag

Article References: Sidra, Dwivedi, D., Faruque, U., Raina, S. I., Sundaram, Siddiqui, L., Prakash, S., & Siddiqui, M. A. (2026). Landslide susceptibility mapping using Frequency Ratio, Random Forest, and XGBoost Models along NH-7 between Nandprayag and Vishnuprayag. Discover Geoscience, 4(1), Article 312. https://doi.org/10.1007/s44288-026-00683-3

Image Credits: AI Generated

DOI: 10.1007/s44288-026-00683-3

Keywords: landslide susceptibility, Random Forest, XGBoost, Frequency Ratio, machine learning, Himalaya, Uttarakhand, NH-7, Chamoli, GIS, remote sensing, disaster risk management

The study, led by Mary Braza of the University of Pittsburgh with Nadine McQuarrie, Claire Battistella, and Delores M. Robinson, published in the journal Solid Earth, tackles a long-standing debate in Himalayan geology. In compressional mountain belts like the Himalaya, the vertical uplift and eventual exposure of rocks at the surface, known as exhumation, can be driven either by sustained fault motion over ramps in the basal décollement, the great gently dipping thrust along which India slides beneath Asia, or by shorter pulses of rapid exhumation when faults break through the thrust stack out of the normal sequence. Which of these drivers dominates, and where, has remained contested across the orogen.

Western Nepal offered a natural laboratory precisely because its topography refuses to fit the standard templates. In central Nepal, the zone of steep slopes and high relief that defines the high Himalaya forms a single band near the Main Central thrust. In western Nepal, that band splits into two distinct zones: one roughly 10 kilometers south of the Main Central thrust, and another roughly 80 kilometers to the south, above the Dadeldhura klippe, a window of ancient thrust sheets preserved above younger rocks. The northern zone carries cooling ages younger than 6 million years, signaling vigorous recent exhumation. The southern zone, by contrast, preserves apatite fission track and (U-Th)/He ages of roughly 5 to 10 million years, which sharply limit how much young exhumation could have occurred there.

That mismatch matters because earlier geophysical and structural studies had proposed active ramps on the Main Himalayan thrust, the basal décollement, at locations between roughly 40 and 120 kilometers south of the Main Central thrust, including directly beneath the Dadeldhura klippe. But the youngest cooling ages in a thrust belt should always sit above the active ramp, because that is where rocks are lifted fastest through the closure isotherms of thermochronometers. Finding 5-to-10-million-year-old cooling ages above the proposed ramps undermined the idea of a long-lived active structure there, and demanded a rethink of the entire cross-sectional geometry.

The team built three alternative balanced cross-sections along the Simikot transect, each restoring the thrust belt to its undeformed state and specifying a different ramp geometry and fault-motion sequence. The first, based on a classic 2006 reconstruction, placed an approximately 7.5-kilometer-thick ramp at about 64 kilometers south of the Main Central thrust. The second, drawing on 2019 work, split the ramp into two smaller steps farther apart. The third was a newly revised section informed by updated geologic mapping, including a reinterpreted Lesser Himalayan duplex whose roof thrust extends from the Main Boundary thrust zone deep into the belt.

Each geometry was pushed through a coupled modeling pipeline. Flexural-kinematic models simulated how the crust bends and deforms as thrust sheets slide over ramps, producing predicted sediment thicknesses and accumulation rates in the Siwalik foreland basin at the mountain front. Those displacement fields then fed Pecube-D, a thermokinematic code that solves the advection-diffusion heat equations to predict cooling ages for four mineral systems: muscovite argon-40/argon-39, zircon and apatite (U-Th)/He, and apatite fission track. Finally, the best models were passed to a landscape evolution model that grew rivers and topography, allowing comparison of predicted elevations and normalized channel steepness indices with the modern, satellite-derived landscape.

The verdict was decisive. The first model managed to match only about 28 percent of the combined cooling-age and basin data, and forced implausibly fast shortening rates of 45 to 63 millimeters per year early on, followed by rates as low as 7 millimeters per year, while predicting basin ages millions of years older than the measured sediment record. The split-ramp second model fared better, reaching roughly 46 percent, but still demanded unrealistic velocity swings and failed to reproduce the young muscovite ages in the north. Both models failed in the same fundamental way: a ramp beneath the Dadeldhura klippe inevitably exhumes rocks there too fast, erasing the old cooling ages that are actually observed.

The revised third model transformed the picture. It placed the modern active ramp of the Main Himalayan thrust roughly 13 kilometers north of the Main Central thrust, directly beneath the youngest measured cooling ages, with a thickness of about 8 kilometers spanning the full Lesser Himalayan stratigraphy. It also added young out-of-sequence thrusting: roughly 12 kilometers of slip on a Ramgarh-Munsiari thrust fault at about 20 kilometers south of the Main Central thrust and 16 kilometers on a splay just 3 kilometers south of it, active between about 6 and 3 million years ago, plus smaller recent displacements beneath the Dadeldhura klippe and near the Main Boundary thrust. This configuration reproduced about 91 percent of the measured thermochronologic and basin constraints, a remarkable fit across 119 cooling ages and 33 sedimentary measurements.

The landscape simulations added an independent check. Motion over the northern ramp and the young out-of-sequence faults generated sharp increases in modeled topography and river steepness precisely where the real, high-relief Himalaya rises in western Nepal, while narrow bands of elevated channel steepness above the Dadeldhura klippe pointed to very recent, low-magnitude uplift on a fault splay cutting the upper Lesser Himalayan units at roughly 80 to 60 kilometers south of the Main Central thrust, a region co-located with modern microseismicity. The modeling also clarified that high topography in thrust belts cannot simply persist passively; it must be actively maintained by fault-driven uplift and then translated southward as erosion grinds it down.

Beyond resolving a regional puzzle, the findings carry broader weight for how scientists read mountain belts and assess seismic hazard. Exhumation rates in western Nepal turn out to be dominated by the size of the active ramp rather than the shortening rate, with brief bursts exceeding 9 millimeters per year during out-of-sequence faulting averaging to long-term rates near 3 millimeters per year. Ramps segment the décollement and may influence where earthquakes nucleate and propagate, so placing them correctly is a matter of practical importance. The study’s message is clear: only models that honor the full record, bedrock cooling ages, detrital signals in the foreland basin, and the shape of the modern landscape, can be trusted to reveal the hidden architecture of the world’s largest collision zone.

Subject of Research: Structural controls on exhumation in the western Nepal Himalaya

Article Title: Structural drivers of exhumation in compressional orogens: examples from western Nepal

Article References: Braza, M., McQuarrie, N., Battistella, C., & Robinson, D. M. (2026). Structural drivers of exhumation in compressional orogens: examples from western Nepal. Solid Earth, 17(8), 1011-1034. https://doi.org/10.5194/se-17-1011-2026

Image Credits: AI Generated

DOI: 10.5194/se-17-1011-2026

Keywords: Himalaya, western Nepal, exhumation, thermokinematic modeling, Main Himalayan thrust, out-of-sequence thrusting, thermochronology, landscape evolution, tectonics, fold-thrust belt, décollement ramp, Solid Earth

The methodology behind the indicator is deliberately simple but technically pointed. The authors divided the globe into 5-degree by 5-degree latitude–longitude cells and, for each cell, summed the country-level World Risk Index (WRI) values attached to every qualifying disaster event. To qualify, an event had to have caused at least 50 reported fatalities, a screening threshold the researchers chose because mortality data are more consistently reported across countries than other EM-DAT variables. Indicators such as the number of people affected or the economic damage of a disaster vary widely in how nations define and estimate them, making cross-country comparison unreliable. Deaths, grim as the metric is, travel better across borders. Applying the threshold to EM-DAT’s 3,217 target events from the study period yielded 344 major disasters: floods dominated the sample at 58.1 percent of events, followed by storms at 22.7 percent, earthquakes at 9.9 percent and mass movements at 9.3 percent.

One of the study’s central technical challenges was spatial standardization. EM-DAT records many events only through textual descriptions of affected places, such as city and village names, and only earthquakes come with precise coordinates for every entry. Floods and storms can sweep across broad, discontinuous areas, complicating any attempt at grid-based aggregation. The researchers therefore applied a rule-guided manual procedure that converts each event into a single representative location anchor, constrained by the geography documented in the original record. When an event spanned multiple administrative units, one anchor was assigned to a single grid cell to avoid double-counting. The authors acknowledge that this point-based approach compresses the spatial footprints of geographically extensive hazards, and they tested its influence directly using the 2022 Pakistan floods, whose EM-DAT description supports locations across nine grid cells.

The choice of the World Risk Index as the contextual weight is central to the design. Published annually since 2011, the WRI expresses risk as the geometric mean of exposure and vulnerability, combining population exposed to hazards with measures of susceptibility, lack of coping capacities and lack of adaptive capacities. By multiplying recent event recurrence with this structural risk context, ARI captures something neither component reveals alone. In diagnostic comparisons across the 153 non-zero cells, ARI correlated moderately with an equivalent grid-level WRI (Spearman rank correlation of 0.698) and with an unweighted count of major events (0.697), but the overlap among the top-priority sets was far lower, with Jaccard similarities as low as 0.148 against the WRI-only reference. In other words, layering a decade of actual catastrophic events onto structural vulnerability reorganizes the upper tier of global priorities in a way that neither index achieves on its own.

The resulting hotspot geography is dominated by Asia. High ARI values concentrate around the Himalayas and along the coastal belts of South and Southeast Asia, where repeated major floods, tropical cyclones and earthquakes have accumulated within a single decade. Many of these clusters extend across national borders, reflecting the transboundary nature of riverine floods and cyclones. The study’s regional zoom-in illustrates the added value of the grid approach: in areas spanning multiple countries, the country-level WRI alone mainly reflects national background differences, while ARI differentiates neighbouring cells according to where events actually accumulated. Within large countries such as China, where the WRI is spatially uniform by construction, ARI still reveals marked subnational heterogeneity, pinpointing internal hotspots that national indices cannot see.

Perhaps the most compelling validation is temporal. The researchers deliberately excluded 2024 and 2025 from the indicator’s construction and then asked where major Asian disasters recorded in those two years actually occurred. Of 44 qualifying events, 24, or 57.1 percent, fell within the predefined global Top-14 ARI cells, compared with only 23.8 percent for the WRI-only reference. Under a strict equal-grid benchmark in an Asia-specific universe of 76 cells, the tie-inclusive Top-10 ARI set contained 22 of the 44 later events, an observed-to-expected ratio of 3.80 at the event level. The concentration was strongest in the most selective upper tail: the Top-5 ARI set captured later events at 5.53 times the equal-grid expectation, the highest ratio among all indicators tested. Unweighted event count became increasingly competitive as the retained set broadened, suggesting that the WRI weighting matters most when only a handful of regions can be prioritized.

Sensitivity analyses reinforce the robustness of the broad pattern. Varying the fatality threshold to 40, 60 or 70 deaths changed the number of non-zero cells, but 12 or 13 of the 14 baseline hotspots remained in the alternative Top-14 sets, with Spearman correlations to the baseline ordering between 0.880 and 0.907. Refining the grid from 5 degrees to 2.5 degrees increased non-zero cells from 153 to 218, yet 85.7 percent of the baseline hotspot cells contained at least one high-ARI finer-resolution child, indicating that refinement localizes the strongest signal within the broader hotspots rather than dissolving them. The Pakistan flood experiment showed that redistributing the event’s full WRI contribution of 26.45 equally across nine supported cells shifted individual ranks slightly but left the global Top-5, Top-10, Top-14 and Top-20 sets unchanged.

Beyond the mapping exercise, the study delivers a substantial practical resource: the 3H Dataset, a compilation of standardized high-resolution remote-sensing imagery assembled for the 14 hotspot grids. High-resolution, sub-metre visible-spectrum imagery is costly and unevenly available, particularly in developing countries where open-data coverage tends to focus on large cities or specific crisis events. Drawing primarily on the RAMP Building Footprint Training Dataset and the DigitalGlobe/Maxar Open Data Program, and acquiring supplementary imagery for two data-scarce grids, the researchers compiled 7,583,094 standardized TIF image tiles of 256 by 256 pixels each, with spatial resolutions ranging from 0.27 to 0.54 metres. All scenes were quality-screened, tiled and annotated with hotspot-grid identifiers, creating a traceable resource for building damage detection, exposure mapping and post-disaster recovery monitoring.

The authors are candid about the framework’s limits. Mortality-based screening emphasizes sudden-onset, high-fatality disasters and underrepresents events dominated by economic losses, cascading impacts or slow-onset processes such as drought. The single-anchor allocation simplifies the footprints of extensive hazards, and country-level WRI values cannot capture within-country variation or rapidly changing local conditions. ARI is also a static, retrospective measure that treats all events in the decade equally, regardless of recency. The researchers point to three directions for future work: temporally weighted formulations that account for event recency and recovery periods, footprint-aware allocation and multidimensional severity measures, and integration of projected hazard changes and socio-economic scenarios to link retrospective accumulation with forward-looking assessment.

Even with those caveats, the study offers policymakers, humanitarian organizations and researchers something existing global indices do not: a transparent, grid-based picture of where catastrophic events have actually piled up most heavily in the recent past, validated against where they continued to strike afterward. As climate change intensifies hazards and populations in vulnerable regions keep growing, tools that can narrow a planet of potential disaster zones down to a defensible shortlist of priority areas are likely to become increasingly valuable. The 14 hotspots identified here, stretching from the Ganges basin to the Philippine archipelago, now come equipped not only with a quantitative rationale for their selection but with the high-resolution imagery needed to study them street by street.

Subject of Research: Global multi-hazard disaster risk assessment using the Emergency Events Database and the World Risk Index

Article Title: Global disaster risk assessment from Emergency Events Database (2013–2023)

Article References: Kong, Q., & Zhu, E. (2026). Global disaster risk assessment from Emergency Events Database (2013–2023). Natural Hazards and Earth System Sciences, 26(10), 4881-4910. https://doi.org/10.5194/nhess-26-4881-2026

Image Credits: AI Generated

DOI: 10.5194/nhess-26-4881-2026

Keywords: disaster risk, EM-DAT, World Risk Index, hotspots, floods, earthquakes, remote sensing, Himalaya, Southeast Asia, vulnerability, multi-hazard, 3H Dataset

The research team, led by Kiker Singh of Kurukshetra University with colleagues from Banaras Hindu University, Netaji Subhash University of Technology and NIIT University, focused on a section of the Lesser Himalaya that is tectonically restless and structurally fragmented. Three major regional thrust faults, the Giri, Chail and Juthogh thrusts, cut across the lower, middle and upper parts of the study area, fracturing the rock and priming slopes for failure. Elevations along the corridor range from roughly 1,425 to 1,700 metres above sea level, and the entire region sits within India’s seismic zone IV, meaning strong earthquake shaking is a realistic possibility. Add intense monsoon rainfall and aggressive road widening, and the ingredients for disaster are all present.

The stakes are not abstract. India accounts for nearly fifteen percent of all reported landslide occurrences worldwide and about thirty-one percent of associated fatalities, and almost 0.31 million square kilometres of landslide-prone territory lies within the Himalayan belt. Globally, landslides cause billions of dollars in economic losses and thousands of deaths each year, and roughly 50,000 people are estimated to have died from rock slope failures during the twentieth century alone. Along the Solan-Shimla stretch, road widening has repeatedly disturbed the natural equilibrium of slopes by removing the toe of hillsides, and since that work began, two lanes of the highway have been completely or partially blocked by landsliding between Solan and Kaithlighat.

To build their map, the researchers started with a landslide inventory: forty-eight landslide locations recorded along the highway using handheld GPS, cross-checked against high-resolution imagery in Google Earth and verified during extensive field campaigns. Each landslide was delineated as a polygon and then converted to point data within a geographic information system. This inventory served as the ground truth against which the model’s predictions could be tested, a step that many older susceptibility studies skipped and one that lends the new work considerable credibility.

The heart of the method is the Analytic Hierarchy Process, a multi-criteria decision-making framework introduced by mathematician Thomas Saaty. AHP asks experts to compare every causative factor against every other in pairs, assigning relative importance on a scale from one to nine. The resulting comparison matrix yields weights through eigenvector analysis, and a consistency check ensures the judgements are internally coherent. In this study the consistency ratio came out at 0.079, comfortably below the 0.10 threshold that signals unacceptable inconsistency. The team then combined the weighted factor layers in ArcGIS to produce a continuous susceptibility index for every patch of terrain along the corridor.

What distinguishes this study from earlier work in the same region is the breadth of factors considered. Previous research had examined slope, curvature, relative relief, aspect, geology, lithology, fault density, distance to road and drainage density. The new analysis adds five previously unexplored variables: the Topographic Wetness Index, which quantifies where water accumulates in soil; the Stream Power Index, which captures the erosive force of flowing water; land use and land cover; geomorphology; and lineament density, a measure of weakness zones in the rock. All terrain-derived layers were generated from a 30-metre Shuttle Radar Topography Mission digital elevation model, while geological and geomorphological data came from the Geological Survey of India’s Bhukosh portal.

The results are striking in their clarity. Slope emerged as the single most influential factor, receiving a weight of about 30.56 percent, with the densest landslide activity concentrated on slopes between 30 and 60 degrees. Distance to road ranked second at roughly 18.68 percent, confirming what engineers have long suspected: cutting into hillsides to widen the highway is a primary trigger of instability. Geology followed at about 9.5 percent, with the Mahi-Krol A and Chhaosa formations, composed of weathered limestone, shale and siltstone, accounting for a third and a quarter of all mapped landslides respectively. Concave slopes, which trap water and raise pore pressure, hosted 62.5 percent of the events, and highly dissected structural hills experienced nearly 73 percent of them. Aspect mattered too: northeast-facing slopes alone recorded about 29 percent of the landslides, likely because uneven solar exposure alters soil moisture dynamics.

When the weighted layers were merged and classified into five categories using natural breaks, the map revealed that 12.83 percent of the study area falls into the very high susceptibility class and a further 21.98 percent into the high class. Moderate susceptibility covers 26.58 percent, while low and very low zones account for 24.02 and 14.57 percent respectively. The validation statistics are what make the map genuinely useful rather than merely illustrative. The model achieved a ROC-AUC of 0.835, placing it in the ‘very good’ range, and a Precision-Recall AUC of 0.943, which is particularly meaningful because precision-recall analysis is less prone to overestimating performance when landslide events are rare compared with stable ground. An F1-score of 0.923 at an optimal threshold of 2.664 confirmed strong classification reliability.

Spatial validation sealed the case. Of the forty-eight mapped landslides, 72.92 percent occurred in the very high susceptibility zone and 25 percent in the high zone, with just one event, or 2.08 percent, in the moderate class and none at all in the low or very low categories. Normalized landslide density climbed steadily with susceptibility class, peaking at 0.821 in the very high zone. In practical terms, the model rarely cries wolf and almost never misses a real hazard, which is precisely the behaviour a planning authority needs when deciding where to install slope protection, drainage works or early warning systems.

The authors are candid about the limitations. An inventory of forty-eight events is modest, expert-driven weighting in AHP inevitably carries subjectivity, and the linear corridor format restricted any detailed rainfall threshold analysis. Future work, they note, will integrate rainfall data to build event-triggered models and apply machine learning approaches with expanded inventories to sharpen predictions further. Even so, the map already offers immediate value: engineers can prioritise slope stabilisation and cut-slope reinforcement along the highest-risk sections, adopt bio-engineering measures such as deep-rooted vegetation and grass turfing in moderate zones, and manage surface and subsurface drainage to keep water out of slope material. For a highway that thousands of people depend on every day, knowing precisely where the mountainside is most likely to move may prove as valuable as any retaining wall.

Subject of Research: Landslide susceptibility mapping along the Solan-Shimla National Highway-5 corridor in Himachal Pradesh, India, using GIS and the analytic hierarchy process

Article Title: Landslide susceptibility mapping along the Solan-Shimla highway using the analytic hierarchy process, Himachal Pradesh, India

Article References: Singh, K., Kumar, S., Khan, I., Kumar, H., & Rawat, A. (2026). Landslide susceptibility mapping along the Solan-Shimla highway using the analytic hierarchy process, Himachal Pradesh, India. Discover Geoscience, 4(1), Article 327. https://doi.org/10.1007/s44288-026-00694-0

Image Credits: AI Generated

DOI: 10.1007/s44288-026-00694-0

Keywords: landslide susceptibility, analytic hierarchy process, Himachal Pradesh, National Highway-5, Himalaya, GIS, remote sensing, Solan, Shimla, slope stability, ROC-AUC, geohazard mapping

The Jaunsari tribe occupies a distinctive place among the communities of the Indian Himalaya. The study describes them as distinct in culture, tradition and language, with a deep and practical attachment to the forest that surrounds their settlements. That attachment is not sentimental but functional: for centuries, the forest has served as the community’s pharmacy, its pharmacopoeia transmitted orally from elders to younger generations. Earlier ethnobotanical work in the region, including surveys of the Jaunsar-Bawar area published in the Indian Journal of Traditional Knowledge and Indian Forester, had already established the richness of this plant-based medical tradition. The new research builds on that foundation at a moment when the ecological and social conditions that sustained the tradition are shifting rapidly.

The methodological approach was deliberately direct. The researchers conducted a household survey across villages in Dehradun district using a structured questionnaire, gathering information on which plants are used, which plant parts are harvested, and which ailments they are deployed against. Household surveys of this kind are a standard instrument in ethnobotany, but their value depends on timing. Knowledge held by a small number of knowledgeable individuals, and passed on only by word of mouth, can disappear within a single generation if the chain of transmission breaks. The authors report with prior informed consent from all participants, and the study notes that anonymity and confidentiality were maintained throughout, in line with institutional and international ethical guidelines for research involving human participants.

The headline finding is the documentation of thirty-six medicinal plant species in active use for the treatment of fever, cough, indigestion, joint pain and skin ailments. That range of complaints is telling. These are not exotic conditions but the everyday health burdens of rural life, and the fact that a substantial proportion of them are still managed with plant remedies indicates that the traditional system remains a living practice rather than a historical curiosity. Each of the thirty-six species represents a distinct bundle of information: the correct season for collection, the appropriate preparation method, the dosage, and the specific symptom profile it addresses. Losing any one of these species, or the knowledge attached to it, removes a node from a network of practice built up over generations.

One of the study’s most technically interesting results concerns which parts of the plants are used. Leaves emerged as the most frequently harvested plant part, followed by roots and then seeds. This hierarchy matters for conservation in ways that are easy to underestimate. Harvesting leaves, in most cases, is a comparatively low-impact practice: a plant can often survive defoliation and regenerate, whereas the removal of roots frequently kills the individual, and seed collection can compromise the next generation’s recruitment. A tradition that favors leaves over roots is, in effect, a tradition with a built-in sustainability bias. The documentation of this pattern gives conservation planners a concrete starting point: protecting the species whose roots are used requires different interventions, such as cultivation programs, than protecting those whose leaves alone are gathered.

The study situates this documentation within the broader context of climate change, which the authors, citing the United Nations Department of Economic and Social Affairs, describe as the greatest challenge facing humanity today. In the Jaunsari context, climatic shifts are reported to be affecting the community in several specific ways: loss of biodiversity, a general decline in traditional ecological practices, and an increasing disconnection of the younger generation from rural life. Each of these pressures interacts with the others. As plant populations shift or decline in response to changing temperature and precipitation patterns, the raw material of the pharmacopoeia becomes scarcer. As young people migrate toward urban employment and formal schooling, the oral transmission chain weakens. The result is a double erosion, ecological and cultural, in which each accelerates the other.

This coupling of ecological and cultural loss is what gives the study its urgency. Ethnobotanists have long argued that indigenous knowledge systems are not merely archives of useful facts but adaptive systems that have co-evolved with local ecosystems. The Jaunsari pharmacopoeia encodes centuries of empirical observation about which plants grow where, when they are most potent, and how they should be prepared. When climate change alters the distribution and phenology of Himalayan flora, the knowledge system must adapt as well, and adaptation requires that the knowledge remain alive in practice. A tradition documented only in a paper archive cannot adjust to a shifting baseline; a tradition maintained by living practitioners can. This is why the authors emphasize documentation as a tool for sustainable preservation rather than as an end in itself.

The Dehradun district setting adds another layer of significance. The region’s flora has been catalogued since at least the 1920s, when Kanjilal and Gupta produced their forest flora of the Chakrata, Dehra Dun and Saharanpur forest divisions, and herbaceous flora surveys followed in subsequent decades. Against this long botanical record, contemporary ethnobotanical surveys can track not only what people know but what remains available to them. The study’s reference list draws on a substantial body of prior regional work, including ethnomedicinal surveys of the Garhwal Himalaya and studies of livelihood adaptation to climate variability in Jaunsar-Bawar, positioning the new data within a research tradition that stretches back decades. The continuity of that record is itself a scientific asset, allowing future researchers to compare present-day plant use with historical baselines.

The authors argue that the findings underscore the importance of climate change mitigation techniques that connect traditional medicinal knowledge systems with contemporary scientific insights, in order to guarantee the sustainable use and preservation of the medicinal flora. In practical terms, this integration can take several forms. Phytochemical analysis can validate and standardize traditional preparations, as has been done for other Himalayan remedies. Cultivation of high-demand species can relieve harvesting pressure on wild populations. Community-based conservation programs can compensate elders for teaching, and young people for learning, the knowledge that formal economic incentives currently undervalue. The study itself did not test these interventions, but its documentation provides the evidentiary basis on which such programs must be designed: you cannot conserve what you have not recorded.

There is also a wider lesson here for how science approaches indigenous knowledge at a moment of global environmental change. The Jaunsari case is local, but the pattern it illustrates is not. Across mountain regions worldwide, communities whose livelihoods depend most directly on ecosystem services are experiencing climate impacts first and hardest, while the knowledge systems they have refined over centuries are simultaneously being lost to cultural change. Studies like this one treat that knowledge as data worth collecting with rigor, using structured questionnaires, informed consent protocols, and quantitative documentation of plant-part usage patterns. The thirty-six species recorded in Dehradun district are, in that sense, more than a list. They are a snapshot of a working system of empirical medicine, taken at the moment when both the plants and the practice face their most serious test in generations. Whether the system adapts or dissolves will depend on decisions about conservation, education and health policy that extend far beyond the villages where the surveys were conducted, but the first step, capturing the knowledge before it fades, has now been taken.

Subject of Research: Documentation of indigenous medicinal plant knowledge of the Jaunsari tribe in Dehradun, India, in the context of climate change

Article Title: Documentation of indigenous knowledge of medicinal plants of Jaunsari’s living in Dehradun, India for sustainable preservation

Article References: Gangwar, P., & Karuna, M. S. (2026). Documentation of indigenous knowledge of medicinal plants of Jaunsari’s living in Dehradun, India for sustainable preservation. Plant Biosystems, 160(4), Article 218. https://doi.org/10.1007/s44473-026-00230-5

Image Credits: AI Generated

DOI: 10.1007/s44473-026-00230-5

Keywords: ethnobotany, medicinal plants, Jaunsari tribe, indigenous knowledge, climate change, Himalaya, Dehradun, biodiversity conservation, traditional medicine, Uttarakhand, household survey, sustainable preservation

The team, led by Ankita Ghoke, Priyom Roy and colleagues, combined three complementary lines of evidence to track the slope’s evolution. Eighteen cloud-free optical images from Indian Resourcesat-2 and 2A satellites, the European Sentinel-2 mission and the high-resolution Cartosat-3 sensor allowed them to map morphological changes between 2013 and 2025. They then applied multi-temporal interferometric synthetic aperture radar, or MT-InSAR, using the Small Baseline Subset technique on Sentinel-1 radar data to measure millimeter-scale ground deformation over four distinct time windows. Finally, field campaigns after the 2024 reactivation documented the structural and geological weaknesses that make the slope so prone to failure. Together, these datasets reveal a landslide that is not a single event but a slow-motion, retrogressive collapse that keeps climbing uphill and swallowing new ground.

The optical record tells a striking story of escalation. In March 2013, the unstable zone was a modest, elongated feature roughly 500 to 600 meters long. A major reactivation in June 2013, coinciding with the disastrous Kedarnath cloudburst that devastated much of Uttarakhand, stretched the slide to about 965 meters and carved the first prominent head scarp while blocking the Shambhu River. Between 2014 and 2016 the failure continued to retreat upslope, damming and re-damming the river. Around October 2018, an entirely new landslide initiated to the southwest, and by late 2019 its scar had merged with the main slide mass, enlarging the failure zone still further. By 2024, four separate head scarps were visible on the slope, evidence of multi-scarp retrogression still underway. The landslide now covers approximately 0.95 square kilometers, and imagery shows that several settlements in its path were destroyed as it grew.

The radar measurements add a crucial dimension: they show the slope moving before it visibly fails. In the first analysis phase, from October 2014 to October 2016, monitoring points on the flanking slopes recorded average line-of-sight velocities of roughly 40 millimeters per year, with deformation accelerating during the 2015 and 2016 monsoons before complete failure and scarp formation. In the second phase, a point tracking the nascent secondary landslide crept at 13.6 millimeters per year before a rapid acceleration culminating in the October 2018 collapse. Later phases recorded velocities between about 5 and 12 millimeters per year, each acceleration episode closely following periods of intense rainfall. The pattern is classic transient landslide behavior: stability, then primary creep, then accelerating motion, then failure, with monsoon precipitation acting as the trigger on a slope already predisposed to collapse by its geology.

That geology is unfavourable in almost every respect. The slope lies near the Main Central Thrust in the seismically active Lesser Himalaya, underlain by highly sheared talc schist, carbonaceous phyllite, quartzite and slate of the Lameri Formation, all draped in thick, loose colluvium. Fieldwork revealed foliation planes dipping at 45 to 50 degrees, parallel to the slope face, creating ready-made planar failure surfaces. The team also mapped transverse cracks, active groundwater seepage channels exuding along those cracks, and linear discontinuities that appear to be faults or joints. Particularly worrying is the village of Kuwari itself, which sits on a previously stabilized portion of the landslide body that field evidence now shows is bulging and deforming anew. A road constructed across the secondary landslide zone in early 2022 has further destabilized that section, a reminder that human activity can amplify an already fragile situation.

Meanwhile, the lake behind the debris dam keeps growing. The first impoundment formed in June 2013 and drained in 2015; a second formed in November 2016, partially drained in October 2017 and re-formed in 2018. From November 2021 onward the water body expanded relentlessly: roughly 885 meters long by April 2022, about 900 meters by October 2022, 1,050 meters by October 2023, and approximately 1,130 meters long and 100 meters wide by October 2024, the largest extent on record. Each enlargement reflects fresh debris choking the valley and raises the volume of water poised above downstream communities. History offers little comfort: previous research indicates that 80 to 90 percent of landslide-dammed lakes fail within their first year, and the Kuwari dam has already been partially breached by the river cutting a narrow channel through the debris.

To assess how stable that dam actually is, the researchers applied two geomorphic indices. The Morphological Obstruction Index, calculated at 3.82, places the dam in the uncertain evolution domain, while the Hydro-morphological Dam Stability Index, computed from the dam volume, upstream catchment area of about 2.46 square kilometers and channel gradient, falls squarely within the instability domain. The dam itself is a type III barrier in the classification of Costa and Schuster, meaning debris has moved both upstream and downstream of the valley and completely obstructs it. With an estimated height of around 40 meters and a surface area near the crest of roughly 10,000 square meters, the structure is an unconsolidated pile of landslide debris, not engineered material, and it is already being eroded by the river flowing through it.

The most dramatic part of the study is the flood simulation. Using HEC-RAS version 6.7, the team modeled a complete overtopping breach of the dam, drawing breach geometry from an empirical relationship derived from historical dam failures. Assuming the full 3.7 million cubic meter reservoir and a 17-meter breach height, the model predicts an average breach width of 48.3 meters forming in just 0.64 hours, less than forty minutes. The resulting flood would race down the narrow valley at a peak velocity of 15 meters per second near the breach, with overbank flooding extending about six kilometers downstream to Choting village, where water could rise up to 10 meters above normal levels and inundate agricultural terraces. Beyond Choting, the flow would remain channelized but still dangerous, reaching the villages of Melkhet, Deosari and Dewal at average velocities of 6 to 8 meters per second, fast enough to threaten anything along the riverbanks.

The authors are careful to note that the simulation is a predictive, end-member scenario that cannot yet be validated, and that its results carry uncertainties in breach width, formation time, erosion rates and dam material properties. But the broader message is hard to escape. The Kuwari landslide is an active, compound hazard in one of the world’s most landslide-prone countries, which a global analysis by Froude and Petley ranked highest in landslide fatalities. The study’s authors call for continuous monitoring, an early warning system for downstream communities, slope stabilization measures, and strict controls on construction across already scarped slopes. As climate change intensifies monsoon extremes across the Himalaya and infrastructure development pushes further into fragile terrain, the integrated approach demonstrated here, pairing satellite radar with optical imagery, field geology and hydrodynamic modeling, may become an essential template for spotting the next Kuwari before its lake bursts.

Subject of Research: Reactivation dynamics of the Kuwari landslide in Uttarakhand, India, and assessment of landslide lake outburst flood hazard using remote sensing and hydrodynamic modelling.

Article Title: Reactivation dynamics of the Kuwari landslide (Uttarakhand, India) and assessment of LLOF hazard using optical imagery, MT-InSAR and hydrodynamic modelling

Article References: Ghoke, A., Gaur, A., Roy, P., Jalan, P., Martha, T. R., & Das, I. C. (2026). Reactivation dynamics of the Kuwari landslide (Uttarakhand, India) and assessment of LLOF hazard using optical imagery, MT-InSAR and hydrodynamic modelling. Discover Geoscience, 4(1), Article 345. https://doi.org/10.1007/s44288-026-00717-w

Image Credits: AI Generated

DOI: 10.1007/s44288-026-00717-w

Keywords: Kuwari landslide, Uttarakhand, Himalaya, MT-InSAR, landslide-dammed lake, LLOF, HEC-RAS, Sentinel-1, dam breach, natural hazards, remote sensing, monsoon rainfall