The terrestrial sediment archives of northern Germany provide important insights into Late Pleistocene climate variability, but continuous and well‐dated records remain rare. In particular, the age and stratigraphic position of the Oerel and Glinde interstadials within the Early Weichselian to pleniglacial transition remain poorly constrained. These interstadials are commonly attributed to MIS 3, but this assignment is uncertain due to limited numerical dating, reliance on biostratigraphic correlation and the limitations of radiocarbon dating near its upper range. This study presents a new, high‐resolution multi‐proxy record (sedimentology, geochemistry, palynology) from two sediment cores at Lichtenberg, Lower Saxony, positioned directly next to a multi‐phased Middle Palaeolithic archaeological site and supported by a robust chronology based on new optically stimulated luminescence (OSL) ages. The record documents the vegetation and landscape dynamics from the late Eemian interglacial into the Weichselian pleniglacial, capturing major climatic transitions and the onset of cold, periglacial conditions. Within this sequence, we identify two previously unresolved climatic oscillations (Dangenstorf I and II), characterised by expansions of Betula nana ‐shrub tundra and increased organic accumulation. Based on pollen assemblages and OSL dating, these oscillations are correlated with the Oerel and Glinde interstadials, which are themselves stratigraphically positioned between the Odderade interstadial and the onset of the Weichselian pleniglacial. Furthermore, our OSL chronology places them firmly within the MIS 5a/4 transition. We assign them to the Greenland interstadials (GI) 20 and 19.2, respectively, rather than to MIS 3 as previously assumed. The Lichtenberg record thus provides a continuous, well‐dated archive that redefines the chronostratigraphy of the Early Weichselian to pleniglacial transition in northern Germany and strengthens its correlation with North Atlantic climate variability.

Pteropods are aragonitic holoplanktonic molluscs whose shells preserved in marine sediments provide a sensitive archive of past seawater carbonate saturation state, intermediate-water ventilation, and monsoon-driven oceanographic change. Here, we present a ∼1.2 Myr record of pteropod assemblages and preservation from sediment core NGHP-17 (1344 m water depth) in the Andaman Sea. Sixteen species were quantified across 180 samples, together with the Fragmentation Ratio (FR), Limacina Dissolution Index (LDX), epipelagic/mesopelagic (Epi/Meso) ratio, and selected planktonic and benthic foraminiferal abundances. Pteropod preservation is episodic, with pronounced preservation peaks during several glacial intervals. In contrast, generally poor and intermittent preservation during MIS 25–14 characterizes much of the Mid-Pleistocene Transition (MPT), although preservation windows occur during MIS 22, 18, and 16. This pre-MBDI decline was followed by the Mid-Brunhes Dissolution Interval (MBDI; MIS 13–11), within which a pronounced MIS 12 preservation peak demonstrates that unfavorable conditions were not continuous. An earlier preservation event occurs during MIS 26. The highest pteropod abundance is observed across MIS 6–5e, with a maximum at ∼125 ka and species maxima spanning ∼137–116 ka, indicating a transient Termination II/deglacial-to-early-interglacial preservation event. The significant inverse relationship between total pteropod abundance and FR supports a substantial preservation control on the record. Overall, the results indicate that pteropod preservation at NGHP-17 was strongly influenced by changes in bottom-water aragonite preservation conditions, within a broader framework of large-scale carbonate-system changes and regional hydrographic variability.

ABSTRACT This study combines historical records, satellite remote sensing data, and inverse modelling to investigate crack formation and calving front evolution of the Fimbul Ice Shelf, Antarctica. Historical maps from the 1950s, together with satellite imagery from 1970s–2026, are used to reconstruct calving‐front positions and track the evolution of crevasses. Principal stress fields obtained from higher‐order three‐dimensional inverse ice‐flow modelling support the fracture analysis. The ice shelf is strongly controlled by ice rises of varying size, which act as pinning points and promote fracture formation. The observed fractures consist predominantly of mode I cracks and wing cracks generated by mode I fractures subjected to shear stresses. Our results indicate that the present‐day calving front of Trolltunga is a remnant of the last major calving event in 1967, subsequently modified by collisions with drifting icebergs. We identify crack formation associated with ice flow past ice rises, occurring both immediately downstream of grounded features and at distances of approximately 8–18 km upstream of the present calving front. Fractures initiated at these ice rises play a key role in shaping the lateral margins of Trolltunga. A small grounded feature controls the shape of its eastern calving front, while the western margin is influenced by wing cracks developing within the western shear margin of Jutulstraumen and by mode I cracks originating near the Apollo Ice Rises. No significant change in fracture formation along the calving front is identified over the past five decades, providing no evidence of substantial present‐day structural change.

ABSTRACT While several marine and terrestrial sediment records provide the general climate boundary conditions for Iberia at a relatively low temporal resolution (often >100 years per sample) during the last several glacial cycles, few high‐resolution proxy records of continental climate have been developed from this region. Speleothems can provide millennia of seasonally to decadally resolved data on vegetation dynamics and hydroclimate variability. In this study we use stable carbon and oxygen isotopes and magnesium concentrations from a stalagmite collected from a cave in southern Portugal to better understand environmental variability during part of marine isotope stage (MIS) 4: ~69–64 ka. Stalagmite growth ceased as temperatures decreased just prior to the start of Heinrich stadial 6, an exceptionally cold period in the North Atlantic region, and growth did not resume after Heinrich event 6, suggesting dry and/or cold conditions in southern Portugal. Integration of these new data with previously published proxy records in western Europe reveals a complex and heterogenous spatial pattern of hydroclimate variability during MIS 4. For example, some regional stalagmite records, including those in Iberia, continued growing during Heinrich stadial 6, suggesting that a fine‐scale heterogeneity in precipitation patterns existed, or that infiltration pathways in these caves remained intact.

Abstract This study investigates the transient adjustment processes of the coupled ocean-atmosphere system to an abrupt change of sulphur dioxide emissions over North America and Europe by performing a set of ensemble simulations using a state-of-the art coupled climate model. The analysis focuses on the transient evolutions that allow an assessment of the sequence of cause and effect in circulation and thermal structure changes in the coupled ocean-atmosphere system. In response to an abrupt increase of sulphur dioxide emissions over North America and Europe, immediate changes in the atmosphere include reduced surface shortwave (SW) radiation and surface cooling in the northern hemisphere (NH), a southward displacement of the Intertropical Convergence Zone (ITCZ), and changes in the subtropical and subpolar jets in both hemispheres. In the ocean, the response of the Atlantic Meridional Overturning Circulation (AMOC) exhibits three timescales: 1) little AMOC change in the first 5 years, 2) rapid intensification of the AMOC, by about 2Sv, in years 6–15, and 3) a relatively steady AMOC state in years 16–30. We elucidate the distinct processes that govern these three stages. The large scale changes in global atmospheric circulation in the tropics and mid-latitudes are similar during the three stages of AMOC response, suggesting a weak feedback of AMOC changes on global atmospheric circulation. We conclude that a wider use of abrupt forcing experiments has great potential to further understanding of the impact of anthropogenic aerosols on the climate system.

Authigenic neodymium (Nd) isotope ratios have been extensively used over the past two decades to reconstruct past changes in the global ocean circulation. Nevertheless, their effectiveness as a water mass tracer is currently strongly disputed based on the potential for modification by non-conservative processes. This study compares authigenic Nd isotope records along a meridional transect in the Atlantic from sediment core sites chosen with careful consideration of sedimentary conditions that are conducive to preserving the original seawater signal, including high carbonate/low lithogenic detritus content, oxic conditions, and exclusion of sites near continental margins and regions impacted by local effects. Records over the last 1.5 Myr shows north-to-south patterns such that, with few exceptions, at each point in time, at nearly every location, Nd isotope ratios are lower toward the north, and higher toward the south. The same relationship is observed in seawater studies of the present-day Atlantic Meridional Overturning Circulation (AMOC) . Moreover, each site shows interglacial-glacial cyclicity, indicating systemic changes through time impacting the entire Atlantic basin. In addition, a comparison of Nd isotope records over the last glacial cycle reaffirms these same meridional trends at high temporal resolution. The coherence of the present-day circulation dynamics and past meridional Nd isotope patterns provide strong evidence that such transects can be used to track temporal changes in the water mass structure and circulation in the Atlantic, if the deep-sea cores are carefully chosen for favorable conditions to preserve the original seawater Nd isotope signature.

ABSTRACT The Moroccan coast provides valuable insight into how large‐scale climate and sea‐level variations have manifested themselves at the regional level. In this paper, we present new findings from shallow marine deposits in the Achakkar Basin near Tangier, Morocco. Data from these deposits record approximately 10 kyr of sea‐level highstand during the Last Interglacial period, deduced from a multidisciplinary study integrating sedimentological, micropalaeontological, palynological and ichnological records. Three facies associations are distinguished in the studied section. The lowest association is interpreted as deposits of migrating subtidal sand dune. The intermediate facies association reflects lower energy conditions in a more distal tempestite lower shoreface setting. The uppermost coarse‐grained interval indicates the transition to an upper shoreface to foreshore zone characterized by significantly increased energy. In addition, the micropalaeontological study shows an abundance of benthic foraminifera. The dominance of benthic forms suggests that the sedimentary environment was dominated by conditions suitable for the establishment of shallow‐water fauna, such as those found in shoreface environments. Species such as Ammonia beccarii , Elphidium crispum and Nonion faba , which are widespread in the lower levels of the sequence, are known for their increased ecological tolerance and ability to survive in unstable, low‐salinity environments. Furthermore, palynological and ichnological evidence supports the micropalaeontological and sedimentological findings, indicating a transition from nearshore to more distal environments, followed by a return to unstable, high‐energy nearshore conditions. These integrated data provide a significant contribution to the understanding of palaeoenvironmental studies within climate‐sensitive nearshore sediments.

Nature Communications, Published online: 19 October 2026; doi:10.1038/s41467-026-78423-x This study models the collapse of ice mélange in glacial fjords in response increasing submarine melt rates. It shows that perennial mélange today is vulnerable to collapse subject to increasing melt, potentially accelerating glacier mass loss.

Abstract Global glacier models typically rely on temperature-index mass balance calculations forced with aggregated meteorological data, limiting their ability to resolve short-term variability. Here, we develop a daily-resolution version of the Global Glacier Evolution Model (GloGEM), which we apply globally and compare against the monthly configuration. Our results show that the daily model more closely reproduces mass balance observations and provides a more detailed representation of glacier runoff. Accounting for short-term meteorological variability leads to systematically greater projected mass loss, with up to 25% greater loss in Svalbard. These differences arise from extreme melt days, precipitation-phase transitions, and surface-type feedbacks not fully represented at monthly resolution. Globally, the daily model yields ice volume losses of 26 ± 10% under low-emission and 47 ± 17% under high-emission scenarios over 2025-2100 (ensemble mean ± 1 StD), resulting in an additional 5-10% mass loss (20-33 mm of global mean sea-level rise) relative to the monthly implementation. Resolving high-frequency forcing, or accounting for its effects through parameterisations, is therefore essential for robust glacier projections, while its omission may systematically underestimate future glacier loss.

We compiled a database of 54 ancient shoreline markers from Iran, Oman, India, the Maldives and the Cocos Islands, dating to Marine Isotopic Stage 5. The records show that the same past ocean level can appear far above today's sea on rising coasts, near today's level on stable coasts, and many meters underwater where land is sinking. This helps separate true ocean change from land motion and improves future sea-level planning.

Geophysical Research Letters, Volume 53, Issue 20, 28 October 2026.

arXiv:2603.28704v3 Announce Type: replace Abstract: Understanding how fast atmospheric variability shapes slow climate variability and sensitivity remains a central challenge in Earth-system science. Recent advances in machine-learned (ML) atmospheric models have demonstrated remarkable skill on weather timescales, but their emergent behaviour in a fully coupled climate system remains largely unexplored. We present early results from a new hybrid modelling framework, in which the ACE2 ML atmospheric emulator is interactively coupled to the NEMO ocean model. We report on a set of 70-year coupled simulations (1950-2020 historical forcing and fixed-1950s control). These experiments represent, to our knowledge, one of the first multi-decadal integrations of a machine-learned atmosphere interacting with a full-depth dynamical ocean. Several historical and fixed-1950s control simulations from the fully dynamic global coupled climate model EC-Earth, which has the same ocean component used in ACE2-NEMO, are also considered for comparison. We assess the behaviour of the coupled system, with particular focus on low-frequency tropical variability and the climate response to greenhouse-gas forcing. Analysis of potentially emergent El Nino-like variability reveals realistic fast timescale air-sea coupling in the tropical Pacific, but the temporal variability is unrealistic, with very low amplitude oscillations; this appears to be due to weak atmospheric feedback in the tropical Pacific. The response to CO2 forcing shows initial agreement with EC-Earth3P but deviates due to reduced downward short-wave radiation in ACE2. These results provide a unique test of physical realism and generalisability for atmospheric emulators and evaluate the possible role of entirely machine-learned components in next-generation Earth system models.

The Northern Pine Island Ice Shelf plays an important role in slowing the flow of Antarctic ice into the ocean, but its changes have been poorly understood. Using satellite observations and environmental data collected over several decades, we found that this ice shelf has thinned, retreated, and lost contact with the seabed. These changes suggest the ice shelf is becoming less stable and highlights the importance of continued monitoring to anticipate potential disintegration.

This study reconstructs hydroclimatic changes from the earlier Late Pleistocene to the mid-Holocene using the 3.88 m BT-1 trench profile in Bayan Tohom Gobi, southeastern Mongolia (45°06′50.6″ N, 112°31′47.1″ E; ∼970 m a.s.l.). Accelerator mass spectrometry radiocarbon and optically stimulated luminescence dating provide discrete chronological constraints for grain-size and geochemical records interpreted using end-member modelling and principal component analysis. Organic-rich fine basin fill deposited during the earlier Late Pleistocene is separated by a sharp contact, potentially representing erosion or non-deposition, from coarse sedimentation within a broadly constrained later-glacial interval encompassing the Last Glacial Maximum. This coarse deposition involved aeolian input, basin-floor reworking, and episodic runoff, followed by finer sedimentation during the deglacial interval. Shallow-water to playa deposition, with its onset broadly bracketed around the Younger Dryas–Early Holocene transition, was accompanied by HCl-soluble enrichment consistent with carbonate accumulation, before renewed fine-grained shallow-water or wetland deposition during the Early to mid-Holocene. The glacier-decoupled BT-1 archive records changes in effective moisture and sediment mobility governed by precipitation, local runoff, potential groundwater input, and temperature-dependent evaporation. Comparison with Lake Ulaan and Orog Nuur highlights the influence of catchment water and sediment sources on the expression of regional hydroclimatic change in Mongolian closed basins.

Nature Geoscience, Published online: 20 October 2026; doi:10.1038/s41561-026-02117-8 Delayed bedrock uplift can allow rapid warming to trigger Antarctic Ice Sheet retreat before classical instability thresholds are reached, lowering critical warming thresholds by up to 0.5 K in East Antarctica, according to coupled model simulations.

arXiv:2610.27793v1 Announce Type: new Abstract: Uncertainty in the terrestrial carbon cycle remains a major constraint in climate projections, partly due to imperfect land surface representations in Earth system models. We propose a data-driven framework for high-resolution land-use reconstruction, treated as a semantic-segmentation task, that combines coarse-resolution scenario data and climate classification with static geophysical features to produce annual land-use maps at 1 km resolution. We introduce TerraNostra-Bench, a change-aware evaluation protocol that separates spatial downscaling from autoregressive temporal rollout and scores models on the rare land-use-change pixels rather than the dominant static background. Using it, we compare convolutional (U-Net) and transformer (Swin) backbones in autoregressive and non-autoregressive schemes against persistence, linear, and shallow convolutional baselines. The autoregressive prior is the most informative input yet accumulates error over long rollouts: autoregressive models lose 5.7-7.2 accuracy points over 13 years, against 2.2-2.8 for non-autoregressive models. This trade-off is governed by the autoregressive choice rather than the backbone. It can be tuned by an inference-time prior-trust threshold that requires no retraining and by the training-time weighting of reconstruction and change objectives; the two levers trace complementary frontiers, with calibration recovering up to 10 accuracy points while keeping change sensitivity above the persistence floor. Extrapolating beyond the training window, backward to 1955-1959 and forward to 2015-2019 under scenario forcing, the models retain most of their in-window skill and show the same trade-off, indicating it is intrinsic to the design. This work provides the methodological foundation and evaluation surface for a forthcoming temporally continuous land-use product.

Nature Geoscience, Published online: 16 October 2026; doi:10.1038/s41561-026-02123-w Forecasts of future Arctic Ocean sea ice loss can be substantially improved using Holocene proxy records by applying a model weighting approach.

arXiv:2610.23216v1 Announce Type: new Abstract: Earth system models underpin the evidence that informs net-zero policy, yet their energy demands are rarely considered when judging model quality. Using archived output from 159 models spanning five generations of the Coupled Model Intercomparison Project (CMIP), we show that model error has fallen at a constant rate of 13.5 per cent per doubling of compute since 1995. Halving CMIP6 error would require 28 times more compute. Improvements are uneven, with sea-level pressure and surface air temperature errors declining 3.5-fold and 2.7-fold since CMIP1, but precipitation error by only 7 per cent. Machine-learning weather models match or exceed operational forecast skill at a fraction of the runtime, but none publishes measured energy, and including training changes how they rank. We propose a skill-energy framework that combines an empirical Pareto front with a ranking score. These findings argue for standardised reporting of energy use and performance ahead of CMIP8.

Proceedings of the National Academy of Sciences, Volume 123, Issue 42, October 2026. SignificanceTipping points are abrupt and often irreversible shifts in complex systems, like the Earth’s climate or fluid flows. We introduce a machine learning framework to forecast future tipping points. While most prior works predict tipping points in ...

Abstract Snow ablation processes have been shown to exert a significant influence on the management of water resources, as well as on the prevention of drought and flood within basin. However, modeling these processes remains challenging due to complex physics and topography. This study develops an improved snow albedo (ISA) scheme for the Noah-MP land surface model that dynamically integrates snow age, depth, and temperature to refine the albedo decay process. We implemented this scheme in WRF model to simulate a severe snow accumulation and ablation event in the Upper Yellow River basin from January to March 2008. Compared to the default Noah-MP (CLASS/BATS) schemes, our improved scheme significantly enhances simulation accuracy. It reduces the mean bias for snow cover fraction (SCF) to 4.19% (from >20%), decreases the surface albedo RMSE by at least 34% (to 0.069), and substantially mitigates the air temperature cold bias, cutting its RMSE by approximately 33.6% (1.1°C). The results demonstrate that a more physical representation of snow albedo is key to improving the surface energy balance and reducing the pervasive cold bias in regional climate simulations over complex terrain. This work provides a validated and effective approach for enhancing snow process modeling in high-altitude regions.

When calving cliffs at the margins of Earth’s ice sheets exceed a critical height, runaway retreat may be initiated in a process called the marine ice cliff instability. However, sufficiently tall ice cliffs have rarely been observed. To address this gap, we measured the characteristics of 15,330 ice cliffs across Earth’s ice sheet margins using laser altimetry. We show that cliff height is highly correlated with mélange thickness, water depth, and crevassing. We found that tall cliffs calve more rapidly than parameterizations assume, but dynamic thinning and retreat into regions with low ice thickness gradients prevent runaway retreat. These observations suggest that model projections may underestimate subannual rates of cliff calving, but that the potential for instability will depend sensitively on additional quantities, not cliff height alone.

ABSTRACT Process‐based depositional models for perennially ice‐covered lakes (PICLs) are important for interpreting palaeolake deposits and their connection to past climate. PICLs are only stable under specific climate states, and unlike other ice‐free or seasonally ice‐covered lakes, PICL deposits generally have bimodal grain sizes: mud introduced by streams and glaciers and moderately sorted medium to coarse sand deposited as disseminated grains, mounds and ridges from the ice cover. Previous work implies that ice properties influence the mud‐to‐sand ratio and sand depositional style. Here we present multi‐year observations from Lake Joyce, Antarctica, demonstrating how ice cover properties influence deposition to help inform interpretations of climate from the sedimentary record. Lake Joyce's ice cover had smooth to rough textures with rougher ice trapping more sediment and forming multi‐grain sediment pods, and smooth ice containing scattered grains. Low‐albedo medium sand grains preferentially melted into the ice over higher albedo grains and those that are finer and coarser. Thicker pods slowed underlying ice melt, whereas thinner pods accelerated it. Grains migrated downwards and accumulated at an equilibrium depth which was shallower than the ice cover thickness. Sediment pods commonly contained inverse grading, resulting from deflation and preferential melting of finer grains into the ice. These observations help inform PICL deposit interpretations: pods in rough ice result in sand mounds and ridges, while scattered grains on smooth ice lead to disseminated grains. Deposits have a narrow grain‐size range due to the preferential melting of medium sand, and inverse grading in mounds mirrors that of pods. Our observations suggest that ice roughness and thickness control where and how much sediment accumulates, its vertical migration through the ice and depositional style. Temporal or spatial variations in ice properties may be preserved as stratigraphic or lateral facies changes. Because ice properties are climate‐dependent, stratigraphic variations in PICL facies may be used as palaeoclimate indicators if found on Earth and Mars.

Glaciers are melting worldwide, but direct measurements are spatially sparse. We developed a new model that combines mechanistic physical rules with machine learning to better estimate yearly gains and losses of ice mass across Europe. By learning local corrections to a physical melt model, the machine learning component improves accuracy and works well on glaciers not used for training. This will help produce more reliable glacier assessments and support future climate studies.

Abstract During the last decade, Antarctic sea ice area has undergone dramatic change, and anomalies have become much larger and longer-lived relative to earlier in the satellite record. These changes in sea ice variance and persistence have led to the suggestion that a regime change may have occurred in the Antarctic system. Using simulations from a coupled Earth system model, we diagnose the factors that give rise to decadal-scale changes in the variance and persistence of Antarctic sea ice anomalies that arise from internal climate variability. The model simulates changes in these properties that are comparable to observed variations. In decades with highly variable and longer-lived (persistent) sea ice anomalies, larger anomalies in total Southern Hemisphere ice area result from a reduced regional compensation of sea ice variations. This is associated with differences in the Southern Annular Mode (SAM) spatial structure and consequent differences in wind-driven ice area anomalies. The higher persistence of sea ice anomalies in these decades results in part from changes in the Weddell Sea where ice variations are more strongly driven by subsurface ocean temperature anomalies. These simulated decadal-scale differences arise from internal variability, suggesting that anthropogenic forcing is not a necessary condition for the increased variance and persistence of Antarctic sea ice anomalies observed in recent decades.

Abstract The increasing interannual variability of Antarctic sea ice near West Antarctica underscores the need to understand how El Niño–Southern Oscillation (ENSO) teleconnections may shape future sea ice changes. Using simulations from the Coupled Model Intercomparison Project Phase 6, this study investigates the projected impacts of near-future ENSO variability on Antarctic sea ice. ENSO teleconnections are found to primarily influence Antarctic sea ice from austral winter through spring, and this influence is projected to strengthen in the near future. We show that the strengthened teleconnection is driven mainly by enhanced tropical forcing with the structural changes in tropical sea surface temperatures and precipitation, with a smaller contribution from changes in the midlatitude jet waveguide. These mechanisms are further supported by idealized nudging experiment using a dry dynamic core model. The strengthened teleconnection deepens the ENSO-induced mid-tropospheric circulation anomaly near West Antarctica, enhancing warm-air advection and the resulting sea ice anomalies. Overall, these changes are projected to increase Antarctic sea ice variability in the near future, highlighting the importance of understanding how evolving ENSO teleconnections may modulate Antarctic climate variability.

Abstract A more accurate radiocarbon dating of short-period climate changes in the Holocene requires to take into account short-term changes in variation of the 14 C concentration in the atmosphere, which could lead to a dating error of up to 300 years. To eliminate this potential problem, Bronk Ramsey’s model was used, based on a comparison of the radiocarbon dates with the calibration curve. Using this model, a corrected (i.e., modeled) age was established for the entire series of previously determined radiocarbon dates obtained from the Nikolsko-Lyutinskoye, Shirinsky Mokh, and Sambalskoye raised bog sections in northwestern European Russia. The deviations of the average annual temperature from its modern value (∆T), presented in previous publications as graphs, were converted into digital form. Using this model, a refined calendar chronology was established. The chronology of short-period climate changes in the Holocene demonstrates the synchronicity of short-term climate fluctuations. In most cases, it agrees well with the chronology obtained in the study of natural objects (lakes, bogs, marine sediments, and polar glaciers), as well as with the advances of mountain glaciers. The chronology of short-period climate changes is compared to the changes in solar activity, established by using concentrations of cosmogenic isotopes: 14 C in tree rings, and 10 Be in polar glaciers. The synchronicity of short-term climate fluctuations and variations of solar activity in the Holocene has been established, and it suggests that changes of solar activity are one of the main causes of short-period climate fluctuations in the Holocene.

Proceedings of the National Academy of Sciences, Volume 123, Issue 42, October 2026. SignificanceTrustworthy uncertainty quantification is key for responsible data science and AI (including data-driven decision-making). Traditional statistical inference methods require specifying an underlying generative model and are often not robust to ...

Mountain snow at high elevations is changing as warming causes more melting and water to linger within it. We use satellite radar images from 2015 to 2024 to track when melting starts and when snow refreezes across the Wrangell and St. Elias Mountains on the border of Alaska and Yukon. We find that melting now reaches nearly all elevations, and water often remains trapped well into fall. Only the highest areas stay fully dry, showing widespread impacts of warming on mountain snowpacks.

Ground temperature data is vital for permafrost and climate research yet data is often fragmented. We created a standardized collection of ground temperatures from over 900 sites across Canada. From 42 published and unpublished sources, we manually verified, cleaned and standardized data with a new software tool. This dataset supports permafrost research on a nationwide scale and can help improve models by acting as a reliable benchmark.

arXiv:2610.19885v1 Announce Type: new Abstract: Inferring climate fields at higher spatio-temporal resolutions from coarse representations is essential for accessing socioeconomic and environmental impacts. The downscaling of modelled climate fields faces a trade-off between data storage constraints, simulation efficiency, and resolution requirements. While many methods for spatial downscaling have been proposed, joint spatio-temporal downscaling has received comparable little attention. When deriving, for example, daily fields from climate datasets that are available only as monthly averages at coarser spatial resolution, it is highly challenging to faithfully reproduce extreme event statistics and spatiotemporal consistency. To address this task, we introduce a generative machine learning approach, using a diffusion model to jointly downscale monthly temperature and precipitation fields into temporally consistent daily fields at higher spatial resolution. We demonstrate that our approach can generate high-dimensional multivariate climate fields at daily temporal and 0.25{\deg} spatial resolution from monthly input fields at 1{\deg} spatial resolution, exhibiting coherent temporal dynamics and physically plausible spatial structures. In particular, by incorporating a conditional guidance technique, we show that a pre-trained diffusion model can be directly leveraged to efficiently generate daily high-resolution fields conditioned on monthly inputs, without requiring additional fine-tuning or retraining. Our probabilistic approach thus provides an efficient and generalizable framework for spatio-temporal climate downscaling, enabling flexible climate fields generation for climate impact assessments and digital Earth simulator constructions.

Abstract Tornadoes cause extreme, localized loss of life and property. Therefore, it is of societal importance to understand how tornado activity may be influenced by climate and climate change. Because climate models do not explicitly resolve tornadoes, tornado-relevant environments are often used to detect tornado trends and project future changes in tornado activity with climate model output. However, when making long-term tornado projections, it is essential to assess how biases in these tornado-relevant environments within climate models affect tornado count estimates. In this study, we developed a tornado count model that uses large-scale monthly environments to predict monthly tornado counts, which is trained with observational environments and reports but is calibrated for use with Coupled Model Intercomparison Project (CMIP) version 6 environments. We found that convective precipitation for most CMIP6 models is too low, and the environment-based tornado count model has seasonal cycle errors, e.g., too high tornado activity in summer. After correcting for these issues, the CMIP6 tornado count model well represents the historical seasonal cycle and spatial distribution of U.S. tornado activity in terms of average and extreme (90th percentile) annual tornado counts. In addition, we estimated recent increases in tornado activity, particularly over the lower Mississippi River Valley, consistent with observational reports. While the spatial pattern of changes in tornado activity from CMIP6 aligns broadly with reports and reanalysis environments, the model with CMIP6 environments estimates a smaller magnitude in recent trends, suggesting most of recent upward trends reflect internal variability. A modest portion of recent upward tornado trends may be driven by external forcing of convective environments.

Science Advances, Volume 12, Issue 42, October 2026.

Glaciers on the Tibetan Plateau are losing mass, yet their turbulent heat exchange with the atmosphere remains uncertain. Using direct observations from the Dunde Glacier (May–October 2023), we analyzed turbulent flux variations at the glacier surface and evaluated five turbulent flux methods. We also showed that during a humid heatwave, turbulent heat input rose sharply and accelerated melt, highlighting the need to improve models under extreme conditions.

The glacier temporarily blocked the Lhende Khola River, then released a destructive debris flow that travelled through the Bhote Koshi and Trishuli corridors into the districts of Rasuwa, Nuwakot, and Dhading in Nepal. This paper relies on secondary data from government disaster management bureaucracies and reporting from the United Nations, multilateral banks, and print and broadcast media. It focuses on the events surrounding the disaster, analysis of causal factors, casualties, damage to infrastructure and the economy, and response efforts by the government and international agencies. On August 29, 2026, the toll in Nepal was reported to be 675 dead with 2,478 missing and over 93,000 affected, while Tibet reported 7 dead and 560 missing persons. The calamity destroyed about 430 megawatts of hydropower capacity, 19 bridges, and 40 kilometres of road, while obliterating the vital Gyirong border crossing for Nepal-China trade. This paper discusses the event in the context of available scientific literature on glacial lake outburst floods (GLOFs) and ice–rock avalanches in High Mountain Asia, highlighting that the same Bhote Koshi corridor faced a similar transboundary glacial lake outburst event in 2016, which points to an issue of recurrent and unmitigated risk. The study finds there are still gaps in cross-border early detection, information sharing across borders, and infrastructure management based on climate risks, and gives some suggestions for governance and engineering. This study provides an early and organised description of a quickly changing disaster which will be useful for investigators, practitioners in civil defence planning, and policy-makers, but it also notes the limits of this investigation using secondary sources. Keywords: glacier collapse; debris flow; glacial lake outburst flood (GLOF); Himalayan cryospheric hazards; transboundary disaster governance; hydropower vulnerability; Nepal; Tibet Autonomous Region; climate change adaptation; Uttarakhand

Science Advances, Volume 12, Issue 43, October 2026.

arXiv:2512.22814v2 Announce Type: replace-cross Abstract: Accurate long-range weather forecasting remains a major challenge for AI models, both because errors accumulate over autoregressive rollouts and because reanalysis training datasets provide limited sampling of the slow modes of variability underpinning predictability. Most AI and all numerical weather prediction models are autoregressive, repeatedly applying short-lead forecasts to reach subseasonal-to-seasonal (S2S) or seasonal lead times, often causing instability and miscalibration. Long-timestep probabilistic models that generate long-range forecasts in a single step offer an alternative, but training on the 40-year reanalysis record leads to overfitting, suggesting orders of magnitude more data are required. We introduce long-range distillation, whereby a long-timestep probabilistic ``student'' model is trained to forecast directly at long-range using a huge synthetic dataset generated by an autoregressive ``teacher'' model. Using the Deep Learning Earth System Model (DLESyM) as teacher, we generate over 10,000 years of simulated climate to train student models across a range of timescales. In perfect-model experiments, the distilled models outperform climatology and approach their teacher's skill while replacing hundreds of autoregressive steps with a single timestep. In the real world, both a diffusion model student fine-tuned on ERA5 and a linear inverse model student with mixed DLESyM-ERA5 training achieve S2S forecast skill competitive with the ECMWF IFS ensemble, with synthetic pre-training providing improvements beyond ERA5-only training. This demonstrates the benefit of synthetic pre-training for both deep learning and classical data-driven methods, and represents the first demonstration that real-world skill can be improved by training on AI-generated synthetic data in volumes far exceeding the observational record.

This study suggests that the dominant periodicity of precipitation shifted from eccentricity-scale (~100–125-kyr) variability to 20-kyr precessional variability during the Late Pliocene, implying that the processes controlling the westward penetration of summer moisture into the basin varied under different background climate states. These findings provide new insight into how background climate conditions may shape the orbital-scale response of precipitation during warm periods.

Science Advances, Volume 12, Issue 42, October 2026.

Journal of Geophysical Research: Earth Surface, Volume 131, Issue 10, October 2026.

Improving mountain snow prediction requires better representation of meteorological uncertainty. This study generates meteorological ensembles accounting for topography, gauge undercatch, and large-scale atmospheric patterns. Driving a physics-based snow model calibrated using a Large Sample Emulator with these ensembles propagates forcing uncertainty into snow accumulation and melt, producing probablistic estimates for data assimilation and hydrological forecasting.

arXiv:2605.15470v3 Announce Type: replace-cross Abstract: Ocean dynamics are inherently chaotic, yet existing machine learning ocean models produce only deterministic forecasts. We introduce Njord, a probabilistic data-driven model for ocean forecasting, applicable to both global and regional domains. Njord combines a deep latent variable framework with a graph neural network architecture, enabling sampling each forecast step in a single forward pass. We apply Njord globally at 0.25{\deg} resolution and regionally to the Baltic Sea at 2 km resolution. To scale to these large ocean grids we introduce K-means cluster meshes that adapt to irregular sea surface geometry. Experiments demonstrate strong performance on both domains compared to deterministic machine learning baselines, while also providing uncertainty estimates from the sampled ensemble forecasts. On the global OceanBench benchmark, Njord achieves the lowest errors on average across upper-ocean variables when evaluated against real-world observations, with the largest improvements in surface temperature prediction.

Cold-arid mountain communities depend heavily on seasonal snow and glacier melt, yet climate-driven changes in cryospheric processes are altering the timing and reliability of water availability. This review critically examines ice stupa technology as an artificial glacier system designed to store winter streamflow as ice and release meltwater during the critical spring irrigation period. The review synthesises evidence on the historical development, physical principles, engineering design, hydrological performance, agricultural relevance, ecological implications, socio-economic dimensions, economics, policy relevance and future research needs of ice stupas. The available evidence indicates that ice-stupa performance is governed primarily by meteorological conditions, surface-energy exchange, hydraulic head and cone geometry. Their compact vertical form can delay ablation, while gravity-fed systems enable operation without pumping energy. Recent advances in automated fountain scheduling substantially improve water-use efficiency and reduce operational failures associated with pipe freezing. However, storage remains primarily village-scale, hydrological performance varies considerably between years and economic, ecological and long-term socio-economic outcomes remain insufficiently quantified. Ice stupas therefore function most appropriately as supplementary seasonal water-storage systems rather than substitutes for natural glaciers or large-scale reservoirs. Their effectiveness depends on careful siting, appropriate engineering, reliable monitoring, community participation and integration with broader catchment-management strategies. Future research should prioritise long-term hydrological monitoring, standardised performance assessment, economic evaluation, ecological measurement and climate-sensitive scaling frameworks.

Abstract Traditional methods for improving tropical cyclone (TC) intensity from climate model outputs or projections have primarily relied on either dynamical or statistical downscaling. With recent advances in deep learning (DL) techniques, it is of interest to examine how DL can provide an alternative approach for enhancing TC intensity and structure retrieval from climate data. Using a common DL architecture based on convolutional neural networks (CNN) and a set of key environmental features relevant to TCs, we show that TC intensity and size can be effectively retrieved from climate reanalysis data without requiring super-resolution enhancement as in previous studies, even when applied to coarse-resolution climate data. This approach allows for retrieving TC intensity metrics and size that are dynamically constrained by the data, rather than estimating these quantities independently. Our results highlight that TC intensity and size are governed not only by TC internal processes but also by local environments during TC development for which DL models can learn and capture. The performance of our DL model depends on several factors such as season, the stage of TC development, or ocean basins, with root-mean-square errors ranging from ≈8–10 ms −1 for the maximum 10-m wind, ≈10–13 hPa for minimum central pressure, and ≈13–21 km for the radius of the maximum wind. Although these errors are better than any direct vortex detection or statistical downscaling methods applied to the same data, their wide ranges also suggest that a 0.5°-resolution climate data may contain limited TC information for DL models to learn from, regardless of model optimizations or architectures. Possible improvements and challenges in addressing the lack of fine-scale TC information in coarse-resolution climate reanalysis datasets are discussed.

China has made considerable efforts in its ice-drilling technology since the 1970s; however, a comprehensive overview of its achievements is still lacking. This paper reviews China’s progress in ice-drilling technology over the past five decades. First, drilling projects in the Third Pole and polar regions are summarized. In the Third Pole region, more than 120 boreholes have been drilled, with the deepest reaching 324 m and the highest on the summit of Mount Everest. In Antarctica, Chinese drilling activities have mainly been concentrated in the area near Great Wall Station and along the traverse from Zhongshan Station to Dome A. Key achievements include a >800 m deep ice-core borehole at Dome A, a 3413 m hot-water borehole at Qilin Subglacial Lake (QSL), and the recovery of two subglacial bedrock cores. Second, the mechanical and thermal ice drills developed in China are reviewed. Before the 2010s, China had only shallow-drilling systems. Since then, a series of ice drills with deep-drilling capabilities have been developed. Future drilling priorities include resuming deep coring at Dome A, exploring QSL, sampling the subglacial bedrock of the Gamburtsev Subglacial Mountains (GSMs), and expanding activities to West Antarctica. Simultaneously, existing drills that are still at the laboratory or early field-testing phase require further refinement and validation, and the development of new drills, such as a deep ice-core drill, a thermal ice coring drill, and a flexible-tube air reverse-circulation drill, should be actively pursued.

The southeastern Tibetan Plateau (SETP), a critical component of the “Asian Water Tower”, hosts a large concentration of maritime temperate glaciers that are highly sensitive to climate change. However, persistent cloud cover, seasonal snow, and supraglacial debris complicate consistent long-term glacier mapping, while the use of static glacier boundaries may introduce temporal mismatches in altimetry-based glacier mass-change estimates. In this study, we developed a multi-source data-fusion framework on Google Earth Engine (GEE) to reconstruct glacier-area changes in the SETP from 2000 to 2025 at five-year intervals and 30 m spatial resolution. Random Forest classifiers were used to integrate Landsat and Sentinel-2 optical imagery, Sentinel-1 synthetic-aperture-radar data, and NASADEM-derived topographic information, supporting glacier delineation under complex surface and atmospheric conditions. The reconstructed glacier area decreased from 8,083 ± 664 km2 in 2000 to 6,228 ± 572 km2 in 2025, corresponding to a cumulative loss of 22.9 ± 3.6% and a mean loss rate of 80.51 ± 10.51 km2 yr−1. The mean area-loss rates were 51.31 ± 6.51 km2 yr−1 during 2000–2010 and 92.99 ± 15.55 km2 yr−1 during 2010–2025, indicating a higher average loss rate in the later period. By combining satellite-altimetry-derived elevation changes from ICESat, CryoSat-2, and ICESat-2 with the reconstructed time-varying glacier areas, we estimated a mean glacier mass-change rate of −6.51 ± 1.22 Gt yr−1 during 2003–2022. Compared with using a fixed 2003 glacier area, the time-varying-area calculation reduced the estimated mass-loss magnitude by 0.63 Gt yr−1, equivalent to 8.8% of the fixed-area estimate. These results indicate that incorporating time-varying glacier areas can reduce temporal mismatch in altimetry-based mass-change estimation and improve the temporal consistency of regional glacier mass-change assessments.

Abstract The climate system displays a seasonal cycle not only in its mean state, but also in predictability and variance. In particular, month-dependent autocorrelation functions (ACFs) reveal seasonal variations in persistence and are used routinely to characterize such phenomena as the spring predictability barrier in El Niño forecasts and the wintertime reemergence of sea surface temperature anomalies. However, detection of such behavior in climate data has often relied on heuristic procedures tailored to specific applications that are difficult to generalize to other applications. This paper proposes a general approach to detecting month-dependent ACFs based on Periodic Autoregressive (PAR) models, which extend standard autoregressive models by allowing parameters to vary with month. Specifically, a likelihood ratio test is used to determine if a PAR provides a significantly better fit than a stationary model with month-invariant parameters. Externally forced variability is incorporated through additional forcing terms. The resulting model provides a self-consistent representation of both internal and forced variability. Applying the test to observational records of sea surface temperature reveals distinct regions with statistically significant month-dependence in their ACFs. These include the tropical Pacific, where the method identifies the canonical spring predictability barrier, and several midlatitude regions displaying the statistical signature of reemergence. The test also identifies a previously undocumented seasonal rebound in the ACF over the Guinea Dome region of the northeastern tropical Atlantic. It is further shown that the widely used Seasonal Autoregressive (SAR) model cannot capture in a parsimonious manner the spring predictability barrier or reemergence.

Science Advances, Volume 12, Issue 42, October 2026.

Science Advances, Volume 12, Issue 42, October 2026.

arXiv:2610.19389v1 Announce Type: new Abstract: Under anthropogenic greenhouse forcing, climate models project a decrease in year-to-year temperature variability in the mid-to-high latitudes and an increase in the tropics and subtropics. Here, we examine year-to-year temperature variability in the CESM Large Ensemble in concert with an idealized stochastic moist energy balance model (SMEBM) with effective surface heat capacity and local feedback parameter specified based on CESM. We find that the zonal-mean CESM variability change pattern largely persists after filtering out the impact of El Ni\~no. The SMEBM largely captures this change pattern. In the SMEBM, changes in the mid-to-high latitudes are driven by changes in the effective heat capacity due to sea ice retreat, combined with changes in anomalous moist static energy convergence modulated by the polar-amplified time-mean temperature change. The CESM local feedback parameter evolves to be less stabilizing in the low latitudes, which causes tropical and subtropical variability to increase in the SMEBM.

arXiv:2610.17896v2 Announce Type: replace Abstract: Rainfall forecasts over the Western Himalaya remain poor, and the relative contributions of the cumulus and microphysics parameterizations to that error are not well constrained.

Science, Volume 394, Issue 6821, Page 311-315, October 2026.

Geophysical Research Letters, Volume 53, Issue 20, 28 October 2026.

Journal of Geophysical Research: Oceans, Volume 131, Issue 10, October 2026.

Journal of Geophysical Research: Atmospheres, Volume 131, Issue 20, 28 October 2026.

Preprint under review for ESSD (discussion: open, 0 comments) Flash floods develop quickly, but the records needed to study them are scattered across many sources. We created TORRENT by linking flood reports, rainfall, river flow, drainage-area information, and field measurements across the contiguous United States. It contains 42 466 reported events and 806 ready-to-use case packages, including 21 carefully checked cases. Its open, machine-readable design will help people and artificial intelligence tools study floods and compare prediction methods.

arXiv:2610.21012v1 Announce Type: new Abstract: In the recent AI Model Intercomparison Project (AIMIP) benchmark, the ACE2.1-ERA5 emulator was shown to underestimate the true near-surface warming trend both during the training period and the out-of-sample extrapolation period. In this work, we present M-ACE2S (Masked-ACE2S), a modified version of the stochastic ACE2S model, which uses variable masking during training and reduces the reliance of M-ACE2S on any single input channel. We show that M-ACE2S improves long-term near-surface air temperature prediction. We also present a novel data normalization strategy called "global mean removal", which, combined with masking, further improves the near-surface temperature forecasts. We show that M-ACE2S outperforms the ACE2S baseline on in-sample data for ERA5 while improving training stability and generalizability of near-surface temperature to unseen forcing when paired with global mean removal.

arXiv:2610.22286v1 Announce Type: cross Abstract: In partially-observed dynamical systems, distinct hidden mechanisms can produce similar observable summary statistics while explaining individual events differently. We introduce causal fingerprints: realization-dependent patterns that track how information about a hidden state is revealed through subsequent evolution. Within a prescribed dynamical model, we compare the filtering posterior, based on observations up to a given time, with the smoothing posterior, obtained after adding an expanding window of future observations. We quantify this retrospective information gain using Kullback-Leibler divergence and track its evolution as the future window of observational data expands. The fingerprint reveals both the strength and temporal reach of hidden influence, which can vary across events and their stages. Because the observed trajectory is held fixed while hidden-state inference depends on the candidate model, fingerprints reveal how competing dynamical mechanisms explain the same observations even when selected statistics agree. In linear Gaussian systems, we derive a characteristic fingerprint timescale for the temporal reach of hidden influence. Calibrated surrogates matching the observable mean, variance, and decorrelation time can still produce distinct fingerprints when feedback memory, hidden-noise memory, or hidden forcing timescales are misrepresented. Nonlinear examples reveal mechanism-dependent signatures for extreme bursts, oscillatory variability, and regime switching, including multiplicative instability versus additive forcing, deterministic nonlinear feedback versus state-dependent multiplicative noise, and intrinsic bistability versus hidden double-well forcing. Causal fingerprints therefore provide a time- and event-resolved diagnostic for asking not only whether a model reproduces observed statistics, but how its hidden dynamics explain individual events.

Science Advances, Volume 12, Issue 42, October 2026.

Proceedings of the National Academy of Sciences, Volume 123, Issue 41, October 2026. SignificanceOxygen and carbon cycling are critical controls on Earth’s climate. Here, we expand on an isotope method using dissolved O2 to estimate the rate of this cycling. The method relies on assumptions about how isotopes are sorted during ...

Journal of Geophysical Research: Oceans, Volume 131, Issue 10, October 2026.

Journal of Geophysical Research: Atmospheres, Volume 131, Issue 20, 28 October 2026.

Abstract We explore the large-scale relations between anomalies of global tropical sea surface temperature (SST) and U.S. precipitation to assess the sources of December–February (DJF) predictability and skill. Canonical Correlation Analysis (CCA) is applied to forecasts from NOAA’s latest seasonal prediction system, the Seamless System for Prediction and EArth System Research (SPEAR). We find that DJF skill can largely be recreated using two SST principal components (PCs) and four precipitation PCs. However, the leading CCA modes based on these PCs are a blend of two climate signals, El Niño-Southern Oscillation (ENSO) and linear trends, which makes them difficult to interpret. We separate the trend and ENSO signals using partial CCA whereby CCA is applied twice: once to data with linear trends removed and once to data with the Niño-3.4 index linearly removed. When both signals are removed, CCA finds no consequential relations. Therefore, ENSO and linear trends alone explain the predictable parts of the winter forecast of tropical SST and U.S. precipitation anomalies. Despite SPEAR model predictions representing both linear and non-linear variability, skill is mostly explained by these simpler linear CCA patterns.

arXiv:2603.28481v3 Announce Type: replace Abstract: The spatial pattern of sea surface temperature (SST) plays a central role in shaping the climate system, yet the influence of land surface temperature (LST) on large-scale SST patterns remains understudied. Using a state-of-the-art coupled climate model, we investigate the climate response to regional abrupt LST warming and time-dependent ramped-warming. We find that regional LST warming can substantially alter the mean-state SST pattern, with the response depending strongly on the location of land warming. In particular, warming over South America strengthens the zonal SST gradient across the tropical Pacific, producing a more La Ni\~na-like mean state. Enhanced land warming increases the zonal contrast in diabatic heating, which excites stationary Rossby wave responses that strengthen alongshore winds and coastal upwelling in the eastern tropical Pacific, providing a dynamical pathway linking regional land warming to changes in the equatorial Pacific mean state. Similar remote SST responses occur for warming over North America, which is accompanied by cooling in the North Pacific, and over Central Africa, which is accompanied by cooling in the tropical Atlantic. In contrast, warming over the Maritime Continent and Tibetan Plateau produces little systematic change in large-scale SST patterns. These results demonstrate that regional LST changes can exert a substantial and geographically dependent influence on remote ocean temperatures. They further suggest that biases in both the mean state and warming trends of LST may contribute to biases in simulated SST patterns in historical simulations, highlighting the importance of accurately representing LST in climate models.

Journal of Geophysical Research: Atmospheres, Volume 131, Issue 20, 28 October 2026.

As a fundamental component of the Earth’s climate system, snow cover critically mediates surface radiative forcing, terrestrial water storage, and large-scale atmospheric momentum exchange. Its long-term trends are a direct fingerprint of global warming, yet the scientific literature on snow cover dynamics has grown exponentially, posing substantial challenges for systematic knowledge synthesis. To map the intellectual landscape of Northern Hemisphere (NH) snow cover dynamics, we conducted a comprehensive bibliometric review of this corpus based on 10,350 peer-reviewed articles retrieved from the Web of Science. The bibliometric analysis reveals a sustained upward trajectory in publication output, with core journals concentrated in Remote Sensing, Journal of Geophysical Research: Atmospheres, and The Cryosphere, and leading research institutions represented by the Chinese Academy of Sciences, the French National Centre for Scientific Research, and the University of Colorado Boulder. Methodologically, this field is shifting toward the integration of high-resolution remote sensing with intelligent algorithms. Geographically, research hotspots are predominantly located in high-mountain areas, with additional focus on typical snow zones across North America, Europe, Russia, Central Asia, and the Tibetan Plateau. Despite ongoing advances in remote sensing and algorithmic techniques, the synergistic integration of multi-source snow data and the operational production of long-term, high-resolution hemispheric snow cover products remain critical bottlenecks that urgently need to be addressed. These findings provide valuable references for optimizing future snow monitoring systems and advancing model development.

Science Advances, Volume 12, Issue 43, October 2026.

Preprint under review for ESSD (discussion: open, 0 comments) A 30-year dataset was made using infrared satellite data from November 1992–December 2022 to track the time, duration, and location of Cloud Mass Meridional Transport events (CMMTs). These events bring substantial heat and moisture poleward to Antarctica. CMMTs occurred more often during austral winter (JJA) and in regions such as Ellsworth Land in West Antarctica and Queen Maud Land in East Antarctica. The average number of CMMTs per year increased while their average duration decreased.

arXiv:2610.19217v1 Announce Type: new Abstract: We present the RemoTAP-Hybrid algorithm for global aerosol products retrieved from Multi-Angle Polarimeters (MAP), and apply it to one-year aerosol retrievals from the SPEXone MAP instrument. For the first time, Multiple Collaborated Neural Networks (MCNN) are used to perform global aerosol MAP retrievals over land and ocean. The combined use of MCNN with a full physics inversion in RemoTAP-Hybrid, significantly improves accuracy and computational speed of SPEXone aerosol retrievals. We validate the retrieved Aerosol Optical Depth (AOD), Angstrom Exponent (AE), Single Scattering Albedo (SSA), Absorption AOD (AAOD), Fine mode AOD (FAOD), Coarse mode AOD (CAOD), and fine mode effective radius (reff_fine) with AERONET data. The validation suggests that aerosol retrievals can be further improved when extending the aerosol description to include a fine non-spherical mode. For AOD, the Root-Mean-Square Error (RMSE) is 0.051 (0.037) over land (ocean), while 78.3% (80.8%) of the retrievals are within the requirement formulated by the Global Climate Observing System (GCOS). The RMSE is 0.186 (0.207) for AE, 0.038 (0.044) for SSA, 0.023 (0.014) for AAOD, 0.047 (0.031) for FAOD, 0.043 (0.027) for CAOD, and 0.021 (0.026) for reff_fine over land (ocean). Compared to the current RemoTAP algorithm, RemoTAP-Hybrid shows improved agreement with AERONET for AOD, AE, and FAOD over land and ocean while the number of valid retrievals is increased by 4% (3%) over land (ocean) and the computational time is reduced by 43% (32%). Finally, we present the first global seasonal maps of aerosol optical and microphysical properties and chemical composition from SPEXone.

In the Hindu Kush region, there is a scarcity of observations on the mass balance of glaciers during the pre-2000 period. Here, using stereo imagery from satellites, we estimate that glaciers in the Eastern Hindu Kush have undergone mass loss at a moderate rate during 1979–2002. This study analyzes the behavior of debris-covered and debris-free glaciers as well as surging and non-surging glaciers. We further discuss individual glaciers and identify the major retreating glaciers in the region.

Preprint under review for ESSD (discussion: open, 0 comments) Weather forecasts are mostly judged on ordinary days, not on the rare heatwaves, cold spells, storms, downpours, and droughts that cause most harm. We built a global daily record of such unusual weather over land since 1981 and compared it with reported disasters. Rarer weather was more often linked to harm, but disaster records miss much of the world. Leading artificial intelligence forecasts often missed extremes, especially strong winds, so forecasts need testing on extremes everywhere.

Abstract This study analyzes a novel nine-member ensemble of convection-permitting regional climate simulations for Germany to assess projected changes in temperature and precipitation extremes under two future global warming levels, +2 and +3 K. The robustness of the climate change signals within the ensemble is also evaluated to ensure the reliability of the results. Throughout the twenty-first century, temperature-related extremes are projected to intensify, with nighttime extremes showing larger increases than daytime extremes. This amplification of nocturnal warming implies elevated risks to human health and ecosystems. While annual precipitation sums in Germany are projected to change only slightly and are mostly not statistically significant, the frequency and intensity of intense and very extreme precipitation events increase substantially. This change signal is particularly evident in summer, despite a predominantly decreasing seasonal mean rainfall. Such contrasting changes indicate a future characterized by generally drier summers interrupted by heavy rainfall events, with implications for both droughts occurrence and flood risk estimates. Temperature-related extremes show high robustness across almost all of Germany, whereas precipitation-related extremes exhibit greater variability, reflecting stronger natural variability and model uncertainty. The high-resolution simulations reveal added value for temperature-based indices, including more detailed spatial patterns of time averages and changes in specific areas, together with reduced uncertainty as indicated by ensemble spread and signal-to-noise ratios. These new climate model projections offer a robust scientific basis and a high potential for developing adaptation strategies tailored to small-scale landscape units.

Abstract Northern Hemisphere snowmelt phenology has shifted earlier over the past 37 years, but Eurasia and North America have followed markedly different trajectories. Using a random forest framework fusing passive microwave brightness temperatures, ERA5-Land reanalysis, and static landscape predictors, we produce annual 6.25 km outputs of main melt onset date (MMOD), snow off date, and snowmelt duration (SMD) across the Northern Hemisphere seasonal snow domain from 1988-2025, trained on more than 13,000 in situ observations. Over the full record, median MMOD and snow off trends are roughly four to five times stronger in Eurasia (~-2 days decade⁻¹) than in North America (~-0.4 days decade⁻¹), while SMD shows little systematic change in either continent, indicating that the dominant signal is a shift in timing rather than melt-season length. A 15-year moving-window analysis shows that this full-period contrast obscures a converging trajectory: Eurasia developed a robust earlier-melt regime by the early 2000s, while North America's trends were weak or even reversed in early decades before becoming more negative in recent windows. Disaggregating by snow regime reveals that this apparent hemispheric convergence is largely an artifact of buffering within area-dominant boreal forest and tundra regimes, where trends weaken or stabilize over time, while smaller seasonal snow regimes, including prairie, montane forest, and ephemeral snow, exhibit stronger, often asymmetric, accelerations toward earlier melt, with the most coherent advances concentrated in Eurasia. These results indicate that hemispheric-scale convergence in snowmelt timing masks substantial, regime-dependent asymmetry, with direct implications for runoff timing and water availability in climatically sensitive seasonal snow environments.

Abstract The annual cycle of precipitation is a distinct phenomenon in most tropical and subtropical land regions. It is normally characterized by amplitude, phase, and length of wet and dry seasons. Previous studies have documented changes in amplitude and phase under global warming, but changes in wet season length are not fully understood. Here, we show that, in the Northern Hemisphere, onset and cessation shift later at a similar pace, resulting in modest changes in wet season length. In contrast, significant changes in wet season length are primarily concentrated in the Southern Hemisphere, where 23 of the 27 CMIP6 models project a shortening of the wet season, with a multi-model mean reduction of 3.70 days. This shortening is accompanied by a larger delay in wet season onset than in cessation. The shortened wet season over the Southern Hemisphere is closely aligned with asymmetric changes in the divergence of atmospheric energy transport between the onset and cessation stages based on the atmospheric energetic framework. Asymmetric changes in the divergence of atmospheric energy are mainly determined by the net energy input to the atmosphere, dominated by the latent heat flux. Future asymmetric changes in latent heat flux are strongly influenced by precipitation-latent heat flux coupling. In the current climate, this coupling is stronger during the onset stage than during the cessation stage, leading to more pronounced changes in latent heat flux during onset relative to cessation. The resulting robust shortening of wet season length could have broad implications for ecosystems and societies.

arXiv:2608.09971v2 Announce Type: replace Abstract: Machine-learning weather models now match or exceed operational numerical prediction in the medium range, but when they are run freely for years they drift, blow up or lose their variability, and dedicated climate emulators are expensive to train. Here we present Rescene, a 0.4 M-parameter wrapper that turns a frozen 1.5$^\circ$, 6-hourly, 69-variable weather model into a stochastic climate emulator. A deterministic slow clock anchors the frozen model to a calendar climatology, and a band-limited AR(1) perturbation, whose spectral envelope is fixed to the observed anomaly spectrum, restores the weather that anchoring removes. A 100-year, 8-member Rescene run shows no detectable drift ($-0.004 \pm 0.014$ K per 100 years), and on the 2022 test year the 16-member ensemble is calibrated (spread/skill 0.89, 0.97 and 0.93 for Z500, T2m and MSLP) with an extratropical synoptic amplitude of 97% of ERA5 on the tuning year. The free run reproduces the geography of day-to-day variability (pattern correlation 0.83 for Z500) and 71% of the observed blocking frequency, where the deterministic wrapper alone has none. Interventions on the frozen operator show a scale-local, approximately linear response in which Z500 perturbations grow upscale 6--14 times more than downscale per 6 h. All constants are fitted on a training year against a rule declared beforehand, and we show that common evaluation shortcuts would have overstated variability and skill. Rescene makes an existing weather model usable for century-long ensembles at negligible training cost.

arXiv:2610.03759v3 Announce Type: replace Abstract: Ocean wave forecasting is essential for maritime safety, offshore operations, and coastal resilience, yet remains challenging due to systematic biases in physics-based models. Physical models, while widely used, rely on approximations and parameterizations that limit their accuracy under complex ocean-atmosphere conditions. To enhance ocean wave forecasting, we propose BridgeCast, within a physics-AI hybrid framework for bias correction. BridgeCast is a probabilistic model based on conditional flow matching (CFM) that learns to transform physical model forecasts into reanalysis-like fields. It treats physical forecasts as corrupted observations and employs a continuous-time generative process to bridge their distribution toward that of reanalysis data. BridgeCast is parameterized by a Transformer-based architecture that enables spatiotemporal modeling, incorporation of exogenous atmospheric variables, and flexible inference via both ordinary and stochastic differential equation formulations. Extensive experiments on real-world datasets demonstrate that BridgeCast consistently outperforms state-of-the-art baselines across regions and forecast lead times.

arXiv:2610.14930v2 Announce Type: replace Abstract: The notion of tipping points and related terminology have so far had inconsistent and sometimes confusing definitions in both scientific and popular literature. While it is difficult for the scientific community to align popular usage of the term tipping point with a precise definition, technical usage of related terms would benefit from greater consistency and rigor. Toward this end, we offer a systematic set of terms to clarify formal communication around potential tipping dynamics, with focus on examples from Earth system dynamics.

arXiv:2603.05570v2 Announce Type: replace Abstract: We present the first application of spectral nudging in a probabilistic ensemble forecasting framework, combining the physics-based ECMWF Integrated Forecasting System ensemble (IFS-ENS) with forecasts from the probabilistic machine-learned AIFS-ENS ensemble. Large scales of virtual temperature and vorticity are relaxed toward the machine-learned forecasts, while mesoscale structures remain governed by the physics-based model. This hybrid ensemble shows substantial improvements in large-scale forecast skill, with gains in predictive skill extended by up to two days in the tropics and by approximately half a day in the extra-tropics relative to IFS-ENS. Despite nudging being applied only to upper-air fields, improvements are also found in several near-surface parameters. Tropical cyclone track forecasts improve significantly, consistent with improved representation of the large-scale steering flow, without degrading storm intensity or ensemble spread. These results demonstrate that spectral nudging can be successfully extended to ensemble prediction and provide an attractive pathway for combining machine-learned and physics-based weather prediction systems.

Science Advances, Volume 12, Issue 42, October 2026.

This study proposes a performance-based method to improve WRF climate simulation evaluation by addressing uncertainty in selecting scheme combinations. The method gives greater importance to difficult-to-simulate variables, such as precipitation and wind speed. Tested in two contrasting climate regions, it identified interpretable and well-performing model settings compared with equal weighting and established statistical weighting methods, including in the independent extreme-year test.

arXiv:2610.19772v1 Announce Type: new Abstract: This work introduces a Process Additive Neural Network (PANN) framework to disentangle compensating biases in atmospheric models, where errors in different physical processes can offset each other and obscure their origins. Using a single-column model, each PANN branch represents the tendency error associated with a separate physical process, and the summed branch outputs approximate the total error relative to a reference tendency. Once trained successfully, the branch outputs are expected to provide process-level error attributions and, when averaged, estimates of process-level bias. Initial experiments focus on cloud-related processes in OpenIFS SCM case studies, where cloud and convection tendencies are strongly coupled and compensation is difficult to diagnose using standard approaches. The aim is not to replace human interpretation of physical processes, but to extend it to regimes where model complexity, process interaction, and compensating errors make direct diagnosis difficult. By providing a physically structured form of machine-learning interpretability, PANN can help connect complex model behaviour with process-level explanations. This work supports both improved understanding of compensating errors and the development of more physically consistent weather and climate models.

arXiv:2502.14148v3 Announce Type: replace Abstract: We describe the user interface, governing equations, and numerical methods underpinning the ocean modeling software called "Oceananigans". Oceananigans development has been led by the Climate Modeling Alliance as part of a larger effort to build a trainable climate model with quantifiable uncertainty. Notably, Oceananigans is written in the Julia programming language, is optimized for GPUs, and can efficiently simulate all scales of ocean motion ranging from millimeter-scale turbulence in a small box to planetary-scale ocean circulation. We argue that this combination of capabilities is enabled by a design that combines a basic, easy-to-optimize, GPU-friendly structured finite volume algorithm exposed behind a high-level, programmable user interface. This design negotiates a dual mandate for highest-possible performance (to support state-of-the-art applications) and enhanced accessibility (to facilitate adoption and development). The dual mandate aims ultimately to accelerate the progress of Earth system science.

arXiv:2601.19265v4 Announce Type: replace-cross Abstract: The exploration of cosmic rays, which are high-energy particles originating from space and the atmosphere, has historically been associated with particle physics and astrophysics. In the last 20 years, these particles have evolved into valuable tools for observing Earth's systems. This review compiles the use of cosmic rays in three primary areas: (1) particle physics and atmospheric processes, which include cosmic- ray-induced cascades, ionization, and their impact on atmospheric chemistry and radiation; (2) geosciences, where cosmogenic radionuclides assist in the dating of geological materials and cosmic-ray neutrons are used for large-scale monitoring of soil moisture and snow water equivalents; and (3) urban science, where cosmic- ray muons are employed for non-invasive subsurface imaging and, when paired with distributed sensors, serve as the basis for smart city monitoring. This review places particular emphasis on integrating these methods with remote sensing and geographic information systems (GIS), which helps close the persistent scale gap between point measurements and satellite observations, thereby enabling three-dimensional digital representations of subsurfaces. This review concludes by discussing data standards, their integration into operational Earth observation workflows, and future research directions.

Geophysical Research Letters, Volume 53, Issue 20, 28 October 2026.

arXiv:2610.20640v1 Announce Type: cross Abstract: Global streamflow forecasting supports flood preparedness, yet data-driven models trained on reanalysis lose skill when driven by operational forecasts, increasingly so with lead time. AIFL addresses this shift with a two-stage transfer-learning recipe and a deterministic LSTM, but it feeds one forecast lead time into the whole input window and gives no uncertainty. We ask how far this open recipe can be extended without changing its recurrent core. KNOU-Flood keeps a single-layer LSTM and the two-stage training and changes the information around it: a longer input window (424 days), lead-aware forecast inputs, GraphCast as a second forecast source next to ECMWF HRES, and a latent-noise ensemble trained with the fair CRPS. We compare against a re-implementation of the AIFL recipe (its code is not public) under identical data, stations, splits and metrics. Our contributions are threefold. First, over three seeds and 2,099 test basins (2021-2023), KNOU-Flood reaches a median KGE' of 0.65 at LT1 and 0.48 at LT10 against 0.62 and 0.34, with significant gains at all ten lead times under two scoring protocols, also against a 1024-unit baseline and on held-out basins (one seed each). Second, identical weights give an LT10 KGE' of 0.18 or 0.34 depending on how forecasts are fed, so much of the lead-time decay reported for AIFL is an input-protocol effect; the probabilistic head contributes most to the gain, followed by the raw GraphCast channels, the longer window and mixed fine-tuning. Third, the ensembles reach lower CRPS than probabilistic versions of the baseline, are nearly calibrated with inference-time dropout, and detect 1.5- and 2-year floods better than all baselines; the benefit for rarer floods depends on the seed, and an exploratory tail-weighted variant improves 10-year and larger events. KNOU-Flood is a controlled extension of an open recipe, not a new architecture.

Proceedings of the National Academy of Sciences, Volume 123, Issue 41, October 2026.

Proceedings of the National Academy of Sciences, Volume 123, Issue 41, October 2026.

Journal of Geophysical Research: Atmospheres, Volume 131, Issue 20, 28 October 2026.

Journal of Advances in Modeling Earth Systems, Volume 18, Issue 10, October 2026.

Abstract Clouds play a critical role in Earth’s radiative balance through their radiative effects. Recent studies suggest that the current planetary energy imbalance is associated with a reduction in global albedo, largely resulting from decreases in cloud fraction. However, the southeastern Pacific Ocean, where maritime stratocumulus clouds dominate off the west coast of South America, shows increasing albedo associated with enhanced low cloud cover. In this study, PATMOS-x v6.0, a satellite-based long-term climate data record derived from the AVHRR instruments aboard polar orbiting satellites, and CERES EBAF are used to investigate changes in cloud fraction and albedo, respectively. Both datasets show increases in anomalies of water cloud amount (1.21%∙decade −1 ) and albedo (0.25%∙decade −1 ), and these variables have been strongly correlated since 2000. Applying the cloud-controlling factors framework using 6 atmospheric variables from the ERA5 reanalysis dataset and ridge regression method indicates that these changes are primarily driven by an increase in estimated inversion strength (0.31 K∙decade −1 ). The slope between water cloud amount and EIS anomalies is 2.81%∙K −1 from Pv6.0 and 2.37%∙K −1 from ridge regression model. The increase in EIS is associated with enhanced lower tropospheric stability, resulting from a decrease in surface temperature (−0.07 K∙decade −1 ) and an increase in atmospheric temperature at 700 hPa (0.24 K∙decade −1 ) over the last four decades. The long-term decrease in surface potential temperature is closely associated with ENSO variability, whereas the increase in potential temperature at 700 hPa is incompatible with an ENSO-only explanation, implying that other thermodynamic or circulation changes are contributing to enhanced lower tropospheric stability.

Nature Communications, Published online: 19 October 2026; doi:10.1038/s41467-026-78785-2 Designing efficient photoreactors remains challenging due to the complex interplay of light transport phenomena, shaped by reflection, scattering and absorption processes. Here, we introduce a workflow that integrates ray-tracing digital twins with multi-objective Bayesian optimization to autonomously design photoreactors in a closed-loop manner, achieving high photon capture and uniform irradiation.

Geophysical Research Letters, Volume 53, Issue 20, 28 October 2026.

arXiv:2610.28219v1 Announce Type: new Abstract: Attribution of observed historical climate evolution to externally forced change and internal variability is fundamental to understanding climate change and evaluating climate models. Yet such attribution remains challenging because the forced response is intertwined with internal variability in observations. Here, we develop a convolutional long short-term memory network (ConvLSTM) to disentangle these two components. Trained and evaluated using large-ensemble simulations from multiple climate models, ConvLSTM consistently outperforms existing model-based, statistical, and machine-learning approaches. Applying ConvLSTM to observations reveals an evolving spatial pattern of the forced response and substantial contributions from internal variability to recent multidecadal climate trends. For several key climate features whose attribution remains debated, ConvLSTM indicates that internal variability amplified the forced reversal of Atlantic interhemispheric temperature asymmetry, dominated the observed La Ni\~na-like warming pattern over the equatorial Pacific and the associated precipitation decline over southwestern North America since 1980, and modulated Arctic amplification around its nearly constant threefold forced value. Comparison between the ConvLSTM-estimated and model-simulated forced responses reveals broadly consistent patterns but also important discrepancies, with climate models simulating stronger Northern Hemisphere warming, weaker subtropical high intensification, and smaller land precipitation declines. Overall, ConvLSTM provides a robust framework for attributing historical climate evolution and evaluating the fidelity of climate models

Abstract Summer extreme heat events pose severe risks to society and ecosystems. This study reveals a significant synergistic effect between the preceding winter negative Northeast Pacific sea surface temperature anomalies (nNEP SSTA), known as the “cold blob”, and spring negative Indo-Pacific Warm Pool SST anomalies (nIPWP SSTA), that intensifies summer heat extremes in Northeast China and Mongolia. When co-occurring, these factors nonlinearly amplify heat events far beyond their individual linear effects. The mechanism for this synergistic effect involves a series of coupled oceanic and atmospheric bridges. The two factors jointly generate spring anomalous westerlies that sustain the nNEP SSTA into summer via the Wind-Evaporation-SST feedback. This persistent summer anomaly triggers eastward propagating Rossby waves energy, establishing a strong local anticyclone and increased atmospheric thickness over Northeast China and Mongolia, which directly increases the temperature of the atmospheric layer and favors extreme heat events. Furthermore, this circulation induces soil moisture deficits, further intensifying the heat through soil moisture feedback. These nonlinear processes are confirmed by General circulation model experiments. These findings highlight the importance of the synergistic influence of these remote SST anomalies for the prediction of summer heat extremes over Northeast China and Mongolia.

arXiv:2610.16714v2 Announce Type: replace Abstract: A potentially super El Ni\~no is forecasted in 2026, raising the need to anticipate land carbon-cycle risks before its impacts unfold. We develop a prospective early-warning framework that uses dynamic global vegetation models driven by an ensemble of ECMWF seasonal climate forecasts and machine-learning burned area predictions. Forecasts indicate a $0.23 \pm 0.21$ PgC reduction in global net biome productivity in the second half of 2026, with hotspots of sink weakening in eastern South America and elevated fire risk in Southeast Asia.

Nature, Published online: 14 October 2026; doi:10.1038/s41586-026-11174-3 Optogenetic stimulation of motor-cortex neuron ensembles shows that precisely timed activity in the primary motor cortex can generate learned neural patterns and movements, demonstrating an active role in skilled motor control.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 10, October 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 10, October 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 10, October 2026.

arXiv:2610.19994v1 Announce Type: new Abstract: High-resolution precipitation datasets are essential for weather forecasting, hydrological modeling, and climate risk assessment, yet observation-based gridded products often lack the spatial detail required for regional applications. Here we present PrecipCorrDiff, a diffusion-based downscaling framework built by modifying the original corrector diffusion based archetecture that incorporates a customized loss function and a multi-scale residual generation strategy to improve the precipitation especially the representation of extremes. PrecipCorrDiff generates 0.0625{\deg} resolution India Meteorological Department (IMD) precipitation fields from 0.25{\deg} gridded data while preserving precipitation statistics and reconstructing fine-scale spatial variability. Validation against independent IMD station observations, comparing the proposed framework with the SRGAN-based model and the original CorrDiff model, indicates that the proposed framework substantially improves the representation of spatial precipitation patterns, rainfall intensity distributions, and extreme rainfall events, while eliminating the systematic negative bias present in the previously developed SRGAN-based model. Across four climatologically distinct homogeneous Indian monsoon regions, PrecipCorrDiff reproduces observed precipitation characteristics with near-zero spatial bias, demonstrating the potential of diffusion-based downscaling for generating physically consistent, high-resolution precipitation datasets for hydrological, meteorological, and climate applications.

arXiv:2610.20405v1 Announce Type: new Abstract: Conventional deep convection schemes remain a primary source of uncertainty in global climate models, yet replacing them with data-driven alternatives while maintaining physical consistency and numerical stability is an open challenge. We present a hybrid configuration of the Community Atmosphere Model version 6 (CAM6), in which a fully connected deep neural network replaces the Zhang-McFarlane deep convection scheme (ZM), while radiation, turbulence, and other cloud processes are left unchanged. The network is trained on output from a storm-resolving global model - the global System for Atmospheric Modeling (gSAM) - using a coarse-graining approach to derive subgrid-scale convective fluxes and tendencies. This data-driven parameterization is ported to CAM6 and found to run stably for more than a decade while conserving energy and moisture. It produces a broadly realistic climatology without retuning the remaining parameterizations. Relative to the default Zhang-McFarlane scheme, the neural network parameterization substantially improves some long-standing model deficiencies. In particular, the biases in light precipitation and extreme precipitation are markedly reduced, and the diurnal cycle of tropical land precipitation shifts from a spurious midday peak over many regions to the observed late-afternoon peak. These results support targeted ML-based convection schemes as a viable path toward reducing systematic model errors in next-generation climate models.

Abstract The Late Miocene was a critical interval in the evolution of the South Asian monsoon system and the Himalayan foreland. This study presents a quantitative paleoclimatic reconstruction based on fossil leaf assemblages from the Lower (10.5–9.2 Ma) and Middle (9.2–7.0 Ma) Siwalik sediments of the Daldale section, western Nepal, using the Climate Leaf Analysis Multivariate Program (CLAMP). The reconstructed mean annual temperatures are 22.2 ± 2.3°C for the Lower Siwalik and 21.5 ± 2.3°C for the Middle Siwalik, with corresponding cold month mean temperatures of 15.4 ± 3.3°C and 15.0 ± 3.3°C. Mean annual precipitation ranges from 228 ± 48 to 238 ± 48 cm, and the annual vapor pressure deficit remains close to 6 ± 2 hPa, indicating persistently warm and humid monsoon‐influenced conditions throughout the study interval. Although several climatic variables exhibit coherent directional trends, including slightly lower temperatures, reduced rainfall concentration during the wettest months, and subtle changes in seasonal moisture availability in the Middle Siwalik, these differences largely overlap within the uncertainties of the CLAMP reconstructions. Consequently, the Daldale record indicates broad climatic stability rather than a major climatic transition between the Lower and Middle Siwalik intervals within the studied section. The study provides the first quantitative paleoclimatic benchmark for the Daldale Siwalik succession and demonstrates that climatic responses within the Himalayan foreland were spatially heterogeneous, with western Nepal maintaining persistently humid conditions despite broader regional environmental changes.

In this study, we investigate the ability to extract information about ice viscosity at the same as basal slipperiness in the inversion procedure of large-scale ice-sheet models. This is an essential part of the initialisation of many large-scale computational ice-sheet models. However, there is a lack of consensus on whether a joint inversion is possible to determine these fields. We present a theoretical framework to better understand the factors influencing the retrieval.

arXiv:2610.22077v1 Announce Type: cross Abstract: Weather prediction is undergoing an important shift, with data-driven AI models outperforming leading numerical weather prediction systems on average at a fraction of the computational cost. Yet, whether this success extends to extreme events remains contested. Earlier evaluations have exposed important shortcomings, but have largely focused on deterministic models and metrics, leaving the capabilities of the latest probabilistic models less understood. We conduct an extensive benchmark of tropical cyclone forecasting, comparing GenCast, FourCastNet 3, and FGN with ECMWF's operational physics-based ensemble, IFS-ENS. We develop a suite of probabilistic metrics and diagnostic analyses to assess cyclone tracking, intensity biases, uncertainty, and the wind-pressure relationship that characterizes storm structure. We find that, despite remaining limitations in intensity and ensemble spread, leading AI models substantially outperform IFS-ENS in cyclone tracking: FGN achieves better probabilistic location scores for 90% of evaluated storms in 2025. Further analyses reveal model-specific failure modes and suggest a lingering influence of pretraining data, pointing to opportunities for targeted improvement. Nonetheless, these results suggest that AI models are capable of capturing physical relationships that remain predictive well beyond typical weather conditions, motivating further investigation into how these capabilities emerge and how to strengthen them.

Proceedings of the National Academy of Sciences, Volume 123, Issue 41, October 2026. SignificanceAs the frequency of global heatwaves continues to increase, heat exposure has emerged as a prominent environmental stressor. Our study investigated sex differences in heat exposure-induced anxiety, focusing on CRH neurons and TLR4, to ...

Proceedings of the National Academy of Sciences, Volume 123, Issue 41, October 2026. SignificanceHow spiking neural networks (SNNs) efficiently control our bodies is still a mystery. Understanding these mechanisms may allow the control of arbitrary dynamical systems with low power consumption devices, such as neuromorphic hardware. Here, ...

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 10, October 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 10, October 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 10, October 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 10, October 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 10, October 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 10, October 2026.

Nature, Published online: 20 October 2026; doi:10.1038/d41586-026-03314-6 Record-breaking avalanches and floods are signs of things to come. Scientists must evaluate and respond to growing risks in mountain regions.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 10, October 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 10, October 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 10, October 2026.

Abstract The Indian Ocean Dipole (IOD) exerts a season-spanning influence on Arabian Peninsula (AP) climate, modulating both autumn (September–November, SON) heat and early-winter (October–December, OND) precipitation. During positive IOD, the southern AP warms in SON via a radiative pathway: anomalous low-level moisture transport enhances the greenhouse effect, strengthening downward longwave radiation and diabatic heating. As conditions cool into OND, the IOD’s influence pivots to precipitation. Positive IOD amplifies OND precipitation through two mechanisms: a low-level moisture conveyor leads to convergence-dominated precipitation along the southwestern coast, while a strengthened subtropical jet interacts with this moistening environment in the northeast. In contrast, negative IOD features weaker, spatially distinct SON cooling centered over the northern AP, primarily from cold-air advection, and broadly suppressed OND precipitation. These seasonal mean shifts would alter the probability of high-impact weather. On synoptic timescales, the enhanced moisture during positive IOD can first radiatively amplify heatwaves and then fuel extreme precipitation within days. This dual fingerprint clarifies the IOD’s role in AP climate variability and provides a physical basis for assessing compound heat-rainfall hazards in a warming world.

Proceedings of the National Academy of Sciences, Volume 123, Issue 42, October 2026. SignificanceMost autotrophic bacteria rely on CO2-concentrating mechanisms (CCMs) to overcome the slow rate and nonspecificity of the enzyme rubisco. These systems elevate cytosolic HCO3−up to 1,000-fold higher than the environmental concentration. This ...

Proceedings of the National Academy of Sciences, Volume 123, Issue 42, October 2026. SignificanceWave impacts are a primary variable driving high rates of coastal erosion and growing flood hazards in the Alaskan Arctic. Yet, existing historical datasets lack the resolution to adequately resolve the coastal wave environment—critical to the ...

Preprint under review for ESSD (discussion: open, 0 comments) We developed a new high-resolution global gravity field model by integrating satellite, marine, terrestrial, airborne, polar, and topographic gravity information. The model substantially improves gravity estimation accuracy in coastal, polar, and data-sparse regions. Accuracy improves by 9 %–24 % within 10 km of the coastline and by approximately 40 % over Antarctica. The dataset can be used for geodesy, oceanography, and other Earth science applications.

Journal of Geophysical Research: Atmospheres, Volume 131, Issue 20, 28 October 2026.

This study improves the Agro-C model to better represent how crops respond to changing soil moisture. The updated model adjusts carbon allocation between roots and leaves according to soil moisture and growth stage, and also accounts for leaf aging under water stress. Tests with field data for maize and wheat showed better predictions of plant growth above and below ground. The results can improve estimates of crop production and soil carbon storage under future drought and climate change.

Most hydrological models tend to underestimate snow over the mountainous southwestern US. This may result from uncertainties in precipitation forcing and/or inadequate representation of snow–vegetation interactions, particularly canopy interception and shading. Through model experiments, we demonstrate how downscaled meteorological forcing and improved vegetation representation can improve snow simulations over the southwestern US.

This research introduces a new method that combines model simulations with satellite observations to attribute the influence of wildfire smoke on an airmass. By dynamically determining the height of smoke plumes, we overcome a key limitation of earlier fixed reception-height approaches. This advancement is crucial for improving our understanding of how wildfire emissions influence cloud formation and the broader Earth's climate system.

Earth's Future, Volume 14, Issue 10, October 2026.

We implemented water stable isotopologues in the non-hydrostatic mesoscale atmospheric model Meso-NH. We validated this isotopic version (Meso-NH-ISO) with a simulation of a well-documented squall line in the Sahel region. In future works, simulations with Meso-NH-ISO will be done on real cases of cyclones or squall lines. The goal is to better interpret and quantify isotopic observations on the field in terms of atmospheric processes that drive development and intensity of those thunderstorms.

Geophysical Research Letters, Volume 53, Issue 20, 28 October 2026.

We developed and tested a way to trace selected cloud particles in a simulation, including how they grow and merge. One run identifies particles that meet chosen conditions, and a second reconstructs only their saved histories. In a three-dimensional test, the saved data were estimated to be one-fifth of what would be needed to keep all particle histories. Larger reductions are possible when fewer particles are selected, making such studies more practical in large simulations.

Ground-based microwave instruments can continuously observe the lower atmosphere, but clouds make their measurements harder to use. We improved a computer model so that it represents how cloud and rain particles affect these observations, and tested it with six months of measurements from two sites in China. The updated model performed better under cloudy conditions without rain, supporting more reliable use of these observations in atmospheric monitoring and weather forecasting.

Air pollution travels across continents, meaning emissions in one region can affect air quality far away. To better understand this, scientists from many groups are planning to run coordinated computer simulations of the atmosphere. By comparing results across models and emission scenarios, the planned study will show how pollution moves between regions and which sources matter most, helping governments design more effective air quality policies.

To enhance atmospheric transmittance accuracy in strong water vapor absorption bands, this study proposes an optimized scheme for the fast transmittance algorithm applied to the Hyperspectral Infrared Atmospheric Sounder-II. It introduces a transmittance threshold for sample selection and a weighted least squares regression with transmittance weighting. Validation against line-by-line models and observations shows significant improvements in forward model accuracy and stability at 6.7 μm.

Accurate soil moisture data is essential for predicting weather. This study examined how observations from three satellites can be combined to improve land-surface simulations. While each satellite helps, their value changes with vegetation type. Merging these data sources gives a more reliable estimate of soil wetness, especially in central and western China. This approach strengthens soil-water monitoring and supports more dependable climate forecasting.

Nature, Published online: 21 October 2026; doi:10.1038/d41586-026-03063-6 Salt-marsh plants can protect inland areas against huge waves — but seemingly only up to a point, and big storms cause progressive damage.

arXiv:2610.29630v1 Announce Type: new Abstract: This paper describes an approach to constructing nowcast and forecast predictive models for the occurrence of high-latitude decameter ionospheric irregularities based on SuperDARN and SECIRA radar data and space weather parameters. The method relies on self-learning pattern recognition in the dynamics of space weather parameters, that influence the irregularities formation. The proposed approach is a data-driven, multiscale method in which the spatial distribution of irregularities is approximated using a neural network of an original architecture, inspired by graph neural networks. The model was trained and tested using data from 11 radars during the period 2010-2025 (SuperDARN: CVW, ICE, ICW, INV, PGR, PYK, RKN, SAS, STO; SECIRA: EKB, MAGW). It is shown that the model achieves optimal performance when using geomagnetic coordinates MLT,MLAT. The optimal set of parameters for the best nowcast and forecast over studied set of parameters is a combination of geomagnetic and solar disturbance indices ($a_{p}$, $K_{p}$, $AE$, F10.7) from ground-based observations, and satellite-measured solar wind and interplanetary magnetic field parameters ($P_{sw}$, $E_{sw}$, $\widetilde{B_{z}}$, $T_{sw}$). Hourly mean values for all the parameters were used in the analysis. The results indicate that acceptable forecast quality is achieved with a lead time of up to 7-9 hours, with the highest quality reached for nowcast (zero lead time). It is shown that the parameters $\widetilde{B_{z}}$ and F10.7 are significant for nowcasting only and can be neglected for other lead times. It is shown that $AE$ and $T_{sw}$ are significant for mid-term forecast (\textgreater 2hours ahead) and can be neglected for short-term predictions. Statistical relationships between the maximum latitude of the equatorial boundary of the irregularities and space weather parameters are also discussed.

Nature, Published online: 21 October 2026; doi:10.1038/s41586-026-11152-9 A multi-tissue signalling system links cold temperature to enhanced longevity and proteostasis in Caenorhabditis elegans.