The Atlantic Meridional Overturning Circulation (AMOC) transports approximately 1.3 PW of heat northward, sustaining the relative warmth of western Europe and influencing precipitation patterns across the tropics and subtropics. Multiple observational and proxy lines of evidence suggest that the AMOC has weakened over recent decades, raising concerns about a potential approach to a critical tipping point beyond which collapse could become irreversible. This study analyzes AMOC projections from eight CMIP6 Earth system models under the SSP5-8.5 scenario, finding a multi-model mean decline of 38 percent by 2100, from 17.7 Sv to 10.9 Sv. Bifurcation analysis applied to three models with freshwater hosing experiments reveals estimated collapse thresholds at 0.28 to 0.32 Sv of anomalous freshwater forcing, with hysteresis widths of 0.12 to 0.18 Sv indicating that recovery from a collapsed state would require substantially greater cooling than the warming that triggered the transition. Two of three analyzed models indicate a high risk of crossing the tipping point by 2100 under unmitigated emissions, with cascading consequences including northern European cooling, southward shift of the Intertropical Convergence Zone, disruption of West African and South Asian monsoons, and accelerated sea-level rise along the North American east coast.

Changes in deep-ocean carbon storage during the last glacial cycle were closely linked to variations in deep-water oxygenation, which reflects the combined influences of ocean ventilation and export production. Yet the relative importance of these processes remains difficult to disentangle. Whilst the Last Glacial Maximum (LGM) and subsequent deglacial changes in Atlantic Ocean circulation have been extensively studied, significantly less is known about the Marine Isotope Stage (MIS) 5a/4 transition, a period of global cooling and rapid glacial inception, and the subsequent full glacial conditions of MIS 4, for which palaeoclimatic records remain scarce. Here we use benthic foraminiferal assemblage records from sediment cores retrieved at bathyal (∼2.65 km) and abyssal (∼4.7 km) depths on the Iberian Margin to investigate changes in export productivity and bottom-water oxygenation between ∼50 and 85 ka. These records are compared with benthic foraminiferal assemblage records from the Iberian Margin spanning the last ∼23ka. We find that bathyal sites are characterised by overall productive, high export flux environments throughout all glacial stages, with assemblages dominated by infaunal foraminifera. In contrast, deeper abyssal sites are characterised by oligotrophic conditions with assemblages primarily composed of epifaunal foraminifera. Peak glacial conditions of the LGM and MIS 4 are represented by a distinct faunal assemblage at both depths, reflecting changes in the quality and seasonality of organic carbon supply rather than enhanced productivity. In contrast, assemblages during the most extreme climatic conditions associated with Heinrich Stadials (HSs) 1 and 6, are distinct from all others, indicating reduced bottom-water oxygenation and major hydrographic perturbations. Assemblages during the MIS 4/HS 6 interval resemble those during the LGM/HS 1 interval, indicating broadly similar benthic conditions during these two important periods of the last glacial cycle. Overall, the reconstructed changes in benthic foraminifera appear to be driven primarily by organic carbon flux, highlighting the dominant influence of changes in food supply, while significant oxygenation changes were largely restricted to Heinrich Stadials. This implies that, despite potential variability in deep-water oxygenation, benthic conditions on the western Iberian Margin were not strongly oxygen-limited during the last glacial cycle, and demonstrates the value of benthic foraminiferal assemblages for disentangling biological and physical controls on past deep-ocean conditions.

Abstract Understanding past relative sea level (RSL) is important for constraining the timing and magnitude of changes in the cryosphere, providing critical benchmarks for models used to project future sea-level change, and supplying a datum for constraining past vertical motion. Past sea-level records from Antarctica can help determine the extent to which small or large changes in ice cover induce a response of the solid Earth in tectonically active settings. Here, we examine raised beaches located at Duthoit Point, Nelson Island, Antarctica. We combine geomorphic mapping and high-resolution topographic surveys to reconstruct RSL change and glacial fluctuations during the Holocene. Our study is the first record from Duthoit Point and identifies sea-level changes associated with up to four Holocene ice advances, likely occurring between ∼7000 and ∼4800 cal yr BP, at ∼2000 cal yr BP, at ∼500 cal yr BP, and during a more recent advance younger than ∼500 cal yr BP. Our findings support assertions that uplift rates are primarily driven by glacial isostatic adjustment and changes in ice extent modulate isostatic responses at centennial to millennial timescales. Improved constraints on Holocene uplift histories provide important benchmarks for evaluating isostatic adjustment models used to reconstruct past ice-sheet change and project future sea-level change.

arXiv:2604.19500v2 Announce Type: replace Abstract: Quantifying and reducing uncertainty in Earth system model parameterizations is essential to improving their reliability in decision-making. Forward uncertainty propagation is used to derive parameter sensitivity but requires physically plausible parameter distributions first be learned from observations. Bayesian inference offers a principled approach but can become ill-posed when observations weakly constrain parameters--a condition difficult to know prior to inference. Addressing this gap, we show that parameter sensitivity results from forward uncertainty quantification can guide a non-iterative strategy for identifying observations informative to Bayesian calibration. We explore both forward and inverse uncertainty quantification for parameterizations of atmospheric turbulence in the Weather Research and Forecasting (WRF) model. To overcome the computational bottleneck of $\mathcal{O}(10^5)$ model evaluations required for both analyses, we leverage Gaussian process emulators trained on several hundred WRF simulations. Using these emulators, we conduct a global sensitivity analysis across observation space, investigating how parameter contributions to output variance depend on quantity of interest, atmospheric stability, time-averaging length, and spatial location. We then introduce nondimensional diagnostic measures that systematically identify regions where a parameter's contribution to output variance exceeds observational noise and its independent effect exceeds interaction effects. We demonstrate that observations from these regions serve as a strong proxy for accurate Bayesian calibration and reduced posterior uncertainty. Through emulator-aided Bayesian inversion with synthetic observations, we show how parameter uncertainty can be systematically reduced by leveraging sensitivity information.

Basal melting of Antarctic ice shelves is a key process driving ice-shelf thinning and dynamic mass loss of the ice sheet. Based on CryoSat-2 SARIn radar altimetry data from 2012 to 2023, this study derives a dataset of surface elevation, elevation change rates, thickness, and basal melt rates for 49 major Antarctic ice shelves within a Lagrangian framework. The results show that the mean basal melt rate of Antarctic ice shelves is 0.52 ± 0.72 m yr−1, with pronounced spatial heterogeneity. Melting is concentrated in the Amundsen and Bellingshausen Sea sectors of West Antarctica, with the Amundsen Sea ice shelves exhibiting the most pronounced melt rates (5.73 ± 0.92 m yr−1). Ice-shelf thickness changes exhibit a strong spatial coupling with basal melting, with regions of rapid thinning closely coinciding with areas of intense melt and predominantly concentrated near the grounding line. In addition, ice-shelf thinning and basal melting display pronounced interannual variability, with strong melt events occurring during 2014–2016 and 2019–2021. This study employs a Lagrangian framework to effectively suppress advection-induced noise, thereby enhancing the reliability of Antarctic ice-shelf change retrievals. The resulting nearly 12-year continuous, high-resolution dataset provides a robust foundation for investigating ice shelf-ocean interactions and assessing the stability of the Antarctic Ice Sheet.

The melting of the Antarctic Ice Sheet contributes to global sea-level rise and widespread ocean freshening. A major driver of this process is the intrusion of warm deep water onto the continental shelf, which enhances basal melting of ice shelves and weakens their buttressing of inland ice. However, the three-dimensional distribution of glacier-derived freshwater remains poorly constrained in East Antarctica. Here we present a hybrid multiple linear regression-neural network (MLR-NN) framework that reconstructs glacier-derived freshwater from dissolved inorganic carbon (DIC) residuals without prescribing fixed freshwater end-member compositions. Applying this framework to hydrographic observations around the Totten and Moscow University Ice Shelves, we estimate freshwater discharge rates of 70 ± 7 and 53 ± 9 Gt yr⁻¹, respectively. The reconstructed freshwater signal extends across the continental shelf to approximately 64°S. We also identify differences in freshwater discharge between observations collected during the 2010s and the 2020 s, although sparse temporal sampling prevents these differences from being interpreted as a robust long-term trend. Our results provide a three-dimensional observational reconstruction of glacier-derived freshwater in this sector of East Antarctica and demonstrate the potential of an end-member-independent hydrographic parameterization for investigating glacier-ocean interactions in data-limited polar environments.

Abstract Satellite altimetry (1993–2024) reveals that the Southern Ocean dynamic sea level (DSL) exhibits a distinct belt-like pattern, with smaller rises south of 50°S and larger rises to the north. This zonal belt is not uniform but strongly asymmetric, with the largest DSL increase in the Pacific sector and weaker changes in the Atlantic and Indian sectors. These pronounced Pacific anomalies are primarily driven by internal climate variability, particularly the phase transition of the Interdecadal Pacific Oscillation (IPO). The IPO during 1993–2024 induces a zonal wind stress curl dipole in the Pacific sector, forcing a distinct belt-like sea level pattern that dominates the observed zonal asymmetry. However, this observed pattern starkly diverges from the Coupled Model Intercomparison Project Phase 6 (CMIP6) multi-model mean. By filtering out internal climate variability, the models project a diametrically opposite zonal asymmetry, with the most robust DSL changes in the Indian and Atlantic sectors, while the Pacific response remains comparatively weak. Single forcing experiments from CMIP6-FAFMIP and an ocean general circulation model indicate that surface heat flux leads this forced zonal asymmetry, while wind stress forcing mainly contributes to zonally symmetric changes. Furthermore, freshwater hosing experiments suggest that buoyancy-forced weakening of the Atlantic Meridional Overturning Circulation (AMOC) enhances the belt-like DSL signal predominantly in the Indian and Atlantic sectors, thereby reinforcing the zonal asymmetry in the forced response.

Abstract The sandy Barrier III of the southern Brazilian coast was formed by sea-level high stands correlated to Marine Isotope Stage (MIS) 5, but little is still known about its behavior and evolution between the last glacial period (MIS 4–2, 71–11.7 ka) and the Holocene (MIS 1). A study on luminescence-dated continental sedimentary successions directly overlying MIS 5c (105.9–92.8 ka) marine deposits shows a wet, partially vegetated aeolian system developed between <104 and >82 ka, succeeded by a dry aeolian system accumulated mainly as transgressive sand sheets (TSSs) on interdune flats between ~82 and 10.5 ka. Protosols interbedded with TSSs indicate episodic vegetation expansion between ∼71 and 46 ka. Sediment accumulation was apparently controlled by autogenic (cannibalization, migration of bedforms) and allogenic (sea-level oscillations, climate) processes. Weathering and pedogenesis promoted changes in color and magnetic susceptibility (MS), and the development of redoximorphic features involving Fe and Mn released from weathered minerals. Strong pedogenesis after ∼10.5 ka, probably related to the Holocene climatic optimum (HCO) and orbitally driven increased seasonality and dryness, produced rubification similar to red beds on the barrier top. This study contributes to the understanding of long-term evolution and sedimentary dynamics of coastal barriers and aeolian systems on glacial–interglacial timescales.

We present an efficient approach for coupling ice sheet and solid Earth models, bringing realistic gravitational and deformational processes into mainstream ice sheet modeling. Earth responses are encoded in precomputed Green’s functions and applied to modeled ice mass change through matrix multiplication. By removing a major computational and technical barrier, the approach could accelerate adoption within the Ice Sheet Model Intercomparison Project (ISMIP) to improve sea level projections.

Interannual variability of surface mass balance over the Greenland Ice Sheet has been increasing, which has been cited as an early-warning signal of ice sheet destabilization under current climatic conditions. A destabilized ice sheet would continue to lose mass and contribute to global sea level rise, even if climate stopped changing in the future. Here, we investigate the mechanisms driving nonstationarity of surface mass balance variability in a high-fidelity surface mass balance reconstruction, and a calibrated simple model. We show that, regardless of the proximity to destabilization, increasing variability in surface mass balance is an expected consequence of atmospheric warming on low-accumulation parts of the ice sheet. Autocorrelation of surface mass balance exhibits spurious multidecadal trends under a wide range of climatic conditions due to the slow response of ice sheets to climatic fluctuations, making this a poor early-warning signal of destabilization. Though these findings do not rule out the potential for destabilization under current or future climate, they provide a conceptual framework for understanding the drivers of surface melt variability.

We present a quantitative hydro-isostatic reconstruction of the Late Pleistocene–Holocene paleogeographic evolution of the Japanese archipelago, with implications for prehistoric human dispersal. Using four ice-sheet history models combined with 252 combinations of Earth rheological parameters (total 1008 models), we simulated regional sea-level changes over the past 50,000 yr. The archipelago-scale results show a maximum sea-level fall of –140 m during the Last Glacial Maximum (LGM), with pronounced spatial variations due to hydro-isostatic adjustments. The Tsugaru and Tsushima straits show that coastal configurations are highly sensitive to hydro-isostatic model parameters. The Tsugaru Strait narrowed to a minimum width of 3.4–9.2 km at the LGM, but no land bridge formed. In contrast, the Tsushima Strait, comprising Korea–Tsushima and Tsushima–Kyushu segments, exhibits strong model-dependent variations: the ICE-6G and ANU models predicted a complete land bridge across the Tsushima–Kyushu segment, reducing the crossing to a single 6.9–17.8 km Korea–Tsushima segment, while the ANU_max and ANU_min models retained water gaps of 4.6–18.4 km (Tsushima–Kyushu) in addition to 18.2–19.8 km (Korea–Tsushima). Variance decomposition analysis revealed spatial partitioning in parameter sensitivity: lower mantle viscosity dominates uncertainty at strait locations, whereas lithospheric thickness becomes increasingly important in terrestrial regions during interglacial periods. Archaeological evidence indicates modern humans reached Japan by ca . 38 ka, prior to the LGM lowstand, implying substantial open-water crossings. The model-dependent paleogeographic variations highlight the importance of ice-sheet history uncertainties in reconstructions of Late Pleistocene human dispersal between continental Asia and the Japanese archipelago. (250 words)

Global warming is often presented as a change in global mean surface temperature, yet its causes, consequences and uncertainties emerge from a coupled Earth system in which the atmosphere, ocean, cryosphere, land and biosphere exchange energy and carbon across widely different timescales. This critical narrative review synthesises evidence from palaeoclimate archives, instrumental observations, Earth-energy inventories, carbon-cycle assessments, attribution studies and process-based modelling to evaluate what is robustly known, where uncertainty remains consequential, and how past climates inform interpretation of present warming without serving as simple analogues. The evidence converges on a physically coherent picture: sustained anthropogenic greenhouse-gas forcing is the dominant cause of contemporary global warming, while aerosols, natural forcing and internal variability modulate its rate and spatial expression. Ocean heat uptake and the positive Earth energy imbalance demonstrate continuing planetary heat accumulation even when surface warming fluctuates from year to year. Fast feedbacks involving water vapour, lapse rate, clouds and surface albedo determine much of the transient-to-centennial response, whereas carbon-cycle, ice-sheet, vegetation and other slow feedbacks increasingly matter on longer horizons. The largest residual physical uncertainties concern aerosol forcing, cloud feedback, the evolution of sea-surface-temperature patterns, carbon-cycle sensitivity and nonlinear changes in major Earth-system components. These uncertainties are asymmetrical: they widen the range of plausible magnitudes and regional outcomes but do not materially weaken attribution of the observed long-term warming trend. Palaeoclimate evidence strengthens inference about climate sensitivity and state dependence, while also showing that boundary conditions and forcing histories limit direct analogy with the modern industrial perturbation. A defensible interpretation therefore combines multiple independent lines of evidence rather than privileging any single temperature record, model ensemble or historical interval. Future progress depends on sustained global observing systems, improved process representation, tighter forcing constraints, better treatment of model dependence and structured integration of palaeoclimate, observations and mechanistic theory.

This paper synthesizes regional paleoenvironmental records in order to evaluate the role of fires and associated erosion during two critical Late Quaternary intervals: the last interglacial (MIS 5e, ~129–116 ka), dominated by natural climatic forcing, and the early Holocene (~11–7 ka), when human activity became increasingly significant. The evidence includes micro-charcoal concentrations, speleothem carbon and strontium isotopes, pollen records, lacustrine and alluvial sediments, and archaeological data from the Dead Sea–Jordan Valley system and the Mediterranean hills and coast. Both periods exhibit pronounced peaks in micro-charcoal abundance synchronous with major indicators of vegetation loss and hillslope soil erosion. During MIS 5e, extreme climatic instability was experienced, with intense lightning-ignited wildfires. These fires coincided with soil stripping, elevated δ 13 C values indicating loss of C3 vegetation, enhanced detrital input to the valleys, and widespread geomorphic instability. A similar but lower-magnitude fire–erosion cascade occurred during the early Holocene, peaking around ~9.5 ka. Although limited anthropogenic burning during the Neolithic cannot be excluded, the spatial extent, synchronicity, and proxy evidence indicate predominantly climatic control, linked to orbital forcing and incursions of dry thunderstorms. Hillslope erosion during both intervals was accompanied by large-scale redeposition of eroded soils in valley bottoms, forming thick alluvial fans and floodplain deposits. During the Neolithic period, such redeposited soils provided favorable conditions for early agriculture and coincided with the concentration of large Neolithic settlements in the Jordan Valley, while contemporaneous abandonment took place in upland areas. It is proposed that Neolithic communities, observing fire-baked clay in recently deposited valley sediments, may have derived the concept of ceramic firing — termed here the ‘natural kiln hypothesis’.

Abstract High-latitude regions are emerging as sensitive and potentially expanding sources of atmospheric mineral dust under contemporary climate warming. Aeolian activity in sub-Arctic landscapes such as Sweden during the last deglaciation may therefore act as a valuable analogue for understanding these future high-latitude dust–climate feedbacks. However, the timing and drivers of postglacial aeolian activity in Sweden remains uncertain. Furthermore, Swedish wind-blown loess deposits are thin and prone to reworking, complicating reliable age determination. Here, we combine high sampling resolution luminescence (quartz OSL, feldspar IRSL) and radiocarbon dating with complementary sedimentological, mineral magnetic and multi-proxy luminescence signal analyses to resolve postglacial–Holocene aeolian activity at Brattforsheden, Värmland. Our analyses demonstrate the value of luminescence sensitivity and portable OSL for assessing age reliability and differentiating between reworked and in situ deposits in thin loess sediments. We show that an intense but short period of postglacial-Holocene aeolian activity occurred in central Sweden immediately following deglaciation, with dust activity persisting for ~ 1500 years after regional ice retreat. We also reveal a later aeolian reworking phase linked to local fire activity at ~ 7 ka, setting new constraints on high-latitude dust dynamics under changing conditions, and implying that even during peak Holocene warmth the landscape was sensitive to aeolian reworking and dust emission could occur.

Abstract Antarctic precipitation is the dominant positive term in the ice sheet surface mass balance, directly influencing its contribution to global sea level rise. In some coastal and ice-shelf regions, episodic extreme precipitation events (EPEs) can account for more than half of the annual snowfall total; yet while individual atmospheric phenomena such as atmospheric rivers (ARs) and cyclones are known to play important roles, their combined effects remain largely unquantified, as most existing studies focus on the impact of individual drivers. Here, we assess how atmospheric phenomena interact and shape Antarctica‘s hydroclimate by applying the Multi-Object Analysis of Atmospheric Phenomena (MOAAP) to ERA5 reanalysis over the 2001–2020 period. The algorithm simultaneously detects ARs, mid-level cyclones, anticyclones, atmospheric fronts, and upper-level jets, enabling quantification of their individual and joint contributions to mean and extreme precipitation in Antarctica within a unified framework. Our results reveal that mean and extreme precipitation are controlled by broadly the same synoptic-scale systems, with cyclones dominating throughout the entire analysis domain. However, ARs play a disproportionately large role at the highest precipitation intensities, consistent with their ability to deliver concentrated moisture streams to the Antarctic coastline. Most importantly, an analysis of the 100 most intense events across the three IPCC Antarctic regions reveals that extreme precipitation events are almost always driven by multiple phenomena, with ARs, cyclones and jets co-occurring in 40–50% of events and forming a mutually reinforcing dynamical system. These findings imply that projections of future Antarctic precipitation extremes based on individual phenomenon trends will be incomplete if the dynamical coupling between these synoptic-scale systems changes under warming.

Surface meltwater ponds on Antarctic ice shelves can trigger ice break-up. Using satellite records of observed ponds along with temperatures, snowfall rates, and melt rates from climate models, we show that ponding occurs more widely than theory predicts. Our calibrated method identifies vulnerable areas more accurately and suggests pond coverage could be over double what standard methods predict by 2100, highlighting the need for careful evaluation of climate thresholds used to predict ponding.

Iceberg mélange is the jumble of icebergs in front of some glaciers that calve into the sea. Some studies suggest mélange might help to control the retreat of glaciers. We studied 3 glaciers in NW Greenland where we used GPS sensors and satellites to track ice movement. We found that glaciers push forward and calve all year, including when mélange and landfast sea ice are present, suggesting mélange is not important in supporting glaciers, but may influence the seasonal calving cycle.

arXiv:2609.10378v1 Announce Type: new Abstract: The potential of crossing climate tipping points (TP) has reached the attention of many researchers and the general public. On the one hand, the basis for this concern is strengthening, with simulations showing that abrupt transitions might occur even for moderate emission scenarios. On the other hand, our understanding of what constitutes such transitions mathematically is becoming more nuanced. This leads to challenges for the fidelity of early-warning signals (EWS), which accompany bifurcations in systems that closely track a slowly changing steady state. Different kinds of out-of-equilibrium dynamics in a rapidly changing climate, as well as chaos, blur the critical tipping threshold and imply limits in predictability. Using EWS to detect loss of local stability from data requires careful evaluation in the high-dimensional system, and may have limited predictive power in a highly multistable climate, where there is more than a single, well-known alternative state. This calls for new methods using non-equilibrium statistical mechanics and dynamical systems to efficiently probe the global stability properties of climate models and observations.

Nature Geoscience, Published online: 07 September 2026; doi:10.1038/s41561-026-02086-y During the last ice age, the Atlantic Meridional Overturning Circulation (AMOC) underwent a series of abrupt reorganizations. The global impact of these events reflects the accumulation of heat in the ocean interior during the weak AMOC mode and its loss during the strong mode, facilitated by radiative feedbacks that altered the planetary energy balance.

ABSTRACT Murtoos and murtooized terrains are new types of subglacial landforms found in the Fennoscandian Ice Sheet (FIS) and Cordilleran Ice Sheet (CIS) regions. LiDAR‐based morphometric analyses have revealed substantial morphometric diversity among murtoos and murtoo‐related landforms, while sedimentological excavations have documented a wide range of architectural elements and lithological characteristics. Most importantly, murtoos are predominately composed of silt/clay‐poor, sandy and gravelly diamictons interbedded with sorted patches of sandy and gravelly sediments. There are two prevailing hypotheses, both suggesting that murtoo formation is related to subglacial flooding episodes during deglaciation. In the FIS region, murtoo genesis has been suggested to be dominated by sediment‐concentrated flows and flooding‐induced sediment transport and liquefaction, and murtoos are considered as depositional landforms. In the CIS region, murtooized terrains are suggested to be formed by subglacial compressional deformation (glaciotectonism), as a consequence of glacial dynamic reorganization and ice flow. In the present study, we analyse and compare the sediment composition and architecture of two morphometrically similar triangular‐shaped murtoos in the FIS region in Finland, located in two different settings. The results demonstrate that murtoo internal structures vary considerably, and these variations cannot be inferred from morphometric characteristics alone. Murtoos may contain different sediment types depending on their depositional environment. They may include a high proportion of sandy to gravelly glaciofluvial sediments with upper‐flow regime (UFR) and lower‐flow regime (LFR) bedforms within their main interior, or they may consist predominately of crudely stratified sandy to gravelly diamicton deposited by subglacial slurries and forming the main sediment type of the murtoo body. The results also indicate that murtoo composition and sediment architecture are dependent on the magnitude of subglacial flooding, which largely relates to their spatial association with subglacial meltwater corridors (SMC). Neither one of the presently studied murtoos exhibits significant features of glaciotectonism, reported from the CIS region. Consequently, murtoos and murtooized terrains may represent equifinal subglacial landforms, and their interpretation requires a combination of regional geomorphological mapping and sedimentological analysis.

We present a high-resolution, quantitative reconstruction of dust provenance in the EPICA Dome C ice core (33.7–2.8 kyr BP) using Rare Earth Elements (REE). Dust was mainly sourced from Patagonia during glacial periods, shifting toward Australia, Southern Africa, and the Puna-Altiplano after 14.5 kyr BP due to sea-level rise and hydrological rearrangement in Patagonia. These changes also reflect major reorganizations of Southern Hemisphere atmospheric circulation.

Antarctic surface mass balance (SMB) is a source of major uncertainty in ice-sheet mass balance estimates, but its evaluation remains limited. We introduce a novel method using continuous, bedrock-fixed GPS vertical displacements as a spatially integrated constraint on low-frequency SMB variability over hundreds of kilometres. Applied across Antarctica, it reveals systematic, region-dependent differences with modelled SMB-driven displacements, guiding SMB model refinement.

arXiv:2609.09247v1 Announce Type: new Abstract: Local heatwave hazard depends on fine-scale air temperature, but ground stations are sparse and reanalysis products such as ERA5 cannot resolve the terrain and land-surface contrasts that shape real heat exposure. We present a conditional diffusion emulator for high-resolution 2-m temperature downscaling, conditioned on static geography, a training climatology, exact-time ERA5 temperature, and solar and temporal features, guided at inference by score-based data assimilation (SDA): a differentiable Gaussian observation likelihood steers the diffusion score toward sparse revealed temperature observations without any retraining. On a controlled 32-case synthetic-grid protocol over AORC, guidance improves hidden-cell reconstruction over both ERA5 and a strong observation-proximal nearest-neighbor baseline once observation density reaches 1\% (RMSE 0.318 vs.\ 0.431~K, winning all 32 cases), while sparser regimes still favor direct interpolation. We further map the full guidance-strength landscape across three observation densities, showing that the optimal strength shifts systematically with density and that over-guiding causes sharp, predictable degradation -- giving a concrete operating recipe rather than a single untuned setting. The resulting fields are intended as a temperature layer for downstream heatwave-hazard products such as threshold exceedance and cumulative heat-burden. The present evidence is a controlled synthetic-grid validation; station-network and held-out-year evaluations are the next steps toward deployment.

We analyzed simulations from Earth system models (ESM) with a constant rate of emissions, zero emissions, and negative emissions of CO2 to quantify the response of land carbon sinks. We found that under positive emissions vegetation in the tropics gained carbon. Under zero emissions and negative emissions most ESMs lost carbon from vegetation in the tropics but gained carbon in mid- and high-latitudes, mostly in soils. Our findings imply that tropical carbon is vulnerable under zero emissions.

Melt rivers from valley glaciers and ice sheets release methane by transporting microbial or geologic methane from beneath glaciers, making it available for atmospheric emission. However, the mechanisms and magnitude of such methane release remain poorly understood due to limited studies. Here, we show how subglacial geology and glacier thermal conditions control methane supersaturation in meltwaters from 19 valley glaciers across central Svalbard. We find that all sampled meltwaters are supersaturated with methane, with peak concentrations up to 425-times atmospheric equilibrium. The carbon isotopic compositions of the methane and the presence of ethane and propane indicate that the methane is largely of thermogenic origin. Ground-penetrating radar surveys show that the extent of temperate ice at the glacier bed plays a critical role in facilitating methane acquisition, particularly in regions underlain by organic-rich, shale-dominated geological formations. This mechanistic insight is relevant for glaciated regions worldwide with organic-rich geology and suggests that accelerated glacier melt may amplify methane mobilization from beneath glaciers.

Journal of Geophysical Research: Oceans, Volume 131, Issue 9, September 2026.

Where and when water is available is a pivotal control on environments. Changes in water availability are frequently invoked to explain past changes in human, flora, fauna, and fire behaviour, especially regarding the transition from the Late Pleistocene to the ostensibly more complex Holocene. Previous syntheses have characterised the Holocene as a period of uniquely elevated hydroclimate complexity—a hypothesis that has significant implications for interpreting past human and ecological responses to climate change. Reconstructions from further back in time tend to be poorer quality, which must be considered when synthesising records. This study analyses 69 natural palaeoenvironmental archives from tropical Australia spanning the last 130,000 years. Archives are highly concentrated near the coast. Temporal sample resolution of archives tends to decrease with age, with no published hydrological records of sub-decadal resolution before 1900 years ago. We analyse a random subset of archives from the well-represented Holocene epoch, equal to the number of archives available in past stages, finding no support for the hypothesis that the Holocene is uniquely hydrologically heterogeneous. We also show records for the well-dated, densely sampled Girraween Lagoon archive do not become more temporally variable in the Holocene. These analyses suggest the trend of reduced resolution in the past could be mistaken as a trend towards greater complexity. To understand how any region fits into the greater climate system, Quaternary science must be explicit about what can and cannot be determined from the data available.

Earth's Future, Volume 14, Issue 9, September 2026.

arXiv:2609.12815v1 Announce Type: new Abstract: We present BEAST, the first-ever Bayesian Swin Transformer for atmospheric forecasting on 0.25$^\circ$ global resolution able to accurately quantify both aleatoric and epistemic uncertainty. To overcome the associated computational bottlenecks, we devise an orthogonal 4D-parallelization scheme that introduces a unique domain-tensor-parallelism strategy and a novel uncertainty parallel method, enabling us to fully leverage GPU capacity and efficiently scale model training. For a 2.4-billion-parameter model, we achieve a peak performance of 3.96 EFLOP/s on 20,480 NVIDIA GH200 GPUs on the JUPITER supercomputer. We train BEAST as a 700-million-parameter model with 96 random weight samples on 384 nodes on 40 years of data for nearly one million gradient updates. This model achieves predictive skill scores competitive with state-of-the-art probabilistic atmospheric AI models and numerical models, and can predict extreme events with exceptional skill, while generating large ensembles 3 to 4 times faster than the current-best AI model. Our contribution unlocks the potential of high-fidelity uncertainty quantification in atmospheric AI models, heralding a new era for AI-based models in climate and Earth system sciences.

Understanding the seasonal expression of temperature signals preserved in terrestrial proxies remains a key challenge in paleoclimatology. Here, we present a high-resolution pollen record from the Greater Khingan Mountains, northeastern China, spanning the last 15.2 kyr, to investigate seasonal temperature evolution since the last deglaciation. Principal component analysis (PCA) of pollen assemblages identifies two distinct climate-related signals. PCA Axis 1 primarily reflects winter temperature variability, with higher values indicating colder winter conditions and stronger influence of the East Asian winter monsoon (EAWM), whereas PCA Axis 2 represents growing-season temperature variability associated with changes in the East Asian summer monsoon (EASM). The reconstructed seasonal temperature records reveal contrasting trajectories since the last deglaciation. Growing-season temperatures increased rapidly during the early Holocene, coinciding with maximum EASM intensity, whereas winter temperatures continued to rise into the middle Holocene. This seasonal contrast resulted in a delayed ecological optimum in the Greater Khingan Mountains. The Holocene Optimum (HO) inferred from vegetation development occurred between ∼6 and 3 ka BP, when temperate forest elements reached their maximum expansion under relatively favorable annual thermal conditions, despite the weakening of summer monsoon intensity after the early Holocene. These results demonstrate that pollen assemblages can preserve separable seasonal climate signals and highlight the importance of winter temperature evolution in regulating vegetation responses at mid-high latitudes.

Earth System Model workflows commonly run on highly congested high-performance computing platforms, meaning that each individual workflow task potentially faces lengthy waiting times in the queues of the schedulers. In this work, we evaluate the task aggregation approach in EC-Earth3 workflows to reduce the queue times and, consequently, the total execution time. The results show up to 23% more simulated years over the same time span, with queuing times reduced by up to a factor of 12.33.

The 60-m deep Plestia1 borehole characterises the most recent fill of the Quaternary Colfiorito basin, an intermontane tectonic depression in the Northern Apennines of Italy. The subsidence rate is controlled by a major SW-dipping, seismogenic normal fault most recently activated (M max = 6) during the 1997-98 seismic sequence. The core has been the subject of lithostratigraphic, paleomagnetic, palynological, tephrostratigraphic and micro-paleontological analyses, which define a sedimentary succession composed predominantly of silts and clays, consistent with deposition in a lacustrine environment. These sediments retain a stable, normal-polarity magnetization. Two tephra layers at 58.67 m and 38.89 m, corresponding to the Maddaloni/X-6 eruption from Campi Flegrei and an undefined eruption sourced at the Latium volcanic complex, have ages of ∼109 ka and ∼86 ka, respectively. A radiocarbon date on ostracod shells extracted from sediments at 8.90 m depth provides an age of ∼26 ka. Pollen analysis shows a succession of vegetational phases consistent with the Late Pleistocene climatic oscillations from MIS 5d to MIS 2, while the ostracod assemblage in the upper part of the succession is consistent with LGM climatic conditions. Overall, the paleomagnetic, radiocarbon, tephra, pollen and ostracod evidence indicates that Plestia1 spans the 110-15 ka interval. A Bayesian age-depth model was developed using these geochronological constraints and biostratigraphic alignments of the Plestia1 pollen profile with well-dated reference records, providing an average sedimentation rate of ∼0.6 mm/yr. Accounting for climatic and sedimentary contributions, this result represents a minimum estimate of the tectonic subsidence rate and implies a similar mid-term fault slip rate.

Abstract Coupled climate models suffer from persistent climatological biases in sea surface temperature (SST). The exact causes of these biases are often difficult to diagnose given the complex ocean-atmosphere interactions underpinning the seasonal cycle within these regions. Here, we investigate the possibility of reducing these biases within both low- and medium- resolution configurations of a fully-coupled climate model (CESM) by globally, and regionally, replacing the net shortwave (SW net ) flux at the surface with climatological values derived from Clouds and Earth's Radiant Energy Systems (CERES)-Energy Balanced And Filled (EBAF). This approach reduces the global mean SST bias. While the SST biases are reduced in several regions, such as the northern hemisphere subtropical gyres and the Pacific cold tongue, biases persist in some regions, notably the Southern Ocean. The persistence of these biases despite the shortwave forcing being correct points to shortcomings in other processes controlling the net surface heat flux, including the representation of surface mixed-layer processes in the ocean component. Additional regional replacement experiments highlight the contribution of both local and nonlocal biases in shortwave forcing towards SST biases.

arXiv:2609.12274v1 Announce Type: cross Abstract: Observations at high frequency have become dramatically more abundant with recent technological advances. Ultra-rapid data assimilation (URDA) has been proposed to exploit them, frequently updating ensemble forecasts at lower cost without re-integrating the forecast model. In this study, we examine the applicability of URDA to a realistic numerical weather prediction (NWP). Specifically, we conducted idealized experiments for the heavy rainfall event of August 2021, using the regional atmospheric model Scalable Computing for Advanced Library and Environment (SCALE-RM). In the experiments, pseudo-observations emulating the Automated Meteorological Data Acquisition System (AMeDAS) were assumed to become available every 10 min. The results show that the threat score for hourly precipitation is generally improved in the forecasts updated by URDA relative to the pre-existing baseline forecasts, through the ensemble-based error covariance. Furthermore, for sea level pressure, temperature, relative humidity, and winds, the root mean square error of the URDA-updated forecasts is reduced relative to that of the baseline forecasts as the forecasts are successively updated by assimilating additional observations. These results indicate the potential of URDA to operate effectively with a realistic NWP model.

Transport processes like rocks carried by ice flow and damage evolution – a proxy for crevasses – are key in ice sheet modeling and should occur without diffusion. Yet, standard numerical methods often blur these features. We explore a particle-based Semi-Lagrangian approach, comparing it to a Discontinuous Galerkin method, and show it can accurately simulate such transport when run at high enough resolution.

The Geoengineering Model Intercomparison Project is a coordinated international model intercomparison effort aiming to provide robust experimental protocols for simulations of various Solar Radiation Modification (SRM) methods. In this manuscript, we describe a list of novel experiments decided by the community, to be run by climate models to better understand sources of uncertainty in SRM. We also show and discuss some preliminary results from a few climate models.

Subfossil pines from the banks of the Drouzet River in the southern French Alps provide a unique Late Glacial archive for reconstructing past solar activity. Dendrochronological analyses produced three floating chronologies together spanning ≈ 700 calendar years. High-precision 14C measurements obtained at CEREGE with the AixMICADAS accelerator mass spectrometer, and calibrated by Markov Chain Monte Carlo wiggle-matching against the IntCal20 absolute chronology, reveal a large and abrupt Δ14C increase of ≈ 30‰ at 14 300 cal yr BP. Carbon cycle model simulations forced with the GRIP 10Be flux record are compatible with this spike, confirming a sharp enhancement in cosmogenic isotope production linked to an extreme solar particle event (ESPE). Lower geomagnetic field intensity and reduced atmospheric CO₂ during the Late Glacial contributed to amplifying the ∆14C signal. Recent three-dimensional atmospheric 14C modelling indicates that this event represents the most powerful ESPE evidenced so far, much stronger than the largest ground-level enhancement of cosmic rays observed over the last 80 years of instrumental data. These results extend the ESPE record beyond the Holocene and demonstrate that subfossil wood has the potential to offer annually resolved evidence of ancient solar events, providing new constraints for radiocarbon calibration and solar-terrestrial interaction studies. This article is part of the Theo Murphy meeting issue 'Radiocarbon and cosmic radiation events'.

In geosciences, data assimilation (DA) combines modeled dynamics and observations to reduce simulation uncertainties. Uncertainties can be dynamically and effectively estimated in ensemble DA methods. With respect to current techniques, the novel Gauss-Hermite High-Order Sampling Hybrid filter (GHOSH) ensemble DA scheme is designed to improve accuracy by reaching a higher approximation order, without increasing computational costs, as demonstrated in idealized Lorenz96 tests.

Geophysical Research Letters, Volume 53, Issue 17, 16 September 2026.

The mid-Holocene decline in elm ( Ulmus ) pollen frequencies is perhaps the most debated and researched pollen-analytical feature of vegetation history in northwest Europe, with various causes or combination of causes proposed to explain it. It has a significant radiocarbon age range in Britain, but its timing in the early stages of Neolithic settlement suggests that subsistence farming activities probably played a role as well as factors such as disease and climate change. With these variables unproven, we have investigated ascospores of the wood-rot fungus Ustulina deusta as an independent proxy indicator of elm tree mortality, as the distinctive spores are often stratified in Elm Decline levels. Inspection of 20 sites in north Britain shows the distinct association of U. deusta with the Elm Decline, although with considerable variation between sites in the amplitude and shape of the U. deusta curve across the Elm Decline levels. U. deusta relies for successful infection on physically damaged trees, and we contend that girdling of elms by early Neolithic farmers to provide clearings for cereal cultivation and cattle grazing provided the conditions for U. deusta infection. Curves for Plantago lanceolata and Sporormiella , other indicators of herbivore presence, mirror the U. deusta curve, supporting this hypothesis. We conclude that the behaviour of the U. deusta curve at the Elm Decline supports early Neolithic animal husbandry and woodland manipulation as a primary cause of the mid-Holocene fall in elm populations.

Shallow littoral environments in dynamic coastal basins undergo complex interactions between natural hydrodynamic regimes and anthropogenic trace element inputs. This study provides an integrated lithostratigraphic, quantitative ecological, and bio-monitoring assessment of modern marine sediments across 15 coastal locations in the Antalya Gulf (Eastern Mediterranean, Türkiye). Pre-Quaternary basement units are unconformably overlain by five distinct Quaternary depositional units. To account for abundance variations in naturally depauperate littoral sites, biodiversity indices (Simpson, Margalef, Pielou) were derived directly from 1 g dry sediment-normalized datasets. Across the coastline, 95 benthic foraminiferal species were identified, with diversity indices peaking in northwestern zones and declining toward the southeast. The Lessepsian immigrant Amphistegina lobifera Larsen, 1976 dominated across most locations, whereas Peneroplis de Montfort, 1808 thrived in high-salinity micro-environments. Severe test discoloration (HCT) directly aligned with dark-colored, Fe-enriched sediments and site-specific trace metal accumulation (e.g., Cr, Ni, Cu, As). However, statistical analysis revealed no significant linear correlation between Fe content and Foraminiferal Abnormality Indices (FAI) ( r = 0.175, p = 0.888 > 0.05, n = 3), confirming that Fe enrichment primarily acts as a geochemical driver of test darkening rather than a direct trigger for structural deformities. Consequently, while benthic foraminifera serve as sensitive bio-indicators of sediment degradation, overall biodiversity indices remain governed by natural littoral hydrodynamics rather than direct trace metal stress. This integrated foraminiferal-geochemical framework offers a transferable tool for coastal quality monitoring across stressed Mediterranean ecosystems.

Focusing on the Xinglong site—one of the region's earliest Neolithic settlements—this study integrates multi-dimensional particle size analysis with multiple climate proxies, including geochemical elements, clay minerals, total organic carbon (TOC) and pollen data, to reconstruct the paleoclimatic sequence since 11,700 cal yr BP and explore human-environment interactions. The environmental evolution in the Bashang region since 11,700 cal yr BP exhibits three stages: (1) Early Holocene (11,700–10,000 cal yr BP): characterized by warm-humid conditions and frequent climatic fluctuations. The stable, low-energy sedimentary conditions created by local topography provided favorable conditions for early human habitation. (2) Early–Mid Holocene (10,000–5000 cal yr BP): generally favorable but interrupted by cold-dry events. The enhanced dynamic events may have pushed settlements from valleys to slopes and promoted mixed farming and hunting-gathering economies. The warm-wet interval 9000–7000 cal yr BP may have supported the flourishing Yumin Culture, whereas aridification after 7000 cal yr BP correlates with weakened cultural continuity and greater human mobility. (3) Late Holocene (since 5000 cal yr BP): persistent cold-dry climate may have contributed to site abandonment, with brief reoccupation during historical dynasties. These findings suggest environmental changes may play an important role in the development and transformation of Neolithic cultures in this region.

This study develops and tests new methods to improve weather forecasts by using radar observations within a modern data assimilation system called the Joint Effort for Data Assimilation Integration. The approach combines information from radar measurements and computer models to better describe storms. Tests with a major U.S. storm show improved prediction of rainfall and storm structure.

Semi-arid and arid regions are largely omitted from long-term geological-scale carbon budgets due to the assumed weakness of silicate weathering. Here we reconstruct annual carbon dioxide (CO₂) consumption by silicate weathering (CO₂(SIW)) using red clay and loess–paleosol sequences from the Chinese Loess Plateau (CLP). We show that variability in eolian mass accumulation rate, rather than intrinsic silicate weathering intensity, primarily controls CO₂(SIW), consistent with persistently low-to-moderate chemical weathering across the CLP. Scaling these results to global eolian-deposited regions reveals that CO₂(SIW) increased from ~3.7 to ~18.8 Tg C yr⁻¹ between 4.0 and 1.0 Ma, followed by a decline to ~13.0 Tg C yr⁻¹ thereafter. This long-term trend broadly coincides with the late Pliocene decrease in atmospheric CO₂. Our findings provide a quantitative, million-year-scale budget of dryland CO₂ drawdown by silicate weathering and highlight eolian-dominated drylands systems as a previously underrecognized component of positive feedbacks in the global carbon cycle. “Red clay and loess-paleosol records from the Chinese Loess Plateau reveal that mass accumulation rate variability drives silicate-weathering CO₂ consumption in arid and semi-arid regions, closely tracking the late Pliocene atmospheric CO₂ decline.”

Journal of Geophysical Research: Oceans, Volume 131, Issue 9, September 2026.

ABSTRACT Studying historical dynamics and compositional changes of ecotones is essential for improving predictions of their responses to anthropogenic global change. However, the small size and geographic isolation of marginal ecotonal stands require retrospective analyses at the local scale. Here, we combine soil macrofossil charcoal analysis (SMCA), a lake‐sediment pollen record, and contemporary forest inventories to reconstruct past dynamics and assess contemporary trajectories of forest stands located at the northern boundary of the boreal‐temperate ecotone (BTE) in the Rimouski hinterland (Lower St. Lawrence region, Canada). Radiocarbon‐dated soil charcoal particles and regional pollen data reveal a delayed local establishment of thermophilous species, such as white pine ( Pinus strobus ), toward the end of the Holocene Climatic Optimum. This lagged expansion of temperate taxa coupled with a decline of boreal conifers led to the development of thermophilous assemblages that persisted into the early Neoglacial despite climatic cooling. Borealization of the landscape did not begin until the late Neoglacial (⁓1500 cal year BP), coinciding with shifts in the fire regime and the arrival of the eastern white cedar ( Thuja occidentalis ). At the stand‐scale, comparisons among charcoal assemblages in buried and surface soil layers with dead and living trees reveal an unprecedented proliferation by trembling aspen, as well as a recent establishment of jack pine likely reflecting anthropogenic disturbances associated with European settlement. However, current size‐class distribution suggests that these recent changes are likely to be transient in the absence of recurrent disturbances. Although predictive modeling remains necessary to forecast future forest trajectories under climate change, our results underscore the value of multi‐proxy reconstructions for disentangling vegetation responses to interacting climate and disturbance drivers.

Nature Climate Change, Published online: 11 September 2026; doi:10.1038/s41558-026-02752-4 Climate extremes influence migration, but their long-term and compounding impacts are understudied. The authors show that associations between climate extremes and migration vary by hazard type, timing, the occurrence of consecutive events and local socioeconomic conditions.

arXiv:2609.05359v1 Announce Type: new Abstract: Long-term climate data exhibit variations composed of internal and forced components. Internal variability arises from natural processes that could occur within a stable climate. Forced variability, on the other hand, reflects climate changes induced, for example, by anthropogenic greenhouse gas and aerosol emissions. Accurately distinguishing between these types of variability is crucial for attributing climate fluctuations and understanding internal variability processes and impacts. In this study, we apply a U-Net convolutional network, a model commonly used in computer vision, to separate forced and internal climate variability from 1950 to 2022 using a multi-model dataset. The dataset includes multiple fields from five single-model initial-condition large ensembles. Cross-validation is conducted by training the U-Net using the data from the ensembles of four models, leaving out the data from one model to assess performance. Validation results yield errors ranging from 0.1{\deg}C to 0.4{\deg}C for the forced variability of local-scale monthly surface air temperature, which is no more than half the magnitude of external forcing. The U-Net achieves better performance than a simple approach based on a fourth-order polynomial trend for estimating forced variability. The error is mainly due to insufficient sampling and poor agreement among the models, as the U-Net underestimates the warming for the model with the highest transient climate sensitivity during validation. Performance is lower for variables like sea-level pressure and precipitation because of their low ratio of forced to internal variability. This framework might be enhanced by incorporating a larger multi-model dataset in the training and validation.

Preprint under review for ESSD (discussion: open, 0 comments) The Artificial Intelligence for Stable Isotope Tracers (AISIT) project unified disparate pan-arctic δ18O and ancillary data into one AI-ready database, harmonised under unified quality control and adhering to Findable, Accessible, Interoperable and Reusable principles. Drawn from repositories, cruises, campaigns and private communications, it holds 45,714 δ18O and 5531 barium measurements from 99 datasets spanning 1967–2025, giving oceanographers and modellers a rich, harmonised resource.

by Franck Hervé Akaffou, Salomon Obahoundje, Arona Diedhiou, Kouakou Lazare Kouassi, Dior Diallo, Ernest Amoussou, Daniel Yamegueu, Eric Antwi Ofosu This study presents a novel five-step ensemble machine learning approach to improve predictive accuracy in assessing climate change impacts on inflow patterns and hydropower generation across seven dam basins in West Africa. The methodology integrates precipitation and temperature using the multi-lag approach. An initial pool of fifteen machine learning models were evaluated, and top-performing models were selected for further refinement through iterative ensemble stacking and weak learner elimination. Historical analysis (1983–2014) utilized CHIRPS and CHIRTS datasets. Future projections employed twelve bias-adjusted CMIP6 models and their ensemble mean (EnsMean) under SSP1-2.6, SSP2-4.5, and SSP5-8.5 for the near (2036–2067) and far (2068–2099) futures. Results showed notable improvements in model accuracy and efficiency across layers, with R² and NSE exceeding 0.6 for all inflow simulations and for selected energy simulations (Bagre, Nangbeto, and Taabo). Projections indicated warming up to 4.5°C and spatially heterogeneous precipitation changes across basins and scenarios, with SSP5-8.5 projecting the most pronounced shifts. Inflow reductions are projected to reach up to 24% at Buyo and 13% at Nangbeto, while hydropower output may decline by up to 19% at Nangbeto and 58% at Taabo in the future. Conversely, Manantali and Taabo are projected to experience inflow increases, and Bagre may see energy gains of up to 42% under SSP5-8.5. These findings highlight heightened vulnerability, as well as contrasting opportunities across the region, underscoring the urgent need for adaptive management strategies, such as enhancing hydropower system resilience, diversifying energy portfolios, and integrating renewable sources to mitigate climate risks. Hydropower managers and policymakers must prioritize proactive measures to ensure energy security and sustainable resource management amid changing climatic conditions.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

arXiv:2609.08241v2 Announce Type: replace Abstract: Despite growing interest in AI-based prediction of climate extremes, event- or label-based AI-ready extreme climate datasets remain limited, constraining efforts to systematically characterize and forecast such phenomena. To address this gap, we present EastAsiaClimateExtremes, an open dataset that provides ERA5/OISST reanalysis-based weekly extreme labels and event-based metrics for anomalously high temperature (AHT), heavy rainfall (HR), and marine heatwaves (MHW) over East Asia, together with analysis workflows hosted on GitHub to facilitate reproducibility and adaptation. The dataset is fully documented and co-registered with ECMWF S2S hindcast outputs on a common spatial grid and temporal framework, thereby enabling direct comparison between reanalysis-derived labels and dynamical model forecasts. This unified dataset serves as a reference framework for East Asian climate extreme research and AI-based subseasonal-to-seasonal prediction. It supports both quantitative characterization of the spatiotemporal occurrence of regional extremes and systematic diagnosis of S2S model skill in reproducing extreme signals. Beyond these immediate applications, the dataset enables a broader range of studies such as extreme event attribution and compound risk analysis. The accompanying analysis workflows characterize the historical statistics of reanalysis-based weekly extremes-including occurrence frequency and mean and maximum intensity-together with their climatological means and trend characteristics, and further assess the skill of the ECMWF hindcast.

Abstract We investigate late Holocene changes in the northern Benguela Upwelling System using a multi‐proxy approach on shallow shelf deposits from the Namibian mud belt. Three sediment cores characterized by organic carbon‐rich, anoxic, and partly gas‐bearing sediments were retrieved along a sampling transect at 23°S in August 2019. A sub‐bottom profile recorded in a sampling campaign in February 2022 enabled stratigraphic correlation of the cores. As a novelty in this area to date, variations in terrigenous input patterns are reconstructed in one gravity core using major and trace element data, including rare earth elements, in a statistical provenance modeling approach. Results indicate a progressive decline in input of Orange River‐derived material over the past 3800 years, attributed to a weakening of the Benguela Coastal Current (BCC). This reduction in northward directed transport energy is reflected in increasing lamination frequency observed in Computed Tomography‐scan imagery and declining Zr/Al ratios toward the present. Sulfidic events observed along the Namibian coastline are interpreted as a direct consequence of the BCC weakening, since decreasing depositional energy facilitated organic matter deposition and degradation, the formation of a shallow sulfate–methane transition zone (SMTZ) and the accumulation of free sulfide and methane gas in the central mud belt area beginning around 1450 years BP. The establishment of sulfidic redox conditions in gas‐bearing sediments is recorded in sedimentary Mo/U ratios exceeding modern seawater Mo/U. This study highlights the strong coupling of oceanographic changes, sedimentation processes, and biogeochemical conditions in the Benguela Upwelling System.

Glacier forelands are well-suited for investigating vegetation development and plant colonization dynamics in previously uninhabited areas. The areas exposed by retreating glaciers represent sites without seed banks; thus, plant colonization almost always represents a true primary succession. Vegetation dynamics relies on diaspores brought in from the surrounding. A common approach for studying long-term vegetation developments is the employment of so-called chronosequences, deriving temporal sequences from spatially different sites of various site ages. In glacier forelands, datable traces of former glacier extents are generally used for site age estimation. In 2009 and 2010, vegetation sampling for ten points in time since deglaciation was carried out in the foreland of Schwarzenbergferner, Ötz valley, Austria. The study area extends between the recent glacier terminus and the terminal moraine left by the Little Ice Age glacier during the maximum extent (around 1850). The vegetation survey is supplemented by a one-year long measurement of soil temperature (1 September 2010–31 August 2011), which allows not only for an assessment of the seasonal course of the topsoil thermal conditions across the chronosequence, but also for the determination of the snow cover duration and the length of the vegetation period. This article summarizes the vegetation development and microclimatic conditions along the chronosequence and discusses the results, considering the interplay between the time since deglaciation and elevation being a microclimatological proxy.

Earth's Future, Volume 14, Issue 9, September 2026.

We used artificial intelligence to reconstruct a complete global temperature record since 1850, including better coverage of the poorly observed Antarctic region since 1961, by filling gaps in historical data and improving understanding of long-term climate change.

One likely consequence of climate warming is the change in glacier meltwater dynamics, which could lead to a shift in the seasonal water supply.Hydrograph separation models are therefore in focus, as they are used to identify contributions of ice and snow meltwater runoff.These models often use electrical conductivity (κ) data, which is easy to measure and inexpensive.To better understand the temperature dependence of κ in hydrograph separation studies, stable water isotopes ( 18 O/ 16 O and 2 H/ 1 H), temperature-compensated and non-compensated κ values, and chloride (Cl) were used to construct multi-component isotope and geochemical mixing models.These models were used to quantify the relative contributions of ice meltwater and snowmelt water to the total discharge of different glacier-fed streams of the Central Himalayas.The results showed that non-temperature-compensated κ values, i.e., widely used in the literature, could lead up to 14% temperature-dependent bias in estimating the meltwater fractions with a temperature variability of 0°C-5°C.As the conservative behaviour of κ can be questioned, δ 18 O-and Cl-based mixing models were also applied.Results from the δ 18 O-κ and δ 18 O-Cl mixing models are in close agreement.These findings suggest that κ values should be referenced to 25°C to accurately estimate meltwater fractions in streams.

arXiv:2604.16238v3 Announce Type: replace-cross Abstract: Decision-makers rely on weather forecasts to plant crops, manage wildfires, allocate water and energy, and prepare for weather extremes. Today, such forecasts enjoy unprecedented accuracy out to two weeks thanks to steady advances in physics-based dynamical models and data-driven artificial intelligence (AI) models. However, model skill drops precipitously at subseasonal timescales (2 - 6 weeks ahead), due to compounding errors, systemic model biases, and the chaotic nature of the atmosphere. To counter this degradation, we introduce probabilistic bias correction (PBC), a machine learning framework that substantially reduces systematic error by learning to correct historical probabilistic forecasts. When applied to the leading dynamical and AI models from the European Centre for Medium-Range Weather Forecasts (ECMWF), PBC doubles the modest subseasonal skill of the AI Forecasting System and improves the skill of the operationally-debiased dynamical model for 91% of pressure, 92% of temperature, and 98% of precipitation targets. We designed PBC for operational deployment, and, in ECMWF's 2025 real-time forecasting competition, its global forecasts placed first for all weather variables and lead times, outperforming the dynamical models from six operational forecasting centers, an international dynamical multi-model ensemble, ECMWF's AI Forecasting System, and the forecasting systems of 34 teams worldwide. These probabilistic skill gains translate into more accurate prediction of extreme events and have the potential to improve agricultural planning, energy management, and disaster preparedness in vulnerable communities.

Abstract The stability of vegetation regimes—the temporal persistence of plant assemblages—has fluctuated markedly since the last deglaciation. While climate is the primary driver of long-term vegetation dynamics, the effects of human land use change remain less clear. Here, we analyze global vegetation regime stability over the past 15,000 years using fossil pollen records and assess the impacts of land use type and intensification. Vegetation stability declined during the climatically variable deglaciation, increased in the early to middle Holocene when climate variability dropped, but declined sharply over the past 2000 years despite no increase in climate variability, with continent-specific differences in timing. Intensive agriculture and urbanism had the strongest negative effects on vegetation stability, followed by extensive agriculture and pastoralism, whereas foraging, hunting, gathering, and fishing had minimal impact. These findings suggest that recent declines in vegetation stability are linked to intensified human land use, highlighting the need to conserve natural land less intensively managed ecosystems.

We present a network of 26 sensors measuring snow water equivalent through cosmic rays across the Italian Alps. We study the application of a parameterisation shared by all the probes. The parameters were defined and validated with data taken during two seasons at 13 sites. We show that the parameterisation gives a good estimate of the water equivalent. This finding can contribute to expand data availability by installing similar probes in sites at high elevations and in inaccessible locations.

Preprint under review for ESSD (discussion: open, 0 comments) This study presents the SYSU SnowDepth dataset (1978–2025), the longest pan-Arctic snow depth on sea ice record covering both first-year ice and multi-year ice. Derived from passive microwave sensors using a new algorithm incorporating GRV(37/19), snow density, and time factor, it was validated against independent observations. Accuracy was high with root-mean-error-square (RMSE) values ranging from ~3 to 17 cm, offering vital support for climate and sea ice modeling.

Science Advances, Volume 12, Issue 37, September 2026.

Abstract Continuous, high-resolution sedimentary records from the Great Basin are rare. By combining physical sedimentology with charcoal analysis to document changes in lake level and biomass burning, we explore the coevolution of climate and fire ecology over the past 7000 years. Elemental data from scanning X-ray fluorescence (XRF) measurements suggest persistently low lake levels during the Middle Holocene (7000–4000 cal yr BP). Two prolonged low lake stands between 4200–3800 cal yr BP and 2600–2400 cal yr BP occurred during widely documented periods of abrupt environmental change throughout the Great Basin. Significant correlations between the first principal component of the XRF record (which explains >75% of the variance) and a bristlecone pine–derived temperature reconstruction suggest warmer temperatures were a major driver of reduced precipitation–evaporation balance. Spectral analysis shows interannual and multidecadal periodicities typical of Pacific Ocean–atmosphere systems have been influential across the region after 4000 cal yr BP. Severe fire events became more commonplace as precipitation increased, highlighting the role of fuel availability on biomass burning in this semiarid region. These results suggest that while near-term warming will likely lead to increased biomass burning, prolonged periods of warmth and aridity will catalyze returns to the fire regimes of the Middle Holocene.

Journal of Geophysical Research: Atmospheres, Volume 131, Issue 18, 28 September 2026.

Geophysical Research Letters, Volume 53, Issue 18, 28 September 2026.

Geophysical Research Letters, Volume 53, Issue 18, 28 September 2026.

Geophysical Research Letters, Volume 53, Issue 18, 28 September 2026.

Journal of Advances in Modeling Earth Systems, Volume 18, Issue 9, September 2026.

Rainfall during the snow season plays an increasingly important role in flood risk as climate warms and extreme events become more frequent. However, a given sized rain-on-snow (ROS) event can yield outcomes ranging from flooding to no runoff, depending partly on the snowpack's antecedent cold content and capillary retention forces. Here, we analyze the seasonal evolution of the snowpack's physical state over a 72-year period to assess long-term changes in its capacity to buffer runoff from liquid water input. We use ERA-5 Land data to force Alpine3D, a distributed snowpack model that tracks the layer-by-layer development of heat, mass, and structural framework of the snowpack throughout the snow season. We test our approach in a large Columbia River headwaters basin in NW Montana, USA. We evaluate cold content and total capillary retention of the snowpack to determine long term trends in Liquid Water Buffering Capacity (LW bc ) as it evolves throughout the snow season. The LW bc of the snowpack exhibited robust long-term declines across all elevation bands, that is driven by reduced SWE along with reduced cold content in late November through late January across all elevation bands, despite high intra- and interannual variability. The largest declines occurred during the Spring period, trending downward across the historical period by 43 % to 80 % depending on the elevation band. The core five weeks of mid-winter showed no trending change of LW bc , and in fact demonstrated an increase in cold content over the 72 years. Our findings demonstrate that changes in the snowpack's ability to buffer runoff, including dependencies on local basin factors related to snowpack seasonality and elevation, are a key component of evolving ROS risk.

ABSTRACT Central Asia (CA) has been warming steadily over recent decades, and the associated changes in climate conditions increase the need to understand how coldwaves and cold spells will evolve in the future. In this study, we examine changes in several coldwave‐related indices in five climate clusters, using Bias‐corrected CMIP6 estimates for the period 1992–2100 under the SSP2–4.5 and SSP5–8.5 scenarios. The present analysis combines clustering, bias correction methods, and trend analysis to accurately capture spatial and temporal patterns. The results indicate a significant reduction in the number of coldwaves (CWN). In the near‐future (2022–2051), CWN is reduced by about 28%–36% under SSP2–4.5 and 31%–39% under SSP5–8.5. Northern Kazakhstan (Cluster 5) experiences a decrease of about −30.7% under the high emission scenario. At the same time, coldwave duration (CWD) increases across the region—by 19%–46% overall and by +67% in Cluster 3 (eastern Kyrgyzstan and Tajikistan) over the period 2100–2071 under SSP5–8.5. The cold season duration index (CSDI) also grows dramatically in this cluster, and it rises by approximately +265.6% in the far‐future. Cold events severity exhibits an alternative pattern. The intensity (CWM) and amplitude (CWA) of coldwave are decreasing on a near‐future basis (approximately −70% and −94% respectively) and largest on Cluster 5 (−47.0°C 2 and −136.2°C 2 ). Nevertheless, there is relative recovery in both indices towards the end of the century, particularly clusters 3 and 5, in SSP5–8.5. This trend points towards fewer but more persistent cold events. The projected changes align with a large increase in Tmax and Tmin. Cluster 1 and 4 have Tmax over 36.5°C and Cluster 1 has the most rapid change (to 35.6°C). Clusters 3 and 5 that have historically been colder and mountainous, also undergo significant warming. All results underline the necessity of specific adaptation efforts, i.e., use of frost‐resistant crop varieties and more efficient heating systems, so that the societies of CA might adjust to changing threats of cold extremes in the future.

This study presents the TPHH (Tibetan Plateau Historical High) dataset, a high-resolution (1/30°) monthly climate dataset for the Tibetan Plateau spanning 1901–2023, featuring 2 m temperature, specific humidity, and surface pressure. By employing a hybrid deep learning framework (FourCastNet), we downscaled coarse historical data through the synergistic mapping of total-field signals and terrain constraints. Validated against independent observations, this physically-consistent dataset bridges the pre-satellite data gap.

ABSTRACT The sensitivity of winter recreation to climate change, especially to warming, is widely recognized but few studies report quantitatively on the observed effects of climate change on ski resorts, in part because consistent and available data directly from ski resorts is scarce. Augmenting strictly modelling‐based studies, we use both numerical modelling and observation‐based estimates of sensitivity in snow depth (HS) and snow water equivalent (SWE) to temperature and precipitation using data from nearby SNOTEL (snow telemetry) and snow course sites for 40 ski resorts in Washington, Idaho, Oregon and California during the ski season. Multiple regression on climate variables then permits statistical projections of future snow depth from projected changes in temperature and precipitation with 20 future climate scenarios (10 global climate models and 2 emissions scenarios, RCP4.5 and RCP8.5). We also use projected future SWE from a hydrology model with climate input from the same 20 climate scenarios. The two methods produce substantially different projected changes at individual resorts but in aggregate the distributions are similar, leading to these broad conclusions: While many resorts indeed face substantial declines in ski‐season snow depth, many of those in the inland Northwest and a few at high elevation are likely to be minimally affected. Mitigating factors include (a) projected increases in winter precipitation over the Rockies that partly offset the effects of warming; (b) low temperature sensitivity at those locations and over high altitudes; (c) lower observed declines and temperature sensitivity for snow in winter compared with spring; and (d) many ski resorts are located in areas of high snowfall and/or span a considerable range of elevations.

Vegetation photosynthesis, quantified as Gross Primary Productivity (GPP), changes rapidly throughout the day. We developed a global hourly GPP dataset called EGO (Eddy covariance site-based Global hOurly) by combining tower observations with advanced artificial intelligence that accounts for causal effects, outperforms existing hourly GPP products and well captures diurnal photosynthesis dynamics. It will provide a reliable foundation for investigating sub-daily ecosystem processes and benchmarking Earth system models.

The global active satellite population has increased from fewer than 3000 objects in 2020 to more than 14,000 by the end of 2025, while filed megaconstellation plans suggest that tens of thousands of additional satellites may be deployed over coming decades. This rapid expansion of low Earth orbit (LEO) infrastructure has raised concerns regarding novel anthropogenic inputs to the upper atmosphere, particularly black carbon from rocket launches and aluminium oxide nanoparticles generated during satellite re-entry. This review synthesises literature published between 2000 and 2026 across atmospheric chemistry, aerosol science, climate dynamics, and hydrology to evaluate the potential pathways through which these emissions may influence precipitation processes and flood risk. The evidence indicates strong support for several upstream mechanisms, including alumina-mediated ozone chemistry, anthropogenic metal accumulation in stratospheric aerosols, and the disproportionately high radiative forcing efficiency of rocket-derived black carbon. However, substantial uncertainties remain regarding the extent to which these atmospheric perturbations propagate through climate and hydrological systems. This review identifies the current state of knowledge, highlights areas of agreement, uncertainty, and contradiction within the literature, and outlines priority directions for future research, monitoring, modelling, and governance. The findings suggest that while satellite-driven changes to precipitation and flood risk remain unconfirmed, the rapid expansion of megaconstellation activity warrants further investigation within integrated Earth-system frameworks.

Carbon isotopic measurement (δ13C) as an indicator of C3 or C4 vegetation preserved in soil organic matter and pedogenic carbonate is an important tool for reconstructing paleoclimate. However, while δ13C values of C4 plants (−14‰) generally indicate warm-season grass, δ13C values of C3 plants can indicate trees (−27‰), cool-season grasses (−27‰), or desert shrubs (−27‰). Additional evidence, therefore, is needed to determine which C3/C4 boundary is indicated: the one between C4 grasslands and forests or cool-season grasslands, or the boundary between C4 grasslands and C3 desert shrublands. Unlike C3 forests and cool-season grasslands, C3 desert shrublands are notorious for having bare ground and high erosion rates. At the Cañada Alamosa Archaeology project area, located near a C4 grassland/C3 shrubland boundary (the Chihuahuan Desert boundary), we tested the hypothesis that δ13C values indicating C3 shrubs accompany periods of erosion-sedimentation, and, in contrast, δ13C values indicating C4 grasses correspond to periods of landscape stability and soil formation. The hypothesis is supported (i.e., failed to be disproven) by δ13C values in both soil organic matter and pedogenic carbonate, which indicate progressively greater amounts of C4 vegetation contributed to the higher δ13C values in the A horizons of buried soils. The hypothesis is also supported by macrobotanical analysis of samples from cultural features that show creosote bush (Larrea) and mesquite (Prosopis) reached their peak concentrations ca. 1200 CE during a period of erosion-sedimentation corresponding to the Medieval Warm Period. Linking the hypothesis to periods of an expanding/contracting Chihuahuan Desert, we make the following inferences: (1) The lowest allostratigraphic unit, termed Alamosa I, was deposited from at least 4000 calibrated (cal.) years before present (yr B.P.) to ca. 3100 cal. yr B.P. and records Chihuahuan Desert expansion during the later stages of (or soon after) the Altithermal. Alamosa I is capped by a well-developed soil with radiocarbon dates indicating landscape stability and desert contraction for at least 1000 yr from ca. 3100 cal. yr B.P to 2100 cal. yr B.P. (2) The overlying and less prominent Alamosa II records a renewed period of alluviation (ca. 2100 cal. yr B.P to 1870 cal. yr B.P.) followed by an estimated 500 yr period of landscape stability associated with desert contraction that enabled the formation of a weakly developed soil. (3) The Alamosa III and Valley Floor alluviation indicates a major desert expansion for ~800 yr (ca. 1370 cal. yr B.P to ca. 570 cal. yr B.P.), including the ~400 yr of the Medieval Warm Period, which was the time when the Pueblo population peaked at the project area with village sites up to 500 rooms. By ca. 600 cal. yr B.P. and through the Little Ice Age, soil formation suggests landscape stability in the project area and a retreat of the Chihuahuan Desert; by this time, Pueblo people had left the project area. The modern period of desertification, given the uncertainties of the radiocarbon ages, began after 1500 CE and before 1800 CE and continues today as the Chihuahuan Desert is expanding yet again.

Abstract The limitations of current Earth system models in accurately representing the propagation of the Madden-Julian Oscillation (MJO) over the Maritime Continent (MC) significantly hinder our ability to make skillful subseasonal-to-seasonal predictions of extreme weather events worldwide. Additionally, our understanding of the intricate interactions between the MJO and the MC remains incomplete. Focusing specifically on Sumatra Island, this study investigates the key processes underlying the damping effect of MC islands on the eastward-propagating MJO based on a series of idealized atmospheric-only global model simulations, which include realistic land-sea distributions and topography of the MC within an aqua-planet environment. The findings suggest that the coupling between land and atmosphere, driven by interactions among clouds, radiation, land surface temperature, and convective instability, plays a critical role in reducing MJO convection over the MC. Notably, incorporating the diurnal cycle of convection in the model significantly enhances this damping effect, due to more active interactions between convection and radiation, consistent with previous modeling studies.

arXiv:2609.12528v1 Announce Type: new Abstract: The deployment of a geoengineering program to cool Earth's climate may be imminent. It is crucial that tools be developed to ensure that such a program would achieve its objectives while minimizing disruption. Here we exploit recently developed differentiable atmospheric models to demonstrate a novel geoengineering control strategy. In the differentiable primitive-equation atmospheric model JAX-GCM we impose a uniform $+4$\,K ocean warming and ask what pattern of sea-surface temperature cooling -- in five ocean-masked zonal bands of prescribed SST forcings whose amplitudes are free -- returns land near-surface air temperature closest to the model's own unwarmed climatology. This idealized set-up represents a cooling pattern that could be delivered physically either by marine cloud brightening or stratospheric aerosol injection. Gradients through chaotic dynamics decorrelate from the true sensitivity beyond the Lyapunov horizon, so we optimize greedily over segments of 8 to 14 days, following receding-horizon control. The learned strategy removes $92.3 \pm 0.4\%$ of the realized land warming across a ten-member ensemble of two-year rollouts, and a three-year run sustains it. If we use the spatial pattern of land temperature as the optimization objective, the distributions of precipitation, evaporation, and specific humidity over land are restored as well, even though they are not included in the objective function. The learned strategy from JAX-GCM replayed in the AI emulators LUCIE and NeuralGCM without re-optimization is successful, suggesting robustness. These promising results demonstrate a strategy for designing optimal climate interventions that can be applied broadly for geoengineering scenarios under consideration.

Glacier velocity is a critical indicator for the prevention and mitigation of glacial hazards, yet its intrinsic link to ice avalanches (IAs) remains insufficiently understood. To elucidate the physical mechanisms connecting glacier velocity and IA, this study analysed pre-failure glacier motion in the Sedongpu Valley using Sentinel-2B remote-sensing imagery. Accordingly, we propose a mechanical pathway in which glacier acceleration increases the deformation rate of ice, reduces its fracture toughness, and thereby promotes IA initiation. Low-temperature four-point bending (4 PB) experiments quantified how ice fracture toughness and crack-tip yield strength (CTYS) decline progressively with increasing strain rate and temperature. The unsupported ice mass was idealized as a cantilever beam containing a single crevasse. Geometric sensitivity analysis indicates that overhang length and crack length are the primary controls on tensile stress concentration, forming the basis for a tensile-failure criterion for IA initiation in hanging glaciers. These results reveal a mechanical pathway whereby glacier acceleration reduces ice crack-growth resistance, thereby promoting IA initiation. Validation against the Eiger hanging-glacier event further shows that glacier acceleration can serve as an external signature of increasing strain rate, fracture-toughness degradation, and progressive damage accumulation. This study provides a new perspective for elucidating the initiation mechanism of tensile-type IAs in hanging glaciers.

Journal of Geophysical Research: Atmospheres, Volume 131, Issue 17, 16 September 2026.

ABSTRACT Palaeoglacier reconstructions based on geomorphological evidence and present‐day topography are essential for deriving equilibrium‐line altitude (ELA) estimates and making robust palaeoclimatic approximations. Current tools to generate 3D glacier reconstructions and calculate ELAs are dependent on proprietary GIS software platforms and associated licences, limiting accessibility. This study presents Q‐GlaRe+, an open‐source QGIS plugin that implements established flowline‐based algorithms and removes proprietary software barriers whilst maintaining scientific accuracy. The toolset incorporates automated workflow logging that ensures reproducibility of glacier reconstructions. Validation against five Swiss Alpine glaciers with known glacier bed topography demonstrates ice thickness estimates consistent with published modelling benchmarks. ELA estimates closely match existing implementations, confirming comparable performance. By eliminating proprietary software dependencies whilst maintaining scientific accuracy, Q‐GlaRe+ supports a broader community use of glacier reconstruction tools and facilitates open data sharing and reproducible research practices.

Water constantly flows between rivers, soils, and aquifers, but capturing this accurately in numerical models is difficult. Detailed models are too expensive for regional use; simpler ones lack the resolution to track groundwater and soil moisture correctly. We built a model bridging this gap using two interchangeable numerical approaches. Tests across six soil types showed near-centimetre groundwater accuracy, making it suitable for ensemble simulations and regional climate-impact studies.

Climate Information Services (CIS) are increasingly recognised as essential tools for enhancing climate change adaptation among smallholder farmers. Yet, the extent to which CIS are available, accessible, usable, and effectively integrated into local farm decision-making in Ghana remains insufficiently understood. This study assessed the effectiveness of CIS in the Mampong Municipality by analysing their availability and accessibility, farmers' perceived impacts on key agricultural decisions, the effectiveness of different dissemination channels, and the determinants of CIS uptake and use. Using a mixed-methods approach involving household surveys ( n = 200) and eight focus group discussions, the study found that CIS are widely available and frequently accessed, with 85.5% of farmers reporting that CIS influenced at least one farming decision. Farmers reported that CIS significantly influenced decisions on planting dates (83.0%), crop choices (82.5%), and irrigation practices (66.0%), though gender differences emerged in accessibility and usability. Radio emerged as the dominant and most effective dissemination channel, but women reported lower usability of CIS delivered through mobile phones and community-based extension services. Probit regression results indicated that household size and peer influence significantly affected CIS uptake, while socioeconomic characteristics, farm size, and climate-risk perception were not statistically significant predictors. Overall, farmers perceived CIS as contributing to improved adaptive decision-making, but its effectiveness is constrained by gendered access barriers, household-level constraints, and social influences. Strengthening dissemination approaches, tailoring CIS content to user needs, and leveraging social and institutional structures can enhance CIS utilisation and improve climate resilience among smallholder farmers.

Science Advances, Volume 12, Issue 37, September 2026.

Journal of Advances in Modeling Earth Systems, Volume 18, Issue 9, September 2026.

We present an open-access compilation of mineralogical and geochemical provenance data for surface sediments across the Arctic Ocean. The Circum-Arctic Sediment PROvenance Database (CASPROD) integrates four provenance proxies, including bulk sediment Sr–Nd isotopes, detrital zircon U–Pb ages, clay mineral assemblages, and detrital mineral compositions, from more than 4000 stations in the Arctic Ocean. All datasets are standardized with consistent metadata and are provided in multiple machine-readable formats.

by Marzie Naserikia, Melissa Anne Hart, Elyas Asadi Shamsabadi, Katrin Meissner, Sanaa Hobeichi, Benjamin Bechtel, Negin Nazarian Urban heat is a growing concern for public health, energy demand, and urban liveability. High-resolution air temperature (T a ) data are needed for developing effective, localised adaptation strategies, but obtaining such data at the city scale remains a challenge as weather stations are often sparse and unevenly distributed. To address this data gap, we developed a machine learning (ML) framework that creates high-resolution, gridded T a maps using crowdsourced observations and diverse geospatial datasets describing the city, and evaluated its accuracy, transferability, and performance under varying data availability conditions. This framework uses satellite-derived land surface temperature (LST), urban form and fabric datasets, and meteorological variables to train a convolutional neural network (CNN) model. The approach was implemented in Sydney, Australia, producing 30 m gridded T a estimates (at ~10:00 a.m. local time) with high accuracy on an independent held-out test set (R² = 0.97; RMSE = 0.91 °C), surpassing previously reported ML performance. We further assessed the generalisation and spatial transferability of this method, showing that the CNN model maintained strong predictive accuracy for unseen locations across the city (R² = 0.91–0.93; RMSE = 1.27–1.44 °C). Model performance also remained stable when the number of T a stations was significantly reduced by ~80%. Performance was more variable for unseen days (R² = 0.66–0.93; RMSE = 1.52–2.60 °C), indicating that ML-based T a mapping requires training data representative of diverse weather conditions. We find that high-resolution city-descriptive datasets are beneficial but not essential, as comparable accuracy was achieved using only globally available predictors. These results indicate that the proposed framework is transferable to other cities, including those in data-sparse regions. The study provides an effective and scalable approach for city-scale air temperature mapping, which is urgently needed for urban heat assessment, climate adaptation, and public health planning.

Nature Climate Change, Published online: 10 September 2026; doi:10.1038/s41558-026-02743-5 Global warming and urban development combine to shape the temperatures people experience in cities. Here I revisit a 2016 attribution study of urban warming in China and consider what its legacy reveals about urban climate knowledge, responsibility and action.

Earth's Future, Volume 14, Issue 9, September 2026.

Ground-level ozone can reduce plant growth, but models represent this damage in different ways. We tested six approaches in the same vegetation model for China during the 2010s. They estimated that ozone reduced plant carbon uptake by 17 % to 27 %, averaging about 21 %. Approaches based on broader observations and leaf traits matched measurements more closely, while using leaf lifespan improved how the model represented the accumulation and decline of ozone damage.

arXiv:2609.10564v1 Announce Type: new Abstract: Recent machine learning weather prediction (MLWP) models have demonstrated remarkable forecasting skill on global reanalysis-based benchmarks. However, their performance remains unclear in challenging environments such as Northern Norway, where narrow fjords and rapidly changing weather result in highly variable local wind conditions. In this case study, we evaluate FourCastNet3 (FCN3), GraphCast, and ECMWF High Resolution Forecast (HRES) for wind speed forecasting using multi-year station observations from Northern Norway, focusing on their relative performance, generalization beyond the training period, and performance under high-wind conditions. Our results show that HRES slightly outperforms FCN3 and GraphCast, with an overall RMSE of 2.89 $\mathrm{m\,s^{-1}}$, compared to 2.96 $\mathrm{m\,s^{-1}}$ for FCN3 and 2.94 $\mathrm{m\,s^{-1}}$ for GraphCast. Notably, the MLWP models maintain comparable performance beyond their respective training periods, with no clear evidence of noticeable degradation. FCN3 performs best under high-wind conditions, although all models substantially underestimate strong winds. Our findings suggest that MLWP has become competitive with NWP for local wind, but further refinements are still needed to capture complex terrain better.

Nature Climate Change, Published online: 11 September 2026; doi:10.1038/s41558-026-02732-8 Heatwaves threaten both physical and mental health, yet physical heat exposure and emotional responses often diverge. The study shows that 84.9% of Chinese cities experienced severe emotional shocks, compared with 52.1% facing severe heat exposure, driven by socioeconomic and environmental factors.

Geophysical Research Letters, Volume 53, Issue 18, 28 September 2026.

Journal of Geophysical Research: Oceans, Volume 131, Issue 9, September 2026.

Geophysical Research Letters, Volume 53, Issue 18, 28 September 2026.

ABSTRACT Statistical parameters derived from grain size data offer insights into grain size distribution (GSD), facilitating the interpretation of depositional environments and sediment transportation history. GSD of glacial sediments is complex, often exhibiting bimodal, trimodal and polymodal distributions, complicating the calculation of statistical parameters. Despite employing established methodologies and computer programs, we observed significant variations in these values. A new approach integrating traditional techniques with modern calculation is necessary to yield more accurate and suitable values for these parameters in high‐energy complex glacial depositional systems. Four programs encompassing six methodologies (Gradistat moments, Gradistat Folk and Ward, QGrain moments, QGrain Folk and Ward, classical manual calculation and GSAT) were used to analyse the same study of grain size data from 17 glacial samples. The variance in the results ranged from 0.01 to 1.95, and in some instances, reached up to 4.93 and 49.34 for the skewness and kurtosis values, respectively. This variability highlights the gaps in our understanding of glacial GSD and complicates the interpretation of the roles of various agents in sediment transport. The considerable variation in all parameters is also evident, along with underestimated values for Folk and Ward calculations in Gradistat and QGrain, which remain less sensitive to the extreme values of grain size data. This indicates the need for a different approach to calculate these parameters which combines traditional methods with recent models to calculate and interpret the GSD of sediments. Herein, we present the Grain Size Analysis Tool (GSAT) to calculate the statistical parameters with high precision.

Glacial lake outburst floods (GLOFs) are destructive cryosphere-related hazards shaped by nonlinear interactions among hydroclimatic forcing, glacier morphology, lake evolution, and geomorphological instability. Reliable precursor screening remains difficult because documented GLOF events are sparse, triggering mechanisms are heterogeneous, and pre-event hydroclimatic signals may appear as subtle changes in temporal organization rather than simple amplitude extremes. This study proposes a hybrid quantum-classical framework for rare hydroclimatic anomaly screening that integrates conditional multivariate sequence generation, quantum expectation-based latent representations, generated-reference temporal mismatch scoring, and supervised rare-event classification. Hydroclimatic dynamics are modeled using conditional Wasserstein generative adversarial networks with gradient penalty (WGAN-GP), temporal convolutional sequence learning, and latent representations derived from parameterized quantum-circuit expectation values. Compared with Gaussian and sign latent baselines under identical WGAN-GP settings, the quantum latent formulation achieved lower errors in cross-feature correlation, Wasserstein distance, Kolmogorov-Smirnov statistic, autocorrelation distance, and power spectral density distance in the present experimental setting. The anomaly-scoring module further identified cross-variable temporal correlation mismatch and autocorrelation deviation as precursor-relevant signals. Because the number of positive GLOF samples is limited, the supervised outputs are interpreted as high-recall anomaly-screening and ranking indicators rather than deterministic forecasts of GLOF occurrence. Explainability and ablation analyses indicate that QLGM-generated-reference temporal mismatch features were more discriminative than compact hydroclimatic and static descriptors within the evaluated models. These results suggest that quantum expectation-based latent representations can reduce hydroclimatic sequence-realism errors under the evaluated WGAN-GP setting and can support generated-reference anomaly prioritization for GLOF risk reduction under severe data imbalance. The findings should be interpreted as empirical evidence for a useful structured latent representation in this setting, rather than as proof of universal quantum advantage.

ABSTRACT This study examines how forest structure, topography and radiative energy affect snow accumulation and ablation at three forested sites in Arizona, USA. Using distributed in situ daily snow depth observations, LiDAR‐derived snow depth maps, radiation measurements and high‐resolution snow model simulations (using the SnowPALM model), we evaluated spatial patterns in peak SWE, snow cover duration, liquid water input and incoming and outgoing radiation fluxes across three forested mountain sites. SnowPALM closely reproduced observed snow dynamics, seasonal subcanopy radiative forcing and strong aspect‐ and canopy‐driven gradients in snow accumulation and persistence. Shaded areas and north‐facing slopes have deeper and longer‐lasting snowpacks, but liquid water input (the liquid water flux delivered to the ground surface from snowmelt and rainfall) can be reduced due to the larger amount of snowpack sublimation (which was enhanced in areas with long snow persistence) relative to incoming precipitation. Furthermore, a semi‐partial correlation analysis showed that, after removing the effect of spatial variability in throughfall on each snowpack metric, the influence of net radiation on peak SWE, snow cover duration and liquid water input was weakest in wet years and strongest in dry years. Overall, these results demonstrate that snowpack evolution in semiarid forests in Arizona arises from the combined effects of canopy and terrain‐controlled accumulation and radiatively driven ablation, with increasing sensitivity to energy inputs under warmer and drier conditions.

Speleothems, such as stalagmites and flowstones, are invaluable archives of past environmental and climatic conditions due to their layered growth and suitability for precise 230 Th/U dating. In tropical karst settings, however, elevated and variable detrital thorium contamination can severely compromise age accuracy, especially for young speleothems. Here, a suite of stalagmites from Áaktun Kóopo' Cave from the Yucatán Peninsula is used as a case study to develop and evaluate a multi-method framework for constraining initial thorium and constructing robust age–depth models. We combine extensive 230 Th/U dating with local isochron analysis, stratigraphic approaches, and annual Sr/Ca layer counting to constrain elevated and initial (230Th/232Th) activity ratios and resolve pronounced chronological inversions. High uranium concentrations in the speleothems (average ca. 1 ppm) allow precise measurements, yet we infer a high and largely unsystematic variability of high initial (230Th/232Th) activity ratios in space and time, with values spanning between 4 and 68, thus far exceeding standard bulk earth values. This variability demonstrates that elevated and temporally variable initial Th may be common in tropical karst settings and that multiple, independent constraints on initial (230Th/232Th) activity ratios are essential for reliable chronologies in such environments. Applying this framework to Áaktun Kóopo' Cave stalagmites yields internally consistent stalagmite age–depth models spanning the past 2.7 kyr, as well as evidence for earlier glacial and interglacial growth phases. These chronologies provide the basis for high-resolution, multi-proxy reconstructions of climatic and environmental changes in the northeastern Yucatán Peninsula during the entire era of Maya cultural evolution, and they illustrate how robust speleothem chronologies can be obtained even in cave systems affected by elevated and highly variable initial thorium.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

arXiv:2609.04632v1 Announce Type: new Abstract: As global warming intensifies, the frequency and intensity of extremely high temperatures are expected to increase. This will impact the production of photovoltaics (PVs), which are increasingly adopted as an effective alternative to replace fossil fuel-based energy sources and reduce CO2 emissions. Furthermore, extremely high temperature days account for a considerable proportion of days with high PV power potential (PVpot). Therefore, this study investigates changes in PVpot on future extremely high temperature days over East Asia, a region with high greenhouse gas emissions and vulnerability to extreme climatic events. The East Asia-averaged PVpot for extremely high temperature days was estimated to decrease across all scenarios and future periods. The East Asia-averaged PVpot for extremely high temperature days was predicted to decrease more substantially toward the late 21st century, with a larger magnitude of decrease expected under the high-carbon emissions scenario compared to the low-carbon emissions scenario. By the mid-and late 21st century, PVpot for extremely high temperature days was projected to decrease in PV hotspot areas, particularly in the regions of northern China and southern Mongolia, by up to -7.2 %. The signs of PVpot projections vary across sub-regions under summer mean conditions, while on extremely high temperature days, PVpot is consistently expected to decrease in all regions. This suggests that extremely high temperatures further intensify the decrease in PVpot. Moreover, under extremely high-temperature conditions, near-surface air temperature has been identified as the primary driver of projected decreases in PVpot among the climate variables considered; its influence is expected to intensify over time, thereby accelerating PVpot decreases under the high-carbon emissions scenario.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

We measured the ice thickness of Hintereisferner, a small valley glacier in the Austrian Alps. This glacier has been observed for decades and is one of the world’s reference glaciers, but its ice thickness has only been sparsely measured in the past. We show that the glacier is thinner than predicted by ice thickness models, which overestimate thickness on small, temperate valley glaciers. Our results will help validate and improve future models and highlight the need for field measurements.

Exposing the technical limits of existing early warning systems after a devastating August 2026 flash flood in Nepal's Trisuli–Bhote Koshi basin, this paper proposes a multi-hazard early warning framework to address rising Glacial Lake Outburst Floods (GLOFs) and avalanche-induced flash floods in the climate-sensitive Himalayas. The framework combines risk prediction, GIS risk mapping, and early warning layers by using Transformer/LSTM deep learning models to integrate six distinct data sources: satellite imagery, glacier temperature, snowmelt rate, rainfall, seismic signals, and river water levels. Additionally, it incorporates Digital Twin technology for real-time basin modeling and "what-if" simulations, demonstrating through the 2026 Nepal disaster case study how lead times can be improved across South Asian Himalayan regions and similar high-risk zones.

Nature Climate Change, Published online: 08 September 2026; doi:10.1038/s41558-026-02737-3 The authors analyse the impacts of the adoption of genetically engineered crops in the USA over 40 years. They show that genetically engineered crop adoption has led to some increases in yield and reduced yield volatility, while also acting as an adaptation against climate change-driven northward shifts.

Proceedings of the National Academy of Sciences, Volume 123, Issue 36, September 2026. SignificanceGlobal projections of climate change impacts on marine invertebrate body size remain scarce. We synthesize data for >23,000 molluscan species to project future changes in the average body size across the global ocean basins. Under future ...

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

This paper presents a flexible workflow using a unified data assimilation framework to evaluate atmospheric composition models. It enables comparison of observations with forecasts of trace gases and aerosols from different models. The system is consistent and adaptable, reducing repetitive work, supporting model validation and observation assessment, and aligning evaluation with operational data assimilation for research and practical applications.

Preprint under review for ESSD (discussion: open, 0 comments) We present a dataset collected with the WProf cloud radar during the CleanCloud Arctic Villum ExpeRiment campaign in northeast Greenland in spring 2024. It includes Doppler spectra, hydrometeor classification, and cloud observations that reveal the microphysical structure of Arctic clouds. The data show that multimodality, where different cloud particle populations coexist, is common and support studies of aerosol–cloud interactions and improvements in weather and climate models.

Abstract Anthropogenic warming is rapidly transforming Northern Hemisphere snow cover, and conventional metrics such as snow mass and season duration fail to capture the full seasonal structure or the pace of ongoing change. Here we introduce a phenology-based framework that characterizes the complete annual snow cover cycle by decomposing it into four temporal features, including snow cover onset, rise to maximum extent, fall from maximum extent, and snow season end, applied to ERA5-Land reanalysis data spanning 1950–2025. Using daily snow cover fraction data across 20°–90°N, we combine feature extraction, trend analysis, and unsupervised K-means clustering to quantify both gradual phenological shifts and nonlinear changes in snow regime area and latitude. We show that snow cover change exhibits strong seasonal asymmetry, with melt-phase timing advancing by 1–4 days per decade across continental Eurasia and Europe, while accumulation-phase trends remain less prominent. Unsupervised clustering identifies six physically coherent snow regimes ranging from low-persistence to near-continuous cover. These regimes reveal accelerating degradation over the 75-year study period, marked by northward shifts of 0.7–2.0° latitude and rapid expansion of low-persistence conditions after 2003. The largest and most statistically consistent changes occur in the timing of spring melt and the duration of continuous maximum snow cover, indicating that losses are driven more by earlier seasonal decay than by delayed autumn onset. These nonlinear regime transitions constitute hemispheric-scale evidence of snow loss consistent with snow–albedo feedback dynamics. This phenology-based framework integrates the internal dynamics of the snow season into a single coherent structure, offering a sensitive and holistic benchmark for evaluating climate-model performance under continued warming and for detecting emerging cryospheric thresholds in a warming climate.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

Preprint under review for ESSD (discussion: open, 0 comments) We present a convective-scale reanalysis of the 'Swabian MOSES 2023' campaign that assimilates 3-months of additional field campaign observations in southwest Germany. It is computed with a 1-hourly assimilation timestep on a 2.2 km grid centred over Germany in moderately complex terrain. In addition to all operationally available observations, it incorporates 6 additional observation types. Among these is an unprecedented number of vertical wind profiles retrieved from 12 Doppler wind lidars.

Abstract Worldwide, lacustrine ecosystems are threatened by eutrophication and deoxygenation. Both processes are driven by the combined effects of human activities and climate change. Understanding the pre‐anthropogenic nexus between warming, eutrophication, deoxygenation, and related feedback is essential to determine how lakes respond to rapid climate change. We investigate the responses of primary producers, stratification, and nutrient cycling in a small Central European lake (Übeschisee, Swiss Plateau) to rapid climate change during Late Glacial times (17–11 ka). After 16.2 ka (Heinrich Stadial; HS1) and during the Bølling/Allerød interstadial, Übeschisee experienced natural eutrophication following summer warming and vegetation development in the catchment. Anoxia events with blooms of sulfur bacteria systematically established during colder periods of HS1 (16.5 ka), the Older Dryas (14 ka), the Gerzensee Oscillation (13.2 ka), and the Younger Dryas (12.7–11.7 ka). Anoxia events also caused reductive dissolution of Mn; labile P was always efficiently recycled. Algal communities responded to anoxia events, but the excursions of community composition were reversible. Ti follows NGRIP Dust, suggesting that Übeschisee provides a continuous record of dust deposition with high rates during colder periods. The paired lake approach with adjacent Amsoldingersee shows that the results are robust and regionally reproducible, and that common drivers influenced both lakes simultaneously.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

Journal of Geophysical Research: Solid Earth, Volume 131, Issue 9, September 2026.

The Northern Patagonian Icefield is the second largest ice mass of South America, located at moderate latitudes. Using an ice-flow model which uses available atmosphere data as input and explicitly models iceberg discharge, we assess its evolution. With present climate, it will lose about 36 % until the year 2200. Climate change strongly accelerates losses and under Paris Agreement compliance it is between 38% and 75% and between 81% and 94% under a business-as-usual fossil fuel scenario.

arXiv:2609.12296v1 Announce Type: cross Abstract: Estimating the impact of observations newly added to an existing numerical weather prediction (NWP) system requires reintegrating the ensemble forecast from the analysis that assimilates the additional observations, which is laborious and computationally expensive. We therefore propose preemptive EFSO (PEFSO; preemptive ensemble forecast sensitivity to observations), which estimates observation impact without model reintegration by invoking a stronger tangent-linear approximation than that used in EFSO. The estimated impact, however, is merely a scalar quantity at a specified verification time, so it is also valuable to obtain the ensemble forecast in which the remaining additional observations are assimilated after denying the detrimental ones. We refer to this procedure as ADD-SEL, but carrying it out requires reintegration. We therefore propose two methods for updating the ensemble forecast without reintegration as approximations to ADD-SEL, collectively termed ADD-SEL-PRE: one recomputes the ensemble transform matrix (i.e., the Kalman gain), and the other modifies only the innovations at lower cost. Experiments with the Lorenz-96 model examine whether both can be approximated without reintegration. PEFSO estimates observation impact comparable to that obtained by EFSO within the range where the tangent-linear approximation remains valid, even under more practical conditions: an ensemble size of 10, the analysis as the reference state, and only the additional observations available. When additional observations are assimilated after denying detrimental ones, both ADD-SEL-PRE methods update the ensemble forecast, giving forecast errors comparable to those from ADD-SEL, particularly for lead times up to two days, where the tangent-linear approximation is expected to hold.

Snowmelt, a vital source of freshwater for a large portion of North America’s population, is increasingly vulnerable to the impact of climate change. Air temperature and precipitation are significant climatic controllers regulating snowmelt dynamics. In this study, we quantified the response of winter snowmelt magnitude to changing temperature and precipitation across North America (Canada and the United States) using gridded, observation-driven data and future projections. Our analysis of far-future conditions (2091-2100) reveals a significant increase (>60%) in winter (November-March) snowmelt in northern latitudes, while southern latitudes are projected to experience a decline (<38%). Projected snowmelt changes are strongly associated with changes in winter temperature, while projected declines are consistent with regions experiencing reduced snowfall under warming climates. Our findings suggest that the probability of an increase in winter snowmelt is high under the warmer and wetter climatic conditions prevailing in northern regions. In contrast, winter snowmelt across southern latitudes is likely to decrease. Plain language summary Snowmelt plays a critical role in sustaining water resources for large populations across North America. As temperature and precipitation are major climatic factors regulating snowmelt dynamics, changes in these variables can affect the magnitude of snowmelt. Given projected climate change, it is essential to assess the vulnerability of snowmelt magnitude to changing temperatures and precipitation to understand future water availability. Our analysis of future winter snowmelt under compound changes in air temperature and precipitation across North America indicates that winter snowmelt could increase by 60% in many regions of Canada. In contrast, in southern areas, the winter snowmelt volume could decrease by as much as 38%. These results suggest that warmer winters are the primary driver of increased snowmelt, while reduced snowfall primarily governs the projected snowmelt declines.

Study region The Yarlung Zangbo River Basin (YZRB), a representative high‑altitude alpine‑glacial basin in Xizang, southwestern China. Study focus To improve the snowmelt runoff simulation accuracy of the EasyDHM (Easy Distributed Hydrological Model) in the YZRB, this study proposes an improved snow cover fraction (SCF)-snow water equivalent (SWE) relationship function. Different from the baseline model adopting traditional empirical formulas, the new function incorporates the nonlinear Sigmoid-based physical mechanisms of snow evolution. By establishing a differential equation with SWE as the independent variable and integrating elevation and slope as dynamic constraints, the model better captures the spatial heterogeneity of snow distribution in complex alpine terrains. New hydrological insights for the region Multi-section validation demonstrates that the improved model significantly boosts runoff prediction accuracy, with the average Nash-Sutcliffe efficiency (NSE) increasing from 0.763 to 0.899 and the percent bias (PBIAS) dropping from 11.29% to 6.25% during the calibration period. The improved module effectively eliminates the premature runoff peaks in May, as found in the baseline model, ensuring precise phase alignment during the critical snowmelt period (March-June). Overall, the improved SCF-SWE relationship function enhances the model’s applicability, providing a reliable tool for snowmelt runoff simulation, water resource assessment, and hydrological forecasting in cryosphere-dominated basins.

Abstract Speleothem oxygen isotope records (δ 18 O s ) from the Asian monsoon region provide valuable insights into past hydroclimate variability, yet their interpretation relies on understanding the controls on modern precipitation isotopes (δ 18 O p ). Although numerous explanations for the spatiotemporal variability of δ 18 O p across this region have been proposed, the primary controls remain unresolved. Here we propose a hypothesis to explain δ 18 O p variability from δ 18 O p records at 46 sites along a transect spanning the Asian monsoon region. Our results show that seasonally low δ 18 O p values migrate gradually westward along the east–west transect. We further demonstrate that the corresponding westward migration of atmospheric updraft, particularly convection, controls δ 18 O p variations along the transect. In contrast to previous interpretations emphasizing monsoon moisture advection, our findings highlight atmospheric updraft as an important control on δ 18 O p variations across the Asian monsoon region. These findings provide an alternative perspective on interpreting the climate signals preserved in Asian δ 18 O s records.

Marine zooplankton transfer energy from microscopic algae to fish and larger predators. Understanding their distribution helps predict how oceans respond to climate change. We developed SeapoPym, a freely available model that simulates zooplankton using ocean temperature and plant productivity. This tool lets scientists test biological hypotheses and estimate parameters from observations.

This study presents a new global leaf turnover model for grasslands in the QUINCY (QUantifying Interactions between terrestrial Nutrient CYcles and the climate system) land surface model. Land surface models often struggle to simulate grassland carbon dynamics and phenology accurately. By allowing environmental conditions to directly control leaf senescence we improve its timing as well as the accuracy of whole-season carbon dynamics across a wide range of climates and grassland ecosystems.

Proceedings of the National Academy of Sciences, Volume 123, Issue 36, September 2026. SignificanceUnderstanding how domestication reshapes brain structure and function remains a fundamental question in evolutionary biology. While the hippocampus—a center for memory and stress regulation—is traditionally thought to undergo “simplification” ...

We developed a new AI framework called Geo-SAN to create 3D geological models. Traditional modeling approaches struggle with complex geological structures and limited data. Our method analyzes relationships based on stratigraphy-aware attention networks between sampling points while incorporating knowledge about how stratigraphical sequence is ordered in nature. This helps the framework understand both the types of lithology and their interface and attitude positions.

We integrate GFDL's (Geophysical Fluid Dynamics Laboratory) land model in the high resolution global atmospheric model SHiELD (System for High-resolution modeling for Earth-to-Local Domains), unlocking new capabilities in simulating new suite of land-atmosphere conditions and extreme hydrological processes such as flooding. A single global simulation captures Hurricane Helene's 2024 landfall, including soil saturation, runoff generation, and river flooding in Western North Carolina, while simultaneously capturing an independent extreme rainfall and flooding event in Nepal.

This study uses a new type of observation of wind and turbulence to investigate the accuracy with which the German weather forecasting model predicts these variables in the lowest 600 m of the atmosphere. The model performs adequately during the day, but struggles with both wind and turbulence at night. This is important for wind energy planning and understanding how airborne particles are transported by the wind. The study suggests ways in which the model could be further improved.

Earth's Future, Volume 14, Issue 9, September 2026.

Solar power plants are expanding rapidly worldwide, but their impacts on local climate remain uncertain. In this study, we developed a coupled model that explicitly represents interactions between solar panels, land surface, and the atmosphere. Evaluation against satellite observations showed improved simulation of land surface temperature over solar farms. Simulations show that solar farms can cool the land surface, warm near-surface air, reduce incoming sunlight, and alter rainfall patterns.

Nature Communications, Published online: 14 September 2026; doi:10.1038/s41467-026-77732-5 A membrane‑free electrochemical alkalization and electrooxidation system for urine valorization enables efficient phosphorus recovery and urea stabilization, providing a scalable route for nutrient recycling and sustainable fertilizer production.