https://arxiv.org/api/JT12+7LyV33+QbBzFZdDh9sMmuo 2026-03-22T16:11:57Z 8090 60 15 http://arxiv.org/abs/2603.05570v1 2026-03-05T14:47:00Z

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

2026-03-05T14:47:00Z 19 pages, 9 figures Inna Polichtchouk Simon Lang Sarah-Jane Lock Michael Maier-Gerber Peter Dueben http://arxiv.org/abs/2603.04395v1 2026-03-04T18:58:27Z

Data assimilation (DA) combines model forecasts and observations to estimate the optimal state of the atmosphere with its uncertainty, providing initial conditions for weather prediction and reanalyses for climate research. Yet, existing traditional and machine-learning DA methods struggle to achieve accuracy, efficiency and uncertainty quantification simultaneously. Here, we propose HLOBA (Hybrid-Ensemble Latent Observation-Background Assimilation), a three-dimensional hybrid-ensemble DA method that operates in an atmospheric latent space learned via an autoencoder (AE). HLOBA maps both model forecasts and observations into a shared latent space via the AE encoder and an end-to-end Observation-to-Latent-space mapping network (O2Lnet), respectively, and fuses them through a Bayesian update with weights inferred from time-lagged ensemble forecasts. Both idealized and real-observation experiments demonstrate that HLOBA matches dynamically constrained four-dimensional DA methods in both analysis and forecast skill, while achieving end-to-end inference-level efficiency and theoretical flexibility applies to any forecasting model. Moreover, by exploiting the error decorrelation property of latent variables, HLOBA enables element-wise uncertainty estimates for its latent analysis and propagates them to model space via the decoder. Idealized experiments show that this uncertainty highlights large-error regions and captures their seasonal variability.

2026-03-04T18:58:27Z 23 pages, 12 figures Hang Fan Juan Nathaniel Yi Xiao Ce Bian Fenghua Ling Ben Fei Lei Bai Pierre Gentine http://arxiv.org/abs/2603.03926v1 2026-03-04T10:35:38Z

State-of-the-art ensemble Kalman filtering (EnKF) algorithms require incorporating localization techniques to cope with the rank deficiency and the inherited spurious correlations in their error covariance matrices. Localization techniques are mostly ad-hoc, based on some distances between the state and observation variables, requiring demanding manual tuning. This work introduces a new ensemble filtering approach, which is inherently localized, avoiding the need for any auxiliary localization technique. Instead of explicitly applying localization on ensembles, the idea is to first localize the continuous analysis probability density function (pdf) before ensemble sampling. The localization of the analysis pdf is performed through an approximation by a product of independent marginal pdfs corresponding to small partitions of the state vector, using the variational Bayesian optimization. These marginals are then sampled following stochastic EnKF and deterministic ensemble transform Kalman filtering (ETKF) procedures, using ensembles larger than the partitions' size. The resulting filters involve the same forecast steps as their standard EnKF and ETKF counterparts but different analysis steps, iteratively adjusting the EnKF and ETKF updates of each partition based on the ensemble means of the other partitions. Numerical experiments are conducted with the Lorenz-96 model under different scenarios to demonstrate the potential of the proposed filters. The new filters' performances are comparable to those of the EnKF and ETKF with already tuned localization, both in terms of computational burden and estimation accuracy.

2026-03-04T10:35:38Z Accepted for publication in the Quarterly Journal of the Royal Meteorological Society Boujemaa Ait-El-Fquih Ibrahim Hoteit 10.1002/qj.70177 http://arxiv.org/abs/2603.03838v1 2026-03-04T08:40:42Z

Global climate projections rely on computationally demanding Earth System Models (ESMs), which are typically limited to coarse spatial resolutions due to their high cost. To obtain high-resolution projections for regions of interest, it is common to use Regional Climate Models (RCMs), which are driven by data produced by ESMs as boundary conditions. While more efficient than running ESMs at fine resolution, RCMs remain expensive and restrict the size of ensemble simulations. Inspired by recent advances in probabilistic machine learning for weather and climate, we introduce a data-driven climate downscaling method based on stochastic interpolants. Our approach efficiently transforms coarse ESM output into high-resolution regional climate projections at a fraction of the computational cost of traditional RCMs. Through extensive validation, we demonstrate that our method generates accurate regional ensembles, enabling both improved uncertainty quantification and broader use of high-resolution climate information.

2026-03-04T08:40:42Z Preprint, first draft Erik Larsson Ramon Fuentes-Franco Mikhail Ivanov Fredrik Lindsten http://arxiv.org/abs/2603.03669v1 2026-03-04T02:50:00Z

The mesosphere and lower thermosphere (MLT), spanning approximately 80--100~km in altitude, is a region of intense dynamical activity where atmospheric gravity waves amplify due to decreasing air density. These waves often undergo breaking, inducing energy and momentum dissipation as well as turbulent mixing. These processes can be described by atmospheric instabilities -- including convective and shear instabilities -- that regulate the vertical coupling between atmospheric layers. Ripple-like structures observed in mesospheric airglow imagery represent small-scale, short-lived signatures of such instabilities. Their occurrence often reflects localized instabilities, critical wave interactions, or wave ducting phenomena. Therefore, detecting and analyzing these features across broad spatial and temporal domains remains a challenge. In this study, we develop a deep learning framework using convolutional neural networks (CNNs) to automatically detect ripple-scale wave structures in all-sky airglow images. Trained on labeled datasets with recognized ripples, the model learns the spatial morphology associated with instability-driven gravity waves and achieves high accuracy in identifying ripple events. Beyond detection, we perform a statistical characterization of the detected ripples to examine their frequency of occurrence, orientation distributions, scales, and geographic and seasonal variability. These statistics are used to infer underlying physical mechanisms, such as localized instability conditions and background wind filtering. This work advances our understanding of instability-driven dynamics in the upper atmosphere through AI-powered detection, while also highlighting the potential of deep learning in scientific research.

2026-03-04T02:50:00Z Jiahui Hu Alan Liu Adriana Feener Jing Li Tao Li Wenjun Dong http://arxiv.org/abs/2603.03611v1 2026-03-04T00:48:27Z

This study examines the decay of Hurricane Melissa (2025) as the storm crossed the mountainous terrain of Jamaica, focusing on changes in inner-core energetics. Using NOAA P-3 reconnaissance observations near the 700 hPa level, integrated kinetic energy within 100 km of the storm center was computed before and after land interaction to quantify vortex weakening. During the approximately four hour period in which the center remained over Jamaica, the storm experienced rapid degradation, including a 48 percent reduction in peak tangential wind, a 58 hPa rise in central pressure, and a 41 percent decrease in integrated kinetic energy. To investigate the mechanisms responsible for this decay, the observations were compared with results from an idealized axisymmetric tangential momentum diffusion model that isolates the effects of vertical turbulent mixing and enhanced surface drag. Despite its simplified physics, the model reproduces many leading order features of the observed weakening, including substantial reductions in tangential wind and a 36 percent decline in integrated kinetic energy. These results indicate that enhanced friction and turbulent mixing associated with extremely rough mountainous terrain can account for a large fraction of the rapid spin down observed during land interaction. Differences between the model and observations suggest that additional processes such as asymmetric dynamics, terrain induced flow distortion, and thermodynamic feedbacks further amplify the weakening of intense tropical cyclones over land.

2026-03-04T00:48:27Z Michael Igbinoba http://arxiv.org/abs/2512.01749v3 2026-03-03T22:19:32Z

Observational data from buoys are of primary importance during the development, calibration, and evaluation of ocean wave models, and these data are also used to make real-time corrections to operational models via data assimilation. By association, systematic inaccuracies in any buoy data are equally important, and thus when two buoy types provide systematically inconsistent information, this is a concern for anyone using an ocean wave model. This report is concerned with the accuracy of the high frequency portion of the ocean wave spectrum commonly observable by buoys, roughly 0.2 to 0.6 Hz. We evaluate four buoy types (two moored, two drifting) using two quantitative measures. The first involves comparing each type with a co-located ocean wave model. The second method involves evaluation of high frequency energy level as a function of wind speed. Both evaluation methods suggest that the Datawell Waverider (DWR) buoys have a strong tendency to report higher energy levels than the other three buoy types. We evaluate high frequency energy level using three different metrics (mean square slope, energy in a band of high frequencies, and spectral density at a single, specific band, 0.4 Hz), and the conclusions are found to be insensitive to the parameter used.

2025-12-01T14:52:47Z technical report, 42 pages W. Erick Rogers Jim Thomson http://arxiv.org/abs/2505.06891v2 2026-03-03T20:49:43Z

We report experimental evidence of an Eulerian-mean flow, $\overline{u}(z)$, created by the interaction of surface waves and tailored ambient sub-surface turbulence, which partly cancels the Stokes drift, $u_s(z)$, and present supporting theory. Water-side turbulent velocity fields and Eulerian-mean flows were measured with particle image velocimetry before vs after the passage of a wave group, and with vs without the presence of regular waves. We compare different wavelengths, steepnesses and turbulent intensities. In all cases, a significant change in the Eulerian-mean current is observed, strongly focused near the surface, where it opposes the Stokes drift. The observations support the picture that when waves encounter ambient sub-surface turbulence, the flow undergoes a transition during which Eulerian-mean momentum is redistributed vertically (without changing the depth-integrated mass transport) until a new equilibrium state is reached, wherein the near-surface ratio between $|\mathrm{d}\overline{u}/\mathrm{d}z|$ and $|\mathrm{d}u_s/\mathrm{d} z|$ approximately equals the ratio between the streamwise and vertical Reynolds normal stresses. This accords with a simple statistical theory derived here and holds regardless of the absolute turbulence level, whereas stronger turbulence means faster growth of the Eulerian-mean current. We present a model based on Rapid Distortion Theory which describes the generation of the Eulerian-mean flow as a consequence of the action of the Stokes drift on the background turbulence. Predictions are in qualitative, and reasonable quantitative, agreement with experiments on wave groups, where equilibrium has not yet been reached. Our results could have substantial consequences for predicting the transport of water-borne material in the oceans.

2025-05-11T08:13:54Z Journal of Fluid Mechanics vol 1029, A6 (2026) Simen Å. Ellingsen Olav Rømcke Benjamin K. Smeltzer Miguel A. C. Teixeira Ton S. van den Bremer Kristoffer S. Moen R. Jason Hearst 10.1017/jfm.2026.11163 http://arxiv.org/abs/2603.03483v1 2026-03-03T19:59:30Z

Near-surface extreme winds profoundly affect human society, yet process-based understanding of their changes under climate forcings remains limited. This study systematically investigates the responses of high (HWE) and low (LWE) wind extremes (10-meter) to climate forcings using a hierarchy of climate model experiments from multiple general circulation models that participated in the Cloud Feedback Model Intercomparison Project. We analyze idealized atmosphere-only aquaplanet (Aqua) simulations and more realistic land-atmosphere (AMIP) simulations to identify robust responses to climate forcings and trace the sources of structural uncertainty. In Aqua simulations, tropical LWE changes exhibit large inter-model spread, which can be traced to dynamically distinct representations of low-pressure systems between models. In contrast, extratropical HWE intensify robustly with surface warming, linked to the strengthening of high-latitude extratropical cyclones. The AMIP simulations confirm the robust intensification of extratropical HWE. The more realistic boundary conditions in AMIP simulations act as a constraint, reducing inter-model spread in tropical zonal means compared to Aqua simulations. A comparison of uniform and patterned 4-K warming experiments suggests that the global magnitude of warming, rather than the specific warming pattern, dominates the large-scale responses of wind extremes. However, regional projections of extreme wind changes, especially over land, remain highly uncertain due to divergences in model physics. Case studies reveal that major disagreements in HWE changes can stem from fundamental differences in representing the type and seasonality of extreme-producing weather systems. Our results underscore that reducing uncertainty in regional wind projections requires constraining the physical representation of weather systems in climate models.

2026-03-03T19:59:30Z G. Zhang M. Rao I. Simpson K. A. Reed B. Medeiros H. -H. Chou T. Shaw http://arxiv.org/abs/2603.03458v1 2026-03-03T19:15:21Z

The El Niño-Southern Oscillation (ENSO) is a fluctuation in sea surface temperature (SST) and pressure across the equatorial Pacific Ocean with a period of 2-7 years. As the largest mode of interannual variability on Earth, ENSO shapes global weather and climate patterns ranging from monsoons in southern Asia to hurricanes in the Atlantic and droughts in South America. Predicting and understanding ENSO's response to greenhouse warming is essential for mitigating the impacts of climate change, yet model ensemble projections are prohibitively expensive to generate across emission scenarios and remain incompletely understood. Here, we use a hierarchy of models to show that ENSO strength undergoes a transient rise followed by a long-term fall under greenhouse warming. An East Pacific energy budget reveals that the initial increase in ENSO variability is due to enhanced upper-ocean stratification, while its subsequent decrease arises from a slowing Walker circulation and stronger surface flux damping. Building on these mechanisms, we derive a linear model which predicts the evolution of ENSO variability from only East Pacific temperature and stratification. We further show that subsurface warming, and therefore stratification, is connected to surface warming with a lag, enabling us to create a lag-linear model that explains $\sim$90\% of simulated changes in ENSO variability from only global mean SST and its history. This efficient predictor can forecast ENSO strength over time in any warming scenario, and reveals that faster emissions lead to stronger peak ENSO variability even with identical total emissions.

2026-03-03T19:15:21Z PJ Tuckman Da Yang http://arxiv.org/abs/2603.02165v1 2026-03-02T18:28:40Z

Particulate matter (PM) is linked to adverse health outcomes, yet the roles of specific PM components and their modification by extreme temperature remain unclear. We examined short-term associations between ten PM chemical components and daily mortality in Greater London (2015-2018). PM components include inorganic aerosols (black carbon from wood burning (BCwb) and traffic exhaust (BCtr), SO4, NO3, and NH4) and organic aerosols (hydrocarbon-like organic aerosol (HOA), biomass burning OA (BBOA), cooking-like OA (COA), more and less oxidized oxygenated OA (MO-OOA and LO-OOA)). We applied quasi-Poisson generalized additive models and weighted quantile sum (WQS) regression to estimate single-pollutant, multi-pollutant, and mixture effects, respectively, and included interaction terms to test effect modification by heat waves and cold spells. All ten components showed positive associations with all-cause mortality in single-pollutant models with stronger estimated risks for respiratory mortality, particularly for NH4, NO3, SO4. In mixture analyses, the WQS index was significantly associated with all-cause mortality (RR = 1.015, 95% CI: 1.006-1.024 per 25th-percentile increase) and showed a marginally significance with respiratory mortality (RR = 1.018, 95% CI: 0.994-1.042). MO-OOA and COA contributed most to all-cause mortality, while BBOA and BC Wood dominated respiratory effects. Heat waves consistently amplified respiratory risks in both single-pollutant and mixture models with little evidence for cardiovascular mortality. Overall, MO-OOA demonstrated harmful associations across outcomes, suggesting potential toxicity link to secondary atmospheric oxidation processes. These findings support source-specific control strategies and highlight the importance of accounting for extreme temperature in air pollution mitigation policies.

2026-03-02T18:28:40Z Xiaolu Zhang Anna Font Anja Tremper Max Priestman Shawn Y. Lee David C. Green Dimitris Evangelopoulos Gang I. Chen http://arxiv.org/abs/2602.01904v2 2026-03-02T12:21:19Z

There have recently been many efforts to create machine learnt atmospheric emulators designed to replace physical models. So far these have mainly focused on medium-range weather forecasting, where these `Machine Learnt Weather Prediction' (MLWP) models can outperform leading operational forecasting centres. However, because of this focus on shorter timescales, many of these emulators ignore the effects of the ocean, and take no ocean variables as inputs. We hypothesise that such MLWP models have learnt a best-guess of the evolution of the atmosphere, by implicitly inferring ocean conditions from atmospheric states, with no access to ocean data. Turning this limitation into a strength, we use it as a means to study the role of the oceans on the evolution of the atmosphere. By exploring how model forecast errors relate to properties of the air-sea interface, we infer what ocean information these atmospheric emulators are able to derive from atmospheric data alone, and what they cannot. This highlights the regions and processes through which the ocean independently influences the atmosphere on fast timescales. We perform this analysis for GraphCast, finding clear relationships between air-sea properties and the forecast errors over the ocean, including clear seasonal effects. We then explore what this reveals about GraphCast's internal representation of the ocean. In addition to understanding real-world ocean-atmosphere interactions, this analysis provides guidance for improving forecast skill and physical realism in MLWP models, and for informing how future machine learning models should use ocean information on short timescales.

2026-02-02T10:09:46Z Bobby Antonio Kristian Strommen Hannah M. Christensen http://arxiv.org/abs/2603.01674v1 2026-03-02T10:03:18Z

The temperature in the transient climate response is lower than the equilibrium temperature for the same amount of forcing. The degree of disequilibrium is not constant in time and depends on various climate parameters. We derive intuition for this by solving the heat equation with a surface temperature feedback for linearly increasing forcing. The surface temperature initially evolves at a slower rate than the corresponding steady state (SS) temperature and it accelerates until quasi-steady state (QSS), when the SS and QSS temperatures evolve in parallel with a constant offset. The offset depends on the rate of forcing and total heat capacity of the system divided by the square of the climate feedback. The timescale over which the climate system approaches QSS depends also on the effective ocean mixing and is order of thousands of years. Over societally relevant timescales (around 100 years), the top-of-atmosphere energy imbalance increases, and the actual temperature moves farther from the steady-state temperature expected for the same forcing.

2026-03-02T10:03:18Z Boriana Chtirkova http://arxiv.org/abs/2603.01539v1 2026-03-02T07:12:12Z

Recent papers by Ansari et al. (2024, Science Advances 10, eadn4650) and Richardson et al. (2025, arXiv eprint 2504.01455) have suggested that global warming of the Martian surface ('terraforming') by 35 K to sustain local habitats above the melting point of water could be achieved through the injection of engineered aerosols into the Martian atmosphere. Using the MarsWRF 3D Global Climate Model, we investigate how artificial warming of Mars through engineered aerosol release would affect the planetary water cycle and the distribution of the major surface ice reservoirs. Within our model framework, every 20 K of global warming induces a tenfold increase in atmospheric water vapour content due to sublimation of H2O ice from the North Polar Cap. This increases the potency of cloud radiative feedbacks which induces nighttime warming (~5-10 K) at low latitudes, but daytime cooling (up to 40 K) in the winter midlatitudes. Water is transferred from the edge of the North Polar Cap to the South Polar Cap and there is minor destabilisation of shallow northern midlatitude subsurface ice. As a result, seasonal sublimation of H2O ice from the South Pole has an increased impact on the global water cycle. These changes persist on Mars at least decades after loading of the atmosphere with engineered aerosols ceases. Our model is limited by the gaps in our knowledge of present-day Martian weather and climate, and of the microphysics and radiative properties of candidate warming agents. Much more data is therefore needed before warming Mars could become feasible.

2026-03-02T07:12:12Z Ashwin S. Braude Edwin S. Kite Mark I. Richardson Alexandre Kling Michael A. Mischna http://arxiv.org/abs/2501.19121v2 2026-03-02T02:54:12Z

Generation of internal waves driven by barotropic tides over seafloor topography is a central issue in developing mixing and wave drag parameterizations for ocean circulation models. Traditional analytical approaches estimate the energy conversion rate from barotropic tides to internal waves using a modal expansion of the wave field. However, this framework becomes inadequate if a background shear flow is present, as singular solutions associated with critical levels emerge. To uncover the distinct roles of regular eigenmodes and singular solutions in tidal energy conversion, this study analytically investigates wave generation over a localized small topography in the presence of shear flow without Coriolis force. Applying horizontal Fourier and temporal Laplace transforms, we identify regions in the topographic wavenumber and forcing frequency space where unbounded energy growth occurs. These regions coincide with the spectrum of an operator governing free wave propagation and consist of discrete and continuous parts, which correspond to regular eigenmodes and singular solutions, respectively. Asymptotic evaluation of the Fourier integral reveals that the far-field response comprises standing wave trains linked to the discrete spectrum and evolving wave packets associated with the continuous spectrum. While the velocity amplitudes of the wave packets decay, their vertical velocity gradients grow during propagation, potentially leading to wave breaking. Finally, we derive a formula for the net barotropic-to-baroclinic energy conversion rate, extending the classical one by incorporating the contributions from both the discrete and continuous spectra.

2025-01-31T13:25:38Z 39 pages, 9 figures Yohei Onuki Antoine Venaille