https://arxiv.org/api/cH1tQ43fY4LNKYxiAyUFkcal5UA 2026-06-13T13:57:29Z 8365 135 15 http://arxiv.org/abs/2510.08107v5 2026-05-15T11:57:00Z

Persistent systematic errors in Earth system models (ESMs) arise from difficulties in representing the full diversity of subgrid, multiscale atmospheric convection and turbulence. Machine learning (ML) parameterizations trained on short high-resolution simulations show strong potential to reduce these errors. However, stable long-term atmospheric simulations with hybrid (physics + ML) ESMs remain difficult, as neural networks (NNs) trained offline often destabilize online runs. Training convection parameterizations directly on coarse-grained data is challenging, notably because scales cannot be cleanly separated. This issue is mitigated using data from superparameterized simulations, which provide clearer scale separation. Yet, transferring a parameterization from one ESM to another remains difficult due to distribution shifts that induce large inference errors. Here, we present a proof-of-concept where a ClimSim-trained, physics-informed NN convection parameterization is successfully transferred to ICON-A. The scheme is (a) trained on adjusted ClimSim data with subtracted radiative tendencies, and (b) integrated into ICON-A. The NN parameterization predicts its own error, enabling mixing with a conventional convection scheme when confidence is low, thus making the hybrid AI-physics model tunable with respect to observations and reanalysis through mixing parameters. This improves process understanding by constraining convective tendencies across column water vapor, lower-tropospheric stability, and geographical conditions, yielding interpretable regime behavior. In AMIP-style setups, several hybrid configurations outperform the default convection scheme (e.g., improved precipitation statistics). With additive input noise during training, both hybrid and pure-ML schemes lead to stable simulations and remain physically consistent for at least 20 years.

2025-10-09T11:44:47Z Helge Heuer Tom Beucler Mierk Schwabe Julien Savre Manuel Schlund Veronika Eyring http://arxiv.org/abs/2605.15699v1 2026-05-15T07:44:16Z

A defining feature of the present-day global overturning circulation (GOC) is the absence of deep water formation in the Pacific, in contrast to the Atlantic. This asymmetry, associated with higher surface salinities in the North Atlantic, is reflected in the Atlantic Meridional Overturning Circulation (AMOC) and the lack of a Pacific overturning (PMOC). A commonly cited explanation is the asymmetry in surface freshwater fluxes, with the Pacific receiving more freshwater per unit area than the Atlantic. Here, we develop a two-basin conceptual ocean model, consisting of a wide and a narrow basin. The model admits three states: sinking confined to the narrow basin, sinking confined to the wide basin, and sinking in both basins. We analyze the (co-)existence of these states as a function of freshwater asymmetry and hydrological cycle strength, defined as the longitudinally symmetric freshwater flux. For a weak hydrological cycle, representative of warm Pliocene-like climate conditions, sinking occurs in both basins, with symmetry breaking only when one basin is sufficiently more evaporative. For intermediate conditions, representative of the present-day climate, the basin with slightly stronger evaporation tends to host sinking, with a stronger preference in the narrow basin. For a strong hydrological cycle, single-basin sinking states are preferred, although a large interbasin freshwater asymmetry is required to uniquely localize sinking. These results provide insight into GOC sinking configurations under past, present, and potential future climates, and show good agreement with a three-dimensional global circulation model.

2026-05-15T07:44:16Z Submitted to Journal of Physical Oceanography Elian Vanderborght Oliver Mehling Henk A. Dijkstra http://arxiv.org/abs/2602.23742v2 2026-05-15T04:34:29Z

The transport of the Antarctic Circumpolar Current (ACC) has been shown to increase with friction. Previous studies explained this counter-intuitive relationship called frictional control based on the eddy geometric parametrizations. They focused on the eddy momentum transfer and eddy energetics. To maintain the balance between wind stress and eddy interfacial form stress, eddy energy must remain unchanged as friction increases; this requires enhanced baroclinicity to compensate for stronger eddy energy dissipation. However, the independence of eddy energy has not been fully verified, and this interpretation assumes negligible barotropic energy conversion. To address this gap, we conduct sensitivity experiments in an idealized stratified reentrant channel with varying linear bottom drag. Numerical simulations show that eddy energy changes substantially with friction. Furthermore, in the high-drag regime, baroclinic energy conversion dominates eddy energy generation, whereas in the low-drag regime barotropic energy conversion contributes substantially. Despite these differences, baroclinicity increases with eddy energy dissipation across all regimes, although the relationship is somewhat weak in the low-drag regime owing to barotropic energy conversion. To explain this phenomenon, we extend the frictional control framework based on the Lorenz energy cycle. A simple scaling argument leads to a generalized frictional control, s~D(E)/τ_w, where s is baroclinicity, D(E) is eddy energy dissipation, and τ_w is wind stress. This framework provides a natural extension of the existing framework and successfully explains the numerical results. These results indicate that eddy dissipation controls the baroclinicity; therefore, properly parameterizing the eddy dissipation rate is essential for representing ACC dynamics in ocean models.

2026-02-27T07:13:09Z Takuro Matsuta Yuki Tanaka Atsushi Kubokawa 10.1175/JPO-D-25-0220.1 http://arxiv.org/abs/2512.14898v2 2026-05-14T15:58:34Z

Long Short-Term Memory (LSTM) models are trained to predict forecast errors for the High-Resolution Rapid Refresh (HRRR) model using the New York State Mesonet and Oklahoma State Mesonet near-surface weather observations as ground truth. When evaluated using mean-absolute-error and percent improvement relative to the HRRR, LSTMs predict precipitation error most accurately, providing, on average, a 48% improvement relative to the HRRR forecast, followed by wind error, providing, on average, a 15% improvement, and then temperature error, providing, on average, a 25% improvement. Precipitation errors exhibit an asymmetry, with overforecast precipitation detected more accurately than underforecast, while wind error predictions are consistent across over- and underforecast predictions. Temperature error predictions are relatively accurate but smoother, with respect to variance, than true observations. This paper describes an overview of LSTM performance with the expressed intent of providing forecasters with real-time predictions of forecast error at the point of use within the New York State and Oklahoma State Mesonets. In practice, the predicted errors can be used to adjust deterministic HRRR forecasts at the point of use, identify locations and variables with elevated uncertainty, and provide supplemental guidance for high-impact decision-making. This research demonstrates the potential of LSTM-based machine learning models to provide actionable, location-specific predictions of forecast error for high-resolution operational numerical weather prediction (NWP) systems. However, model performance is variable-dependent, and the approach relies on the availability of dense mesonet observations, which may limit applicability in data-sparse regions.

2025-12-16T20:22:41Z This manuscript is a preprint and has been submitted for peer review to the Weather and Forecasting journal. The content is subject to change based on the outcome of the peer-review process and should not be considered final or definitive. Copyright in this Work may be transferred without further notice David Aaron Evans Kara J. Sulia Nick P. Bassill Chris D. Thorncroft Jay C. Rothenberger Lauriana C. Gaudet http://arxiv.org/abs/2601.21151v2 2026-05-14T14:21:49Z

Recent machine-learning approaches to weather forecasting often employ a monolithic architecture in which distinct physical mechanisms-advection (long-range transport), diffusion-like mixing, thermodynamic processes, and forcing-are represented implicitly within a single large network. This is particularly problematic for advection, where long-range transport typically requires expensive global interaction mechanisms or deep stacks of local convolutional layers. To mitigate this, we present PARADIS, a physics-inspired global weather prediction model that enforces inductive biases on network behavior through a functional decomposition into advection, diffusion, and reaction blocks acting on latent variables. We implement advection through a Neural Semi-Lagrangian operator that performs trajectory-based transport via differentiable interpolation on the sphere, enabling end-to-end learning of both the latent modes to be transported and their characteristic trajectories. Diffusion-like processes are modeled by depthwise-separable spatial mixing, whereas local source terms and vertical interactions are handled via pointwise channel interactions, yielding a physically structured operator decomposition. Evaluated on ERA5 benchmarks, PARADIS achieves competitive deterministic forecast skill, with particularly strong short-lead performance, while preserving substantially better spectral fidelity and forecast activity during medium-range rollouts.

2026-01-29T01:20:21Z Carlos A. Pereira Stéphane Gaudreault Valentin Dallerit Christopher Subich Shoyon Panday Siqi Wei Sasa Zhang Siddharth Rout Eldad Haber Raymond J. Spiteri David Millard Emilia Diaconescu http://arxiv.org/abs/2605.03646v3 2026-05-14T07:28:43Z

Turbophoresis in inhomogeneous turbulent flows leads to the formation of large-scale nonuniform particle number density distributions of inertial particles. This effect is associated with an effective drift velocity directed toward regions of lower turbulence intensity and proportional to the particle Stokes time and the spatial gradient of the turbulence intensity. In the present study, turbophoretic transport is experimentally investigated in air flows generated by one-grid and two-grid oscillating turbulence systems. The flow velocity field and particle spatial distribution are measured using Particle Image Velocimetry. To isolate the effect of particle accumulation due to turbophoresis from that associated with mean fluid flow, the measured particle number density of inertial particles is normalized by the corresponding distribution obtained for noninertial tracer particles under identical flow conditions. The measurements show preferential accumulation of inertial particles in regions of lower turbulence intensity, consistent with the expected behavior of turbophoretic transport.

2026-05-05T11:20:11Z 12 pages, 14 figures, revtex4-2, revised paper E. Elmakies O. Shildkrot N. Kleeorin A. Levy I. Rogachevskii http://arxiv.org/abs/2605.14426v1 2026-05-14T06:18:53Z

Reliable precipitation monitoring is essential for disaster risk reduction, water resources management, and agricultural decision-making. Multi-source satellite observations, particularly the combination of geostationary infrared and passive microwave measurements, have become a primary means of precipitation detection. Traditional multi-source satellite precipitation estimation methods remain computationally inefficient, and many deep learning methods lack the flexibility to incorporate new sensors without retraining the full model. Here we introduce PRISMA (Precipitation Inference from Satellite Modalities via generAtive modeling), a plug-and-play latent generative framework for multi-sensor precipitation estimation. PRISMA learns an unconditional precipitation prior from IMERG Final fields and constrains it through independently trained, sensor-specific conditional branches, allowing new observation sources to be incorporated without retraining the generative backbone. Applied to FY-4B AGRI infrared and GPM GMI microwave observations, PRISMA improves Critical Success Index by up to 40.3% and reduces root-mean-square error by 22.6% relative to infrared-only estimation within microwave swaths, while also improving probabilistic skill and maintaining an average inference time of about 37 s. Independent rain-gauge validation across China confirms consistent gains, and typhoon case studies show that microwave conditioning restores eyewall and spiral rainband structures, reducing storm-core mean absolute error by up to 42.3%. PRISMA thus provides an extensible and efficient framework for multi-sensor precipitation estimation.

2026-05-14T06:18:53Z Yunfan Yang Haofei Sun Xiuyu Sun Wei Han Xiaoze Xu Xingtao Song Jun Li Zhiqiu Gao Wei Huang Hao Li http://arxiv.org/abs/2605.14317v1 2026-05-14T03:27:54Z

Extreme precipitation causes severe societal and economic damage, and weather control has long been discussed as a potential mitigation strategy. However, to the best of our knowledge, perturbation-based interventions for weather control using data-driven weather forecasting models have not yet been explored. While adversarial attacks also generate perturbations that alter forecasts, they aim to exploit model artifacts and do not account for physical plausibility. In this paper, we propose a gradient-based guidance framework for precipitation-reduction interventions through diffusion sampling in diffusion-based weather forecasting models. Instead of directly perturbing atmospheric states, our method steers the diffusion sampling trajectory, enabling precipitation reduction while maintaining consistency with the atmospheric distribution. To assess physical plausibility, we evaluate from three perspectives: (i) vertical and variable-wise perturbation profiles, (ii) latent-space trajectory deviation, and (iii) cross-model transferability. Experiments on extreme precipitation events from WeatherBench2 demonstrate that our method achieves effective precipitation reduction while yielding more physically plausible interventions than adversarial perturbations.

2026-05-14T03:27:54Z 12+7 pages, 7+2 figures Ayumu Ueyama Kazuhiko Kawamoto Hiroshi Kera http://arxiv.org/abs/2605.11968v2 2026-05-13T17:46:08Z

Modern Earth System Models (ESMs) operate on horizontal scales far larger than typical cloud features, requiring stochastic subcolumn generators to represent subgrid horizontal and vertical cloud variability. Traditional physically-based generators often rely on analytical cloud overlap paradigms, such as exponential-random decorrelation, which can struggle to capture the complex, anti-correlated behavior of non-contiguous cloud layers. In this study, we introduce a novel two-stage machine learning subcolumn generator for the GEOS atmospheric model, utilizing a Conditional Variational Autoencoder combined with a Generative Adversarial Network (CVAE-GAN) and a U-Net architecture. Trained on a merged CloudSat-CALIPSO height-resolved cloud optical depth dataset, the ML generator creates 56 stochastic subcolumns representing cloud occurrence and optical depth profiles. Evaluated against the established Räisänen, the ML approach accurately reproduces bimodal cloud overlap distributions, significantly reduces biases in grid-mean statistics, and halves the root-mean-square error in ISCCP-style cloud-top pressure and optical thickness joint histograms. The improvements brought by our deep generative models translate into more accurate offline radiative transfer calculations, reducing the global-mean shortwave top-of-atmosphere cloud radiative effect bias by a factor of three. Provided that the generator can be accelerated on CPUs, this offers a practical pathway to reduce structural errors at the cloud-radiation interface.

2026-05-12T11:19:57Z Dongmin Lee Lazaros Oreopoulos Nayeong Cho Daeho Jin http://arxiv.org/abs/2605.13654v1 2026-05-13T15:13:09Z

We report the experimental characterization of free-surface deformations generated by three-dimensional homogeneous and isotropic turbulence. Using Fourier transform profilometry in a jet-forced turbulent tank, we perform spatiotemporal measurements of the surface elevation field over a wide range of turbulence intensities. The standard deviation of surface deformations scales linearly with subsurface velocity fluctuations. The spectra of surface deformations highlight the coexistence of two mechanisms: transient coherent structures (e.g., upwelling) contributing to the low-frequency, large-scale spectral components, and a passive response to subsurface turbulent pressure fluctuations responsible for the power-law spectral scaling. The wavenumber and frequency spectra of surface deformations exhibit similar power-law exponents (-2.5), suggesting the advection of turbulent structures at the free surface. We develop a linear response model based on the transfer function from the free surface to turbulent pressure fluctuations, incorporating wave-turbulent damping. The model successfully predicts the main features of the turbulent surface: spatiotemporal spectrum shape, similar spectrum power-law exponents (-7/3), and dominance of passive response over wave generation. These findings provide new insights into free-surface turbulence in regimes where turbulent velocities remain below the surface-breaking threshold.

2026-05-13T15:13:09Z Michaël Berhanu Eric Falcon http://arxiv.org/abs/2605.13417v1 2026-05-13T12:13:13Z

The dynamical basis of stratospheric polar vortex predictability remains unclear, particularly the relative roles of persistence, structural variability, and cross-level coupling. Here we provide a state-resolved and quantitative framework using eigen microstate theory applied to ERA5 geopotential height fields, enabling attribution of predictability to dynamically coherent circulation states via a mesoscopic Granger-causality approach. We show that short-term predictability is dominated by persistence of the leading stratospheric state, whereas extended predictability arises from higher-order stratospheric structures and tropospheric variability. These contributions exhibit strong lead-time dependence and become more distributed during sudden stratospheric warming events. Our results unify SPV predictability within a multimodal, state-resolved framework and provide a physically interpretable pathway for improving subseasonal-to-seasonal forecasts.

2026-05-13T12:13:13Z 16 pages, 4 figures Shuo Yang Dan Zhao Tingting Xue Chunhua Zeng Yongwen Zhang Xiaosong Chen http://arxiv.org/abs/2603.28785v2 2026-05-13T06:30:00Z

Micrometeoroids enter Earth's atmosphere at hypervelocity speeds and experience rapid coupling between drag, heating, radiation, melting, ablation, and deceleration. This paper develops a reduced threshold model for the thermal survival boundary of spherical micrometeoroids. The model uses free molecular drag, an exponential atmosphere, projected-area heating, full-sphere radiative cooling, and a surplus-heat ablation rule at the melting temperature. The switching surface $T=T_m$ is treated as a Filippov/complementarity surface. Sustained melting occurs when the local heating-to-radiation ratio exceeds unity. Under the additional Allen--Eggers assumptions of constant radius, constant entry angle, negligible gravity during the main heating interval, and constant transport coefficients, this threshold yields the classical approximate survival scaling $r_0^{\rm crit}\sim v_0^{-3}$. An exact radius-loss identity is obtained along the prescribed Allen--Eggers trajectory, and a perturbative stability estimate explains when this expression approximates the full reduced model. The inverse problem is formulated through a transfer matrix from pre-atmospheric entry bins to observed survivor bins. Entry bins with zero survival probability lie in the survivor-only null space and require external information for reconstruction. The framework gives a compact analytical description of threshold entry survival and identifies the information lost when only surviving particles are observed.

2026-03-19T22:21:42Z 9 pages, 7 figures, 2 tables Md Shahrier Islam Arham Prasun Panthi Min Heo http://arxiv.org/abs/2603.10305v3 2026-05-12T18:57:37Z

Machine learning models can represent climate processes that are nonlocal in horizontal space, height, and time, often by combining information across these dimensions in highly nonlinear ways. While this can improve predictive skill, it makes learned relationships difficult to interpret and prone to overfitting as the extent of nonlocal information grows. We address this challenge by introducing data-driven integration kernels, a framework that adds structure to nonlocal operator learning by explicitly separating nonlocal information aggregation from local nonlinear prediction. Each spatiotemporal predictor field is first integrated using learnable kernels (defined as continuous weighting functions over horizontal space, height, and/or time), after which a local nonlinear mapping is applied only to the resulting kernel-integrated features and optional local inputs. This design confines nonlinear interactions to a small set of integrated features and makes each kernel directly interpretable as a weighting pattern that reveals which horizontal locations, vertical levels, and past timesteps contribute most to the prediction. We demonstrate the framework for South Asian monsoon precipitation using a hierarchy of neural network models with increasing structure, including baseline, nonparametric kernel, and parametric kernel models. Across this hierarchy, kernel models achieve near-baseline performance with far fewer trainable parameters, indicating that much of the relevant nonlocal information can be captured through a small set of interpretable integrations when appropriate structural constraints are imposed.

2026-03-11T01:05:16Z Presented at Climate Informatics 2026 (14 pages, 5 figures, 1 table) Savannah L. Ferretti Jerry Lin Sara Shamekh Jane W. Baldwin Michael S. Pritchard Tom Beucler http://arxiv.org/abs/2605.12311v1 2026-05-12T15:57:26Z

Fundamental inorganic chemistry shows that increasing concentrations of atmospheric CO2 will have no harmful effect on organisms that live in the natural waters of the Earths, and may well benefit them. Alkalinity and dissolved CO2 give high buffering capacity to most natural waters and minimize the change of pH from external influences. For example, doubling the atmospheric concentration of CO2 from 430 ppm to 860 ppm would reduce the pH of representative sea water at a temperature of 25 C from pH = 8.18 to pH = 7.93. This change is comparable to diurnal pH changes in biologically productive surface waters, due to photosynthetic fixation of dissolved inorganic carbon during the day and respiration at night. The change is also less than the variations of pH with latitude, longitude and depth in the oceans. This paper includes a quantitative review of the carbonate chemistry of seawater and freshwater, the buffering capacity, the Revelle factor, the transport of calcium carbonate in ground water, the formation of flowstone, and the classic use of limewater to detect gaseous CO2. The paper concludes with a brief review of those parts of chemical thermodynamics that are involved in ocean acidification.

2026-05-12T15:57:26Z W. A. van Wijngaarden P. Ridd M. Cornell W. Happer http://arxiv.org/abs/2605.21509v1 2026-05-12T12:39:52Z

We use a newly commissioned rapid expansion aerosol chamber (REACh) facility to study the homogeneous nucleation of water vapor to form liquid droplets. We perform high-speed measurements to track the partitioning of water into vapor and droplets throughout the expansion process, including tunable diode laser absorption spectroscopy (TDLAS) to access the vapor concentration and in-line holography to track the size and concentration of nucleating droplets. We retrieve the peak saturation ratio achieved in each expansion from the TDLAS measurements in combination with adjusted thermocouple temperature readout. We monitor the number of nucleated droplets and their subsequent growth as a function of saturation ratio, and observe the onset of homogeneous nucleation of water vapor occurring at a threshold saturation ratio near $S=5$, in agreement with prior literature and classical nucleation theory. The trends we observe in average diameter and droplet concentration suggest that warm air pockets near the chamber walls inhomogeneously mix with cold air at the center of the chamber following expansion. Active forced mixing with fans yields more spatially uniform temperature readings across the chamber, but also significantly broadens the droplet size distribution. Our results demonstrate the capability of TDLAS and holography techniques to track both water vapor and liquid water in the high saturation ratio environments necessary for the homogeneous nucleation of droplets. Our findings also reveal that droplet nucleation and growth dynamics are highly sensitive to turbulence.

2026-05-12T12:39:52Z Cole R. Sagan Gwenore F. Pokrifka Samuel M. Koblensky Martin A. Erinin Ilian Ahmed Nadir Jeevanjee Luc Deike Marissa L. Weichman