physics.ao-ph

0

physics.ao-ph 2026-07-03

Cloud feedback sustains Southern Hemisphere storm tracks

by Or Hadas

Storm Track Self-Reinforcement Through Cloud Radiative Effects

Shortwave effects from midlatitude clouds reinforce temperature gradients when summer insolation is uniform, strengthening activity in late

abstract click to expand

Traditionally, midlatitude storm tracks are viewed as being driven by meridional temperature gradients maintained by differential solar heating. Yet in the Southern Hemisphere, storm activity remains strong even when the summertime insolation gradient nearly vanishes. Here, we show that storm-track cloud radiative effects play a major role in maintaining the Southern Hemisphere storm activity. Satellite observations reveal that sunlight reflected by midlatitude clouds in early summer creates a substantial meridional gradient in surface heating, despite the nearly uniform summer insolation. Idealized aquaplanet simulations then show that shortwave cloud radiative effects reinforce meridional sea-surface temperature gradients, thereby strengthening storm activity primarily during late summer and autumn, while longwave cloud effects partly offset this response. To interpret these results, we develop a simple theoretical model linking storms, clouds, and sea-surface temperature gradients. The model reproduces the simulated seasonal response and identifies two emergent cloud properties that control the feedback strength: the maximum attainable cloud albedo and the sensitivity of cloud cover to storm activity. Together, these findings indicate that cloud radiative feedbacks are key to maintaining the thermal gradients that sustain storm activity. More broadly, they reveal a strong coupling among storms, clouds, and the ocean spanning distinct spatial and temporal scales.

0

physics.ao-ph 2026-07-03

Gravity variations correlate 0.79-0.87 with tropical wind flux

by M. S. Adiaha, V. O. Chude +5 more

Mathematical Exploration of Earth Gravitational Field Impact on Seasonal Wind Flux in a Tropical Region

2010-2020 data from Nigeria links Earth's gravitational changes to seasonal wind patterns via mathematical analysis.

abstract click to expand

The Earth's gravitational field exerts a significant influence on atmospheric dynamics, including the behavior of seasonal wind flux, defined by periodic variations in wind speed and direction. While temperature gradients and Earth's rotation are established drivers of wind patterns, the role of gravitational forces in modulating these processes remains poorly understood. This study investigates the mathematical relationship between gravitational variations and seasonal wind flux in Nigeria, a region of pronounced climatic variability and varied wind patterns. Utilizing Navier-Stokes equations for atmospheric dynamics, Fourier decomposition for seasonal wind flux analysis, and Pearson correlation coefficients for gravitational-wind interactions, we analyze 2010-2020 meteorological data, alongside gravitational field measurements from the GRACE (Gravity Recovery and Climate Experiment) satellite. Results show significant annual fluctuations in average wind speed (5.1-5.6 m/s) and gravitational variations (9.60-9.95 mGal), with an inverse relationship observed in certain years, suggesting a coupling between atmospheric dynamics and gravitational forces. Seasonal wind flux exhibits a distinct sinusoidal pattern, peaking mid-year and declining toward year-end, consistent with Nigeria's monsoon climate. Correlation coefficients between gravitational variations and wind flux range from 0.79 to 0.87, indicating a strong positive relationship. These findings underscore the importance of gravitational forces in modulating wind patterns and highlight the potential for integrating gravitational data into climate models, thereby enhancing accuracy of weather forecasting and renewable energy planning. This study provides a foundational framework for further exploration of gravitational influences on atmospheric processes, with implications for global climate science and sustainable energy strategies.

0

stat.AP 2026-07-03

IRT model extracts rider skill and condition difficulty from binary outcomes

by Fabio Carucci

Inverse Suitability: Identifying Condition Difficulty and Rider Skill from Behavioural Outcomes via Continuous-Item Response Theory

Continuous-item formulation recovers skill at r=0.96 and improves Brier score by 0.33 over expert curves on synthetic cohort of 80 riders.

abstract click to expand

Suitability scoring for outdoor activities (kitesurfing, paragliding, ski touring) maps environmental conditions to a go/no-go verdict via expert-defined curves. These curves conflate two distinct quantities: the intrinsic difficulty of a condition and the skill of the person facing it. We introduce Inverse Suitability, a continuous-item Item Response Theory (IRT) model that identifies both from behavioural outcomes alone. Each outcome is a triple (rider r, condition metric x at site s, binary outcome y); we model P(y=1) = sigma(a (theta_r - delta(x, s))), where theta_r is latent rider skill, delta(x, s) is a latent difficulty function anchored to a physics-derived expert curve as its prior, and a is a discrimination parameter. The formulation is strictly more general than a single suitability curve, which it recovers exactly when skill is integrated out under the population distribution. Parameters are estimated by marginal maximum likelihood with Gauss-Hermite quadrature; identification holds when the rider-by-condition incidence graph is connected, with a documented single-curve fallback otherwise. We validate via synthetic recovery: on a reference cohort (80 riders times 30 outcomes) the model recovers latent skill at r = 0.96, locates the difficulty minimum within 3 units of ground truth, and improves held-out Brier Skill Score by +0.33 over the expert-curve baseline. The recovered difficulty function defines a measurable, site-level construct, an intrinsic difficulty atlas, that existing meteorological observation networks do not capture. All results reproduce from a single command on synthetic data, requiring no proprietary observations.

0

physics.ao-ph 2026-07-03

Three ocean regions back credible 2026 East Asia heat forecast

by Xiaolei Liu, Jingzhi Su

Is the 2026 East Asia Summer Extreme Heat Forecast Credible?

Positive SSTAs in western Pacific, K-KE, and North Atlantic point to high skill and above-average temperatures across China, Korea, and Japa

abstract click to expand

The limited predictive skill of forecasts makes it difficult for decision-makers to act decisively. Advance assessment of real-time forecast credibility can strengthen decision-makers' resolve and confidence to act. Such an assessment can draw on real-time observations of large-scale background signals. This study evaluates how credible the 2026 East Asia summer temperature forecast is. Enhanced predictability of East Asia summer temperature can be indicated by the synergistic forcing of sea surface temperature anomalies (SSTAs) across three key oceanic regions: the tropical western Pacific, the Japan Sea-Kuroshio-Kuroshio Extension (K-KE), and the North Atlantic. Based on the latest observational data and model predictions, the SSTAs in these three regions maintain positive anomalies, which suggests that East Asia's summer temperature forecast skill will stay at a relatively high level in the coming summer. Based on the predictions, the following summer is expected to feature pronounced positive temperature anomalies over central and eastern China, the Korean Peninsula, and Japan, which may trigger regional droughts and place severe strain on power supply networks.

0

physics.ao-ph 2026-07-03

SamudrACE shows biases in Indian monsoon simulation

by Bijit Kumar Banerjee, Devabrat Sharma +7 more

A Deep Learning Earth System Model Simulation of Indian Monsoon Intraseasonal and Interannual Variability

Direct comparison to observations identifies systematic errors in intraseasonal and interannual patterns

abstract click to expand

With the data-driven artificial intelligence/machine learning (AI/ML) models having demonstrated their ability to extend the prediction horizon of large-scale weather at a fraction of computational cost of numerical weather prediction models, a pertinent question is, could these models do the same for sub-seasonal to seasonal (S2S) prediction? A key challenge in developing a S2S prediction system is the requirement for a coupled ocean-atmosphere Earth system emulator that can stably simulate the observed intraseasonal and interannual variability with fidelity. In the rapidly evolving field of AI/ML weather models, such a deep learning 3D ocean-atmosphere coupled model has become available, called SamudrACE. With our interest in developing an AI/ML S2S model for Indian monsoon, here we examine the extent to which SamudrACE faithfully simulates Indian monsoon intraseasonal and interannual variability. Compared to observation, we found biases in SamudrACE's simulation of monsoon intraseasonal and interannual variability. Our systematic documentation and analyses of these biases provide a useful benchmark for improving not only SamudrACE but also coupled emulators in general and could fast track the development of a deep learning 3D global S2S prediction system.

0

astro-ph.EP 2026-07-02

SKA-LOW to map lightning initiation with radio waves

by Brian M Hare, Sjoerd Bouma +16 more

Unveiling the Mysteries of Lightning: Exploring its fundamental Physical Processes with SKA-LOW

Wide bandwidth and sensitivity will capture the faint signals marking how flashes start and spread.

abstract click to expand

Lightning is a surprisingly poorly understood phenomena. It consists of a wide variety of complex processes such as initiation, propagation, connection to ground, even emission of high-energy radiation. However, due to the extreme challenges in observing lightning at fast time scales, small spatial scales, and behind obscuring clouds, these processes are not well understood. In the past, interferometers such as the LOFAR radio telescope have provided unique insight and discoveries into the physics of lightning. The new SKA-LOW being built in western Australia will provide unrivaled spectral bandwidth and sensitivity, which will be combined with high resolution resulting from large antenna baselines. We will use SKA-LOW to observe lightning in order to explore its fundamental plasma physics, such as how it initiates and propagates. SKA's high bandwidth will allow us to test how lightning emits VHF radiation, giving tremendous insight into precisely how the plasma behaves. SKA's sensitivity will allow us to explore extremely faint lightning processes, such as the very first radio emission from a lightning flash. Here, we detail the lightning physics that can be explored with SKA, as well as the observation strategy needed explore such physics.

0

stat.AP 2026-07-01

Argo data produces OHC maps with correlated uncertainty

by Thea Sukianto, Mikael Kuusela +4 more

Locally stationary Argo ocean heat content estimates: Modeling, validation and uncertainty quantification

Locally stationary Gaussian process yields 2004-2022 anomaly fields and validated error ensembles from temperature profiles.

abstract click to expand

Argo profiling floats measure seawater temperature and salinity in the upper 2000 meters of the ocean. These floats are uniquely capable of measuring the global Ocean Heat Content (OHC), a quantity that is of central importance for understanding Earth Energy Imbalance. Yet, producing Argo-based OHC estimates with reliable uncertainties is statistically challenging due to the complex structure and large size of the Argo dataset. Here we present an end-to-end mapping and uncertainty quantification framework for Argo-based OHC estimation using state-of-the-art methods from spatio-temporal statistics. The framework is based on modeling vertically integrated Argo temperature profiles as a locally stationary Gaussian process defined over space and time. This enables us to produce computationally tractable OHC anomaly maps based on data-driven decorrelation scales estimated from the Argo observations. Our modeling choices are validated using statistical cross-validation, which demonstrates the importance of including a climatological time trend in the mean field and accounting for time in the covariance function. We quantify the uncertainty of these maps using local conditional simulation ensembles, a novel approach that leads to principled spatially and temporally correlated uncertainty quantification. A new paired cross-validation technique is presented to validate these uncertainties. The mapping framework is implemented in an open-source codebase that is designed to be modular, reproducible and extensible. To demonstrate the mapping and uncertainty quantification capabilities of this approach, we present new Argo OHC maps with uncertainties for 2004-2022 and report on various downstream climatological estimates and their uncertainties.

0

physics.ao-ph 2026-07-01

AI emulator runs global 4.9 km atmosphere after 17 days of training

by Zeyuan Hu, Akshay Subramaniam +10 more

Scaling Storm-Resolving Atmospheric AI Simulation to the Entire Planet

STRATA produces stable 24-hour rollouts at storm-resolving scales with 50 times the energy efficiency of physics models.

abstract click to expand

Kilometer-scale convection shapes precipitation extremes, tropical organization, and cloud feedbacks, but most global atmospheric models approximate these processes at 25-100 km resolution. Global storm-resolving physics models resolve convective systems explicitly, but at a cost -- roughly one MWh per simulated day on exascale supercomputers -- that limits long-duration simulation. We introduce STRATA (Storm-resolving Tile-based autoRegressive Atmosphere Transformer Architecture), the first autoregressive AI emulator for global storm-resolving atmospheric dynamics. STRATA is trained on the highest-resolution atmospheric dataset yet used for global AI emulation: 17 days of SCREAM physics-model output at 4.9-km resolution (~25 million grid cells) sampled every 10 minutes. Our central premise is that on 10-minute timescales atmospheric dynamics are predominantly local, so training on small spatial tiles trades scarce global temporal samples for abundant local spatial samples and enables global rollout via overlapping-tile blending. STRATA combines 3D patch embedding and local 3D neighborhood attention, a novel Stereographic Rotary Position Embedding (StereoRoPE) for grid-invariant encoding, and a pixel-space de-aliasing decoder that suppresses patch-scale rollout artifacts. An iso-FLOP scaling study reveals that km-scale emulation requires ~10x more FLOPs per grid point than coarse-resolution AI weather models, consistent with the higher information density of convective-scale dynamics. Trained on only 17 days of data, STRATA produces stable 24-hour global rollouts with realistic km-scale dynamics across diverse regimes, though large-scale biases develop with lead time. It achieves 48 simulation days per megawatt-hour -- about 50 times better energy efficiency than the SCREAM physics model -- and 741 simulated days per wall-clock day at 512 H100 GPUs. Code and dataset are publicly available.

0

physics.ao-ph 2026-06-30

Flux sampling resolves cyclone genesis rates over three orders of magnitude

by John S. Schreck, William Chapman +2 more

Conditional Tropical Cyclogenesis Rates via Rare-Event Sampling in a Neural Weather Emulator

A neural emulator paired with rare-event techniques computes formation probabilities that direct ensembles miss at feasible sizes.

abstract click to expand

We couple Forward Flux Sampling (FFS), a non-equilibrium rare-event technique from statistical mechanics, to a neural weather emulator (SDL-WXFormer, 1{\deg} grid spacing) to estimate conditional tropical cyclogenesis rates, or how often a tropical cyclone achieves a hurricane-level central pressure, without modifying model dynamics. Tropical cyclogenesis rates vary by orders of magnitude across regimes, yet direct ensemble sampling cannot resolve this variability at operationally feasible ensemble sizes. FFS decomposes the rare disturbance to mature cyclone intensification path into a flux through an initial interface pressure and a product of conditional crossing probabilities across four intermediate interface pressures. We use the 1{\deg} emulator because FFS requires O(10^4) model trajectories per initial condition, and because the model's calibrated stochastic layers provide the necessary exploratory spread. Applied to 98 Atlantic basin initial conditions spanning 21 August - 8 October 2022, FFS resolves genesis rates spanning nearly three orders of magnitude, capturing a seasonal cycle qualitatively consistent with observations. A self-consistency check comparing FFS rates to independent direct-sampling rates yields a mean ratio of 1.03 +/- 0.15 across all initial conditions. Computational enhancement factors range from 3X (most active environment) to 140X (most suppressed), with a geometric mean of 14X. Three case studies illustrate the physical diagnostics the method provides: the rate-limiting step is initial tropical organization for the Earl environment, uniformly high crossing probabilities for the Fiona precursor environment, and a compound barrier at the final intensification stages for the Ian environment. More efficient emulators would enable application of FFS to finer resolutions.

0

physics.ao-ph 2026-06-30

Log-ratio rule sets best timing to sample rare weather events

by Justin Finkel

Routes to rare events with optimally timed perturbations: a Tent Map is all you need

In the Tent and Logistic maps the advance split time equals the log of rarity over perturbation size, replacing ad hoc choices in extreme-ev

abstract click to expand

Extreme weather events are difficult to understand for the same reason that they are dangerous: they happen rarely, catching victims unprepared when they do occur and scientists unable to assess risks confidently, given such limited precedent to learn from in the real world and high computational expense to simulate more examples. Rare event sampling (RES) algorithms seek to reduce this expense by forcing simulations more directly towards the extremes and then compensating for that forcing in statistical analysis. But the performance of RES hinges on several hyperparameter choices which are ad hoc in practice, and must be better understood if RES is to be broadly useful. This paper addresses one particular parameter, the \emph{advance split time} (AST), which prescribes when to perturb a simulation to split off the most informative possible ensemble of alternative extreme event scenarios. We prescribe the optimal AST as the time it takes for an initial perturbation to amplify into the size (inverse rarity) of the extreme event being targeted. For the Logistic and Tent maps, two archetypal examples of one-dimensional chaos, we rigorously derive and express the rule as a simple log-ratio between perturbation size and event rarity. The pair of examples also illuminates where the rule breaks down, and subsequently, we generalize the rule into a maximum-entropy criterion that solidifies recent heuristic and empirical results. Despite the idealized setting, our results deliver theoretical clarity that can anchor future developments of principled RES methods applicable to real-world, high-impact weather and climate extremes.

0

physics.ao-ph 2026-06-29

Generative models outperform on precipitation downscaling

by Neelesh Rampal, José González-Abad +35 more

CORDEX-ML-Bench: A Benchmark for Data-Driven Regional Climate Downscaling -Experiment Design and Overview

Benchmark across three regions shows they capture extremes better; historical-only training underestimates future change signals.

abstract click to expand

Machine learning (ML) has emerged as a cost-effective approach to complement dynamical downscaling for producing high-resolution regional climate projections. However, the absence of standardised training and evaluation protocols, applied consistently across multiple domains, continues to hinder meaningful model intercomparison. We introduce CORDEX-ML-Bench, a benchmark aligned with CORDEX, which constitutes the first phase of a community initiative to advance data-driven downscaling toward operational readiness, and complement future dynamical downscaling efforts under CMIP7. The framework targets downscaled daily maximum temperature and precipitation to ~10 km resolution (20x increase) across three pilot regions; European Alps, New Zealand, and Southern Africa. Using a perfect-model experimental design, we evaluate 40 ML configurations developed independently, spanning traditional ML, convolutional U-Nets, vision transformers, graph neural networks, and generative models based on diffusion, flow matching, and generative adversarial networks. Models are trained under two experimental periods, an empirical-statistical downscaling pseudo-reality (historical period only) and Emulator (historical and future periods) -and are evaluated against a core set of metrics developed specifically for assessing downscaling skill. Generative models consistently outperform deterministic approaches for precipitation, better capturing fine-scale variability and extremes. For temperature, the generative advantage narrows and deterministic architectures remain competitive. Models trained solely on the historical period systematically underestimate future climate-change signals while those additionally trained on a future period perform better. These findings raise concerns about historically trained models widely used in an operational setting, underscoring the need for rigorous extrapolation testing.

0

physics.ao-ph 2026-06-29

Curved mean temperature profile creates temperature skewness

by Keiko Kircher, Cristi Proistosescu +1 more

Analytical Model for the Higher Order Moments of Midlatitude Atmospheric Temperature Distributions

Symmetric advection from a nonlinear north-south temperature field reproduces observed higher moments and their decline under warming.

abstract click to expand

Observed distributions of atmospheric temperature are non-Gaussian. Therefore, moments beyond variance are necessary in determining the frequency of extreme temperature events. Here we propose a simple kinematic model for atmospheric mid-latitude temperature variability based on symmetric advection from a non-symmetric background temperature profile. We then use this model to derive analytical expressions for the higher order moments of temperature distributions. Our results show that nonzero skewness and kurtosis arise due to the nonlinearity of the time-mean meridional temperature profile. The analytical model matches an idealized Held-suarez atmospheric model, indicating nonlinearity of time-mean temperature in latitude is the dominant contribution to nonzero skewness and kurtosis in synoptic temperature variations. Model analysis further shows decrease in higher order moments due to climate change come roughly equally from changes in mixing length and changes in the background temperature profiles.

0

physics.ao-ph 2026-06-29

Two-step filter beats EnKF on nonlinear observations in Lorenz tests

by Zixiang Xiong, Feng Bao +4 more

A Two-Step Ensemble Score Filter for Data Assimilation in Partially Observed Systems

EnSF-LR applies nonlinear score updates to observed components then uses prior covariance to correct unobserved ones, lowering full-state RM

abstract click to expand

Data assimilation blends model forecasts with observations to estimate the evolving state of complex dynamical systems, but sparse observing networks remain challenging because unobserved state variables are not directly constrained by observations. In this work, we introduce the Ensemble Score Filter with Linear Regression (EnSF-LR), a two-step filtering method for partially observed nonlinear systems. At each analysis time, EnSF-LR first applies the Ensemble Score Filter (EnSF) to update the observed state components using a nonlinear score-based analysis update. It then computes the resulting observed-state analysis increments and maps these corrections to the unobserved components through the ensemble-based prior covariance matrix. The latter amounts to the same linear regression mechanism used by Ensemble Kalman Filters (EnKFs). We evaluate EnSF-LR using the Lorenz-63 and 40-dimensional Lorenz-96 systems with sparse linear and nonlinear observations. The method is compared with the original EnSF and with the classical stochastic EnKF. In the linear-observation experiments, EnSF-LR produces accuracy comparable to the EnKF baseline while substantially reducing error relative to the original EnSF. In the nonlinear-observation experiments, EnSF-LR achieves lower full-state root-mean-square error than both the original EnSF and the EnKF reference. These results suggest that hybridizing score-based and EnKF analysis schemes provides an effective strategy for assimilating sparse and nonlinear observations.

0

physics.ao-ph 2026-06-29

Fully-developed seas follow k to the -2.5 wave spectrum tail

by Hannah Hata Williams, Michael E. Mueller +1 more

Observations and empirical functions for the ocean surface wave spectrum

Using peak wavenumber and wave height the form fits field data and shifts modeled boundary-layer roughness by amounts seen in real wave chan

abstract click to expand

Accurate parameterizations of ocean wave spectra are necessary in a wide array of disciplines including coastal, ocean, and naval engineering as well as in the study of wave interactions and ocean-atmosphere momentum flux. Many such applications use spectrum parameterizations based on temporal data collected well over a half century ago. The development of spatial wave measurement techniques that can accurately capture a larger range of scales allows us to revisit the question of how best to represent an ocean wave spectrum in a variety of ocean wave conditions. We discuss two commonly used wave spectrum parameterizations through a comparison to data collected in field campaigns studying fetch-limited, fully-developed, and mixed sea conditions. We discuss a spectrum parameterization for fully-developed seas that has a $k^{-2.5}$ (or $\omega^{-4}$) dependence on the wavenumber (or angular frequency) in the tail as opposed to the $k^{-3}$ (or $\omega^{-5}$) dependence seen in other frequently-used parameterizations. With knowledge of the peak wavenumber $k_p$ and significant wave height $H_s$, alongside the wind speed, fully-developed conditions can be well-represented. We then compare the impact of using different wave spectrum parameterizations through a Large Eddy Simulation (LES) study of Marine Atmospheric Boundary Layers (MABLs) over the sea surface and find that changing the parameterization used results in variations in the equivalent roughness akin to significant changes in wave conditions.

0

physics.ao-ph 2026-06-29

Natural processes mask human CO2 emission signals regionally

by Yogesh Bali, Darja Cvetković +8 more

Investigation of regional variations in CO₂ growth rates : Integrating Emission Inventories and Atmospheric Observations

Atmospheric data integration shows 2020 reductions not reflected due to biosphere and transport dominance over local emissions.

abstract click to expand

Atmospheric carbon dioxide (CO2) growth rates reflects the combined influence of anthropogenic emissions, biospheric carbon exchange, and climate variability. While climate mitigation is primarily evaluated using bottom-up emission inventories within political boundaries, there is a need to validate these emission reductions using atmospheric measurements. Here, we present a global top-down analysis of atmospheric CO2 growth rates using CAMS atmospheric CO2 reanalysis, EDGAR anthropogenic emissions, GOSIF dataset and the Southern Oscillation Index (SOI) as a measures of biospheric activity, to quantify the relative influence of human and natural drivers. We find that atmospheric CO2 growth rate varies substantially across space and time but is dominated by natural carbon-cycle processes and global background trends. Anthropogenic emission signals are frequently masked by natural variability, making regional top-down detection of human emission changes difficult. The COVID-19 emission reductions in 2020, despite occurring during a neutral ENSO year, were not consistently reflected in regional atmospheric CO2 growth rates, highlighting the dominant roles of biospheric dynamics and atmospheric transport. Using unsupervised clustering and persistence analysis, we identify five characteristic carbon-cycle regimes. Spatial averaging removes much of the regional variability, leaving large-scale climate as the dominant control in most regimes. The active biosphere is the main exception, where strong biogenic signals persist, underscoring the critical role of tropical forests in shaping atmospheric CO2 variability.

0

physics.ao-ph 2026-06-29

Gulf Stream front shifts affect jet and storms only below 50 km

by Luca Famooss Paolini

Ocean-atmosphere interaction at the Gulf Stream sea surface temperature front: variability and impacts on midlatitude atmospheric circulation

High-resolution runs match observed anomalies; NAO leads front changes by 2-3 years with non-stationary covariance after 1972.

abstract click to expand

Sea surface temperature (SST) gradients associated with western boundary currents affect the atmospheric circulation across a range of spatial and temporal scales. Yet, several aspects of ocean-atmosphere interactions linked to oceanic fronts remain unclear. This PhD thesis analyses such interactions for the Gulf Stream SST front (GSF). The first part assesses the atmospheric response to the interannual GSF meridional shifts and its dependence on model horizontal resolution, using ERA5 reanalysis and atmosphere-only simulations forced by observed SST. Results show that the response is strongly resolution dependent, with only simulations finer than 50km resembling observed anomalies. Locally, diabatic heating near the GSF is mainly balanced by vertical motion and transient eddy heat transport. At large-scale, the GSF shifts is associated with a homo-directional shift in the North Atlantic eddy-driven jet and storm track, mediated by changes in low-level baroclinicity. The second part assesses the North Atlantic Oscillation (NAO)-GSF interaction and the mechanisms through which the NAO forces the GSF shifts on decadal timescale, using atmosphere and ocean reanalyses. The NAO and GSF covary on decadal timescales only during 1972-2018. This non-stationarity is also reflected in their lead-lag relationship: the NAO leads the GSF shifts by 3 years during 1972-1990 and by 2 years during 1990-2018. The lag is interpreted as the joint effect of the fast response of wind-driven oceanic circulation, the lagged response of deep oceanic circulation, and the propagation of Rossby waves. However, Rossby wave propagation is evident only before 1990, suggesting that its non-stationarity may explain the different NAO-GSF time lag before and after 1990. Overall, the thesis improves understanding of GSF variability and its role in North Atlantic and extratropical climate variability.

0

physics.flu-dyn 2026-06-26

Framework generalizes internal wave spectrum beyond hydrostatic limits

by Leticia Fabre-Lima (1), Jeffrey Early (2) +3 more

Toward a Universal Framework for the Internal Gravity Wave Spectrum

Wavenumber-space vertical modes from arbitrary stratification improve vertical energy and boundary representations over GM theory

abstract click to expand

The Garrett-Munk (GM) spectrum has long provided a canonical model of the oceanic internal gravity wave field. However, it relies on hydrostatic assumptions and idealized stratification that limit its applicability where non-hydrostatic dynamics, vertical boundary effects, or non-monotonic stratification are important. Here we develop a generalized framework for the internal wave spectrum based on non-hydrostatic vertical modes formulated in horizontal wavenumber-vertical mode space. Energetic orthogonality among wave modes requires that such a formulation be cast in horizontal wavenumber space rather than frequency space. In this formulation, the deformation radius associated with each vertical mode provides a proxy for distinguishing hydrostatic and non-hydrostatic regimes. Vertical modes are obtained numerically from the fixed-K Sturm-Liouville problem, allowing arbitrary stratification and multiple turning depths. Combined with a generalized spectral function, the formulation yields expected distributions of horizontal kinetic, vertical kinetic, and potential energy as functions of depth, frequency, and horizontal wavenumber. Example applications illustrate departures from GM theory associated with boundary effects and non-hydrostatic dynamics, including improved representation of vertical variance and high-frequency vertical kinetic energy, while reproducing observed features of horizontal wavenumber spectra.

0

physics.ao-ph 2026-06-26

A-optimality in model space minimizes EnKF forecast error variance

by Takumi Saito, Shunji Kotsuki

Sparse Sensor Placement for Reducing Forecast Errors in Ensemble Kalman Filtering

Sparse sensor placement is extended from analysis errors to forecast errors by deriving Fisher information matrices and a fast greedy algori

abstract click to expand

Designing efficient observation networks for reducing forecast errors is a fundamental challenge in numerical weather prediction. Data-driven sparse sensor placement (SSP) and ensemble-based data assimilation via the Ensemble Kalman Filter (EnKF) have each addressed this challenge independently, yet their mathematical connections have not been systematically formalized. This study presents a unified theoretical framework integrating SSP and EnKF through optimal experimental design, providing new theoretical and algorithmic results. While conventional SSP methods aim to reduce analysis errors, this study extends the SSP to target forecast error reduction by using a tangent linear model approximated by an ensemble forecast. We derive the Fisher information matrices in the ensemble and model spaces for the EnKF, and clarify the mathematical interpretations of A-, D-, and E-optimality in terms of forecast error reduction. A-optimality in the model space minimizes the mean forecast error variance; D-optimality is ill-defined in the model space due to rank deficiency and is therefore formulated in the ensemble space, where it maximizes the Shannon information content of assimilated observations; and E-optimality in the model space minimizes the worst-case forecast error variance. We further propose a fast greedy algorithm for selecting observation locations under A-optimality in the model space, avoiding matrix inversion at each greedy step and substantially reducing computational cost. Numerical experiments using the Lorenz-96 model support these theoretical findings. Among the three optimality criteria, A-optimality in the model space most consistently reduces forecast spread and root-mean-square error, and yields stable incremental improvements consistent with post-assimilation observation impact diagnostics.

0

physics.ao-ph 2026-06-26

Pre-training closes data gap for diffusion models on ENSO

by Lluis Palma, Vincent Verjans +3 more

Learning Climate Variability from Scarce Data with Diffusion Models: A Test Case for ENSO

Fine-tuning after CMIP6 pre-training reproduces observed statistics better than LIMs despite only 700 real samples.

abstract click to expand

Diffusion models are increasingly applied to climate emulation, but whether they capture the correct modes of variability remains unclear, a concern amplified by data scarcity at longer timescales. We investigate this using synthetic tropical Pacific SST fields from Linear Inverse Models (LIMs), whose known low-order structure bypasses the overlapping and confounding modes of real observations. With sufficient training data, our model recovers the correct structure of both Gaussian and non-Gaussian LIMs, including ENSO's Eastern/Central Pacific asymmetry. Yet an ablation study on the number of monthly training samples reveals that the 700 observations in ERSSTv5 fall an order of magnitude short of the 7,000 samples needed for convergence, and that not all diffusion parameterisations recover the correct low-order structure. Pre-training on CMIP6 with a learned model embedding, followed by fine-tuning on scarce observations, closes this gap, reproducing observed statistics more faithfully than both Gaussian and non-Gaussian LIMs.

0

physics.ao-ph 2026-06-26

Entropy of eigen-microstates tracks polar vortex breakdown

by Dan Zhao, Yongwen Zhang +4 more

Spectral condensation in a finite nonequilibrium atmospheric transition

Sudden stratospheric warmings condense to few states, enter high-entropy competition, then recondense into a weak vortex.

abstract click to expand

Order parameters are difficult to define in high-dimensional nonequilibrium systems that lack a Hamiltonian, a thermodynamic limit or an observed control coordinate. Here we show that such transitions can be diagnosed from the spectrum of occupations over data-derived eigen-microstates. We combine Eigen Microstate Theory with a Marchenko--Pastur random-matrix baseline to isolate an emergent sector, whose entropy quantifies competition among statistically significant collective states. As a finite atmospheric realization, we analyse 51 sudden stratospheric warmings in ERA5. The event-aligned ensemble undergoes spectral condensation, decondensation and recondensation: a polar-vortex state dominated by a few eigen-microstates gives way to a high-entropy regime of competing emergent states before selecting a reorganized weak-vortex state. A stochastic wave--mean-flow model, in which upward wave-activity flux provides a reduced control coordinate, reproduces the same entropy maximum, collapse and top-down timing. These results identify emergent-sector entropy as an order-parameter-like, state-based spectral diagnostic for non-Hamiltonian transitions and place polar-vortex breakdown within a broader class of finite nonequilibrium phase reorganizations.

0

physics.ao-ph 2026-06-26

ML model matches cell frequency near cold fronts

by George Pacey, Stephan Pfahl +1 more

Modelling convective cell occurrence in proximity to cold fronts using extreme gradient boosting

It correctly places peak activity at the surface front using ERA5 data but underestimates counts there.

abstract click to expand

Machine learning is emerging as a valuable tool for convection-related applications such as post-processing numerical weather prediction output, improving understanding of convective storm climatology and potentially improving existing convective parameterization schemes. In a rapidly developing field, it is vital to assess the strengths and limitations of machine learning approaches across different applications. Here, a probabilistic model is developed using a convective cell dataset as ground truth and predictors primarily from ERA5. The model's ability to reproduce the convective cell climatology at different regions relative to cold fronts (i.e. post-frontal and pre-frontal) is assessed during the warm-season in Germany. The optimal number of features (predictors) is selected using a feature elimination strategy. Overall, the optimised model exhibits high skill in reproducing the spatial and temporal cell frequency at different regions relative to the front. While the highest cell frequency is correctly identified near the surface front, the model underestimates the actual cell count in this region. Feature importance analysis shows that the model depends most heavily on CAPE to make its predictions. Additionally, the time of day predictor is key for accurately capturing the diurnal cycle of convective cells on both sides of the cold front. The study highlights both the advantages and the limitations of data-driven models, offering valuable insights for future data-driven climate and weather prediction models.

0

physics.ao-ph 2026-06-25

Diffusion model samples sea state from five-day wind history

by Jiarong Wu, Bertrand Chapron +1 more

Sampling sea state using a diffusion model

Replaces spectral wave integration with direct sampling to produce probabilistic forecasts and derived quantities at reduced cost.

abstract click to expand

Sea state prediction is essential for operational maritime applications and coupled earth system modeling, yet current spectral wave models remain computationally prohibitive for many use cases, including online coupling to climate simulations and making probabilistic (ensemble-based) predictions. While deep learning has recently demonstrated strong performance in weather forecasting, existing AI-based wave models are predominantly deterministic and largely limited to bulk variables such as significant wave height, leaving probabilistic sea state estimation largely unexplored. In this work, we propose a diffusion-based generative model for global sea state estimation that conditions on a relatively long history (5 days) of global wind forcing. This generative model directly samples the complex conditional distribution of sea state without autoregressive time-stepping. Unlike prior approaches, our framework naturally extends beyond bulk variables to estimate partition-related variables and derived quantities, such as Stokes drift and mean square slope. Trained on a 30-year global WAVEWATCH-III hindcast, the model achieves substantial computational acceleration compared with numerical spectral models while delivering skillful predictions and a calibrated ensemble spread for the bulk variables. Our results suggest that diffusion-based sea state sampling offers a promising path toward probabilistic wave forecasting and efficient coupling of sea state information into broader earth system models.

0

math-ph 2026-06-25

Bracket product builds thermodynamic QG ocean model

by Francisco J. Beron-Vera, Erwin Luesink

A new formulation of metriplectic dynamics with an application to quasigeostrophic ocean modeling with advected quantities

Multiplying two skew-symmetric brackets creates a four-bracket that conserves energy and generates entropy with flexible irreversibility.

abstract click to expand

A general formulation of metriplectic dynamics is presented, where the metriplectic four-bracket is constructed by multiplying two skew-symmetric brackets. The new formulation is then used to introduce irreversibility in a generalized two-dimensional (2D) quasigeostrophic (QG) upper-ocean model involving advected quantities, with the thermal QG model as a special case. By construction, the resulting dynamics ensure the conservation of internal energy and the generation of entropy, in accordance with the first and second laws of thermodynamics. Our metriplectic dynamics formulation allows for a flexible specification of irreversibility, ranging from a type that results in nearly material conservation of potential vorticity to the representation of realistic forcing and dissipation in 2D QG ocean modeling with advected quantities.

0

cs.LG 2026-06-25

Aurora latent space tracks seasons more than storms

by Emma Kasteleyn, Ana Lucic

Does Aurora Encode Atmospheric Structure? Latent Regime Analysis and Attribution

Analysis shows the model attends to vertical atmospheric layers, with masking tests confirming their importance for forecasts.

abstract click to expand

ML foundation models are able to emulate atmospheric dynamics accurately and efficiently but operate as opaque ``black boxes''. We investigate the internal representations of the Aurora model using spatially pooled PCA and layer-wise relevance propagation (LRP). We find evidence that Aurora's latent space is primarily organized by seasonal cycles, whereas extreme storm events do not form a linearly separable cluster. LRP indicates that the model attends to features consistent with the 3D vertical structure of the Great Storm of 1987. Perturbation tests show masking relevant regions degrades forecasts $3.31\times$ more than random masking. These findings suggest that Aurora learns meteorological coherence and vertical structure without explicit instruction.

0

physics.flu-dyn 2026-06-25

Enthalpy Lagrangian yields energy-conserving moist rotating shallow water model

by Colin J. Cotter, Darryl D. Holm +1 more

Variational derivation of a moist thermal rotating shallow water model

Latent heat alters buoyancy which feeds back into the vertically integrated flow

abstract click to expand

We introduce a new energy-conserving, moist shallow water model with thermal stratification and rotation. The model is derived from a variational principle, using a Lagrangian expressed in terms of enthalpy. In this model, the latent heat from phase transitions modifies the buoyancy dynamics, which in turn feeds back to alter the vertically integrated hydrodynamic motion. Finally, we generalise this moisture parameterisation to non-hydrostatic Green-Naghdi equations.

0

physics.ao-ph 2026-06-25

Lightning-weighted loss boosts heavy rain forecasts

by ChangJae Lee, Heecheol Yang +1 more

Event-Aware Loss Design for Forecasting of Convective Precipitation and Lightning

Multiplying observed strikes into the training loss improves skill at intense thresholds over standard deep learning and physics models.

abstract click to expand

Accurate forecasting of high-impact weather, specifically extreme precipitation and lightning, remains a significant challenge in numerical weather prediction (NWP) due to the complexity of atmospheric microphysics. While deep-learning models have shown promise in large-scale forecasting, they often suffer from systematic under-prediction of rare, high-intensity events and localized convective showers when optimized with conventional loss functions like Mean Squared Error (MSE). This study proposes an Event-Aware multi-task deep-learning post-processing framework designed to improve the representation of convective processes by leveraging lightning observations. The model jointly predicts precipitation amount, rainfall probability, and lightning occurrence using a shared-backbone Patch-cGAN (Conditional Generative Adversarial Network) architecture. To address the rare event problem, we introduce a lightning-informed loss-weighting strategy that element-wisely multiplies the MSE component by a spatial weight map derived from observed lightning strikes, forcing the model to prioritize accuracy in convective regions during training. Evaluations conducted over the Korean Peninsula during the 2025 Summer demonstrate that our framework outperforms standard AI benchmarks and conventional NWP models, particularly at intense rainfall thresholds (40 mm/6 h). Furthermore, the model exhibits superior skill in predicting lightning compared to conventional lightning parameterization and instability-index-based methods. These results indicate that integrating physical event indicators into the loss formulation effectively guides models to learn the meteorological signatures of deep convection, offering a pathway toward more reliable extreme weather forecasting.

0

physics.ao-ph 2026-06-25

ML models match physics skill at sub-seasonal leads

by Catherine de Burgh-Day, Chen Li +5 more

Evaluation of medium range machine learning models for sub-seasonal prediction

GraphCast and FourCastNetV2, built for medium-range use, equal ensemble-mean skill at short leads and member skill at longer leads.

abstract click to expand

The performance of two machine learning (ML) atmosphere models - GraphCast and FourCastNetV2 - is evaluated in the context of sub-seasonal prediction, including their ability to represent key climate drivers of variability, namely the Madden-Julian Oscillation and the Southern Annular Mode. Model skill is assessed over both a 38-year hindcast period and a 2.5-year hindcast period. The longer period overlaps with the training windows of the ML models but provides a larger sample for robust evaluation, while the shorter period is independent of the ML model training period. This dual evaluation illustrates a compromise approach to the problem of insufficient independent data for evaluation of the models for sub-seasonal prediction. The ML models are compared against the Bureau of Meteorology's physics-based seasonal prediction system, ACCESS-S2, for the 38-year period, and a more recent physics-based coupled model for the shorter hindcast period. Across the two evaluation periods, both ML models have surprisingly good skill for sub-seasonal timescales, given they were designed for forecasting on medium range timescales. In general, the ML models are as skilful as the physical model ensemble mean at shorter lead times and comparable to the physical model ensemble members at longer lead times.

0

physics.ao-ph 2026-06-24

Machine learning forces overhaul of weather forecasting practices

by Peter Dueben, Peter Bauer +3 more

Machine learning is revolutionizing weather forecasting -- the next step is a change in how we work

The value chain from model coding to service delivery must adapt to new digital tools while preserving reliability and expertise.

abstract click to expand

Following the success of machine learning in producing weather predictions with competitive skill compared to complex traditional systems, this article shifts attention from forecast output to the working practices that make prediction systems possible. We argue that machine learning and recent digital technologies will reshape the forecasting value chain: how models are coded and developed, how observations and Earth-system data are exploited, how data and computing are managed, how systems are verified, and how information is created, evaluated and turned into services. We discuss six non-exhaustive areas in which agentic software engineering, open and compressed data, shared verification workflows, interactive computing and generative methods may make modelling, evaluation and service creation faster, more interactive and more widely accessible. These changes will require weather and climate centres to adapt their infrastructures, data stewardship, trust and quality-assurance frameworks, skills and service delivery while maintaining scientific understanding, operational reliability, human expertise and their public-service role.

0

physics.ao-ph 2026-06-24

New trigger boosts spread in tropical precipitation ensemble

by Mark R. Muetzelfeldt, Robert S. Plant +6 more

An observationally constrained probabilistic trigger for organized deep convection in an NWP ensemble

Observationally constrained activation of convection scheme reduces underdispersion while preserving scale improvements in NWP.

abstract click to expand

A novel stochastic parametrization scheme representing organized convection is described. The effects of mesoscale convective systems (MCSs) are represented in an observationally constrained manner, by probabilistically triggering an MCS scheme in regions of enhanced environmental total column water vapour. In combination with the probabilistic trigger, patterns with given spatiotemporal scales determine where and when the scheme is active. Our scheme builds on the multiscale coherent structure parametrization (MCSP), which represents the top-heavy heating structure associated with MCSs. The original and new MCSP schemes are tested in a numerical weather prediction (NWP) ensemble. Both MCSP schemes improve the spatiotemporal scales of tropical precipitation compared to a control. When the spread-error relationship of tropical precipitation is analysed, the new scheme successfully boosts spread compared to original MCSP, improving the underdispersion of the ensemble seen with the original MCSP.

0

physics.ao-ph 2026-06-24

Lag sync with ENSO extends monsoon forecasts past chaos limit

by Vladimir García-Morales, Devabrat Sharma +3 more

Why is Seasonal Climate Predictable Beyond the Limit of Deterministic Predictability set by Chaos?

Coupled oscillator model produces correlated time series at 18-month leads where deterministic predictability ends.

abstract click to expand

The Earth's climate is an ensemble of interacting, spatially extended oscillatory media ('climate systems') whose slow-changing averages coexist with chaotic, high-frequency weather fluctuations in quasi-equilibrium. The limit of deterministic predictability (LDP) for any climate system is determined by its fastest-growing errors. However, recent findings show that the Indian Summer Monsoon Rainfall (ISMR) can be predicted up to 18 months in advance-far beyond its LDP. Using a model of two interacting oscillatory media, we show that this extended predictability arises from lag synchronization between ISMR and its predictor, the Global El Nino-Southern Oscillation (G-ENSO), to which it is strongly coupled. We introduce complex order parameters representing the internal dynamics of the two climate systems. Their spatiotemporal evolution is governed by coupled Complex Ginzburg-Landau Equations, producing aperiodic yet strongly correlated time series at long lead times. Our findings have far-reaching consequences in advancing seasonal prediction across climate systems.

0

physics.ao-ph 2026-06-23

ML microphysics scheme runs stable decade-long climate sims

by Ellen Sarauer, Mierk Schwabe +4 more

From stable online coupling to decade-long climate simulations: A machine learning parameterization for cloud microphysics in ICON

It matches the graupel scheme's climate performance and removes two tuning parameters.

abstract click to expand

The representation of cloud microphysics and its nonlinear character and scale-dependence is a remaining source of uncertainty in Earth system models (ESMs). Here, we develop and couple online a machine learning (ML)-based cloud microphysics parameterization with the Icosahedral non-hydrostatic modeling framework (ICON). The primary challenge is achieving numerically stable, long-term online coupling when transitioning from training with km-scale data to application in coarse-scale simulations, where the coupled system encounters atmospheric states and feedbacks not seen during training. The training data is obtained from a global convection-permitting ICON simulation at 5 km resolution. The ML microphysics scheme uses a two-stage design: a classifier to identify active grid cells and a regressor to predict cloud microphysical tendencies. Physical constraints such as enforcing mass positivity and overshoot prevention prove essential for numerical stability in the coupled system. We demonstrate that achieving stable online coupling requires enforcing physical constraints and careful dataset curation, and that strong offline performance alone is insufficient. The coupled model maintains numerical stability over decade-long simulations with a performance in reproducing the observed climate comparable to the classical graupel scheme. The ML-based scheme eliminates two microphysics-specific tuning parameters of the classical graupel scheme, though systematic improvements in long-term mean-state biases are not yet realized. This study demonstrates that stable, decade-long climate simulations with an ML-based cloud microphysics scheme trained on convection-permitting data are feasible, providing a foundation for future hybrid ESMs.

0

physics.ao-ph 2026-06-23

Dataset friction framework complements FAIR without being predicted by it

by Emma Pidduck, Umberto Modigliani

The Dataset Friction Framework: measuring user-facing friction as a complement to FAIR

Support ticket analysis shows the two measures disagree in both directions and inform different aspects of data service design.

abstract click to expand

Open research data services have matured to the point where the cost of sustaining them at scale has become a primary design constraint, driving providers to make deliberate choices that may reduce user convenience to keep the service viable. The FAIR (Findable, Accessible, Interoperable, Reuseable) principles describe whether a dataset is well stewarded, and FAIR compliance is often treated as a proxy for usability. FAIR does not capture the cost to a user of finding, accessing, interpreting, and applying a dataset. We introduce the Dataset Friction Framework (DFF) as a complement to FAIR, directly addressing usability. DFF measures user-facing friction across six dimensions, distinguishing engineered friction (deliberate data provider design choices that sustain a service) from accidental friction (defects that require remediation). The framework is validated against 18,556 support tickets from the European Centre for Medium-Range Weather Forecasts (January 2024 to May 2026), which serves 280,000 registered users. Restricting the analysis to tickets raised by external reporters reduces the corpus by 12.3%, but every dimension's internal-staff share falls below this baseline -- confirming that the reported friction signals are genuinely user-facing. We then assess three real datasets across three providers and show that FAIR compliance and DFF friction can disagree in both directions: a 92% FAIR-compliant dataset can still carry substantial friction, and a 42% FAIR score can be an artefact of anti-scraping policy rather than poor stewardship. The two measures are non-redundant and jointly informative: FAIR compliance does not predict DFF friction in either direction. This constitutes the first large-scale empirical application of the framework; cross-institutional validation is identified as the immediate next step.

0

astro-ph.SR 2026-06-23

Sound speed gradient damps acoustic waves in solar tachocline

by D. Tsiklauri

Phase-mixing of acoustic waves with applications to solar tachocline

Phase-mixing supplies the bulk dissipation missing from classical models of low-ℓ p-modes above 3000 μHz.

abstract click to expand

We adapt the \textit{magnetohydrodynamic} wave phase-mixing paradigm [Tsiklauri et al. (2003)] to investigate \textit{acoustic} wave propagation and damping in media where transverse sound speed gradients exist. Using an analytical model, we recover previous harmonic wave and Gaussian pulse evolution solutions now controlled by the spatial gradient of the local \textit{sound speed}. We discover a scaling law governing a Harris current sheet-like pulse evolution: under developed-stage phase-mixing, the peak envelope of such pulse amplitude scales with propagation distance as a new power-law $\max(P_1) \propto x^{-9/2}$. Applying our model to the solar tachocline directly resolves the 26-year-old helioseismic mystery of low-$\ell$ global $p$-mode linewidth anomalies observed by BiSON above $\nu \approx 3000\,\mu\text{Hz}$. We demonstrate that the sound speed gradient forces rapid, non-turbulent energy damping directly in the shear zone, providing the exact high-efficiency bulk dissipation needed to account for the missing energy sink deep within the solar tacholine. Our model provides the exact damping rates that classical, homogeneous models severely underestimated in the past. Finally, our results provide actionable design strategies for engineering compact stealth coatings or meshes to achieve enhanced acoustic signature suppression from moving bodies immersed in a fluid.

0

physics.ao-ph 2026-06-22

Extreme cyclones rise only in the North Atlantic

by Ivo Welch

Basin-Specific Intensification of Tropical Cyclones

The full chain from aerosol changes to temperature contrasts to more category 3+ storms holds solely in that basin, so global IPCC claims re

abstract click to expand

This paper uses ADT v9.0 HURSAT v07b data to measure tropical cyclone (TC) activity in four basins from 1990-2024. It is only in the North Atlantic basin that relative sea-surface temperature contrasts and the number of extreme (category 3+) tropical cyclones increased. And it is only there that the complete chain held -- from annual aerosol pollution to sea-surface temperature contrasts, and in turn from those contrasts to the number of cyclones. Conversely, associations are inconsistent, insignificant, or counterintuitive in the three Pacific Ocean basins, especially insofar as annual temperature contrasts did not associate with annual cyclone activity. (Where appropriate, our analysis controls for aerosol changes and the El Nino-Southern Oscillation.) Worldwide, it is no longer justified for the IPCC to hold that "It is likely that the global proportion of Category 3-5 tropical cyclone instances ... ha[s] increased globally over the past 40 years."

0

physics.ao-ph 2026-06-22

Tropical models turn chaotic with sub-day Lyapunov times

by Stéphane Vannitsem, Jonathan Demaeyer

Emergence of Chaos in the Tropical Atmosphere: Study of the Weak Temperature Gradient System

Reduced-order vorticity equations under realistic forcing show chaos emerging, contradicting assumptions of high tropical predictability.

abstract click to expand

The atmospheric tropical belt is believed to be more predictable than the extratropics. This question is revisited here by exploring the emergence of chaos in reduced-order model versions of the vorticity equation under the weak temperature gradient hypothesis, which provides a good description of the large-scale tropical atmosphere. The analysis reveals that under fairly realistic divergence forcing amplitudes, chaos may emerge, sometimes with Lyapunov time scales of less than a day. This result contrasts with the idea of a predictable tropical atmosphere, and opens important questions on the effective origin of predictability in the Tropics.

0

physics.ao-ph 2026-06-22

Temperature gradient and friction control superrotation transition

by Corentin Herbert

Factors governing the existence of an abrupt transition to superrotation in an idealized GCM

In an idealized model, these two quantities decide whether the shift to equatorial westerlies is abrupt or gradual.

abstract click to expand

Some numerical simulations of very warm climates suggest that the Earth's atmosphere may undergo a transition to a state of equatorial superrotation, where the zonal-mean zonal wind in the tropics is westerly. However, major uncertainties remain about the circumstances under which such a transition could happen. A natural first step towards reducing these uncertainties is to better understand the dynamical processes involved in the transition in idealized setups. However, simple numerical experiments have reported very different responses to tropical diabatic heating in different models, with both a continuous and an abrupt transition to superotation. In this paper, we investigate the mechanisms controlling the nature of the transition. We show that in an idealized Held-Suarez framework, it is governed by both the meridional temperature gradient and the bottom friction coefficient. These two parameters control a competition between two feedback mechanisms: a positive tropical wave-jet mechanism, and a negative feedback mechanism related to absorption of extratropical waves near their critical latitudes in the tropics.

0

physics.ao-ph 2026-06-22

Asymmetric storm tracks shift blocking upstream under polar warming

by Michele Filippucci, Stephen Thomson +2 more

Zonal asymmetries control the response of atmospheric blocking to Arctic warming in an aquaplanet experiment

Reduced carrying capacity crosses the blocking threshold only when zonal asymmetries exist, displacing the frequency maximum upstream.

abstract click to expand

In recent years a weak but robust response of mean midlatitude circulation to Arctic amplification (AA) has emerged from modeling experiments. However, open questions remain about the mechanisms linking such circulation differences to weather extremes in the midlatitudes. In this study we investigate such mechanisms and the importance of zonal asymmetries in shaping the atmospheric blocking response to AA. We perform idealized aquaplanet simulations in two configurations: a zonally symmetric setup and a zonally asymmetric experiment featuring a localized midlatitude storm track. For each configuration, we examine the response to AA by imposing an anomalous surface heating in the polar region. In the zonally symmetric configuration atmospheric blocking increases uniformly with AA from mid to high latitudes. In the asymmetric configuration, the response is more complex; instead of a zonally uniform response, we observe an upstream displacement of the blocking maximum, which sits at the exit of the localized storm track. We interpret these changes through the lens of the Traffic Jam theory by diagnosing the carrying capacity of the midlatitude flow. In both configurations, the zonally averaged increase in blocking is primarily driven by a weakening of the zonal winds, which reduces the Doppler-shifted Rossby wave group velocity and, in turn, decreases the flow carrying capacity. While the reduction in carrying capacity has similar characteristics in the two configurations, in the asymmetric case it leads to an upstream shift of blocking frequency as a direct consequence of the threshold behavior of blocking onset that lies at the core of the Traffic Jam theory. This mechanism, which has received limited attention so far, highlights the importance of mean circulation characteristics in shaping the blocking response to external forcing such as Arctic warming.

0

physics.ao-ph 2026-06-22

Network retrieves cloud variables from six solar satellite channels

by Stefano Franzoni, Christopher Bülte +3 more

Using Distributional Regression Networks to Retrieve Cloud Properties from Solar Satellite Channels for Data Assimilation

The backward operator yields unbiased, well-calibrated estimates with realistic covariances, potentially easing assimilation into weather mo

abstract click to expand

Satellite observations in the solar spectrum (including visible and near-infrared channels) offer high-resolution information on clouds and atmospheric properties valuable for data assimilation. While forward operators for a direct assimilation of solar images have become available recently and a first visible channel is already used operationally, their assimilation remains challenging due to strong non-linearities, ambiguities and high inter-channel correlations. This study addresses two central questions: what is the potential impact of assimilating multiple solar channels jointly, and can observed reflectances be transformed into physically meaningful, uncertainty-quantified variables better suited to assimilation than the raw reflectances themselves? As a proof of concept, we assess the joint information content of six solar channels from the Flexible Combined Imager (FCI) onboard Meteosat Third Generation and introduce a novel "Backward Operator" (BO) for probabilistic retrievals of cloud-related variables. The BO is implemented in a machine learning approach as a distributional regression network that is trained on synthetic images from a NWP regional model run and produces multivariate Gaussian estimates of total optical thickness, column cloud fraction, ice fraction, and effective radii of water and ice. The BO predictions are unbiased and well-calibrated, with realistic, situation-dependent and non-trivial covariance structures. The retrieved variables can be overall usefully constrained. Despite strong inter-channel correlations, combining multiple channels yields substantial performance improvements. As the BO does not require prior information, is consistent with an existing forward operator, and yields cloud variables more linearly related to the NWP model state, assimilating these variables could be a viable alternative to direct reflectance assimilation.

0

physics.ao-ph 2026-06-22

ENUFFT compacts orographic spectra by 25-60% on irregular grids

by Tridib Banerjee, Felix Jochum +1 more

Elastic Non-Uniform FFT (ENUFFT) spectral reconstruction of irregularly sampled orography on unstructured grids

Direct Fourier computation without interpolation keeps energy error at 14-24% and allows flow-dependent mode selection for mountain-wave mod

abstract click to expand

Subgrid-scale orography remains a leading source of uncertainty in numerical modeling because terrain spectra must be recovered from irregularly sampled elevation data and then reduced to a flow-dependent launch budget for parameterizations. Existing approaches are limited either by assuming regular samples on rectangular grids or by fitting coefficients whose truncation and regularization effects become embedded in the spectrum. None achieves dynamic, flow-dependent truncation. This study introduces an Elastic Non-Uniform Fast Fourier Transform (ENUFFT) framework that computes local Fourier coefficients directly from irregularly sampled orography on unstructured grids, without interpolation or fitting. It combines a type-1 NUFFT with local windowing, quadrature weights, and a new Elastic Mode Selection (EMS) algorithm for retaining a local flow-dependent subset of modes. ENUFFT is compared with the strongest relevant existing method in a monochromatic and a real Alpine terrain test. In both cases, it recovers peak amplitude and direction comparably while significantly compacting the spectra (monochromatic ~25%, Alpine ~60%). It also satisfies the Parseval condition more closely with its spectral variance (energy) deviating from reference by ~14-24% versus ~500-122,000% for the existing method. Its EMS is additionally tested in a mountain-wave test where it reduces the launch spectrum by >=75% while keeping launch-power loss <=7%. Along with better compute scaling, ENUFFT is thus a computationally efficient, physically interpretable framework that can make Fourier-based orographic source descriptions practical for spectral-budget-aware parameterizations.

0

physics.optics 2026-06-22

Boundary-conformal scheme restores fourth-order convergence for faceted scatterers

by Zihua Wu, Yu Xiong

Boundary-conformal integration for the invariant-imbedding T-matrix method: high-order convergence for faceted particles

Closed-form azimuthal terms, exact panel splits at tangency loci, and square-root substitution remove staircasing in the invariant-imbedding

abstract click to expand

The invariant-imbedding T-matrix method (IITM) is a standard tool for light scattering by large, sharply faceted, non-axisymmetric particles (atmospheric ice crystals and mineral dust) where the surface-based extended boundary condition method loses accuracy. Its accuracy is limited by "staircasing": the dielectric contrast of a faceted particle is integrated across boundaries that cut the quadrature grid, so standard quadrature converges at low algebraic order. We show that this non-smoothness has a single geometric origin, the tangencies of the integration sphere to the faces and edges of the particle, which produce jumps, kinks, and half-integer branches according to the tangency type, in all three integration directions. A boundary-conformal scheme removes them using closed-form azimuthal coefficients, panel splitting at the analytically known tangency loci, and a square-root substitution $x \mapsto x_c + t^2$ that absorbs the half-integer branches. For a hexagonal prism the azimuthal integration becomes exact and the zenithal and radial directions recover spectral and fourth-order convergence; because the construction depends only on the contact geometry, it extends to any convex polyhedron, demonstrated on the solid hexagonal bullet (a faceted ice habit with tilted faces). The zenithal crossing is a square-root branch rather than a kink, so the established interval-splitting alone gives only $\mathcal{O}(N^{-3})$, while the radial step removes the half-integer edge branch that caps the Riccati recurrence on faceted particles. The convergence orders are fixed by the local contact geometry and verified size-independent up to $k\,r_{\max} = 20$; what grows with size is the resolution needed to reach each asymptotic regime, not the order.

0

physics.flu-dyn 2026-06-19

Theoretical and data-driven ocean eddy closures are linked

by Laure Zanna, Pavel Perezhogin

Towards bridging the gap between data-driven and theoretical turbulence closures in stratified flows

Analytical and data-driven methods connect models for 10-100 km eddies that control momentum and energy redistribution in stratified flows.

abstract click to expand

Turbulence closure models are essential for solving the equations of motion in realistic systems, where fully resolving all relevant scales of motion is computationally infeasible. Developing turbulence closures remains one of the most challenging problems in fluid dynamics. Specifically, the Navier-Stokes equations, when filtered to isolate large-scale motions, introduce new terms representing the influence of subgrid-scale turbulent stresses. These terms, which can only be computed directly by resolving the turbulence itself, therefore lead to the closure problem: we must add new equations or introduce assumptions to relate the unresolved scales of motions to the resolved flow. Here we consider the closure problem for oceanic flows, i.e., stratified, Boussinesq, incompressible, in a rotating frame of reference. In particular, we focus on a closure for ocean mesoscale eddies, which have horizontal scales of 10-100km and are key to the redistribution of momentum, energy, and tracers in the ocean. In particular, mesoscale eddies can reinject energy and momentum into the large-scale flow through an inverse energy cascade. Here, we explore a range of theoretical and data-driven ocean mesoscale closures and examine their connections using analytical and data-driven methods. This note aims to bridge the gap between novel methods from artificial intelligence (AI) and machine learning and theoretical fluid dynamics to address significant challenges in the physics of turbulence.

0

physics.ao-ph 2026-06-19

Machine learning reconstructs microphysical rates from standard outputs

by Miriam Simm, Tom Beucler +1 more

PRecover 1.0: Process Rate Recovery with Machine Learning

PRecover recovers most rates accumulated over 10 minutes or less using cloud variables, with uncertainty estimates and cross-domain transfer

abstract click to expand

Comprehensive information on cloud microphysical process rates from numerical simulations allows for better understanding of precipitation formation pathways and aerosol-cloud interactions. However, resource limitations often make it impractical to include all microphysical process rates in the model output, limiting in-depth analyses. To address this shortcoming, we introduce PRecover, a data-driven post-processing approach to recover microphysical process rates that are not stored during runtime from standard output of a numerical weather prediction model. In particular, we train random forests, gradient boosting models, and feed-forward neural networks to recover microphysical process rates from a two-moment bulk microphysics scheme in the ICOsahedral Nonhydrostatic (ICON) model. We use cloud variables as input, obtained from high-resolution simulations in a limited-area setup over Europe. Warm-rain and ice microphysical process rates are recovered with a two-step classification-regression approach for both instantaneous and accumulated process rates. As a physics-based baseline, we assess whether process rates can be directly recalculated from stored ICON output variables. Accurate recalculation is possible for process rates such as accretion and self-collection but not for the autoconversion, rain melting or heterogeneous ice nucleation rate. Using PRecover, we successfully recover most of the process rates that are accumulated over output time steps of 10 minutes or less, but the values are increasingly difficult to recover for rates accumulated over longer accumulation intervals. To quantify predictive uncertainty, we provide calibrated prediction intervals through conformalized quantile regression. We demonstrate spatial transferability of the models with two case studies over different regional domains and simulation settings unseen during training.

0

physics.ao-ph 2026-06-19

Stratification intensification makes Gulf Stream meander chaotically

by Lennard Miller, Antoine Venaille +2 more

Enhanced Gulf Stream Path Variability Under Intensified Stratification

Eddy-resolving models show steady eastward path replaced by vigorous meanders independent of AMOC and wind changes

abstract click to expand

Increased upper-ocean stratification is an unavoidable consequence of global warming and will strongly impact the structure of ocean currents. Using a high-resolution ocean model, we show that intensification of stratification leads to the loss of coherence of the Gulf Stream Extension, replacing its steady eastward path with vigorous, chaotic meanders. This regime shift persists independently of changes in the Atlantic Meridional Overturning Circulation and surface wind forcing. Enhanced meandering under intensified stratification also proves to be a robust feature across both idealized and realistic ocean models that resolve mesoscale eddies, but is not captured by coarse-resolution models that parameterize eddies. The presented findings therefore highlight the need for improved representations of oceanic turbulence in climate projections.

0

eess.IV 2026-06-19

Bidirectional model links profile retrieval to radiance simulation

by Jingdong Shen, Fu Wang* +5 more

SIMBA: ABidirectional Retrieval Forward Simulation Framework for Modeling FY-4A GIIRS Hyperspectral Infrared Radiances Toward NWP Applications

Cycle consistency and Mamba module let SIMBA beat one-way baselines on FY-4A GIIRS temperature, humidity, and radiance tasks.

abstract click to expand

Hyperspectral infrared observations are an important data source for numerical weather prediction (NWP) because they provide rich information on the vertical structure of atmospheric temperature and humidity. However, most existing deep learning methods mainly focus on one-way retrieval from radiances to atmospheric profiles, while the reverse radiance simulation process and the consistency between atmospheric state space and radiance observation space are insufficiently considered. In this study, we propose SIMBA, a unified bidirectional retrieval-forward simulation framework for FY-4A GIIRS hyperspectral infrared radiance modeling toward NWP applications. The framework jointly performs atmospheric profile retrieval and radiance reconstruction, introduces a cycle-consistency constraint to strengthen the coupling between the two processes, and employs a bidirectional Mamba state-space module to capture long-range dependencies along pressure levels. Using collocated FY-4A GIIRS observations and ERA5 reanalysis data, the proposed method is evaluated for temperature retrieval, specific humidity retrieval, long-wave radiance reconstruction, and medium-wave radiance reconstruction. Experimental results show that SIMBA outperforms several representative deep learning baselines across both retrieval and reconstruction tasks, while ablation experiments confirm the contribution of the bidirectional design and cycle-consistency mechanism. These results demonstrate that the proposed framework is effective for joint atmospheric profile retrieval and hyperspectral infrared radiance modeling, and suggest potential for future Jacobian-related analysis and NWP-oriented extensions.

0

physics.ao-ph 2026-06-19

Entropy production peaks let ENSO variance peak in boreal winter

by Yuki Yasuda, Tsubasa Kohyama

A Stochastic-Thermodynamic Constraint on the Seasonal Phase Locking of the El Ni\~no-Southern Oscillation

The TUR bound on variance tendency relaxes when irreversibility maxima occur in autumn and late winter.

abstract click to expand

We investigate the seasonal phase locking of the El Ni\~no-Southern Oscillation (ENSO) in a linear stochastic recharge oscillator (SRO), a damped oscillator with additive noise and a time-dependent growth rate. Phase locking is reflected in the seasonality of the variance of the sea surface temperature anomaly (SSTA). In general, energy drives such a change, whereas entropy governs whether it occurs; phase locking is thus subject to both an energy- and an entropy-based constraint. We quantify this entropy-based constraint using a thermodynamic uncertainty relation (TUR), a fundamental inequality in stochastic thermodynamics. The TUR constrains the tendency of the SSTA variance by the partial entropy production rate, which is dominated by the ratio of forward and backward transition probabilities and quantifies the irreversibility of SSTA transitions. The growth rate governs this irreversibility: its extrema occur in boreal autumn and late winter, and the entropy production rate peaks at both times. These peaks relax the TUR constraint on the tendency of the SSTA variance, so that the variance itself can peak in boreal winter, consistent with observed ENSO phase locking. Conversely, when irreversibility is insufficient, ENSO cannot grow or decay. If this irreversibility were interpreted as dissipated energy, the constraint on ENSO growth and decay would require this dissipation to be exported from the equatorial Pacific. A more realistic model is needed to test this hypothesis and to further explore the physical connection between entropy and dissipated energy.

0

physics.ao-ph 2026-06-18

Conformal prediction guarantees coverage for AI weather forecasts

by Anna Asch, Raphael Rossellini +2 more

Rigorous uncertainty quantification of probabilistic AI weather forecasts with conformal prediction

Applied to leading models, it enforces statistical reliability on temperature and precipitation predictions without harming other metrics.

abstract click to expand

Probabilistic weather forecasting is undergoing rapid transformation with artificial intelligence (AI). In traditional numerical weather prediction, computing power can limit how well ensemble forecasts approximate the unknown statistical distribution of future states. AI models facilitate larger ensembles and are trained with probabilistic considerations, ideally leading to better uncertainty quantification. Forecasts from these state-of-the-art models are often considered well-calibrated. However, here we show that the statistical coverage of such models, the ultimate measure of calibration, can struggle, especially on extreme events. To address this shortcoming, we employ conformal prediction, a class of statistical methods that mathematically guarantees coverage under no distributional assumptions, unlike previous post-processing techniques. We apply online conformal prediction to temperature and precipitation forecasts (including extremes) of three leading global weather models, GenCast, NeuralGCM, and AIFS-ENS, ensuring calibrated uncertainty at no expense to other probabilistic metrics. This post-processing method can be applied to any forecasting model.

0

physics.ao-ph 2026-06-18

Models underestimate land-ocean warming contrast behind trends

by Benjamin O. Johnson, Maria Rugenstein

A Land-Sea Contrast Pattern in Surface Temperature and Atmospheric Circulation Trends in Recent Decades

Observed faster land warming produces pressure and Pacific patterns that historical simulations miss except in early CO2 quadrupling runs.

abstract click to expand

Spatial patterns in observed climate trends remain poorly understood. Here we argue that a warming of land relative to ocean has shaped observed surface temperature and atmospheric circulation trends, including the negative Inter-Decadal Pacific Oscillation (IPO)-like tendency across the Pacific basin. Observed and modeled trends display an overall decline in sea level pressure over the faster-warming land relative to ocean, with a spatial pattern that resembles the seasonal cycle and the response to land heating in idealized climate model experiments. Coupled climate model simulations with historical forcing underestimate the land-sea warming ratio. It is only in the early response of abrupt CO2 quadrupling climate model simulations that climate models are able to recreate the observed land-sea warming ratio, in which case a strengthening of oceanic surface highs and a negative IPO-like surface warming pattern over the Pacific comparable to observed trends are seen. We propose that discrepancies between modeled and observed trends in many climate variables may be explained by the underestimation of the land-sea warming ratio by climate models. Determining the cause of this discrepancy has the potential to constrain projections of future climate change as the underlying mechanism causing climate models to underestimate the land-sea warming ratio discrepancy will set the persistence of this problem.

0

physics.ao-ph 2026-06-18

One optimized scenario beats six standard ones for climate emulators

by Christopher B. Womack, Shahine Bouabid +5 more

Optimal scenario design for climate emulation

Gradient updates from a simple model reshape training data to improve skill on unseen pathways with less total computation.

abstract click to expand

As deep learning for physical systems continues to grow in popularity, efforts to improve generalizability have primarily focused on designing architectures that embed physical constraints. However, for machine-learning surrogate climate models (emulators), we show that the low structural diversity in existing scenarios commonly used to generate training data places a ceiling on predictive skill. Here, we examine whether training datasets themselves can be optimized to improve generalization. We introduce a method to create datasets that produce emulators capable of generalizing to new, structurally different scenarios absent from the training data. We use a differentiable Simple Climate Model (SCM) to calculate the sensitivity of emulator loss to perturbations in the training data, iteratively updating the training data to maximize emulator skill. For an SCM, training on one scenario optimized in this fashion outperforms an emulator trained on six standard ScenarioMIP pathways. We achieve this higher predictive skill despite training on a smaller dataset, finding that our emulator successfully isolates distinct physical behaviors of different climate forcing agents (e.g., greenhouse gases vs. aerosols) without single-forcing runs. We then demonstrate that scenarios optimized using an SCM, when used to drive an intermediate-complexity climate model, produce a training dataset that yields a more skillful emulator than training on ScenarioMIP outputs. Our results suggest that, in the compute-constrained environment of running full-scale climate models, generating a small number of dynamically rich scenarios provides greater marginal value for emulation and characterizing system responses than expanding the suite of traditional emissions pathways.

0

physics.ao-ph 2026-06-18

ML model matches IFS on medium-range forecasts from observations alone

by Ewan Pinnington, Peter Lean +9 more

AIFS-DOP: End-to-End Medium-Range Weather Prediction from Observations Alone with Machine Learning

Trained only on 40 years of gridded data with no NWP input, it reaches competitive scores when checked against real observations.

abstract click to expand

We introduce the Artificial Intelligence Forecasting System for Direct Observation Prediction (AIFS-DOP). AIFS-DOP is trained on a 40-year harmonized dataset of gridded observations, without using numerical weather prediction (NWP) reanalysis or model data. The resulting model is competitive with ECMWF's Integrated Forecasting System (IFS) when scored on a one year period of forecasts across 2021/2022. This progress on Direct Observation Prediction represents the first time that a data-driven model, trained solely on observations, is competitive with the IFS at medium ranges for several key upper-air and surface headline scores, when verified against observation data.

0

cs.LG 2026-06-18

Hybrid LSTM-ViT doubles skill predicting precipitation forecast errors

by David Aaron Evans, Jay C. Rothenberger +3 more

A Hybrid LSTM--Vision Transformer Architecture for Predicting HRRR Forecast Errors

Vertical profiles from profilers raise accuracy over baseline LSTM on HRRR errors at short leads and during PBL activity.

abstract click to expand

Forecast errors in high-resolution numerical weather prediction (NWP) systems are often linked to unresolved planetary boundary layer (PBL) processes, convection, terrain-induced circulations, and other vertically structured atmospheric phenomena. Previous work demonstrated that Long Short-Term Memory (LSTM) networks can successfully predict forecast errors in the High-Resolution Rapid Refresh (HRRR) model using mesonet observations, but we believe performance degradation is linked to periods of complex vertical atmospheric evolution. To address this limitation, we develop a hybrid LSTM-Vision Transformer (LSTM-ViT) framework that combines temporal sequence learning from surface observations with atmospheric profiles from the New York State Mesonet profiler network. The LSTM-ViT framework is trained to predict HRRR hourly precipitation, 10 m wind speed, and 2 m temperature forecast errors at individual mesonet stations. Across all three predictors, incorporation of profiler-derived atmospheric structure improves forecast error prediction skill relative to the baseline LSTM architecture, with the largest gains occurring at shorter forecast lead times and during periods of enhanced PBL activity. Improvements are particularly pronounced for precipitation forecast error, where the LSTM-ViT framework achieves approximately a twofold increase in predictive skill relative to the baseline LSTM while better capturing convectively driven error evolution and reducing degradation associated with PBL processes. These results demonstrate that combining temporal sequence learning with vertically informed attention mechanisms provides a physically meaningful pathway for improving forecast error prediction in operational NWP systems. Our research offers forecasters enhanced guidance regarding model bias and forecast confidence.

0

physics.ao-ph 2026-06-18

Atlantic gradient index shows wider nonlinear scaling than regional SST

by Sebastián Jaroszewicz, Nahuel Mendez +2 more

Multifractal Dynamics of Tropical Atlantic SST Indices: Nonlinear Scaling Structure and Episodic Statistical Association with ENSO Variability

TASI spectrum width reaches 0.72 versus 0.27-0.34 for others; shrinks during major El Nino events with 15-18 month lagged ENSO link but no d

abstract click to expand

The Tropical Atlantic exhibits complex sea surface temperature (SST) variability driven by internal ocean-atmosphere interactions and remote climate forcing. We perform a comparative multifractal analysis of three SST indices, South Atlantic Tropical (SAT), Tropical Southern Atlantic (TSA), and the Tropical Atlantic SST Gradient Index (TASI), using weekly data from 1981 to 2025. Multifractal Detrended Fluctuation Analysis (MFDFA) reveals robust scale-dependent behavior in all indices. TASI displays a substantially broader multifractal spectrum (Delta h about 0.72) than SAT (0.27) and TSA (0.34). Surrogate-data tests show that multifractality in SAT and TSA is mainly explained by linear autocorrelations, whereas TASI contains an additional nonlinear contribution associated with phase correlations. To investigate temporal variability, we introduce a moving-window MFDFA framework that tracks the evolution of multifractal width. Significant reductions are observed during the major 1997-1998 and 2015-2016 El Nino events, indicating a suppression of multiscale variability under extreme Pacific forcing. Lagged correlation analysis reveals a significant negative association with the Oceanic Nino Index at delays of 15-18 months, consistent with known Atlantic-Pacific teleconnections. However, Granger causality and Transfer Entropy tests do not detect significant causal links, suggesting an episodic rather than persistent relationship. Lagged multifractal cross-correlation analysis further reveals scale-dependent inter-basin coupling. These results demonstrate that time-dependent multifractal measures provide a useful framework for characterizing nonlinear Atlantic variability and identify TASI as a dynamically distinct index whose scaling properties contain information not captured by regional SST indices alone.

0

cs.LG 2026-06-18

3D vertical coupling improves ML emulation of stratospheric warmings

by Oskar Bohn Lassen, Simon Driscoll +3 more

Investigating Inductive Biases for Machine Learning Emulation of Sudden Stratospheric Warmings in Idealised Isca Simulations

In paired Isca runs, architectures with explicit layer connections cut errors during active events, though low error does not guarantee corr

abstract click to expand

Machine-learning emulators are increasingly used for weather prediction and have the potential to extend skill on subseasonal-to-seasonal timescales by learning dynamically important sources of predictability. A key challenge is whether the models can exploit predictability anchors, such as stratospheric variability, that influence tropospheric circulation beyond short lead times. We test how architectural inductive bias affects emulation of sudden stratospheric warming (SSW) dynamics using paired idealised Isca simulations that differ only in an imposed wave-2 heating perturbation. Across convolutional, transformer, and graph-based architectures trained for one-step prediction, model differences are modest when the stratosphere is dynamically quiet but widen substantially when SSW-like variability is active. Our results identify explicit three-dimensional vertical coupling as a key inductive bias for machine-learning emulation of stratospheric dynamics. However, Eliassen-Palm flux diagnostics show that low forecast error does not guarantee physically faithful wave-mean-flow interaction, with coherent errors remaining in stratospheric wave-driving structure.

0

physics.ao-ph 2026-06-12

Microphysics schemes differ in storm details but match poorly to observations

by Abraham I. Roseman (1), Falko Judt (2) +12 more

Comparison of Two Operational Microphysics Schemes Across Various Regional-MPAS Simulations

TEMPO spreads precipitation widely while NSSL concentrates it, yet both underperform in mesoscale organization and stratiform rain across MP

abstract click to expand

Accurately representing convection and precipitation remains a persistent challenge for Numerical Weather Prediction (NWP) models due to biases in convective initiation, storm organization, and rainfall distribution, particularly in subtropical/tropical environments. This study evaluated how microphysics parameterizations influence convective organization and precipitation using hindcasts with the Model for Prediction Across Scales - Atmosphere (MPAS-A) on a variable-resolution mesh down to 1-km resolution. Two operational microphysics schemes, National Severe Storm Labs (NSSL) microphysics and Thompson-Eidhammer Microphysics Parameterization for Operations (TEMPO), were examined across three subtropical/tropical regions during boreal summer under strongly- and weakly-forced regimes. Both schemes captured the general timing and placement of convection, but differed in storm structure and rainfall distribution. TEMPO produced more numerous, weaker convective cores with earlier, more widespread precipitation and cooler surface conditions, while NSSL favored fewer, stronger cores and updrafts with more cloud water, ice, and graupel hydrometeors, though less snow, and more spatially concentrated, intense rainfall. Despite these structural differences, both schemes diverged more from observations than from each other, producing scattered convective cells with minimal mesoscale organization and insufficient stratiform precipitation. The simulations also exhibited regime-dependent errors, with rainfall under- (over)-represented in strongly- (weakly)-forced regimes and forecast skill notably lower in the latter. Improving representation of localized precipitation processes remains essential for capturing convection across a wider range of scales and regimes. Future work should target microphysics evaluation across regimes and regions, with process-level improvements reducing convective biases.

0

physics.ao-ph 2026-06-11

Agentic system completes climate workflows without human input

by Dmitrii Pantiukhin, Boris Shapkin +3 more

CMIP-Forge: An Agentic System that Retrieves, Computes, and Self-Reviews Climate Science

Retrieval from thousands of papers plus code guardrails and independent review let the system run full analysis pipelines on live archives.

abstract click to expand

The Coupled Model Intercomparison Project Phase 6 (CMIP6) has generated thousands of peer-reviewed publications documenting model configurations, evaluation procedures, emergent constraints, and projection uncertainties. As the community transitions toward CMIP7, efficiently extracting and operationalizing this unstructured knowledge alongside live data analysis represents a critical bottleneck. Here we present CMIP-Forge, a hybrid retrieval-augmented generation (RAG) and autonomous analysis system that bridges the gap between scientific literature and Earth System Grid Federation (ESGF) data archives. The system pairs a curated corpus of 6,581 CMIP6-related open-access publications (101,828 indexed chunks) with an agentic pipeline in which a tool-augmented worker plans and executes Python workflows over live climate data, while a panel of independent reviewer models audits its methodology end to end. CMIP-Forge introduces a multi-layered Defense-in-Depth architecture that enforces physical and methodological invariants through executable mechanisms: Abstract Syntax Tree (AST) static analysis, audited scientific primitives, and an autonomous adversarial peer-review protocol. We demonstrate the system's capabilities through end-to-end autonomous research pipelines spanning atmospheric teleconnections, ocean dynamics, regional extremes, and global warming projections. An agentic analysis system grounded in peer-reviewed literature, constrained by automated code guardrails, and audited by an independent adversarial review loop can complete complex climate-research workflows autonomously. The same experiments expose concrete failure modes of the review loop (sycophantic regression, REVISE verdicts that are never resolved, and the submission of stub code for review), each diagnosable from the immutable telemetry and provenance record released with the article.

0

physics.ao-ph 2026-06-11

Alpine campaign instruments 30 sites to track multi-scale air transport

by Manuela Lehner, Claudia Acquistapace +45 more

The TEAMx Observational Campaign

Year-long effort with aircraft and UAS targets gravity waves, convection, and turbulence across four areas.

abstract click to expand

As part of the international research programme TEAMx (multi-scale transport and exchange processes in the atmosphere over mountains - programme and experiment) a one-year long measurement campaign, the TEAMx Observational Campaign (TOC), was conducted between 2024 and 2025 in a north-south transect through the Alps. Building on the dense operational measurement network in the Alps, the TOC was designed to collect long-term atmospheric observations over the highly complex Alpine terrain. During two six-week long Extended Observational Periods, more than 40 research institutions came together to instrument about 30 sites in the four target areas of the TEAMx domain and study different transport processes, from gravity waves to orographic convection, thermally driven flows, and turbulent exchange. In addition to a suite of ground-based in-situ and remote-sensing instruments, observational activities included airborne measurements with up to three research aircraft and multiple UAS. This paper gives an overview of the science goals and the TOC design, together with preliminary analyses that highlight the potential of the collected dataset.

0

cs.LG 2026-06-11

U-Net turns coarse scenario data into high-resolution land cover maps

by Amirpasha Mozaffari, Marina Castaño +8 more

Scalable Deep Learning Framework for Global High-Resolution Land Use Reconstruction

Framework extends annual global reconstructions to periods without observations using static geophysical features.

abstract click to expand

Uncertainty in the terrestrial carbon cycle remains a major constraint in climate projections, partly driven by the uncertainties affecting the land surface representation and variability in Earth system models. To address this limitation, we present a data-driven framework AI4Land, for generating high-resolution historical reconstructions and future projections of key land surface variables. The framework follows a two-phase approach using a U-Net architecture. In the first phase, which is the focus of this work, it reconstructs annual land use and land cover by integrating coarse-resolution scenario data with static geophysical features. In a planned second phase, the resulting high-resolution maps will be used to predict dynamic biophysical variables, particularly leaf area index, at finer temporal scales. Trained on Earth observation data, the models learn to reproduce spatially explicit and physically consistent land surface patterns, extending temporal coverage to periods lacking direct observations. AI4Land was developed and trained on MareNostrum5, demonstrating how GPU-accelerated HPC infrastructure enables global-scale climate AI pipelines. The final product is a suite of open-source emulators designed for real-time coupling with digital twin platforms, such as those developed under the Destination Earth initiative. By delivering realistic and evolving land surface conditions on demand, this work aims to reduce critical uncertainties and improve the predictive power of next-generation climate simulations.

0

physics.ao-ph 2026-06-11

Harmonized cubes align satellite data for 48 cities into ML-ready form

by Jonathan Starfeldt, Maria J. Molina +4 more

Urban Heat MiniCubes: An AI-Ready dataset for urban heat research

Urban Heat MiniCubes collocates Landsat, Sentinel, GOES and microwave observations on one grid to cut preprocessing steps for urban heat stu

abstract click to expand

Urban heat is amplified by impermeable surfaces and heterogeneous built environments, yet street-level variability remains difficult to quantify because multi-sensor observations are rarely available in consistent, analysis-ready form at the necessary spatiotemporal scales. We present "Urban Heat MiniCubes," a publicly available, FAIR-oriented dataset designed for machine learning applications in urban heat research. The dataset provides harmonized 90 x 90 km gridded data cubes for 48 cities in the Western Hemisphere spanning 2022-2023, with variables reprojected and collocated to a common grid to reduce preprocessing (e.g., reprojection, resampling, and spatiotemporal alignment). Urban Heat MiniCubes includes two complementary modalities: (i) higher-spatial-resolution, lower-frequency observations from Landsat 8/9 (e.g., surface reflectances) and Sentinel-1 (e.g., synthetic aperture radar backscatter), and (ii) higher-temporal-frequency, coarser observations from GOES-R (e.g., longwave infrared brightness temperatures) and a microwave land surface temperature product. We document variables and metadata and provide technical assessment using inter-variable analyses and autoencoder-based reconstruction-error summaries across pixel classes (e.g., water and cloud). Potential use cases and limitations are also discussed.

0

physics.ao-ph 2026-06-10

LLM ports ocean model from Fortran to GPU in weeks

by Nikolay V. Koldunov, Suvarchal K. Cheedela +4 more

An Ocean Model Ported by a Large Language Model: Experience and Lessons from FESOM2 (Fortran to C to C++/Kokkos)

Two-stage literal translation with validation keeps five-year statistics and delivers 1.6-3.7x speedup on A100 nodes.

abstract click to expand

Large language models (LLMs) can translate and modify source code, and have been shown to do so for codes of different complexity. Whether they can port a complete, production geophysical model to a different language without degrading its physics has not been established. We demonstrate that LLM-assisted code translation can preserve the physics of a complete production ocean model while moving it into a modern performance-portable form. We report our experience using an agentic LLM coding assistant, directed by domain experts, to port the FESOM2 unstructured mesh ocean--sea-ice model (about 74000 lines of core Fortran) first to C and then to C++/Kokkos for performance portability across CPUs and GPUs. We describe the practices that proved necessary, what worked and what did not, and the failure modes that we encountered. Three practices mattered most: translating in two stages that separate reproducing the numerics (Fortran to a clean C reference) from introducing parallelism (C to Kokkos); requiring a strictly literal translation in which the assistant was not permitted to ``improve'' the source; and validating each stage against an acceptance criterion suited to it. The C port reproduces the original Fortran at the level of long-term simulation statistics over five years. The Kokkos port is bit-for-bit identical to the C reference on CPU and statistically close on GPU over multi-year runs. On eddy-rich meshes up to 7.4 million surface vertices a single A100 GPU node runs 1.6--3.7 times faster than a CPU node, reaching the 1-2 simulated-years-per-day required for production integrations. The result is more than a single GPU port: by following a clear validation procedure, an LLM moved a full Fortran ocean model into another language and onto accelerators while preserving its physics in a matter of weeks.

1 0

0

cs.LG 2026-06-10

New metrics test if ML weather models obey physics

by Emma Kasteleyn, Timo Maier +4 more

PhysMetrics.Weather: An Evaluation Framework for Physical Consistency in ML Weather Models

Conservation, spectral and dynamical checks quantify realism beyond pixel error scores for operational decisions.

abstract click to expand

Machine learning weather prediction (MLWP) models have achieved impressive forecasting performance at a small fraction of the computational costs required for traditional physics-based methods. However, they are primarily (1) data-driven and (2) evaluated using pixel-wide error metrics (e.g., RMSE), so there are no guarantees that their forecasts are consistent with known physical laws. We introduce PhysMetrics$.$Weather, an evaluation framework that assesses the physical realism of MLWP models across three types of metrics: conservation, spectral, and dynamical. By quantifying physical realism, this tool guides the development of physics-informed architectures and helps evaluate whether MLWP models are reliable for operational use. Our framework is available on Github at https://github.com/Emmakast/PhysMetrics.Weather.

0

physics.ao-ph 2026-06-09

Fossil fuels contribute over 3x more remote ozone than biomass burning

by Chaoqun Ma, Hang Su +1 more

Deep learning reveals a stronger fossil fuel influence than biomass burning in shaping remote tropospheric ozone

Deep learning resolves tracer-model mismatch and identifies the dominant source in remote troposphere.

abstract click to expand

Tropospheric ozone (O3) is a key greenhouse gas and atmospheric oxidant, yet its sources in the remote troposphere remain strongly debated. Observation-based tracer analyses suggest that O3 attributed to biomass burning is much greater than that from fossil fuel sources (by a factor of ~2-10), contradicting state-of-the-art global models. Here we show that this discrepancy primarily arises from the strong sensitivity of tracer methods to differences in tracer lifetimes, especially after extended transport to the remote regions. To resolve this discrepancy, we develop a deep learning (DL) framework that synthesizes global observations and chemical transport model simulations. The DL approach accurately infers source contributions and reveals that fossil fuel emissions contribute over three times more O3 to the remote troposphere than biomass burning. Our findings underscore that phasing out fossil fuels remains the most powerful lever for mitigating remote tropospheric ozone.

0

math.ST 2026-06-09

Kling-Gupta regression scales OLS coefficients by variance factor

by Hristos Tyralis, Georgia Papacharalampous

Kling-Gupta linear regression

Predictions match observed response variance while sharing mean and correlation with ordinary least squares results.

abstract click to expand

Although the Kling-Gupta efficiency ($\mathrm{KGE}$) is widely adopted for model evaluation in hydrology, its properties as a statistical estimator remain unexplored. Investigating these properties is necessary because parameter estimation and forecast evaluation are inherently linked. To address this, we formalize the negatively oriented Kling-Gupta loss $L_\mathrm{KG} = (1 - \mathrm{KGE})^2$ within an extremum estimation framework (equivalent to maximizing $\mathrm{KGE}$) and analyze its behavior in multiple linear regression. We establish explicit formulas for the parameter estimates, showing that Kling-Gupta linear regression scales the ordinary least squares (OLS) coefficient vector by a variance-inflation factor governed by the sample variances and covariances of the predictors and the response. We show that Kling-Gupta linear regression predictions replicate the sample variance of the response on the training set, in contrast to the variance reduction inherent to OLS, while both estimators maintain the sample mean of the observations and achieve the same sample correlation between the predictions and the response. We show analytically that no single estimator can simultaneously maximize both the Nash-Sutcliffe efficiency $\mathrm{NSE}$ and $\mathrm{KGE}$: the OLS estimator attains the maximum possible $\mathrm{NSE}$ but not the maximum $\mathrm{KGE}$, while the Kling-Gupta estimator maximizes $\mathrm{KGE}$ at the cost of $\mathrm{NSE}$. We prove the almost sure convergence of the Kling-Gupta estimator to well-defined population limits and express those limits algebraically. Furthermore, we evaluate the training and test set performance metrics for both estimators, demonstrating that for each estimator the metrics on the training set and on an independent test set converge asymptotically to identical limits (though the limits differ between OLS and Kling-Gupta regression).

0

math.NA 2026-06-09

19-mode Koopman backbone extracted from Pacific SST

by Paula Lorenzo-Sanchez, Matthew J. Colbrook +1 more

Residual Pseudospectra Reveal a Physics-Informed Koopman Backbone for Tropical Pacific Variability and ENSO Prediction

Residual minimization on ERA5 and HadISST records yields compact structure that improves 8-18 month ENSO forecasts.

abstract click to expand

Tropical Pacific sea-surface-temperature (SST) variability spans interacting timescales, with the ENSO as its dominant interannual expression. Yet the dynamical structure organizing this variability and underpinning extended-range predictability remains difficult to extract from high-dimensional observations. Koopman operator learning offers spectral coordinates for nonlinear dynamics, yet finite geophysical records often produce dense, sampling-sensitive spectra whose physical content is ambiguous. We show that this apparent redundancy reflects coherent operator-level structure. Combining kernel Extended Dynamic Mode Decomposition with residual minimization and pseudospectral analysis, we use the Koopman eigenvalue relation as a physics-informed consistency test to organize learned spectra. Applied to ERA5 and HadISST tropical Pacific SST anomalies, the residual landscape identifies 19 robust residual-minimum frequencies with coherent spatial modes that persist across products and sampling realizations. Together, these modes define a compact Koopman backbone spanning low-frequency modulation through quasi-biennial components, including ENSO-band variability. The surrounding spectral cloud is structured by integer powers and nonlinear combinations of this backbone, forming a residual-ordered Koopman hierarchy. The backbone reconstructs substantial Nino3.4 variance and enables skillful out-of-sample forecasts, with greatest gains at 8-18-month leads. By embedding dynamical consistency into physics-informed operator learning, the framework turns opaque spectra into robust, interpretable and predictive representations of tropical Pacific variability.

1 0

0

physics.ao-ph 2026-06-09

Low-cost sensors revise Swiss urban heat warnings after bias fixes

by Martí Bosch, Moritz Burger

Revisiting urban heat indices in Switzerland using low-cost measurement networks

Corrected city data show rural AWS miss warnings but some LCDs overstate them; raw LCDs still beat non-urban stations.

abstract click to expand

Urban populations are increasingly exposed to extreme heat events such as heatwaves, which can be exacerbated in cities due to the urban heat island (UHI) effect. With the aim of developing adaptation strategies, recent years have seen a growing interest in deploying high-resolution measurement networks using low-cost devices (LCDs), which enable the evaluation of intra-urban temperature distribution and its impacts at an unprecedented spatial resolution. However, the reliability of LCD measurements has been called into question, especially regarding potential overheating due to inadequate radiation shielding. In this study, we develop a statistical method to correct temperature biases based on short-wave radiation using a generalized additive model (GAM) and then apply it to LCD measurements in the urban climate networks of the cities of Bern, Lausanne, Neuchatel and Zurich (Switzerland). To that end, we first calibrate the correction procedure to the LCD models used in each city using an intercomparison field study, in which the LCD models are collocated next to a professional automated weather station (AWS) operated by MeteoSwiss in the rural surroundings of Bern. Then, we evaluate how these corrections can influence two climate indices, namely the number of tropical nights and the number of heat warnings issued in each city according to MeteoSwiss heat warning system. The findings suggest that the current AWS underestimate the heat warnings, whereas some LCD models likely overestimate them due to radiative errors. Nevertheless, uncorrected LCD measurements still provide a more reliable estimate of urban temperatures than AWS located outside urban settings. The insights can guide selection of LCD models for new monitoring networks and support the application of model-specific radiative bias corrections to existing LCDs, enabling more accurate assessments of heat and its impacts.

0

physics.ao-ph 2026-06-08

Climate networks flag AMOC edge state via equatorial links

by Laure Moinat, Reyk Börner +3 more

Climate network characterization of the AMOC edge state

Normalized degree centrality detects cross-equator teleconnections as the circulation nears the collapse boundary in model runs

abstract click to expand

The Atlantic Meridional Overturning Circulation (AMOC) has been identified as a tipping element in the Earth system. Under the current climate change scenarios, it is urgent to develop robust methods for determining the probability of future AMOC transitions. Recent studies using an Earth System Model of Intermediate Complexity (EMIC) have revealed the importance of an AMOC edge state, located on the boundary of the attraction basin of the collapsed state, in AMOC transitions. Here, we provide a characterization of this edge state through climate networks, using instantaneous temporal correlations between geographical locations to define the network links. We apply the climate network analysis to a set of EMIC simulations with CO$_2$ forcing according to an intermediate climate change scenario (SSP2-4.5) that exhibit qualitatively different AMOC responses as a result of interaction with the edge state. We show that network measures, specifically the normalized degree centrality, reveal the presence of teleconnections across the equator as the AMOC approaches the edge state. A similar result is obtained for an Earth System Model (ESM) simulating AMOC collapse or recovery, suggesting that climate networks could be used to detect the onset of an AMOC tipping event in ESMs.

0

cs.LG 2026-06-08

Dual decoding improves extreme detection in 3D hydrometeor forecasts

by Dandan Chen, Yaqiang Wang

Physics-Guided Dual Decoding and Spectral Supervision for Global 3D Hydrometeor Prediction

PredHydro-Net uses unidirectional modulation and wavelet spectral matching to outperform GFS and deep learning baselines on 72-hour global r

abstract click to expand

While global data-driven models excel at predicting continuous atmospheric variables, three-dimensional hydrometeor forecasting remains challenging due to the zero-inflated, long-tailed distributions of these variables. Standard deep learning optimization often yields overly smooth forecasts, attenuating extreme events and spatial textures. We propose PredHydro-Net, a physics-guided dual-decoding framework that mitigates this smoothing. To resolve multi-variable optimization conflicts, it employs a decoupled architecture where macroscopic thermodynamic and dynamic fields unidirectionally modulate hydrometeor generation. By integrating wavelet-based frequency decoupling, spectral amplitude matching, and adversarial training, the model achieves a favorable trade-off between quantitative accuracy and spatial fidelity. In a 72-h global evaluation, PredHydro-Net outperforms both spatiotemporal deep learning baselines (Earthformer and PredRNNv2) and the operational Global Forecast System (GFS) in extreme-event detection and spectral representation. Furthermore, it demonstrates strong climatological consistency with Global Precipitation Measurement (GPM) satellite retrievals. The model reasonably reproduces the three-dimensional cloud structures in extreme weather events, such as Hurricane Ian. Feature attribution confirms its dependence on physical precursors such as relative humidity and wind convergence, offering a robust, physics-informed approach to long-tailed atmospheric prediction.

0

physics.ao-ph 2026-06-08

Continuous-time model cuts ocean forecast time by factor of four

by Qinghui Chen, Zekai Zhang +3 more

KFTD: Koopman-Fourier Time-Differentiable Network for Continuous Ocean Spatiotemporal Forecasting

Koopman embedding and Fourier interpolation remove repeated sampling while adding physical constraints directly in training.

abstract click to expand

Accurate oceanic forecasting is critical for climate monitoring and disaster early warning. However, ocean spatiotemporal forecasting encounters the double challenges of modeling complex dynamical systems and ensuring computational efficiency. We present Koopman Fourier Time-Differentiable (KFTD) Network, a time continuous twostage paradigm that decouples interpolation from prediction to achieve efficient and scalable spatiotemporal modeling. We map complex nonlinear dynamics into the Koopman linear space and exploit Fourier analysis to enable continuous time interpolation at arbitrary sub-steps. A lightweight residual network consumes the high fidelity intermediate states to yield the final forecast. Unlike diffusion models, KFTD eliminates multi step noise sampling and directly evolves the system in continuous time, yielding a 4 computational speedup. We further introduce a DPP Loss that supports arbitrary PDE constraints in an endtoend manner, breaking the physical consistency bottleneck of pure data-driven approaches. Empirical results on four ocean datasets confirm that our continuous time framework reduces MSE by an average of 5.6% (up to 12.7% for SST) and improves efficiency over MCVD by 76.25%.

0

physics.ao-ph 2026-06-08

Independent SST-CO2 data lets emulator handle AMIP +4K and 4xCO2

by Spencer K. Clark, Troy Arcomano +9 more

Disentangling the effects of sea surface temperature and CO₂ in global machine learned weather-climate emulators

Prior versions failed on these cases because training data kept the two drivers correlated; new random-CO2 runs break that link.

abstract click to expand

While previous versions of the Ai2 Climate Emulator (ACE) have been trained with CO$_2$ as a forcing, they are only accurate within a narrow range of scenarios, for example climate over the last 80 years forced by observed sea surface temperature (SST), sea ice, and CO$_2$ (AMIP), or equilibrium or near-equilibrium climates with CO$_2$ concentrations ranging from 1x to 4x that of the present day. Attempting to simulate climate forced by AMIP SST perturbed by +4 K or the response to an abrupt quadrupling of CO$_2$, results in unphysical behavior. We attribute this to these models being trained on datasets where the SST and CO$_2$ are correlated, limiting their ability to accurately learn their separate effects. In this study we introduce a new class of "random-CO$_2$" reference simulations where the SST and CO$_2$ are prescribed to vary independently. Trained on a balance of AMIP, equilibrium-climate, and random-CO$_2$ data, and including a total energy conservation constraint for improved interpretability, we present a more data-efficient model that not only accurately emulates its reference model in scenarios in which previous models excelled, but also scenarios like AMIP +4 K and slab-ocean-coupled abrupt 4xCO$_2$ where they did not. Limitations are that it has simplified or prescribed representations of other Earth system components like the ocean, land, and sea ice; does not expose other known climate drivers as forcings; and relies solely on physics-based model output for training data, inheriting the biases relative to observations thereof. Each of these represent opportunities for future work.

0

astro-ph.EP 2026-06-08

Wind shifts meteorite ground positions by median 143 m

by Hadrien Devillepoix, Martin Cupák

Freo Doctor: Atmospheric Modelling for Meteorite Falls and Spacecraft Re-Entries

Model differences exceed the 100 m uncertainty from fireball tracking and must be included in recovery predictions.

abstract click to expand

How much does the wind affect the path of meteorite falls? We finely model the lower ~30 km of the atmosphere using Weather Research and Forecasting open source tools at 1 km spatial resolution. Models initialised at different times give different results, which can be used as a proxy for uncertainty. We find that in most cases the differences on the ground positions are significant: median shift for a 1 kg meteorite is 143 m, doubling to 307 m for a 10 g rock, though these vary by over an order of magnitude between events. The differences wind model choice makes on the ground are significantly larger than the typical uncertainty on meteoroid state vector obtained from bright flight observations of the fireball (<100 m), and should be taken into account when predicting meteorite free-fall path to the ground. Unsurprisingly the cases where we see the largest differences coincide with documented extreme weather events. We also find that high spatial resolution models (1 vs. 3 km) tend to perform better. We have successfully used these models to guide field teams to the location of 12 fallen meteorites after fireball observations. We release as open data 1107 models we have calculated for 302 meteorite fall events and spacecraft re-entries around the world.

0

physics.ao-ph 2026-06-08

AI multi-agent system automates WRF-Chem mechanism validation

by Haoluo Zhao, Hongchun Zhang +7 more

TianJi-Environ: An Autonomous AI Scientist for Atmospheric Environmental Research

TianJi-Environ converts hypotheses into simulations and judges evidence completeness for ozone and PM2.5 cases.

abstract click to expand

As atmospheric environmental prediction continues to improve, interpretable validation of pollution mechanisms and feedback processes has become a main challenge in atmospheric chemistry. Yet mechanism validation based on complex numerical models still relies heavily on expert knowledge: mechanistic hypotheses must be operationalized into executable experiments, and model outputs must be organized into traceable evidence. We present TianJi-Environ, an auditable AI Scientist for atmospheric-chemistry mechanism validation. TianJi-Environ establishes the first WRF-Chem-based multi-agent framework that autonomously drives complex atmospheric-chemistry simulations, converting mechanistic hypotheses into executable configurations, testing experiments, and evidence criteria. Using ozone response and particulate-matter feedback as two representative examples, we demonstrate TianJi-Environ's capability for mechanism validation. In a summertime ozone case over the North China Plain, the system detects directionally consistent aerosol-radiation-interaction signals in shortwave radiation and boundary-layer height, but judges the evidence for ozone response to NOx control to be incomplete. In a wintertime PM2.5 case over the Guanzhong Basin, it localizes the unsupported link to insufficient propagation from black-carbon perturbation to particulate response and missing diagnostics of vertical absorptive heating. These results show that TianJi-Environ makes expert-driven mechanism validation explicit, structured, and auditable, offering a reproducible paradigm for multi-agent systems coupled with complex atmospheric-chemistry models.

0

nlin.CD 2026-06-08

Delayed observations capture Loop Current extension variability

by Francisco J. Beron-Vera, María J. Olascoaga +1 more

Loop Current Extension as an Effective Delayed Dynamical System

Compact maps from altimetry data forecast better than persistence at 30-90 days, with no extra info from channel flows.

abstract click to expand

The Loop Current is the dominant circulation feature of the Gulf of Mexico and exhibits pronounced variability associated with northward extension, retraction, and eddy shedding. Despite decades of study, the extent to which this variability admits a reduced dynamical description remains unclear. We investigate this question using delayed-coordinate representations constructed from satellite-altimetry observations of Loop Current extension. Ridge regression, multilayer perceptron forecasting, and Sparse Identification of Nonlinear Dynamics (SINDy) are applied to learn delayed evolution maps from the extension time series. Forecast skill consistently exceeds persistence at lead times of 30--90 days while requiring only a small number of delayed coordinates. Ridge regression reveals saturation with delayed-state dimension, indicating that much of the predictive information is contained within a compact representation. Neural-network forecasts provide modest additional improvements, while delayed SINDy identifies sparse evolution maps involving intraseasonal memory scales, from approximately two weeks to a few months, that remain stable under recursive iteration. Physical diagnostics associated with Yucatan Channel inflow, Florida Straits outflow, gateway geometry, and northern Caribbean vorticity contain predictive information but do not provide additional independent state information once the delayed Loop Current state is included. These results support the interpretation of Loop Current extension as an observable evolving on an effective low-dimensional delayed dynamical system. A substantial fraction of the predictable variability can be reconstructed from a small number of delayed observations and represented through compact delayed evolution maps.

0

physics.ao-ph 2026-06-05

MODIS estimates cyclone central pressure to 4.3 hPa RMSE

by Jinghuai Yao, Chi Yan Kwok +3 more

MODIS Thermal Infrared Sounding (MOTIS): Estimating Tropical Cyclone Central Pressure from Warm-Core Anomalies

Warm-core anomalies from infrared soundings give accurate Pc for intense storms and can cut best-track uncertainty in half when direct obser

abstract click to expand

This study presents a novel framework for estimating the central sea-level pressure ($P_\mathrm{c}$) of tropical cyclones (TCs) using infrared radiometers. We leverage the long-overlooked combination of high spatial resolution and sounding capability of the Moderate Resolution Imaging Spectroradiometer (MODIS) to measure warm-core anomalies in TC eyes. We develop the MODIS Thermal Infrared Sounding (MOTIS) framework, which performs instrument-specific preprocessing and estimates $P_\mathrm{c}$ using multiple linear regression. MOTIS yields $r^2 = 0.945$ and RMSE = 4.3 hPa for high-intensity TCs with observed clear eyes (mean $P_\mathrm{c} = 937$ hPa), outperforming all existing methods for intense TCs. We construct a dataset of 3288 (1082 clear-eye) MOTIS estimates from 2002 to 2025 and demonstrate its potential to improve the quality of Best Track $P_\mathrm{c}$, roughly halving uncertainties in the absence of pressure observations. Although MODIS is nearing the end of its mission, the MOTIS framework could be extended to next-generation geostationary sounders to provide accurate real-time $P_\mathrm{c}$ estimation for high-intensity TCs.

0

quant-ph 2026-06-05

Spatial modes reach quantum limit on turbulence radius estimate

by A. Hrebeniuk, M. Klen +3 more

Quantum-limited estimation of atmospheric turbulence via spatial mode decomposition

In the weak-field regime, mode decomposition outperforms direct imaging when the aperture is smaller than the coherence radius.

abstract click to expand

We establish the ultimate precision limit for estimating the optical spatial coherence radius (Fried parameter) within a quantum metrological framework. In the weak field regime, we show that spatial-mode decomposition -- originally introduced for superresolution imaging -- enables substantially more precise estimation than conventional direct imaging when the receiver aperture is smaller than the coherence radius.

0

physics.ao-ph 2026-06-05

Fixed rules turn satellite fire spots into tracked events

by Ronan Paugam, Jean-Baptiste Filippi +11 more

Leveraging MTG-FCI fire observations for event-based fire behavior monitoring from near-real-time operation to seasonal analysis

One algorithm follows individual wildfires from 10-minute detections for both immediate response and seasonal analysis.

abstract click to expand

Wildfire monitoring and suppression require timely information on fire behavior, including fire energy release and rate of spread, to support operational decision-making and resource allocation. Active fire products from the Flexible Combined Imager (FCI) aboard the geostationary Meteosat Third Generation (MTG) satellites provide 10-min observations over Europe and Africa. Deriving fire behavior information from these observations requires associating individual hotspot detections into coherent fire events. We present a Fire Event Tracker (FET) algorithm that performs spatio-temporal clustering of hotspot detections from the LSA-SAF FCI active fire product. The algorithm assigns persistent identifiers to fire events and updates their geometry, fire radiative power, and rate of spread at each 10-min interval. The same parameterization is used for both near-real-time and retrospective processing. FET was applied retrospectively to the Mediterranean FCI hotspot archive of 2025 and operationally in two near-real-time contexts: wildfire monitoring in Portugal and support of the 2025 SILEX airborne campaign within the EUBURN project, where besides fire monitoring, FET products were also used to initialize coupled FOREFIRE-MesoNH simulations for plume forecasting. Results show that event-based clustering of FCI active fire detections provides a consistent description of fire evolution, enabling both tactical wildfire management and high-frequency seasonal fire analyses.

0

physics.ao-ph 2026-06-04

Surface pressure spectra recover stratification parameters

by A. V. Kochin

The relationship between atmospheric stratification and internal wave processes

Frequencies of internal gravity waves inverted from ground records match radiosonde ascent data.

abstract click to expand

The atmosphere is a resonant system and its oscillation spectrum is determined by the spatial distribution of parameters. For example, the frequency of internal gravity waves depends on the vertical temperature gradient. Therefore, the study of the spectra of internal wave processes can be used to estimate the spatial distribution of atmospheric parameters. The work is aimed at detecting wave fluctuations in the atmosphere and calculating atmospheric parameters based on the measured spectra. The rate of ascent of the radiosondes was used as a reference information, which was compared with the spectra of pressure fluctuations at the surface. The stratification parameters were calculated based on the frequency of internal gravity waves and showed good agreement with the upper-air sounding data.

0

astro-ph.EP 2026-06-03

Entry angle controls bolide infrasound detection

by Miro Ronac Giannone, Elizabeth A. Silber

The Role of Source Geometry and Atmospheric Propagation in Global Bolide Infrasound Detectability

Steeper trajectories with lower energy deposition reach global arrays more reliably than shallow high-altitude ones across the 623-event sam

abstract click to expand

Global infrasound monitoring provides a persistent means of detecting energetic bolide atmospheric entries, complementing optical observations and extending coverage over remote regions. We present a global assessment of the physical factors governing bolide infrasound detectability by correlating 623 bolide events reported by the Center for Near-Earth Object Studies between 2007 and 2025 with waveform data from the International Monitoring System. We identify 311 events with confirmed infrasound detections, corresponding to a detection rate of approximately 50%, substantially higher than inferred from earlier surveys, reflecting both the maturation of the global infrasound network and advances in automated, multi-frequency array processing. Analysis of flight parameters shows that infrasound detectability is selective rather than uniform across the bolide population. Detected events are preferentially associated with steeper entry angles and lower-altitude energy deposition, while shallow, high-altitude trajectories are less consistently observed. Very high-energy events remain detectable regardless of geometry, but for the more common lower-energy regime, observability depends on specific combinations of entry parameters and propagation conditions. This geometric dependence persists across comparable energy ranges and atmospheric conditions, indicating that entry angle exerts a primary control on detectability, with energy and propagation acting as secondary modulating factors. These results provide new physical constraints on bolide-atmosphere interactions and improve interpretation of global infrasound observations for planetary defense and atmospheric-entry studies.

0

cs.LG 2026-06-02

NTK features yield 31-37% sharper weather uncertainty

by Jose Marie Antonio Miñoza, Rex Gregor Laylo +1 more

Scalable Uncertainty Quantification for Extreme Weather Forecasting via Empirical Neural Tangent Kernels

Intervals adapt to cyclone severity using only a matrix-vector product at inference, unlike fixed conformal baselines.

abstract click to expand

Deep learning weather models now match numerical weather prediction accuracy while running orders of magnitude faster, but produce deterministic forecasts without uncertainty estimates, a critical gap for high-stakes decisions during extreme weather events. This paper proposes Neural Tangent Kernel-based uncertainty quantification (NTK-UQ) using last-layer empirical features. Theoretical analysis predicts that UQ quality is architecture-dependent through two mechanisms. First, a variance collapse mechanism explains when UQ fails: when the eigenvalue truncation rank approaches the effective rank of the feature space, the GP correction term consumes nearly all prior variance, destroying discrimination between tropical cyclones and routine conditions; architectures with concentrated spectra (spectral operators) require aggressive truncation ($k \leq 10$), while attention-based models tolerate full-rank computation. Second, decomposition performance depends on the non-Gaussian, heavy-tailed structure of extreme weather: Independent Component Analysis exploits higher-order statistics (kurtosis, negentropy) to isolate heavy-tailed extreme-event features, achieving higher discrimination than singular value decomposition, which captures only second-order variance. A data-driven selection rule chooses ICA or SVD from the feature eigenspectrum concentration ratio, correctly prescribing the superior decomposition for all four evaluated architectures. Compared to split conformal prediction (the natural post-hoc baseline), NTK-UQ achieves 31--37\% sharper prediction intervals at 90\% coverage, and uniquely produces \emph{adaptive} intervals that scale with extreme event severity, which conformal prediction cannot achieve by construction. The framework requires no retraining; inference-time uncertainty requires only a single matrix-vector product per sample.

0

stat.AP 2026-06-02

Autoencoders yield probability ratios for 2003 heatwave under future warming

by Frieder Loer, Maybritt Schillinger +1 more

Probabilistic storyline attribution using machine learning

Temperature distributions conditional on circulation predict intensity rise to 32.1°C and likelihood increase by factor 3.2 by 2053

abstract click to expand

A fundamental goal in climate attribution is to estimate how forced climate change contributes to observed extreme weather events. The storyline attribution method compares an observed weather event, conditional on its atmospheric dynamic state (i.e., atmospheric circulation), in the current, 'factual' climate to an event with very similar circulation conditions in a hypothetical, 'counterfactual' climate. However, physical climate models cannot directly transfer these storyline counterfactuals across different climate forcing states. Statistical and machine learning techniques may overcome this limitation; yet, emulating circulation-conditional extreme events under different climate states is challenging. Here, we demonstrate distributional autoencoders (DAEs) as a versatile method for generating climate counterfactuals. They model the full distribution of spatially resolved European temperature fields conditional on the atmospheric circulation state and the mean global warming level. These distributions allow for deriving meaningful conditional probability ratios, which is a particular advantage of the DAE-based storyline approach. We train DAEs on fully coupled climate model simulations and we evaluate the modelled distributions across different factual and storyline-based counterfactual climate model simulations. In an illustrative case study, we revisit the 2003 European heatwave and we generate counterfactuals for a hypothetical `2003-like European heatwave' using ERA5 circulation, which we hypothesize to occur a quarter century (2028) and a half century (2053) after 2003. The conditional intensity would increase from 29.3 {\deg}C in 2003, to 30.3 {\deg}C and 32.1 {\deg}C in 2028 and 2053, respectively and conditional probability ratios would be 2.1 and 3.2 when compared to 2003.

0

physics.ao-ph 2026-06-02

Interior heatwaves grow longer and stronger by late century

by Md Tasim Ferdous, Reda Snaiki +1 more

Multiscale Dynamics of Heatwave Persistence and Intensity Under Climate Change

Simulations link the shift to redistribution of activity into lower-frequency modes with weaker damping in continental regions.

abstract click to expand

Climate change is expected to increase heatwave risk, but exceedance frequency alone cannot explain why some regions show stronger amplification in event persistence. This study develops an integrated event-dynamical workflow to diagnose changes in warm-season heatwaves and link them to coherent, multiscale structures of temperature variability. Heatwaves are identified over southern Canada using a fixed historical 90th percentile threshold (2001-2010 reference, 15-day moving window) and a minimum-duration criterion. Events are summarized using frequency (HWF, HWN), persistence (HWMD, HWD), and intensity (HWI, HWM) metrics. The daily mean temperature field is analyzed using multiresolution dynamic mode decomposition (mrDMD). Event and dynamical perspectives are connected through heatwave-conditioned mode participation ratios and spatial alignment analyses between mode-energy footprints and gridded heatwave metrics using hotspot overlap and Spearman rank association. The workflow is applied to CORDEX-NAM12 regional simulations (CRCM5 downscaling of CanESM5) under SSP5-8.5 for 2016-2025, 2051-2060, and 2091-2100. Results show a clear shift toward persistence-dominated heatwave regimes in the continental interior. By the late century, increases in seasonal heatwave days are accompanied by much longer events, with regional HWMD reaching about 26.66 days/event and HWD about 69 days, together with stronger above-threshold intensity, with HWI reaching about 6.88 K. Dynamical diagnostics indicate a redistribution of dominant activity toward lower-frequency levels and weaker effective damping in interior regions, while coastal and maritime regions show smaller changes. Heatwave-relevant low-frequency modes remain active during long events and align with persistence and intensity hotspots, supporting a process-informed interpretation of regional heatwave amplification under climate change.

0

physics.flu-dyn 2026-06-02

Breaking dissipation scales with crest steepness and duration after threshold adjustment

by Rui Cao (1, 2) +5 more

Breaking-induced energy dissipation of surface gravity waves at varying scales and co-flowing wind stresses

Lab measurements show scale and wind effects act through earlier breaking with less lean, producing a new relation for fractional energy los

abstract click to expand

Breaking-induced energy dissipation is studied for individual unsteady breaking waves using laboratory measurements of unidirectional surface gravity wave groups across a range of wave scales and wind stresses. A refined framework to estimate breaking-induced dissipation $\Delta E_{br}$ is proposed that accounts for background dissipation from non-breaking processes. Using this framework, we show that variations in wave scale primarily influence breaking energetics, such as fractional dissipation $\Delta E_{br}/E_0$ and dissipation rate $\epsilon_b$, by modifying the breaking onset threshold. Also, co-flowing wind systematically reduces both $\Delta E_{br}/E_0$ and $\epsilon_b$ relative to unforced conditions, as wind-forced waves break earlier with reduced crest forward-leaning. Exploiting the crest-front steepness at incipient breaking $\mathcal{S}_{\text{front}}(t_b)$ to characterise breaking onset and local crest geometry, we formulate a scaling for $\epsilon_b$ based on this local measure. This then yields $\Delta E_{br}/E_0 \propto \beta^{*}\,\mathcal{S}_b\,(\tau_b/T_b)$, where $\beta^{*}$ is crest forward leaning, $\mathcal{S}_b$ local steepness, and $\tau_b/T_b$ non-dimensional breaking duration. This scaling highlights the important roles of crest asymmetry and breaking duration in setting the breaking energy dissipation. Finally, we consider the breaking strength parameter $b$ by assessing existing steepness-based scaling laws, and relate $b$ to $\mathcal{S}_{\text{front}}(t_b)$, yielding an approximately linear dependence once the breaking-onset threshold is considered.

0

physics.ao-ph 2026-06-01

Airborne proton accelerator proposed to form clouds and trigger rain

by Orfeu Bertolami

On an Airborne Proton Accelerator for Enhancing Cloud Formation or Inducing their Precipitation

Following CERN findings, the device could cool regions and regularize precipitation patterns.

abstract click to expand

We argue that an airborne proton accelerator is an interesting tool for weather control. Following the findings of the CLOUD experiment at CERN, one expects that a beam of protons, likewise cosmic rays and other aerosols, can enhance the formation of low-altitude clouds, allow for tailor made cooling of overheated areas and induce the precipitation of high-altitude clouds that trap solar radiation reflected from the ground. The proton accelerator can also be used to mitigate droughts, regularise precipitation and avoid that it takes place through large and harmful storms.

0