https://arxiv.org/api/X1TNQKbxwg3dhTsVPzXMkixAkSs 2026-11-12T03:43:29Z 8941 15 15 http://arxiv.org/abs/2611.12614v1 2026-11-09T15:56:45Z

Most operational near-nadir geophysical model functions (GMFs) are empirical and fitted one radar band at a time, and do not explain the difference between Ku and Ka band. This paper derives a backscatter model in which one local description of the sea surface serves both bands. The near-nadir Kirchhoff integral is written in terms of the elevation difference between two surface points separated by a horizontal lag. The surface then enters through the structure functions of that difference, and the scattering kernel selects the lags each band samples, about 2.5-1.4 cm at Ku and 1.0-0.5 cm at Ka band. Geometric optics is recovered as the quadratic, isotropic limit, without a spectral cutoff. With a logarithmic second-order structure function, one wind-dependent log-slope sets both the incidence shape and the Ku-Ka shape difference, and band transfer reduces to evaluating one curve at a second lag. Same-footprint dual-frequency radar data show that one band alone does not determine this transfer. The log-slope is the level of the short-wave saturation range and grows about as the square of the wind speed; a physical short-wave spectrum reproduces this growth but overestimates the level. Dual-frequency radars thus measure the centimetre-scale roughness level, although they do not separate a scale-dependent structure function from non-Gaussian surface statistics. A companion paper delivers and evaluates the resulting GMF, KuKa-Mod.

2026-11-09T15:56:45Z 15 pages, 6 figures, 4 tables. Submitted to IEEE Transactions on Geoscience and Remote Sensing. Part I of two companion papers. Model release: doi:10.5281/zenodo.24084870 Alexey S. Mironov Frédéric Nouguier Bertrand Chapron http://arxiv.org/abs/2611.12515v1 2026-11-09T14:54:47Z

It has been a long-held paradigm that the Atlantic Meridional Overturning Circulation in coupled climate models is too stable, underestimating the possibility of a collapse. A recent analysis of extended simulations to year 2300 in the Coupled Model Intercomparison Project, CMIP6, demonstrated a partial AMOC collapse where the deep circulation completely vanishes. This collapse was simulated by at least one model in all emission scenarios, and it occurred in every model forced by the highest emission scenario when extended to year 2300. Given the modelling community's previous emphasis on stability in climate models, in this perspective, we explore the alternate possible mechanisms by which the AMOC in CMIP6 might have become too unstable. This is motivated by the lack of evidence of downward trends in the AMOC in the observational records, and a subset of CMIP6 models displaying too high climate sensitivity and too strong sensitivity to aerosol removal, both of which could predispose them toward a spurious collapse. For these reasons, it is critical to evaluate the mechanisms by which CMIP6 models simulate AMOC collapse, to be able to develop programs further investigating the trustworthiness of simulated AMOC collapse scenarios.

2026-11-09T14:54:47Z 15 pages, 4 figures Sybren Drijfhout Alejandra Sanchez-Franks http://arxiv.org/abs/2607.02213v3 2026-11-09T12:52:39Z

Midlatitude storm tracks play a central role in Earth's climate by transporting heat from low to high latitudes. Their strength is therefore set by the meridional temperature gradients, traditionally linked to differential solar heating. This framework predicts a substantial seasonal reduction in storm activity, as meridional insolation gradients vanish during summer. Yet in the Southern Hemisphere, storm activity decreases by only about 40% from its seasonal maximum. Here, we show that cloud radiative effects contribute to the seasonal maintenance of storm activity by reinforcing the temperature gradients that sustain it. 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 captures the seasonal response and demonstrates the importance of two emergent cloud properties: the maximum attainable cloud albedo and the sensitivity of cloud cover to storm activity. Together, these findings indicate that cloud radiative feedbacks help maintain the thermal gradients that sustain storm activity on seasonal timescales. More broadly, they reveal a strong coupling among storms, clouds, and the ocean spanning distinct spatial and temporal scales.

2026-07-02T14:21:46Z Or Hadas http://arxiv.org/abs/2608.30795v2 2026-11-09T12:30:13Z

End-to-end weather forecasting systems produce skillful global gridded and station forecasts directly from raw Earth observations, replacing the numerical weather prediction pipeline, including data assimilation, at a fraction of its cost. These systems are deterministic and issue no uncertainty. Here we render the Aardvark Weather model probabilistic by attaching one stochastic mechanism to each component: learned, input-dependent noise at the observation encoder, aimed at the uncertainty inherited from the observing system, and Monte Carlo dropout in the processor, aimed at the uncertainty of the learned dynamics. The resulting nested ensemble attributes forecast spread to the two sources through a law-of-total-variance decomposition, cross-checked by withholding observation streams. Probabilistic finetuning significantly improves the mean forecast, by 11% on average across variables and lead times. The ensemble is close to calibrated against ERA5 through the medium range, is at least as accurate as the deterministic model at weather stations while beating it in relative economic value at every lead time, and remains considerably behind the operational ECMWF IFS ensemble. The encoder branch behaves as observation-driven uncertainty. Component-attributed uncertainty makes end-to-end forecasts more transparent, a step toward observation-driven digital twins of the atmosphere.

2026-08-31T13:49:00Z 25 pages, 16 figures Rodrigo Almeida Noelia Otero Jost Arndt Simon Baur Wojciech Samek Jackie Ma http://arxiv.org/abs/2611.12249v1 2026-11-09T11:37:17Z

Earth-system AI models approach kilometre-scale resolution, but halving grid spacing quadruples the grid-point count and end-to-end training cost. We introduce One2N, a recursive incremental framework extending a coarse model through $N$ levels of cross-resolution residual correction. At each level the parent is frozen and excluded from backpropagation, so its activations and optimizer states are never retained; only a tier-specific module is trained. A single refinement level on the NeuralGCM-0.7$^\circ$ baseline yields the two-level One2N-0.07$^\circ$, trained on one 80 GB GPU. Against HRES-0.07$^\circ$ and ERA5-0.25$^\circ$ analyses it matches its parent's tropospheric skill while reducing upper-stratospheric errors and resolving finer structure. Tropical-cyclone track errors nearly match the parent's and are lower than Pangu-Weather-0.25$^\circ$ at medium range; RMSE is lower than that model's on 93.46% of common targets. One2N thus turns an existing forecasting system, or an inexpensively trained coarse baseline, into a reusable foundation for high-resolution development, breaking the resolution barrier.

2026-11-09T11:37:17Z 32 pages, 12 figures and 7 tables, including Supplementary Information Yanyan Tan State Key Laboratory of Tibetan Plateau Earth System, Environment and Resources University of Chinese Academy of Sciences, Beijing, China Xiangjun Tian State Key Laboratory of Tibetan Plateau Earth System, Environment and Resources University of Chinese Academy of Sciences, Beijing, China Can Wang College of Mathematical Science, Beijing Information Science and Technology University, Beijing, China Hongqin Zhang Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China Yinhai Luo Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China Min Zhao State Key Laboratory of Tibetan Plateau Earth System, Environment and Resources Zhe Jin Institute of Carbon Neutrality, College of Urban and Environmental Sciences, Peking University, Beijing, China Xingchao Yang State Key Laboratory of Tibetan Plateau Earth System, Environment and Resources Zaifei Wang State Key Laboratory of Tibetan Plateau Earth System, Environment and Resources Banglin Zhang College of Meteorology and Oceanography, National University of Defense Technology, Changsha, China Aiwen Wang College of Mathematical Science, Beijing Information Science and Technology University, Beijing, China Jing Zhou State Key Laboratory of Physical Oceanography, Institute of Oceanographic Instrumentation, Qilu University of Technology Tao Wang State Key Laboratory of Tibetan Plateau Earth System, Environment and Resources Baosheng An State Key Laboratory of Tibetan Plateau Earth System, Environment and Resources Xuliang Zhuang State Key Laboratory of Tibetan Plateau Earth System, Environment and Resources University of Chinese Academy of Sciences, Beijing, China Shilong Piao Institute of Carbon Neutrality, College of Urban and Environmental Sciences, Peking University, Beijing, China Tandong Yao State Key Laboratory of Tibetan Plateau Earth System, Environment and Resources http://arxiv.org/abs/2607.28093v2 2026-11-09T08:18:06Z

Accurate Surface Solar Irradiance (SSI) estimation is increasingly important for photovoltaic energy monitoring and forecasting. The recently introduced Meteosat Third Generation (MTG) satellite constellation provides imaging data with higher spatial resolution compared to the Meteosat Second Generation (MSG) satellite constellation, but its benefits for machine-learning-based SSI retrieval are not well established. We compare multi-resolution convolutional neural network models for 10-minute SSI retrieval over Northern Europe (Estonia) using either MSG/SEVIRI or MTG/FCI satellite imagery. We evaluate model performance against ground-based pyranometer measurements from eight Estonian meteorological stations using site-based cross-validation with a temporally independent test year and multiple training seeds. Model performance is also compared with the SARAH-3 physics-based satellite SSI product. The FCI-only model significantly outperformed the SEVIRI-only model under overcast and cloudy conditions, reducing RMSE by 5.1 W/m^2 and 6.5 W/m^2, respectively. However, under partly cloudy or clear skies, there was no statistically significant difference in RMSE between the FCI-only and the SEVIRI-only models. In all-sky conditions, the RMSE of the FCI-only model was 5.1 W/m^2 lower on average, but this difference was not statistically significant after Bonferroni correction. Compared with physics-based SARAH-3, the FCI-only model yielded skill scores of 28% under overcast conditions, 19% under cloudy conditions, and 17% overall. Furthermore, both models underperformed SARAH-3 in clear-sky conditions. These results show that higher-resolution MTG/FCI imagery improves CNN-based SSI retrieval when clouds dominate irradiance variability, but also indicate that higher resolution alone is insufficient to address clear-sky limitations in machine-learning-based SSI retrieval.

2026-07-30T12:05:34Z Gordei Pribõtkin Piia Post Velle Toll http://arxiv.org/abs/2611.11918v1 2026-11-09T05:44:00Z

Integrating machine learning weather prediction (MLWP) models into classical data assimilation (DA) workflows requires that they accurately evolve ensemble covariances. We evaluate this capability in a two-way coupled DA system, where an ensemble variational (EnVar) component evolves the mean and an ensemble Kalman filter (EnKF) evolves the covariances. The system is cycled over one month and assimilates a wide range of real-world observations, including microwave radiances. Using the MLWP model GraphCast for the EnKF and the physics-based MPAS model for the EnVar yields ensemble spread and covariances comparable to a reference configuration in which MPAS evolves both components. The resulting EnVar mean achieves consistently lower RMSE against GFS analyses, except near the model top, where spurious correlations degrade performance. These results demonstrate that MLWP models can propagate covariance information in a realistic DA cycling system and provide a proof of concept for their integration into future operational ensemble systems.

2026-11-09T05:44:00Z 22 pages, 7 figures Simon Toedtli David Ahijevych Wei Wang Chris Snyder http://arxiv.org/abs/2611.12999v1 2026-11-08T15:36:31Z

Surface shortwave radiation (SWR) can change within minutes as clouds evolve, but operational forecasts often trade spatial detail and update frequency for forecast range. We present a numerical-weather-prediction-independent, observation-to-forecast framework that converts sequences of Himawari-8 Level-1 multispectral imagery and solar geometry directly into 24 gridded SWR fields at 10-min intervals for the next 4 h. A controlled six-model experiment separates the effects of satellite information from those of temporal self-attention. In the 0-1 h window, the full-multispectral attention model achieves a mean absolute error of 37.9 W/m2 and a root mean square error of 63.7 W/m2, reductions of 21.0% and 19.5% relative to a historical-SWR baseline. Improvements extend across nine subregions and include weaker spatial bias and better structural consistency. Temporal attention does not improve forecasts from historical SWR alone, but becomes effective when supplied with multispectral cloud and thermodynamic signals. Direct multispectral observations therefore provide the precursor information required for high-frequency, spatially resolved SWR nowcasting.

2026-11-08T15:36:31Z Tianyu Xie Zhaoliang Zeng Zhehao Liang Ke Gui Xutao Zhang Yaqiang Wang Dabo Guan Huizheng Che http://arxiv.org/abs/2611.11335v1 2026-11-08T12:21:59Z

Paleoclimate proxies show that abrupt Greenland changes were followed by Antarctic responses within decades, suggesting a fast atmospheric pole-to-pole teleconnection involving shifts in the Intertropical Convergence Zone (ITCZ) and reorganization of extratropical jets. Here, we examine how the mean ITCZ position modulates this interhemispheric teleconnection. Using an aquaplanet slab-ocean model, we apply northern high-latitude forcing in climatologies with different mean ITCZ positions. The southern high-latitude response is about twice as large when the mean ITCZ is in the Southern Hemisphere. The teleconnection is therefore directionally asymmetric. The asymmetry arises in the subtropical jet response per degree of ITCZ shift, because, even with a comparable Hadley cell response, the angular momentum carried into the upper branch depends on the distance of the ITCZ from the equator, so that an ITCZ shift toward or away from the equator reinforces or offsets the jet response. Our results suggest that the southward-displaced glacial ITCZ made Arctic-to-Antarctic transmission more efficient than today, whereas the present-day ITCZ favors the reverse.

2026-11-08T12:21:59Z Yechul Shin Jong-Seong Kug http://arxiv.org/abs/2611.11248v1 2026-11-08T09:04:23Z

Halving the floating-point precision of an ocean model can make it substantially faster, but whether the rounding errors stay negligible over climate time scales has rarely been tested. We implement single precision in the global ocean-sea-ice model FESOM2 and compare 60-year hindcasts with double-precision controls. A first implementation lost salt steadily, accumulated a heat-content deficit of 129 ZJ, and shifted major ocean fronts. Controlled experiments trace the drift to the tracer update: small increments added to a large salinity value are rounded away, and the increments themselves are biased. Accumulating tracer tendencies and layer thicknesses in 64 bits, and carrying the rounding error of each tracer update into the next, removes the drift for at most 8% additional runtime. Single- and double-precision simulations then differ in heat content and mean state about as much as two double-precision runs with slightly perturbed initial conditions, sea ice differs by at most 0.2%, and he salinity difference is four orders of magnitude smaller than before. In the coupled model AWI-ESM3, single precision adds no detectable heat source or sink over 60 years. Single precision speeds up the 60-year hindcasts by a factor of 1.35, and the model by up to 1.66 on meshes of up to 5 km resolution; on GPUs, a C++/Kokkos version runs up to 1.27 times faster and needs 43% less memory. Whether reduced precision is safe for climate integrations is decided by long-term budgets, not by short tests.

2026-11-08T09:04:23Z Suvarchal Kumar Cheedela Nikolay V. Koldunov Sergey Danilov Jan Streffing Dmitry Sidorenko Thomas Jung http://arxiv.org/abs/2611.11162v1 2026-11-08T06:09:27Z

A common method for diagnosing and correcting model bias in weather and climate prediction is based on analysis increments: the difference between the analysis and the forecast at each data assimilation cycle. Their time average is used to estimate and correct the bias, and machine learning is increasingly used to map the forecast state to the increments, giving a state-dependent correction. Despite its widespread use, the method lacks theoretical foundations. We first analyse a Kalman filter whose forecast model has a constant bias. We show what the averaged increment measures, and prove that iterating the correction removes all of the bias seen through the observations; for linear dynamics, no method can identify the rest. We then extend the analysis to state-dependent biases. For linear dynamics, each increment records the model error over the filter's memory, as seen through the observations. We use this to determine what the increments measure when averaged over a long period, averaged over periods with similar slowly varying conditions, and used without averaging, as in machine-learned corrections. Numerical experiments confirm the theory for a linear model and suggest that its main conclusions carry over to the nonlinear Lorenz '96 model, including for a correction learned by a neural network.

2026-11-08T06:09:27Z 15 pages, 5 figures Eviatar Bach http://arxiv.org/abs/2611.11066v1 2026-11-08T01:50:25Z

Earth system models underpin climate projections, but their cost limits how many scenarios, ensemble members and impact pathways can be explored. Machine-learning weather models forecast as skillfully as physics-based models at a fraction of the cost, yet coupled to an ocean they have produced weak El Nino variability or have learned it from climate-model output, and none carries the climate through rivers to the people exposed to it. Here we present Rescene-ESM, an Earth system model whose coupler, YAHO, yokes a frozen 20.8-million-parameter AI weather model to physically formulated ocean, sea-ice, land, river, carbon and energy components constrained by observations. Adding the couplings one at a time shows what each contributes: over a climatological ocean the AI atmosphere varies from year to year by a third of the observed amount over the oceans, and an interactive mixed layer, an equatorial wave ocean, the residual of the observed sea surface temperature (SST) equation as stochastic forcing and the land's moisture flux raise this step by step and bring El Nino and its teleconnections, leaving the mean climate almost unchanged. The full model simulates about 1,000 years per day on one graphics processing unit, holds the observed climate for a thousand years without drift, reproduces the observed 1979-2022 global warming, carries forced change with a physically based structure around the frozen network (transient climate response 1.60 K), routes runoff to 212 river gauges with a median seasonal-cycle correlation of 0.79 and, driven by emissions, separates the heatwave exposure of an SSP1 and an SSP3 world by a factor of 4.7 in 2081-2100, three-quarters of it through the warmer climate.

2026-11-08T01:50:25Z 47 pages, 6 figures, 11 Extended Data figures; Supplementary Information (20 figures, 9 tables) included Minjong Cheon Kaoru Kakinuma http://arxiv.org/abs/2611.10882v1 2026-11-07T18:07:19Z

Assessing climate risks requires projections of daily temperature distributions, including rare and spatially widespread extremes. Climate simulations are costly and biased, while correcting them toward observations risks suppressing inter-model differences. We train a single machine-learning emulator, a conditional diffusion model, on seventeen CMIP6 models and ERA5 reanalysis to generate spatially coherent daily temperature fields conditioned on model identity, date, and CO2-equivalent concentration. At sampling time, diffusion guidance reduces model biases by steering each emulated climate-model distribution toward observations, with its strength selected through perfect-model experiments using withheld simulated futures. Observational influence decreases with projection horizon and forcing. Across seventeen synthetic targets, guidance more than halves median biases in the 1st and 99th percentiles of daily temperature, improves probabilistic scores and reliability, and reduces median annual global-mean temperature RMSE by 32-57% across three emission scenarios. In a detailed benchmark, guidance also outperforms established bias corrections on daily-field probabilistic scores. Efficient sampling generates large ensembles without further climate simulations. For example, our framework corrects CMIP6's roughly threefold historical overestimate of extreme-heat coverage, yet projects that concurrent extremes will affect nearly half of global land on a typical day in the 2080s under SSP3-7.0.

2026-11-07T18:07:19Z Francesco Immorlano Elijah Tavares Felix Draxler Juan Nathaniel Violette Launeau Duncan Watson-Parris Padhraic Smyth Pierre Gentine Stephan Mandt http://arxiv.org/abs/2611.04892v2 2026-11-07T09:09:04Z

Artificial intelligence (AI) based ocean state forecasts now approach or surpass those produced by operational ocean forecasting systems as measured by conventional pointwise metrics. However, these metrics may favor spatially smoothed forecasts and cannot determine whether forecast variables remain physically coherent. Forecast evaluation should therefore move beyond overall pointwise accuracy toward assessing forecast trustworthiness. We propose a multidimensional framework comprising overall pointwise accuracy, physical consistency, and distributional consistency. Overall pointwise accuracy summarizes forecast and reference agreement over all valid samples; physical consistency evaluates dynamical balances using model outputs alone; and distributional consistency evaluates distribution tails and multiscale spatial structures. Three AI models were deployed as representative testbeds in a unified operational environment. This comprehensive assessment shows that lower overall pointwise errors cannot preclude substantial dynamical imbalances and suppressed fine-scale variance. Conversely, higher pointwise errors may arise from spatial displacement of coherent structures rather than a poor representation of their amplitude. The results demonstrate that the evaluation of AI ocean forecasting models should be based on the multi-dimensional assessment of forecast trustworthiness, rather than solely relying on a single metric. The study provides a preliminary framework for AI model intercomparison, which is also informative for targeted model refinement and operational implementation.

2026-11-03T13:39:32Z Ziqing Zu Ye Yuan Tiejun Ling Zhi Chen Xiang Li Zhiqiang Li Qiang Wang Jiangjiang Xia Jiuke Wang Yongjun Jia Shinan Wang Zhi-Yuan Zhang Renbo Pang Xiaojuan Lin Yu Zhang Huier Mo Ang Li http://arxiv.org/abs/2604.20222v5 2026-11-07T07:10:33Z

(Abridged) On 13 September 2025 at around 22 UTC, a localized tornado outbreak affected Camarines Norte in the eastern Philippines. The event developed within an atypical easterly severe weather regime, with warm, moist southeasterly low-level flow and modest low-level shear ahead of an approaching easterly wave. Convective initiation was focused along the upstream segment of the inverted trough, within a vorticity convergence zone that may have enhanced low-level rotation. MUCAPE reached 2167 J kg$^{-1}$, LCLs were below 200 m, and modestly curved but strongly streamwise hodographs favored supercells and tornadogenesis. At least five vortices were identified, three of them tornadoes from semi-discrete supercells. The Magang tornado, confirmed by a ground damage survey and Doppler radar, was rated IF2.5 along a $\sim$1.4 km path; the radar-identified Cahabaan ($\sim$2.1 km path) and Napilihan tornadoes were rated IF1, and the remaining two vortices IF0. The near-simultaneous Magang and Cahabaan storms were close enough that inflow-outflow interactions may have aided tornadogenesis. Dual-polarization radar revealed Z$_\mathrm{DR}$ and K$_\mathrm{DP}$ columns, a tornadic debris signature co-located with the Magang velocity couplet, and a bounded weak echo region in the Cahabaan supercell. To our knowledge, this is the first well-observed tornado outbreak with simultaneous tornadic supercells documented in the Philippines within an easterly severe weather regime.

2026-04-22T06:10:58Z 44 pages, 18 figures, 4 tables. Accepted for publication in Springer Nature's Earth Systems and Environment (ESEV). The supplementary damage analysis can obtained by downloading the source files of this preprint (.zip) Generich H. Capuli