SDECast is a Neural Stochastic Differential Equation framework designed for continuous‑time probabilistic weather forecasting. Unlike traditional autoregressive models that operate at fixed temporal resolutions and can accumulate errors at short time steps, SDECast learns stochastic dynamics directly in physical space without repeated SDE simulation during training. Experiments on a simulated geophysical flow demonstrate that SDECast captures meaningful drift dynamics and reproduces continuous‑time behavior, while large‑scale tests show skillful probabilistic forecasts up to five days ahead at hourly resolution.
arXiv:2512.22814v2 Announce Type: replace Abstract: Accurate long-range weather forecasting remains a major challenge for AI models, both because errors accumulate over autoregressive rollouts and be...
The paper introduces a new diffusion-based approach for probabilistic weather forecasting that incorporates auxiliary conditional denoising tasks. By predicting future states from both clean and corrupted contexts, the model reduces ambiguity and learns conditional distributions across noise levels using a proper scoring rule. Experiments demonstrate improved one‑step forecasting, especially at longer horizons, and better calibration in high‑dimensional global weather scenarios.
The paper introduces Climate Physics Dynamic Matching (ClimPhyDM), a variational, simulation‑free framework that blends an advection‑type physics prior with data‑driven components for weather forecasting. It leverages deep generative models to capture complex dynamical systems while preserving underlying physical structure. On the ERA5 benchmark, ClimPhyDM outperforms existing methods such as ClimODE and GB‑DM, achieving lower error over extended horizons and demonstrating improved temporal stability and resistance to error accumulation, all while training on a single modest 12 GB consumer GPU.
arXiv:2607. 21080v1 Announce Type: new Abstract: Long-horizon weather forecasting is a fundamental challenge in atmospheric science, for which autoregressive Deep Learning Weather Prediction (DLWP) has emerged as the primary paradigm.
arXiv:2406. 14399v4 Announce Type: replace Abstract: The development of Time-Series Forecasting (TSF) models is often constrained by the lack of comprehensive datasets, especially in Global Station Weather Forecasting (GSWF), where existing datasets are small, temporally short, and spatially sparse.
arXiv:2606. 02663v1 Announce Type: cross Abstract: Recent advances in machine learning have produced probabilistic weather forecasting models comparable to state-of-the-art numerical weather predictors.
arXiv:2610.09770v1 Announce Type: cross Abstract: Deep learning has made rapid advances in weather forecasting: autoregressive models trained on atmospheric reanalyses now rival dynamical models acro...
GenONet introduces a Spatio-Temporal U-DeepONet architecture that serves as a generator in a GAN framework for high‑resolution precipitation nowcasting up to three hours ahead. By learning continuous‑time precipitation dynamics with a Deep Operator Network and enforcing physics through a moisture‑conservation loss, the model produces sharp, physically consistent forecasts that outperform baselines, especially for high‑intensity events and longer lead times. Ablation studies confirm the added value of the physics‑informed regularizer and the synergy of operator learning with adversarial training.
arXiv:2610.03759v3 Announce Type: replace-cross Abstract: Ocean wave forecasting is essential for maritime safety, offshore operations, and coastal resilience, yet remains challenging due to systemat...
Just Weather Scoring (JWS) is a single‑stage, end‑to‑end diffusion model that forecasts precipitation directly in radar space, eliminating the need for separate compression or deterministic components. By combining Masked Asynchronous Diffusion with a scoring‑rule objective, JWS adapts diffusion training to high‑dimensional spatio‑temporal data and enables few‑step generation. On SEVIR and MeteoNet benchmarks, JWS achieves state‑of‑the‑art probabilistic forecasting while using fewer parameters and delivering over 17× faster inference than existing methods.
arXiv:2601.21151v3 Announce Type: replace Abstract: Machine-learning approaches to weather forecasting often employ a monolithic architecture in which distinct physical mechanisms, such as advection,...
The paper proposes a method for predicting future snow water equivalent (SWE) across the Western United States by first removing spatial correlations using a Gaussian Process-based linear transformation, then training a long short-term memory (LSTM) neural network on the decorrelated data. This separation of spatial and temporal components improves predictive accuracy compared to baseline models. Additionally, the authors incorporate conformal prediction to provide distribution‑free uncertainty estimates for SWE forecasts.