# Weather forecasting

5 related articles

Subscribe to updates

arXiv Nov 8, 2026
Weather downscaling no longer just tunes the backbone: PROLA splits a fixed rank budget between principal and orthogonal directions, cutting Z500 error to 50.58
Mapping coarse-resolution history directly to future high-resolution weather trajectories requires forecasting, spatial downscaling and temporal densification at once, yet existing low-rank fine-tuning does not separate reusing pretrained principal directions from correcting outside them under a fixed rank budget. PROLA splits that budget into a principal branch and an orthogonal branch and rescales rank-1 optimizer updates with gradient signals; adapting Aurora on WeatherBench 2 with 3.92M trainable parameters, it reaches a 3-hour Z500 RMSE of 50.58, below AdaLoRA's 53.40.
Mapping coarse-resolution history directly to future high-resolution weather trajectories requires forecasting, spatial downscaling and temporal densification at once, yet existing low-rank fine-tuning does not separate reusing pretrained principal directions from correcting outside them under a fixed rank budget. PROLA splits that budget into a principal branch and an orthogonal branch and rescales rank-1 optimizer updates with gradient signals; adapting Aurora on WeatherBench 2 with 3.92M trainable parameters, it reaches a 3-hour Z500 RMSE of 50.58, below AdaLoRA's 53.40.
Minseo Yoon（First author）
PROLA Predictive Spatiotemporal Weather Downscaling Parameter-Efficient Fine-Tuning Weather forecasting
arXiv Nov 9, 2026
One2N pushes a 0.7° forecast model to 0.07°: four weeks on one 80 GB GPU, tropospheric skill intact
As Earth-system AI models approach kilometre scales, halving grid spacing quadruples the grid-point count and the end-to-end training cost. One2N splits resolution refinement into successive increments: at each level the complete parent predictor is frozen and only a cross-resolution residual correction module is trained, so the parent's activations and optimizer states are never retained. One refinement level on the NeuralGCM-0.7° baseline yields One2N-0.07°, trained for about four weeks on a single 80 GB A800 GPU; against HRES-0.07° analyses it keeps the parent's tropospheric skill while reducing upper-stratospheric errors and resolving finer structure.
As Earth-system AI models approach kilometre scales, halving grid spacing quadruples the grid-point count and the end-to-end training cost. One2N splits resolution refinement into successive increments: at each level the complete parent predictor is frozen and only a cross-resolution residual correction module is trained, so the parent's activations and optimizer states are never retained. One refinement level on the NeuralGCM-0.7° baseline yields One2N-0.07°, trained for about four weeks on a single 80 GB A800 GPU; against HRES-0.07° analyses it keeps the parent's tropospheric skill while reducing upper-stratospheric errors and resolving finer structure.
Peking University Chinese Academy of Sciences Qilu University of Technology Beijing Information Science & Technology University National University of Defense Technology
Yanyan Tan（First author）
One2N NeuralGCM Residual Learning Swin Transformer Weather forecasting
arXiv Nov 10, 2026
MarsNet learns the Martian atmosphere from observations alone: 4.2 K near-surface temperature error and the best surface pressures of four products
The Martian atmosphere has been described by reanalyses that assimilate spacecraft retrievals into general circulation models, so where observations are sparse their fields largely reflect the model. MarsNet trains a two-hour forecast and a per-window observation update jointly inside a fixed physical structure, using no reanalysis or model field as input or target. On MGS radio-occultation temperatures assimilated by no dataset, it matches or improves on the three reanalyses in the lowest three scale heights, with a near-surface RMSE of 4.2 K against 4.4–5.4 K, but is less accurate aloft.
The Martian atmosphere has been described by reanalyses that assimilate spacecraft retrievals into general circulation models, so where observations are sparse their fields largely reflect the model. MarsNet trains a two-hour forecast and a per-window observation update jointly inside a fixed physical structure, using no reanalysis or model field as input or target. On MGS radio-occultation temperatures assimilated by no dataset, it matches or improves on the three reanalyses in the lowest three scale heights, with a near-surface RMSE of 4.2 K against 4.4–5.4 K, but is less accurate aloft.
Sebastian Mengwall（First author）
MarsNet Data Assimilation Martian Atmosphere Weather forecasting
MIT News - Artificial intelligence Oct 14, 2026
Two-to-six-week forecasts stop being a coin flip: a postprocessing layer doubles accuracy and beats climatology for the first time
A machine-learning postprocessing layer that learns to correct errors in existing forecasting models doubles the accuracy of a gold-standard model over the two-to-six-week subseasonal range and, for the first time, consistently beats climatological averages.
Microsoft
Subseasonal Forecasting Bias correction Weather forecasting
Journal of the European Meteorological Society. Sep 16, 2026
Machine learning is revolutionizing weather forecasting — the next step is a change in how we work
This article by Peter Dueben, Peter Bauer, Oliver Fuhrer, Nikolay Koldunov and Jørn Kristiansen argues that, following the success of machine learning in producing weather predictions with competitive skill compared to complex traditional systems, attention should shift from forecast output to the working practices that make prediction systems possible, and 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; it discusses six non-exhaustive areas in which agentic software engineering, open and compressed data, shared verification
This article by Peter Dueben, Peter Bauer, Oliver Fuhrer, Nikolay Koldunov and Jørn Kristiansen argues that, following the success of machine learning in producing weather predictions with competitive skill compared to complex traditional systems, attention should shift from forecast output to the working practices that make prediction systems possible, and 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; it discusses six non-exhaustive areas in which agentic software engineering, open and compressed data, shared verification
Peter Dominik Dueben（First author）
Weather forecasting Agentic software engineering

You’re all caught up

Weather downscaling no longer just tunes the backbone: PROLA splits a fixed rank budget between principal and orthogonal directions, cutting Z500 error to 50.58

One2N pushes a 0.7° forecast model to 0.07°: four weeks on one 80 GB GPU, tropospheric skill intact

MarsNet learns the Martian atmosphere from observations alone: 4.2 K near-surface temperature error and the best surface pressures of four products

Two-to-six-week forecasts stop being a coin flip: a postprocessing layer doubles accuracy and beats climatology for the first time

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