AM-IA brings together news, research papers and applications relating to artificial intelligence published around the world into a single feed — and makes them available in your European language at the click of a button.
Scroll to browse · Click to filter
The programme focuses on local projections, hybrid models and their evaluation. The presentation, updated in October, emphasises data and partnerships.
AI emulators are already rivalling physical models in weather forecasting; when it comes to climate, they will need to explicitly represent forcings and withstand conditions that have never been observed.
From the very first bulletin, the National Hurricane Centre based its forecasts on WeatherNext 3 rather than on physical models, and announced a Category 2 storm when the system did not even have a name.
Using fifteen years of Météo-France radar data, a team is comparing U-Net, Transformers, GANs and diffusion models to produce detailed precipitation maps. Deterministic models consistently underestimate heavy rainfall.
Trained on EC-Earth3 simulations covering the period 1950–2166, the emulator reproduces daily variability, seasonal cycles and the warming trend in temperatures.
Guided by the geometry of buildings, a flow-matching model generates realistic 3D turbulent fields – with an error of less than 3 per cent in mean wind speed – to help cities adapt to climate change.
Weather and climate models are being developed faster than our capacity to evaluate them. The authors propose five priorities and three areas of focus for the coming decade, including a programme dedicated to their reliability.
Based on more than twenty years of NOAA observations, the model reconstructs continuous wind profiles from sparse probe measurements, and can extend surface winds observed by satellite into three dimensions.
Its components fit together like building blocks, from metre-scale simulations to global climate projections, and can be calibrated using a gradient on a GPU.
NOAA’s aerosol forecasts capture the fires observed 1.5 days earlier. A graphical model predicts their global evolution over 1–7 days and reduces the error by between 24 and 43 per cent.
With around 28,000 parameters and the ability to be trained on a CPU, this model simultaneously predicts five surface variables and outperforms all others in the 60 comparisons carried out across six continents.
A new benchmark test asks researchers to build a stochastic ENSO model from real observations within six hours; six out of twelve systems outperform the benchmark.
When tested with Claude, GPT, Gemini and Qwen on a microphysics model, the LAFT agent-based framework produces accurate translations up to 357 times faster than serial Fortran.
Researchers have added a generative layer to the smaller version of Aurora, which is approaching state-of-the-art performance on aggregate metrics and rivals the larger version.
In Muhos and Kajaani, land clearing and earthworks must be halted by 23 October, whilst the environmental impact assessments are being completed.
Linus Vogt and Laure Zanna generate oxygen fields for models that do not calculate them, and suggest that biases in the oxygen-poor region of the Pacific are overestimated.
A paper accepted for publication in *Noise & Health* finds that noise from hyperscale data centres – often a low-frequency hum between 70 and 140 Hz – is still not measured accurately enough to assess its effects on health.
Google’s largest electricity contract on the PJM network combines an increase in capacity equivalent to 11 nuclear reactors with a long-term supply agreement, in response to the regulations imposed on data centres.
The protocol treats physical, hybrid and purely data-driven models equally, testing them against ten years of observations from after 2015 and against past climates.
A global comparison covering a ten-day forecast period shows that AI-based weather models are better at pinpointing the location of heavy rainfall associated with atmospheric rivers, although they do not accurately estimate the volume of that rainfall.