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
A study applies post-training quantification to two global forecasting AI models, DLWP and FourCastNet, to reduce computational load and energy consumption without compromising short-term forecasts.
Rather than using a series of fixed-time steps that lead to accumulated errors, SDECast models the atmosphere on a continuous-time basis and produces hourly probabilistic forecasts for up to five days.
By predicting only the slow-changing components of the climate, this model – based on the JEPA architecture – matches the CEPMMT reference set and outperforms it on several criteria in weeks 5 and 6.
Trained on more than 300,000 VIIRS detections, the model identified 26 out of 51 historical fires before existing systems did, compared with 6 for the operational product.
The angle at which waves strike the coast affects the marine surge. Simulated using 14 climate models, it has changed little to date but will change significantly by 2099 under the highest-emission scenario.
A feasibility study is examining whether small, targeted disturbances are sufficient to alter Aurora’s trajectories – the first step towards the idea of controlling a chaotic system such as the atmosphere.
A study published in *Energy & Fuels* predicts that emissions will increase more than fourfold by 2030 and suggests that gas-fired power stations fitted with carbon capture technology could help to limit the impact.
Tested near Chicago, the model predicts water levels across 4.2 million 30-metre cells up to one day in advance, around 150 times faster than a hydrodynamic model.
A conditional variational autoencoder replicates the Holton-Mass model, including the rare transitions between strong and weak vortices, and its latent space reveals a discernible physical structure.
This data-driven weather forecasting model rivals the Swiss operational numerical models for up to five days, but it still smooths out thunderstorms too much and underestimates certain local gusts.
Researchers have published the world’s first corrected and downscaled dataset designed to measure the concrete impacts of climate tipping points at +2 °C.
Over 15 billion dollars for a campus powered off-grid by the neighbouring gas-fired power station, without waiting for a connection to the grid. Commissioning is scheduled to begin in 2028.
Tested over a seven-year period at the Canada Glacier, a GeoAI framework developed by the University of Wollongong combines field surveys, drone imagery and satellite data to monitor polar vegetation.
The agreement with Constellation covers 690 MW and supports $3 billion worth of works, as well as a 20-year extension of the two reactors at Maryland’s only nuclear power station.
Using CASPER, researchers have shown that the error in a downscaling model depends on the climatic discrepancy with its training data far more than on the volume of that data.
The penalty imposed on DataOne in Vineland highlights the downside of on-site electricity generation by AI data centres: air pollution and noise affecting local residents.
Across 47 buoys and 37 years of measurements, LSTM, PatchTST, Mamba2 and the like are all on a par, and perform worse than simple persistence for the top 1 per cent of the most extreme waves.
Trained on 209 years of simulation data, it predicts ocean currents more accurately than any of the benchmark models, but not temperature or salinity, which depend on the atmosphere.
Tornadoes cause an average of more than 50 deaths and $100 million in damage per year in the United States. The network produces its probabilistic forecasts much faster than a physical model.
The data set harmonizes satellite observations of fire, atmospheric reanalysis, vegetation, fuel and topography, with a progression of the front observed every twelve hours.