While every week we read news of extreme climate events, and on many occasions destructive and absolutely unprecedented, meteorological sciences try to improve even more their predictive models to anticipate these natural phenomena and not only control their damage but also save lives.
A new approach that take advantage of the advantages of Artificial Intelligence (AI) in conjunction with information from physics has just opened brand new and promising doors.
This is an experiment that was published in the prestigious magazine Nature and? combines the results of machine learning models from large volumes of information with traditional physics-based methods.
This hybrid model, named NeuralGCMcould be a significant advance in the field of meteorology and climatology.
Developed by researchers from Google DeepMindMIT, Harvard University and the European Weather Forecast Center, NeuralGCM proved to be more accurate in short term weather forecasts (that is, for the next ten days) than traditional systems while offering interesting projections of climate conditions for a decade, with more precision than current extended-range models.
Perhaps the key is in the amount of data it can handle: it is a model that was trained with decades of historical meteorological data from different parts of our planet and with physics-based descriptions of large-scale weather patterns.
This combined methodology distinguishes it from other models based purely on Artificial Intelligence that are being developed these days by companies like NVIDIA and Microsoft.
Despite the dimension of the information used, one of the main advantages of NeuralGCM is its efficiency and portability. While traditional forecasting models require hours of processing on supercomputersthis open source model can run relatively fast on a laptop connected to the Internet.
These promising results, however, are not an indication that we will soon see its implementation: there is still a long way to go before NeuralGCM and its competitors join official systems because it is an extremely sensitive area that moves billions of dollars a year.
The next efforts will be concentrated on proving that their forecasts are reliablewhile those who use them must understand their strengths and weaknesses for each sector.
Thus, while traditional weather and climate models will not be replaced immediately, the integration of AI technologies appears to be inevitable and if carried out with the necessary caution. could revolutionize our understanding and prediction of atmospheric phenomena.
In times where climate change is constantly creating challenges, you can improve our ability to prepare and respond to what is to come.
link sbobet sbobet88 sbobet judi bola online