Tech Desk 
Dec 08, 2024
Google has released GenCast AI, a smart, artificial intelligence-assisted weather forecasting model. It efficiently utilises a diffusion-based approach, specifically tuned for Earth's geometry, to provide precise weather predictions.
Citing Nature, ProPakistani noted that GenCast utilises a diffusion-based approach tailored to Earth's geometry to deliver better results.
In addition, the Alphabet-owned tech titan trained this advanced model on four decades of historical data from the European Centre for Medium-Range Weather Forecasts (ECMWF) archives. GenCast AI is set to outperform traditional systems, including ECMWF’s ensemble (ENS), particularly for forecasts extending beyond 25 days.
GenCast efficiency test
To analyse its efficiency, Google conducted a test by running 1,320 forecasts from 2019 using GenCast, which was trained on data up till 2018. The results were remarkable, as GenCast surpassed ENS in up to 97.2% of cases, with 99.8% precision for predictions surpassing 36 hours.
Google GenCast’s applications expand beyond disaster management, as this powerful model is also capable of predicting wind speeds near wind farms and more, delivering profound insights for reusable energy planning.
Moreover, GenCast AI can effectively generate 15-day forecasts in just eight minutes on Google Cloud TPU v5 in less time.
This breakthrough marks a significant move forward in the evolution of weather forecasting. Google also partnered with different weather agencies and researchers worldwide to improve GenCast.
