L’Artificial intelligence seems destined to revolutionize various sectors, including that of forecast.
Apparently, Google is working on a new one AI model capable of carrying out analyzes and consequent weather forecasts in a short time and with impressive precision.
The project is the work of Google DeepMind And his name is GraphCast. According to the Mountain View giant, this service will be able to predict the weather with 10 days’ notice. When compared to today’s weather simulation standards, GraphCast looks set to completely change the way we think about forecasting.
The first and evident test bench, in this sense, was theuragano Lee which hit Canada last September. The model proposed by Google accurately predicted its arrival in Nova Scotia with nine days in advancewhile more traditional weather systems drew the same conclusion just six days earlier.
Weather forecast for 10 days in less than 60 seconds: this is the potential of GraphCast
Even when it comes to data processing speed, GraphCast offers satisfactory results. In less than a minutein fact, the system is capable of develop a 10-day forecast.
The results achieved by AI are the result of the many data that fed its databases. The model, in fact, was able to benefit from forty years of historical data regarding meteorological studies.
Not only that: the databases in question are destined to be enriched with further data over the next few years. This means that GraphCast will become even more powerful and accurate in the medium to long term.
The project conducted by Google DeepMind also includes a philosophy open source. This means that it is possible for scientists and meteorologists to adapt this technology to different contexts. On a practical level, it is believed that the AI model in question could become a formidable weapon for prevention extreme weather phenomena and severely limit its impact on the population.