The work was published in the journal Nature Communications .

Using neural networks, scientists reconstructed historical data on extremely warm and cold days and nights. The method used was called CRAI (Climate Reconstruction Artificial Intelligence). It demonstrated high accuracy in analyzing climate anomalies even for data-poor regions, such as rare weather stations in the polar regions or Africa.

“To test the model, we used archival data covering the period from 1901 to 2018. During the work, rare and poorly documented climate events were identified, such as the heat wave of 1911 or the extremely cold winter of 1929. Previously, these phenomena were mentioned sporadically due to insufficient information from weather stations. CRAI made it possible to reconstruct a complete picture of these events, including their spatial and temporal characteristics,” the study says.

Scientists emphasize that the introduction of AI into climate science can become a key tool for global analysis of extreme events and their changes. Applying CRAI to data helps to significantly improve the accuracy of reconstructions and forecasts, especially for regions with limited observations.

https://www.zakon.kz/tekhno/6458405-iskusstvennyy-intellekt-obnaruzhil-neizvestnye-ekstremalnye-klimaticheskie-yavleniya.html#bounce