Weather forecasting is important for various sectors, including agriculture, military operations, and aviation, as well as for predicting natural disasters like tornados and cyclones. It relies on predicting the movement of air in the atmosphere, which is characterized by turbulent flows resulting in chaotic eddies of air.
However, accurately predicting this turbulence has remained significantly challenging owing to the lack of data on small-scale turbulent flows, which leads to the introduction of small initial errors. These errors can, in turn, lead to drastic changes in the flow states later, a phenomenon known as the chaotic butterfly effect.
To address the challenge of limited data on small-scale turbulent flows, a data-driven method known as Data Assimilation (DA) has been employed for forecasting. By integrating various sources of information, this approach enables the inference of details about small-scale turbulent eddies from their larger counterparts.