Deep learning-based prediction of atmospheric turbulence toward satellite-to-ground laser communication

Atmospheric turbulence is one of the key factors that affect the stability and performance of satellite-to-ground laser communication (SGLC). Predicting turbulence could provide a decisive strategy for the SGLC system to ensure communication performance and is thus of great significance. In this Letter, we proposed a hybrid multi-step prediction method for atmospheric turbulence. In the proof-of-concept experiment, we collected Fried parameters (representing turbulence strength) along the SGLC link continuously for more than 3 months at the Miyun satellite ground station, near Beijing, China, and then trained the model for prediction. The favorable experimental results illustrate that the proposal can achieve 4-h prediction of turbulence Fried parameter at a resolution of 10 min, with performance increase of 7.54%, evaluated by mean absolute percentage error (MAPE).