Fig. 1. The meso-scale numerical prediction of atmosphere turbulence. (a) Results comparison between the model of Cerro Paranal station site and the measured \begin{document}$ {C}_{n}^{2} $\end{document} profiles^[33]; (b) WRF modeling system flow chart^[25]; (c) Comparison of WRF and measured data at Gaomeigu

Fig. 2. Prediction results of global spatial distribution of atmospheric coherence length (Fried parameter) ^[50]

Fig. 3. Results of the MetNet model for forecasting precipitation over the United States^[85]

Fig. 4. Atmospheric optical turbulence prediction through deep learning. (a) BP neural network^[87]; (b) AGA-BP neural network^[89]; (c) Results of measured data, Polar WRF prediction, AGA-BP neural network prediction^[89]

Fig. 5. TsVMD-AR model structure diagram^[98]

Fig. 6. Turbulence forecasting results of the TsVMD-AR model and the other schemes^[98]. (a) Observed atmospheric turbulence results (blue), and forecasting results through TsVMD-AR (red), SARIMA (green) and WRF (grey); (b)-(d) 2D histograms of observed atmospheric turbulence result verse forecasting results of TsVMD-AR(b), SARIMA(c) and WRF(d)

Fig. 7. Comparation of evaluation criteria of TsVMD-AR, SARIMA and WRF models^[98]