Abstract:Aircraft turbulence poses great danger to safe aircraft operation and affects the comfort of passengers.Therefore,improving the accuracy of aircraft turbulence forecast is of great significance to reducing injury and property loss,and for many years has been the focus of aviation meteorological research.With the enrichment of observation means and progress of science and technology,great progress has been made in the prediction of aircraft turbulence.In this paper,the main methods of aircraft turbulence prediction used throughout the world,especially the latest application of artificial intelligence(AI),are reviewed from the perspectives of both qualitative and quantitative forecast.Then,on this basis,the main problems present in the application of AI to aircraft turbulence prediction and key research directions in the future are summarized as follows:(1)open sharing of aircraft real data,and integration and construction of multi-source turbulence data;(2)interpretability and physical mechanism of aircraft turbulence prediction model built based on AI;and (3)ensemble forecasting of aircraft turbulence based on AI.Finally,the development ideas of the third generation of AI in the field of meteorology are proposed.The future third generation of AI in meteorology may be based on deep learning,integrating mathematical logic reasoning algorithms,and based on “knowledge” (physical mechanisms,experience of forecasters,or basic principles of weather and climatology) and “data” (model data and observation data).That is,it may combine the advantages of the schools of connectionism and symbolism to carry out weather forecasting and climate prediction work.Of course,the third generation of meteorological AI must have characteristics such as higher prediction accuracy,self-learning,and self-adaptation (such as adapting to changes in the rule of meteorological data caused by climate change),and strong explanation in its mechanism.