Application of optical flow technique to short-term rainfall forecast for some synoptic patterns in Vietnam

Heavy rains can lead to human, economic, and ecological disasters with large-scale consequences. There are now many precipitation forecasting systems that use radar products with different algorithms and radar image extrapolation techniques to provide very short-range forecasting (up to one to three hours), such as McGill algorithm for precipitation nowcasting by Lagrange extrapolation (MAPLE), short-term ensemble prediction system (STEPS), and short-range warning of intense rainstorms in localized systems (SWIRLS). Most of the above models use radar image extrapolation techniques with various track the radar echo area or rain area. The optical flow can do it, and it is a popular technique that can be easily and efficiently applied in practice. The study objectives are: (i) to apply different optical flow techniques using radar data from available libraries such as Rainymotion and SWIRLS to quantify rain forecast for the next 1 to 6 h, and (ii) to evaluate which optical flow method is the most suitable for Vietnam’s weather conditions and the superiority of the optimal parameter set in the selected one. The study’s results show the following: (i) changing the parameterizations of the ROVER_HKO method using different weather conditions in Vietnam to forecast accumulated rainfall total for the entire rainfall events gives better results than the other methods, and the ROVER_VN method gives the best results when rain is a result of combined synoptic patterns; and (ii) the performance of quantitative precipitation forecasting (QPF) using a performance diagram with light and moderate rainfall thresholds is also better captured in a combined local and global optic flow method than in either type of optical flow.