基于光流法的终端区对流临近预报系统的实现与应用

Implementation and application of convective nowcast system in terminal area based on optical flow method

利用Sparse SD、Sparse、Dense及Dense Rotation 4种光流算法,以乌鲁木齐机场多普勒雷达组合反射率基数据为外推数据源,实现了乌鲁木齐机场终端区组合反射率临近2 h外推预报,并构建了外推产品与闪电、自动气象站、航迹、航路等数据融合显示的WebGIS系统,为战术阶段的航空用户决策提供及时、丰富的临近预报产品。客观检验表明,外推预报的Dense、Dense Rotation算法在前60 min的预报具有较好参考价值,可为战术阶段运行决策、飞机合理绕飞雷雨等危险复杂天气、重要天气复盘等提供数据支撑。

By using four optical flow algorithms,Sparse SD,Sparse,Dense,and Dense Rotation,and using the Doppler radar combined reflectance base data of Urumqi Airport as the extrapolation data source,a 2 h extrapolation prediction of the combined reflectance in the terminal area of Urumqi Airport was achieved.A WebGIS system was constructed to integrate the extrapolation products with lightning,automatic weather stations,tracks,routes,and other data for display,providing timely and rich proximity prediction products for tactical aviation user decision-making.The objective test shows that the Dense and Dense Rotation algorithms used for extrapolation forecasting have good reference value in the first 60 min of forecasting,and can provide data support for tactical operational decision-making,rational flight of aircraft around dangerous and complex weather such as thunderstorms,and important weather reviews.