Two new techniques of dynamical weight blending and trend evolving superimposing have been proposed based on radar extrapolation, coordinating with numerical weather prediction as environment field. In dynamical weight blending, three ways to calculate weight are used: sine weight, hyperbolic tangent weight and realtime scrolling weight. A convective process of Shanghai and surrounding areas on 21 June, 2009 has been used as an example to analyze and examine. The result shows that after blending, the prediction results are both improved compared to radar extrapolation and numerical weather prediction. The blending results of sine weight and hyperbolic tangent weight in dynamical weight blending are more close to observations. It shows that dynamical weight blending and trend evolving superimposing are effective and feasible and of great value to the study of shortterm forecast and nowcasting.