The radar quantitative precipitation estimation (QPE) technology is applied to gain the high resolution rainfall information at time and space scales, and the latter is taken as the input of hydrological model to improve the precision of flood forecast. The paper takes the Bailianhe Basin as research area, and applies the grouping relationship of Z and I to transform the radar reflectivity into the precipitation, and then uses the Kalman filtering method to adjust the precipitation transformed from radar data using the precipitation detected by rain gauge as the criterion. And then, we input different precipitation data into the Xinanjiang model to make the flood forecast test. The results show that if the radar QPE without revision is taken as the input of hydrological model, the forecast precision is not ideal, but when the radar QPE is revised, the forecast precision gets improvement.