Precipitation Nowcasting Based On Blending Weather Radar With Numerical Weather Model And Its Applicaiton

Nowcasting can realize the weather forecast with high quality in the short forecast period in the future.It is possible to extend the forecast period of runoff forecast by couping Nowcasting with hydrological model,which can realize the forecast and early warning of flood forecast.Especially under the climate change,the summer rainfall concentration and extreme events in the northern mountainous areas of China are becoming more and more prominent,which always lead the floods with thecharacter of short prediction period,strong emergency,concentrated peak volume and serious disasters.This paper analyzes and evaluates the performance of precipitation forecast in different forecast periods based on weather radar extrapolation and numerical weather forecast with data assimilation improvement.This study further points out that the fusion of "learn from each other's strengths and complement each other's weaknesses" is the main wat to realize the 0?6h seamless rainfall forecast;and analyzes the weight of the two approaches in the 0-3h forecast period under different fusion methods Heavy change,the change trend of different types of rainfall fusion intersection is discussed,and the response relationship of fusion method with the change of forecast period is expounded.Further,a variety of adjacent forecasts are used as the input of hydrological model to extend the forecast period of hydrological forecast.Radar-rain gauges merging methods have been widely used to produce high quality precipitation with fine spatial resolution by combing the advantages of the rain gauge observation and the radar quantitative precipitation estimation(QPE).Different merging method implies a specific choice on the treatment of radar and rain-gauge data.In order to improve their applicability,significant studies focus on evaluating the performances of the merging methods.In this study a categorization of the radar-rain gauge merging methods is proposed as:(1)Radar bias adjustment category;(2)Radar-rain gauge integration category;and(3)Rain gauge interpolation category for a total of six commonly used merging methods,i.e.,MFB,RIDW,CCok,FBRK,RK and KED.Eight different storm events are chosen from semi-humid and semi-arid areas of Northern China to test the performance of the six methods.Based on the leave-one-out cross validation(LOOCV),conclusions are obtained that the integration category always performs the best,the bias adjustment category performs the worst and the interpolation category ranks between them.The quality of the merging products can be a function of the merging method that is affected by both the quality of radar QPE,and the ability of the rain gauge to capture small-scale rainfall features.Secondly,in order to solve the problem that rainfall in mountain area is easily affected by terrain,PBN algorithm which can realize different height stratification and pixel grid tracking extrapolation,is selected to carry out radar extrapolation and proximity prediction of mountain area weather,with an input which is modified by radar QPE of rainfall station based on ked method.The evaluation of the two-dimensional evaluation index system show that PBN algorithm has the best performance in the rainfall field with uniform time distribution,especially in the one hour prediction period;for the field with uneven time distribution,the performance is general,and its effective prediction period is not more than three hours.At the same time,the WRF model based on different physical parameterization combination scheme is used to simulate 8 rainfalls,to explore the optimal physical parameterization scheme,and to assimilate data from different sources relying on wrf-3dvar.The results show that the scheme of assimilating GTS data in outer region and radar emissivity data in inner region is the best.Blending Nowcasting can be regarded as a post-processing step of prediction information.The purpose of this blending is to obtain a seamless 0-6h proximity prediction by extrapolating PBN and WRF prediction.Its "seamless" is usually used to represent consistent prediction,rather than considering the location,lead time or prediction program.In this research,two blending methods are set up:dynamic weight time effect matching method based on prediction effect and recognizer method based on prediction intensity of extrapolation revision model.The main difference between two methods lies in the change of fusion weight brought by different blending methods.In this paper,the performance of the two blending methods in the weight trend change,intersection change and 0?3H forecast period of 8 rainfall events is studied.The results show that the weight change of the blending method is a comprehensive combination of the extrapolation prediction and model prediction results.In the prediction period,the extrapolation prediction weight changes from large to small,and in the rainfall duration,the initial weight is significant,and the weight decreases with the increase of the duration.The intersections of the two blending methods are all concentrated in the 2-h prediction period,and the intersection time of the time-dependent matching method is significantly delayed by about 0.5 h compared with the recognizer method.The recognizer method uses extrapolation prediction to revise the model prediction,which improves the quality of the model prediction rainfall intensity to a certain extent,and advances the intersection of the blending.For the blending effect,for the different types of rainfall fields selected in this study,the time effect matching method of 1H prediction period is always better than the recognizer method,and the higher the extrapolation quality is,the better the performance will be;for 2h prediction period,the two blending methods of spatiotemporal uniform rainfall basically have the same performance,and for the fields of spatiotemporal and spatiotemporal uniformity,the recognizer method is better.With the help of the advantages of model prediction,the method of recognizer is better than that of time-dependent matching,especially in the case of uneven distribution of rainfall types.The high-resolution rainfall data and semi distributed Hebei model are coupled to simulate the runoff data.The parameterized scheme is calibrated.The scheme is applied to the hydrological forecast driven by Nowcasting forecast,and its runoff forecast ability in the 0?3H forecast period is analyzed.The results show that the higher the quality of rainfall data,the higher the accuracy of runoff simulation,and high-resolution rainfall data can better drive the grid Hebei model.The parameterized scheme of runoff simulation is applied to the runoff prediction of a variety of near prediction driven semi distributed Hebei models.The results show that with the increase of the prediction period,the fusion of Nowcating can better achieve the runoff prediction in the three-hour prediction period.