Application of optical flow radar extrapolation precipitation products for forecasting sudden heavy rainfall
The floods occurring in small and medium-sized rivers in complex hills are strong suddenness and short confluence time.The precipitation short-term proximity prediction product with high accuracy and long-term timeliness is particularly cru-cial for flood forecasting models.Taking the 1000-year return period catastrophic flood of Yahekou Reservoir in Henan Province in September 2021 as an example,the TS score and forecast deviation of radar extrapolated precipitation prediction products based on the improved optical flow method are examined for the 0~1 h and 0~3 h precipitation forecasts during this flood using the three source fusion grid point real precipitation data provided by the National Meteorological Information Center.The results showed that:① The improved optical flow method is more accurate in 0~1 h hourly precipitation forecast.When the accumulated rainfall is below 50 mm,the TS score ranges from 0.45 to 0.85;when the rainfall is between 50 mm and 70 mm,the TS score is between 0.35 and 0.70,when the rainfall is above 70 mm,the TS score is between 0.25 and 0.35.The high TS score of rainfall above 50 mm shows the superiority of the improved optical flow method in extreme heavy rainfall prediction.②In 0~3 h precipi-tation forecast,when the accumulated rainfall is less than 50 mm,the TS score is 0.55 to 0.85,when it is above 50 mm,the TS score is between 0.35 and 0.75.This precipitation forecast product not only has good prediction effect on extreme precipitation,but also has a prediction time up to 3 hours,providing a longer decision-making time for flood control scheduling.③ Compared with the grid point precipitation of the fusion real situation,the precipitation forecast product of the improved optical flow method in 0~3 h is relatively close to the precipitation below 20 mm,and the average absolute error is within 10 mm.The error gradually increases when the rainfall exceeds 20 mm;with the increase of rainfall,the average error,average absolute error and root mean square error increase gradually.④The improved precipitation prediction products of optical flow method in 0~3 h have good pre-diction performance for extreme heavy precipitation of small influence range,large precipitation intensity,long duration and large accumulated precipitation.The research results can provide a more reliable precipitation input forecast for flood forecasting model.
extreme heavy rainfallprecipitation forecast productnowcastoptical flow methodradar extrapolation productYa-hekou Reservoir