Research on the correction method of precipitation forecast based on long short memory neural network
Numerical forecasting systems are the current method for precipitation forecasting,but due to the uncertainty of initial values and numerical forecasting itself,there are certain errors in precipitation forecasting.The article is based on the precipitation forecast products of automatic weather station observation data and numerical forecast system output.Establish and train a precipitation correction model using Long Short-Tern Memory(LSTM)algorithm.Experimental results show that this correction method is of great significance for the correction of small rainfall magnitude results in numerical forecasting systems,thereby improving the accuracy of precipitation forecasting and providing strong technical support for meteorological forecasting operations.
neural networkprecipitation correctionnumerical prediction system
NETL
NSTL
万方数据
