Design And Forecast Application Of Meteorological Data Analysis System

With the development of meteorological data collection technology and the perfection of the meteorological center algorithm,numerical weather prediction accuracy has been steadily improving.Facing today's society,however,the demand for high precision weather forecast and the weather forecast from various meteorological,center patterns is not able to satisfy the requirement of people.So each region of the meteorological department need to considerate their own circumstance to forecast the weather phenomenon in order to improve forecast accuracy by reference to the superior guidance forecast the future of this area.However,the regional meteorological department is limited by equipment itself,so it is difficult to obtain the local meteorological data which has a guiding significance.According the above reasons,based on Zhejiang province meteorological observatory forecast work of actual investigation and combined with the country issued by quality inspection requirements,"the weather forecast data analysis system" in Zhejiang province is designed and implemented to solve the practical problems encountered in the process of meteorological forecast.And through the analysis of the result data of the system,the weather prediction module is designed and implemented based on neural network method.This module will use weather forecast results as a new forecast model added into the data analysis system to provide more data so as to support the work for weather forecast.The main work content is as follows:1.According to the data requirements on the weather forecast of the Zhejiang province meteorological observatory,with the analysis of the existing meteorological data structure,a meteorological data warehouse is designed,which can achieve high query efficiency in the large time span.And the implementation of the data warehouse make the annual data statistical time reduced to seconds from hundreds of seconds.2.A meteorological data analysis system is designed.The C#,HTML,JavaScript and T-SQL languages are used to develop the data analysis system which owns the ability of mining,quality inspection and auxiliary prediction,along with the characteristic of abundant and intuitive data expression forms.The application of this system can effectively improve the prediction accuracy of forecasters.3.The data mining model is built by neural network to predict the reference data of whether information data.The model is built by BP neural network method.The process is to l earn the historical data to forecast each reference error of the prediction model.And a new reference data as a new reference model for forecasters is coming out according to the results of the prediction.Compared with the data from other models,this kind of reference data can result a large improvement on the forecast accuracy and stability.The numerical forecast analysis system is used in Zhejiang province meteorological observatory practice proved that the method proposed can well meet the needs of numerical forecast team on prediction error control.The reference data calculate by the learning data model not only has the accuracy of forecasts,provides quality inspection functions,but also powerful data support for the numerical forecast team.