Research On Precipitation Similarity Medium-term Prediction Based On Meteorological Historical Data

With the development of society and economy and the improvement of people's living quality,the role of medium-term precipitation forecast in daily life is becoming more and more significant.For a long time,people have accumulated a very large and rich historical weather data,which contains many levels of all kinds of situation field and all kinds of ground observation data.The similar forecasting method which makes full use of meteorological historical data to forecast the future weather has gradually become a more and more used weather forecasting scheme in business.In this thesis,a multi-level similarity prediction scheme is proposed.The scheme uses the global historical situation data obtained by the European Meteorological Center as the data source,and carries out similarity matching step by step under the conditions of similar time,similar space and multiple similar factors.Finally,the similarity results are obtained.In terms of similar time,the thesis adopts the time range of year by year and close to date to avoid the historical data of different seasons to participate in the similar calculation.In the aspect of similar space,the thesis designs a three-layer space matching process from large to small and from sparse to dense centering on the key research area.In the aspect of multi-factor screening,the influences of various situation fields including height field,temperature field,humidity field,wind field and sea level pressure field on precipitation formation are mainly considered.Based on the above three aspects of similarity matching,the thesis will optimize the traditional similarity degree of separation algorithm,give a similarity matching algorithm based on continuous degree of separation weighting,and select the best parameters through experiments.Compared with the traditional single-moment similarity dispersion algorithm,which only considers the similarity of static single moment weather state,the continuous deviation weighting algorithm can better represent the similarity of dynamic weather process.In this thesis,a total of 115200 times of historical situation and real data in the past 40 years are used to verify the effectiveness of the continuous dispersion weighting algorithm.On this basis,according to the actual needs of meteorological disaster warning center of Liaoning Province was designed and implemented the middle similar precipitation forecast system,the system mainly includes real-time data monitoring,data processing,similar example query,weather,live the quantitative calculation and drawing system shows six modules.The multi-stage precipitation forecast ability and comprehensive disaster early warning ability of the system are tested and evaluated through practical operation,which provides a reference for the mid-term precipitation forecast of meteorological disaster early warning center in Liaoning Province.