Study On Integration Technology Of Rolling Weather Forecast

With the development of society, the progress of science and technology and the increasing of weather forecast service demands, weather forecast accuracy and the degree of precision is highly demanded. The integrated forecast synthesizes the forecast results of different forecasting methods about the same elements, which can obtain a better forecasting conclusion than the single forecasting method. It forecasts on numerical weather forecasting reliability, it is the important step that weather forecast developes into probability forecast, and it provide more accurate forecasting information.Particle Swarm Optimization(PSO) is a global random search algorithm based on swarm intelligence. It realizes searching for the optimal solution about complex space through collaboration and competition between individuals. Its characteristics include simple concept, easy implementation, high accuracy, fast convergence, strong ability to find better solutions and strong global searching ability on the nonlinear and multimodal problems. It has been widely concerned in scientific research and engineering practice.The thesis according to the weight distribution method of integrated forecast, using a good global optimization characteristics of the PSO optimize its weight factor in order to better concentrate on the advantages of various forecasting models. Taking the weather forecast of Tianjin region as example, the integrated forecasting system which includes the diagram of data import, integrated model generation and integrated result inquiry based on PSO was established and the integration of several forecasting models was realized.Firstly, the background, the current domestic and foreign status, meaning of the system development are introduced in this thesis. Secondly, the integrated forecasting techniques and PSO are specified. In addition,analysised and researched on that PSO was applied to multi-mode integration of weather forecast. Meteorological data are analyzed and interpolation. Thirdly, the requirement analysis of system, database and detailed diagram design are described in the thesis.Besides, development platform and language of the system are presented in brief. Finally, the system is tested and the some aspects of the system to be improved are pointed out. It analyses the integrated results and makes comparison with integrated forecast of Genetic Algorithm(GA).