The Study Of Component-based Four-dimensional Variational Data Assimilation System And Its Framework Implementation

Initial value is one of the key factors to improve numerical weather forecastingeffect,so data assimilation techniques take the important role in numerical weatherforecasting research. Four-dimensional variational dataassimilation technique (4DVAR)is one of the most effective methods, since it can use observations of different time,different areas, different properties as an whole to calculate initial value that meets theconsistencywithmodels.According to the characteristics of 4DVAR, this thesis analyzes the calculationflow and data streamfor 4DVARsystem.The characteristic of module in the system isobvious and component technique is appropriate for the system. Based on the study ofsoft component technology, this thesis analyzes the requirement of constructing4DVAR system by distilling common features and distinguishing the variablerequirement for 4DVAR system, and establishes the framework model for 4DVARsystem, further more, brings the process for constructing 4DVAR system forward byuse of the software component technology, then designs acomponent base for 4DVARsystem in brief and makes some research about the assembly of component based4DVAR system. In the end, the thesis constructs the framework for component based4DVAR system within ESMF and performs experiments by use of the WeatherResearch and Forecasting(WRF) 3DVAR system. The application demonstrates thatcomponent based 4DVAR system can improve the quality in software reuse byabstractingelements from system and wrapping them as components,thus it is useful toimprovetheefficiencyandqualityinthe softwaredevelopment.