In order to increase the availability of automatic weather station data, a data quality control algorithm for automatic weather stations based on association rules mining is proposed.Taking the relationship between relative humidity and wind direction and speed changes as the example, some possible association rules among observation items are summarized to guide the installation of the expert knowledge rule base that is suit for different areas and climate characteristics.The goal is to perfect the data quality control system for automatic weather stations and provide more accurate meteorological observation data for weather forecasts and meteorological service.The dependability of the expert knowledge rule base with certain practical value is verified.