The research and development of an operational objective analysis (OA) system are conducted based on the Cressman linear interpolation method. Several methods for solving the problems encountered in the operation of the OA system are presented: (1) the stations of low creditability are automatically detected, so that their negative effect on the computation of analytic field is minimized; (2) a postprocessing step is conducted on the isopleths so that they agree with observation data, without added or subtracted high/low value centers; (3) an adaptive smooth method is applied to the analytic field so that the smallarea centers can be preserved and meanwhile isopleths are still smooth. The feasibility and validity of the proposed methods are verified by the experimental results, which show that the analytic results of the OA system agree well with observation data, and the smoothness of the analytic isopleths is comparable to that drawn manually by forecasters.