A regional atmospheric and oceanic coupled model named Aviation Shortterm Regional Climate Model (ASRegCM) is developed and nested into the shortterm climate prediction model CGCM_1.1 from the National Meteorological Center，China Meteorological Administration, which forms an operational prediction system ASRCPOS for aviation climate prediction. 20-year experiments from 1983 to 2002 were conducted using ASRCPOS. Numerical results show that ASRCPOS can improve prediction accuracy as compared with global scale prediction system of CGCM_1.1, and with earlier results by statistic approach. ASRCPOS makes great progress in anomalous sign prediction accuracy particularly for the spatial continuous variables such as surface temperature. It also raises the prediction accuracy of the anomalous magnitude of spatial discontinuous variables such as precipitation. The present numerical experiments show there are still lots of works to be done to improve the prediction accuracy of ASRCPOS, particularly for the climate durations of the abnormal temperature, and for the forecasts with time scale below one month.