Abstract: The rapid update cycle (RUC) based on the rapid update cycle data assimilation system and high resolution meso-scale numerical prediction system is widely applied at home and abroad. Using high frequency update cycle assimilation analysis based on dense meso-scale observations, RUC can provide high quality initial field for high resolution numerical model to produce refined numerical forecasts.Beijing rapid update cycle of assimilation forecast system (BJ-RUC) is established in 2007 and applied to operation in 2008 in China. The refined weather forecast products of BJ-RUC are important reference in daily weather forecast operation for forecasters. BJ-RUC is designed based on WRF model and WRF three-dimensional variational data assimilation system. Recent observations, including conventional observation from GTS global exchange, automatic weather station (AWS) observation and the high spatial and temporal resolution unconventional observations, are assimilated at intervals of three hours using WRF three-dimensional variational data assimilation system to produce a new estimation of the atmospheric state, which will be used as WRF model initial field. And then, the high-resolution WRF model outputs are used for short-range weather forecast.Using hourly surface observations of Beijing AWS from May to September during 2008-2010 the operational forecasts of BJ-RUC for the air temperature, relative humidity, 1 h rainfall, 6 h rainfall, the wind speed and direction are analyzed in details through objective verification methods. The operation performance is evaluated preliminarily based on the analysis.Results show that the surface meteorological elements of BJ-RUC system are well consistent with AWS measurements. The forecast results of 2 m temperature are too high, its error range is-1.5-1.5℃, and namely higher in the morning and night while lower at noon; results of 2 m relative humidity are too small, its error range is-25%-0, and bigger during the day while smaller during the night; forecast results of wind speed are too big, especially obvious in the afternoon, and the error range is 0.6-1.2 m·s^-1; the forecasting performance of the 6 h accumulated rainfall judgment is good, and the TS is 0.4; the stability of BJ-RUC is not good at the beginning of forecast time, and the forecasting performance is unstable with the increasing of the forecast range. In general, the forecasting surface information of BJ-RUC has a superior performance within 12 hours, which is very useful for short range weather forecast.