Abstract: Due to the spatial heterogeneity in watershed and the complexity of hydrologic processes,calibration can become difficult.Compared with manual calibration,autocalibration could increase the efficiency of calibration process and improve the objectivity and creditability of the calibrated results.SWAT(Soil-Water-Assessment-Tool) is a physically based model at watershed scale.An autocalibration-sensitivity analysis procedure was embedded in SWAT version 2005(SWAT2005) to optimize model parameters.The objective of this study was to evaluate the effectiveness of the autocalibration-sensitivity analysis procedure in a meso-scale watershed River xiangxi,the largest tributary of Yangtze River in Hubei part of Three Gorges Reservoir.The calibration period and validation period were 1970-1974 and 1976-1986,respectively.The sensitivity rank of pre-define parameters by SWAT2005 were obtained based on observed monthly flow for 5-year calibration period at Xiangshan gauging station.The parameters of CN2,CANMX,CH_K2,ESCO,SURLAG,BIOMIX and ALPHA_BF were selected for autocalibration.Monthly discharge was simulated based on the optimal parameter sets for both calibration and validation periods,and compared with observed values for effectiveness evaluation.The results showed that slight difference exist between the mean and standard deviation for monthly discharge during calibration and validation period.The Nash-Sutcliffe efficiency(Ens) was 0.70 and 0.77,and the coefficient of determination(R2) was both all 0.80 during calibration and validation period.Overall,the autocalibration-sensitivity analysis procedure in SWAT2005 was useful as evidenced by statistical measure in discharge simulation in River Xiangxi,and could be used for other watershed for hydrologic modeling.

Key words: Hydrologic modeling, Watershed, Autocalibration, SWAT2005, River Xiangxi