Hydrological modeling of River Xiangxi using SWAT2005: A comparison of model parameterizations using station and gridded meteorological observations

Gridded climate data sets are widely used in the analysis, modeling and forecasting of the consequences of climate change. The objective of this study is to compare the impact of different climate datasets (station vs. gridded) on the parameterization of a hydrological model (developed using SWAT2005) of the River Xiangxi, the largest tributary of Yangtze River in the Hubei part of the Three Gorges Reservoir. Climate data used in this study derive from two sources: point daily observations from the Xingshan meteorological station (STN) and gridded (0.5° × 0.5°) monthly observations of the CRU TS3.0 global dataset (CRU) downscaled to daily data using a weather generator. Data from 1970 to 1974 were applied for sensitivity analyses and autocalibration and subsequently validate hindcasts over the period 1976-1986. Despite there being only slight differences in mean annual precipitation (1003 mm vs. 1052 mm) between STN and CRU, the data differ more in their estimates of the number of rain days (136 vs. 112) and wet days standard deviation (11.75 mm vs. 18.49 mm). The mean, maximum and minimum temperatures from CRU are all lower than those from STN. SWAT parameter sensitivity analysis results show slight differences in the relative rank of the most sensitive parameters, with the differences mainly caused by the lower temperature and more intensive rainfall in CRU relative to STN. Autocalibrated parameters showed very similar values, except for the surface runoff lag coefficient which is higher for the CRU dataset compared to that derived from the STN dataset. Statistic results for discharge simulated based on CRU compared rather well with that based on STN CRU as evaluated using the standard statistics of the Nash¿Sutcliffe efficiency, coefficient of determination, and percent error. The sensitivity analysis and autocalibration tool embedded in SWAT2005 is a powerful utility in hydrological modeling of the River Xiangxi, and the CRU dataset appears to be appropriate for application to hydrological modeling in this case, thus providing a good basis for climate change studies.