Abstract: Abstract: Snowmelt runoff is an important component of streamflow in the arid and semi-arid watersheds. It is often simulated by a snowmelt runoff model (SRM), an empirical temperature-index model with 3 main variables, such as air temperature, precipitation and snow cover area (SCA) and some additional deterministic parameters describing the basin characteristics. However, the SRM model only considers the effect of elevation on snowmelt and does not consider the effect of aspect and slope on the mountainous area. In order to introduce this model to the Urumqi River basin located at the northern slopes of Tianshan mountain, Xinjiang where the data acquisition was difficult, in this study, we improved the traditional SRM with topographic factor and validated the reliability of the enhanced snowmelt runoff model in simulating the daily runoff. The topographic factors of aspect and slope were introduced in the traditional SRM model. The degree-day factor in the traditional SRM model was improved by adjusted temperature based on aspect and slope. The adjusted values were obtained from previous studies on relationships between air temperature and aspect. Finally, the enhanced SRM included the aspect, slope and elevation. The Urumqi River basin covers an area of 1 073.64 km2, ranges in elevation from 1 683 to 4 459 m with the average elevation of 3 066 m. Three hydro-meteorological stations were available in this region. The watershed was classified into 5 elevation zones for traditional SRM and further into 14 zones by the aspect and slope for the enhanced SRM. Meteorological and hydrological data were collected daily from 3 hydro-meteorological stations located in the watershed, and the SCA was extracted from satellite images of the moderate resolution imaging spectroradiometer (MODIS).The precipitation data from Daxigou station was used to represent the most area of this region based on the relative high correlation between precipitation and runoff. In addition, the data from Yingxiongqiao and Yuejinqiao stations were also included in the low latitude area since the Daxigou station was located in the high elevation area. The degree-day factor and snow runoff coefficient and rain runoff coefficient were obtained for different elevation zones from literatures and empirical formula. The traditional and enhanced SRM models both were used to simulate the daily snowmelt runoff during the snowmelt season of spring and summer in 2007 with limited hydro-meteorological data. For the model validation, the daily runoff for the spring and summer snowmelt seasons from 2005 to 2006 was selected. The results showed that the enhanced models could well simulate daily snowmelt runoff in the mountainous catchments. After including the aspect and slope in the calculation of snowmelt water and numbers of degree-day, the enhanced SRM considering topographic factors performed better than the traditional SRM. The 3-year average of nash-sutcliffe R2 was increased from 0.77 for the traditional SRM model to 0.80 for the enhanced SRM, and the root mean square error (RMSE) was decreased from 5.7 m3/s for the traditional SRM to 5.35 m3/s for the enhanced SRM. The relative error of the total runoff amount was decreased from 4.17% for the traditional SRM to 2.16% for the enhance SRM. It can be concluded that the enhanced SRM with topographic factors proposed a new method to improve the daily snowmelt runoff simulation with the better performance and has a high potential to simulate snowmelt runoff in an arid mountainous watershed with sparse data.