The hydrological model has developed from simple conceptual model into the complex distributed physical model in the long evolution stage. The climate model of meteorological subject has developed fast in recent years, meanwhile, it has improved the development of hydrology. This paper carried out researches from the perspective of meteorology and hydrology development. After analyzing previous studies, we found that although meteorological and hydrological models have reached to a perfect stage in their respective development, the advantage of coupling and learning from each other has not been exploited. On the basis of ex - ploring the development of atmosphere, land surface and hydrology coupling model, this paper chose XJLDAS atmosphere-land surface coupling system, CLM3.5 public land surface model and Rapid confluence model as the key coupling objects respectively. With the above coupling systems, we simulated the runoff process of Xinjiang Jingbo River Basin. Through analysis, it is deemed that XJLDAS+CLM3.5+RAPID model can reappear the annual distribution of land surface runoff well. However, due to the selection of land surface parameterization schemes, there are some biases of runoff in the study area. In the research, furthermore, we found that there exists some differences between runoff simulation results and observed values. As a result, when taking large scale hy - drological simulations, it is necessary to consider improvements on model based on geographic features (such as geological structure, land surface cover and so on), so as to reappear the utmost large scale runoff process