Landscape evolution model is the basis for quantitative research on landscape evolution, but it involves complex processes and numerous model parameters. In order to alleviate the difficulty of parameter identification and improve the reliability of model parameter setting, identifying the key driving mechanisms and parameters of the landscape evolution model has become the focus of current research. In this study, the Kumarak River Basin in the Tarim River Basin of Xinjiang was selected as the study area, and the landscape evolution model of the study area was established based on the LE-PIHM (Landscape Evolution-Penn State Integrated Hydrologic Model). The parameter sensitivities of the LE-PIHM model were identified by the global sensitivity analysis method, Sobol method. The surface elevation of three typical landforms in the study area was used as the objective function for the sensitivity analysis, and the results showed that the key model parameters for the landscape evolution of the mountainous areas in the study area were the weathering rate of bare bedrock (P0), the fitting coefficient for bedrock weathering equation (α) and the rate of tectonic uplift (U), which were mainly related to the weathering effect of the bedrock and the tectonic uplift effect. Meanwhile, in the plain area of the study area, the model parameters have high sensitivity except for aquifer porosity (ng), and the elevation change in the plain area is mainly controlled by the interaction between the parameters. In the riverine area of the study area, the sensitivity of the model parameters is similar to that in the plain area, but the soil porosity (ns) and aquifer horizontal hydraulic conductivity (KHg) have significantly stronger sensitivities, which are related to surface watergroundwater interactions and aquifer permeability properties, respectively.