Abstract: Physical models have been developed in the past for simulations of microwave land surface emissivity (MLSE), yet the application of such physical models are very limited.This study focuses on the Land Microwave Emissivity (LandEM) model in the Advanced Radiation Transfer Simulation System (ARMS) and attempts to improve soil dielectric constant and rough surface reflectance models. Corresponding MLSE simulations are conducted and evaluated based on ground MLSE observations. The main input parameters to LandEM are also optimized. Taking the Qinghai-Xizang plateau region as an example, three semi-empirical models of rough surface reflectance and one empirical model of emissivity are selected to form nine MLSE simulation schemes. Using three sets of MLSE products retrieved from satellite observations, the above nine different simulation schemes are evaluated, and the optimal scheme is selected to simulate the MLSE in the Qinghai-Xizang plateau region. The results show that the introduction of Mironov's room temperature soil dielectric constant model to the LandEM model can improv the simulated MLSE. The introduction of the Zhang frozen soil dielectric constant model further compensates for the shortcomings of the original frozen soil dielectric constant estimation. The MLSE simulated by introducing the Chen-Weng rough surface reflectance model has also been effectively improved. For MLSE simulation in the Qinghai-Xizang plateau region, the Qp model is more suitable for bare soil rough surfaces, while the Chen-Weng model is more suitable for vegetation rough surfaces. The improved LandEM model and optimized simulation scheme can effectively simulate horizontal polarization MLSE and its spatial distribution in the Qinghai-Xizang plateau region with a spatial resolution of 0.10° as well as the vertical polarization MLSE. However, the model cannot effectively simulate spatial differences in vertical polarization MLSE.