Non-photosynthetic vegetation (NPV) can significantly impact the magnitude of wind erosion. However, most wind erosion models did not take NPV into account, which led to large uncertainties in wind erosion simulation. To reduce these uncertainties, the effects of NPV on wind erosion should be considered in wind erosion simulations. Herein, we collected the hyperspectral and fractional coverage (fNPV) data of NPV from the Mu Us Sandy Land (MUSL). Through constructing a model between the normalized difference tillage index (NDTI) and fNPV, the fNPV values in the MUSL from 2014 to 2017 were estimated by Landsat 8-OLI images and were used to improve the combined vegetation factor (COG) in the Revised Wind Erosion Equation (RWEQ) model to improve this model. Then, the improved RWEQ model was employed to simulate the wind erosion process of the MUSL during these years. The results showed that the mean values of the fNPV in the MUSL from 2014 to 2017 were approximately 2.71 times higher than those estimated by NDVI data (MOD13Q1). The improved RWEQ that considering the NPV significantly improved the precision of the simulation results, as validated by measured data. Compared with the wind erosion modulus (WEM) without NPV, the decreased values caused by NPV were 130.48 t/km2/a (annum), 91.79 t/km2/a, 85.51 t/km2/a and 93.76 t/km2/a from 2014 to 2017, respectively, and the rates of decrease in wind erosion in the corresponding year were 26.52 %, 16.9 %, 21.47 % and 31.33 %, respectively. We believe that integrating NPV monitoring technology into wind erosion models could significantly improve the accuracy of wind erosion simulation, and this study provides new insight into wind erosion modelling, which would be of interest to scholars in the fields of wind erosion and dust emission.