Grassland, as an important part of the ecosystem in the Qinghai-Tibet Plateau, plays an ecological indicator role. However, during the non-growing season, it generally didn’t monitor or observe alpine grass in winter. It could be a great gap to develop the methods of grassland monitoring and its application in winter. PROSAIL, a physical radiation model, can quantitatively describe the relationship between various vegetation parameters and canopy reflectance spectra. In this study, the latest version of the PROSAIL-D model and ground observed data were applied to explore the thresholds of critical range for 10 parameters of withered grass affected by reflectance spectrum. Based on the reflectance spectrums and the corresponding character’s parameters of withered grass that were obtained in the field, 15 000 possible withered grass spectrums were simulated by the PROSAIL model. Compared to the difference of reflectance spectra between withered grass and green grass observed in winter and summer respectively, it is found that a clear difference displayed on visible and near-infrared bands and with a significant linear in 4001 300 nm spectral range for withered grass in winter in Qinghai-Tibet Plateau. On that basis, we proposed a method to distinguish the withered and green grass using the difference between red and green spectral reflections. It can be considered as a withered grass spectrum while the difference is greater than 0. Furthermore, a dataset of potential withered grass spectrum was established by two-steps identification from 15 000 possible spectrums based on the methods described above. The potential withered grass spectrums are correlated closely to the observed spectrums with a whole range of 4002 500 nm, and the R² of all the simulated spectrum lines was between 0.904 and 0.994. By EFAST method and global sensitivity analysis, the brown pigment, carotenoid, anthocyanin, leaf structure and hot spot were identified as non-sensitive factors that respond to the withered grass spectrum. Finally, PROSAIL model was run again in OFAT (One Factor at a Time) with 99% confidence interval as the criterion and cosine distance as the evaluation function. The parameter threshold intervals of the sensitive factors of withered grass are estimated as: leaf area index of 0.20.89, chlorophyll content in 01.29 μg·cm^-2, the average leaf angle between 11°90°, equivalent water thickness from 0.000 1 to 0.005 cm, dry matter content within 0.0080.05 g·cm^-2.The results provide some important parameters and further understanding of grass characteristics in winter, and it will strongly promote the application in remote sensing monitoring.