In order to analyze the atmospheric Rayleigh-Brillouin scattering spectrum and obtain the vertical temperature profile of the atmosphere from 0 km to 16 km, a high spectral resolution temperature LiDAR (HSRL) based on a Fizeau interferometer combined with a photomultiplier tube array was used. The temperature retrieval ability and system error under different seasonal environments were analyzed theoretically. The results of the influence of different atmospheric models on temperature profile fitting was obtained by using atmospheric Rayleigh-Brillouin scattering spectra combined with Tenti S6 model for fitting. The results indicate that, using the US76 atmospheric model combined with the Tenti S6 model, after 10000 pulse integrations, the maximum random error within the range of 0 km to 16 km is 1.1 K. Using the temperature and pressure field provided by weather research & forecast (WRF) model to replace the temperature and pressure field of US76 model for simulation, the random error varies slightly with seasons, not exceeding 0.65 K in spring, 0.98 K in summer, 0.59 K in autumn, and 0.63 K in winter. The evaluation results of detection capability of HSRL system by using WRF mode are better than the evaluation results by using US76. If different weather types are selected, this method can also be extended to evaluate the random error of HSRL in different seasons. The advantages of using the WRF model for HSRL evaluation are reflected. This study provides a reference for evaluating the working ability of the HSRL temperature measurement system in different seasons.