Abstract: In order to improve the assimilation effect of Fengyun-3 (FY-3) Microwave Humidity Sounder-2 (MWHS-2) data in numerical models over the complex terrain region of the eastern Tibetan Plateau and enhance the initial humidity field and the forecast accuracy, this study conducts sensitivity experiments on surface height detection thresholds using the Weather Research and Forecasting Data Assimilation (WRFDA) assimilation system. The original surface height detection threshold (800–1500 m) for satellite data assimilation was adjusted to 2000–4000 m to retain more MWHS-2 observations. The applicability of MWHS-2 data assimilation over the Tibetan Plateau was then evaluated. The results show that the number of assimilated MWHS-2 observations over the Tibetan Plateau increased by 2–3 times compared with the original scheme after adopting the new surface height threshold. The assimilation of MWHS-2 observations based on Three-Dimensional Variational Assimilation (3Dvar) effectively reduced the bias of initial field, making the simulated brightness temperature closer to observations. The assimilation of MWHS-2 observations adjusted the initial humidity fields in Qinghai and southern Xinjiang in the northern Tibetan Plateau, and the satellite observation data served as a supplementary observation for sparse conventional stations in the Tibetan Plateau. The new scheme reduced relative humidity forecast biases in the mid-to-lower troposphere (below 400 hPa). With the extension of the forecasting time, the data assimilation scheme with a 4000m height threshold exhibited superior performance.