Abstract: Radar wind observations, such as Doppler weather radar radial velocity and wind profiler data, are critical for initializing numerical weather prediction (NWP) models due to their high spatiotemporal resolution. However, their quantitative contribution to China Meteorological Administration Mesoscale Numerical Weather Prediction Assimilation System (CMA-MESO) hasn't yet been systematically evaluated. To address this gap, analysis sensitivity to observation (ASO) method is adopted for quantitative evaluation. Assimilation results from CMA-MESO system at 8 analysis times throughout 2024 are analyzed, incorporating wind profiler zonal and meridional winds, Doppler weather radar radial winds, aircraft reports, satellite-derived winds, and other conventional observations. Quantitative analysis are conducted through energy norm change in analysis increments, aggregated by variable type, altitude, season and analysis time, and their contributions are evaluated. Results demonstrate that wind observations contribute most significantly to the reduction of analysis error in CMA-MESO assimilation system. Based on these findings, the impact of wind variables from radar and other observation on the analysis innovation is analyzed. Collectively, radar wind observations reduce analysis errors by 23.72%, ranking third behind satellite winds (28.45%) and aircraft reports (28.49%). Temporally, contributions from radar exceed that of aircraft reports at 1800 UTC and 2100 UTC and during autumn seasons. This outcome is attributed to the stable error reduction induced by radar observation (root mean square error no greater than 0.63 m·s^-1 vs. aircraft's 5.46 m·s^-1), which compensates for the reduced availability of aircraft reports during convective periods. The combined contribution of Doppler weather radar and wind profiler to analysis error reduction is most pronounced at 700-800 hPa layer, where their total contribution reaches 4.1 times that of aircraft reports. Within 800-900 hPa layer, Doppler weather radar exhibits the greatest contribution, exceeding that of the other two observation types. Doppler weather radar's per observation contribution is constrained by its single-variable (radial wind) limitation, with high-impact stations concentrated in central-eastern China. In contrast, wind profilers exhibit higher per observation contribution than Doppler weather radar, particularly in eastern coastal regions and Beijing-Tianjin-Hebei Area. The spatial distribution of wind profiler contributions is influenced by both the volume of observations and per observation contribution efficiency, whereas that of Doppler weather radar is predominantly correlated with the volume of observational. It is confirmed that radar wind observations substantially optimize the initial fields of CMA-MESO system through two key advantages: Temporally stable error reduction, and a dominant impact in the mid-to-lower troposphere (700-900 hPa), which is critical for refining NWP initial fields.