This study develops a novel stepwise-clustered climatic factors downscaling method and evaluates the dynamics of temperature, precipitation, and wind speed across 11 stations. An ensemble modeling framework, enhanced through Bayesian Model Averaging, is employed to improve prediction accuracy and reduce uncertainty in long-term projections. Temperature projections reveal a strong upward trend across all cities, with temperate regions showing pronounced seasonal oscillations, while tropical cities display stable patterns. Precipitation exhibits a general upward trend, with time series indicating increasing variability and more frequent high-intensity episodes. Wind speed projections, although less pronounced, show stability in cities like Shanghai and Tokyo, while variability is more evident in Beijing and Bangkok. The ensemble model, generated through Bayesian Model Averaging of four CMIP6 GCMs (ACCESS-CM2, CanESM5, IPSL-CM6A-LR, and MRI-ESM2-0), improves predictions, particularly for temperature and precipitation, by reducing biases and narrowing uncertainty ranges. However, wind speed remains challenging due to its localized and transient nature. Uncertainty analysis shows narrower bands for temperature and precipitation, while wind speed uncertainties are more significant, especially in cities like Shanghai and Hanoi. These findings emphasize the importance of ensemble approaches in improving climate projections and underscore the need for tailored strategies to address regional climate impacts. This research provides critical insights into regional climate dynamics, supporting the development of robust adaptation and mitigation strategies for future climate scenarios.