Advancing Drought Projections and Trend Analysis Through Redundancy‐Controlled Multi‐Model Climate Simulations

Among natural hazards, drought stands out for its complexity and severe impacts on ecosystems and societies. Global Climate Models (GCMs) play a crucial role in projecting future climate variability and associated extremes. However, the presence of biases and inter‐model correlations within GCM outputs often compromises the precision of drought representation. This research proposes a new ensemble weighting method, termed Less Redundant Medially Weighted Aggregation (LRMWA). The procedure of LRMWA is based on the consideration of Mutual Information (MI) and model‐to‐model deviations in the simulated precipitation data. In the application, we used precipitation simulations from 18 global climate models participating in CMIP6, considering three projected future scenarios over the Tibetan Plateau. Results indicate that LRMWA consistently outperforms alternative approaches with lower average RMSE (1.59) and RAE (1.98), reflecting improved error reduction and spatial robustness. Furthermore, the proposed framework extends the drought standardization procedure through weighted probabilistic modeling across multiple time scales, enhancing drought characterization under changing climatic conditions. In addition, steady‐state drought probability statistics demonstrate the prevailing of Normal drought overall. However, mean state probabilities also highlight the increase in drought risk from low to high scenario especially over the longer time scale. In summary, the study contributes to advancing robust weighting schemes for GCMs, improving drought assessment, and supporting informed decision‐making for effective drought management.