A novel Bi-weight Mid Correlation Coefficient Divergence (BMCCD) approach for multi-model ensemble-based drought assessment

Drought is a complex natural disaster that has persisted for decades. It significantly impacts ecosystems, water resources, and agricultural sustainability. Global climate models (GCMs) are widely recognized as forecasting tools for climate processes. However, variations among the GCMs limit the reliability of individual models. To overcome this limitation, the Multi-Model Ensemble (MME) approach provides a more robust framework compared to single-model analyses. Therefore, the main objective of this study is to introduce a novel “Bi-weight Mid Correlation Coefficient Divergence (BMCCD)” weighting scheme that surpasses existing methods in efficiency and reliability. The performance of BMCCD is compared with the traditional Simple Model Averaging (SMA) and the more recent weighted ensemble (WE) approaches. Results reveal that BMCCD has the highest average correlation value of 0.749 with the referenced data. Moreover, the mean error value of BMCCD, which is 1.332, is the least among all three approaches. The data aggregated using BMCCD was then utilized to develop the Standardized Bi-weight Divergence Index (SBDI), which serves as a key tool in this study. The BMCCD-aggregated data was projected for the period from 2015 to 2100 using linear regression under three different future scenarios. The standardized projected data were analyzed across seven time scales and three Shared Socio-economic Pathways (SSP) to evaluate the long-term characteristics of drought. The results indicate that extreme drought (ED) and extreme wet (EW) events have low probabilities under all SSP scenarios. However, despite their low probability, these high-impact events necessitate attention from policymakers when designing strategies to mitigate future risks.