A novel generalized multi-source weighted ensemble scheme of precipitation simulations under global climate models for improving future drought characterization

Abstract

Drought is a major climatic hazard with severe socioeconomic and environmental consequences, including water scarcity, agricultural losses, ecosystem degradation, and increased wildfire risks. Therefore, continuous and accurate drought monitoring is essential for reducing its adverse impact and supporting climate adaptation policies. In recent years, the use of simulated data of precipitation under global climate models (GCMs) is common in climate monitoring modules. However, inherent uncertainties and biases in individual GCMs necessitate the development of robust ensemble strategies to improve climate projections. Consequently, effective drought monitoring and future risk assessment rely on accurate ensemble of precipitation simulations under GCMs. This study introduces a novel generalized multi-source weighted ensemble scheme-hybrid distance-based weighting scheme (HDWS) to enhance precipitation simulations under GCMs. In HDWS, three statistical factors-1) point-to-point deviation, 2) Mahalanobis distance, and 3) regression analysis-are jointly integrated to derive weights for the ensemble of multiple GCMs in precipitation simulation. Application of HDWS is based on 18 GCMs for the Tibetan Plateau Region of China. The performance of HDWS is assessed against the conventional simple model averaging (SMA) approach using mean absolute error (MAE) and normalized relative absolute error (nRAE). The results of the comparative analysis indicate that HDWS significantly outperforms SMA. Furthermore, the study utilizes the ensemble data to estimate the standardized precipitation index (SPI) and assesses drought characteristics at 1-, 3-, 6-, and 12-month timescales.