Abstract
Drought regionalisation remains sensitive to index selection, temporal scale and clustering methodology, especially in hydroclimatically transitional regions. This study introduces a multi‐index multi‐scale meteorological drought regionalisation framework based on long‐term historical precipitation. The new approach starts with calculations of two standardised meteorological drought indices, Standardized Precipitation Index (SPI) and China‐Z Index (CZI), at 1‐, 3‐ and 12‐month time scales. Then, Principal Component Analysis (PCA) is applied for dimensionality reduction, followed by K‐means clustering to delineate homogeneous drought regions. To ensure methodological robustness, cluster validation is performed using Variance Ratio Criterion (VRC). The proposed approach was demonstrated using monthly precipitation data from 28 climatological stations (1946–2023) distributed across Serbia. The results indicated the presence of three statistically stable drought regions largely governed by topography and continentality gradients. Mountainous western and southern Serbia exhibit high cross‐scale stability (> 80% station persistence), whereas lowland northern regions show moderate sensitivity to index selection. SPI‐based regionalisation demonstrates stronger interannual variability representation, while CZI produces smoother regional drought signatures due to distributional assumptions. Crni Vrh and Zlatibor climatological stations in the mountainous regions of southwestern and central Serbia are representative stations with scale‐resistant drought characteristics. The most detailed spatial resolution that identified six distinct homogeneous sub‐zones is produced by the proposed approach across all time scales. The VRC across all index‐scale combinations is used to validate the stability of the identified regions. The proposed stability‐enhanced framework improves the reliability of drought regionalisation and provides a transferable methodology for climate transition zones under non‐stationary hydroclimatic conditions.