Analyzing Drought Vulnerability with Clustering: A Study of Southeast Türkiye Using Multiple Drought Indices

Abstract Droughts are complex and costly natural hazards with significant impacts on water resources, agriculture, and ecosystems. This study investigates drought patterns in Southeast Türkiye by clustering regional drought behaviour using four widely applied indices: the Standardized Precipitation Index (SPI), Standardized Precipitation Evapotranspiration Index (SPEI), Palmer Drought Severity Index (PDSI), and self-calibrated PDSI (scPDSI). Hierarchical clustering based on Dynamic Time Warping (DTW) and Euclidean Distance (ED), optimized using the Silhouette and Elbow methods, identified two primary clusters across 29 stations for all indices. Cluster validation using the Davies–Bouldin (DBI) and Calinski–Harabasz (CHI) indices shows that DTW generally outperforms ED for indices incorporating precipitation and evapotranspiration dynamics. DTW yields lower mean DBI values for SPEI-12 (1.04 vs. 1.08) and scPDSI (1.20 vs. 1.35) and higher CHI scores in five of eight clusters. Stronger separation is observed for scPDSI (CHI ≈ 400 vs. 327–372 under ED) and SPI-12 Cluster 2 (397 vs. 268), while ED performs better for PDSI (mean DBI = 0.99 vs. 1.21). Sen’s slope and Mann–Kendall analyses reveal significant drying trends across all indices and clusters. Cluster 2, largely representing high-altitude, snow-covered regions, shows steeper drying for SPI-12, SPEI-12, and scPDSI. These areas are increasingly vulnerable due to declining snowpack and a shift in the spring season from March–May to April–June, disrupting water availability and agricultural activities. Overall, the findings highlight the importance of region-specific drought management strategies, particularly for Southeastern Anatolia, where effective water management is vital for sustainability and climate adaptation.