Document Type : Research Article
Authors
Department of Physical Geography, Faculty of Humanities, Tarbiat Modares University, Tehran, Iran
Abstract
This study investigates the structural evolution and regime shifts of droughts in Iran from 1958 to 2022 using a three-dimensional (3D) clustering framework. The analysis employed high-resolution (~4 km) climate data from the TerraClimate dataset. Drought clusters were identified as continuous spatiotemporal events by applying thresholding to the Self-Calibrating Palmer Drought Severity Index (scPDSI), a widely used measure of drought intensity, and extracting connected components within the data cube. To refine the results, a minimum-size filter was applied, with the optimal value determined through a geometric optimization method based on curvature analysis (cos C method). The findings reveal a substantial transformation in Iran’s drought patterns, particularly after 2000, when short, sporadic episodes gave way to more persistent and spatially extensive drought systems. Among the five drought thresholds tested, the moderate level (scPDSI = -2.0) provided the most balanced trade-off between event frequency and spatial coverage. In parallel, a minimum cluster size of 16 voxels was identified as the geometric filter threshold, effectively reducing noise from small, transient clusters while preserving significant drought events and enhancing structural coherence. Results show that, since 2000, drought clusters have become larger, longer-lasting, and more spatially synchronized. Moreover, regions in central, western, northeastern, and northwestern Iran-previously less affected by severe drought-have emerged as new hotspots, experiencing marked increases in both frequency and duration. By integrating 3D clustering with threshold optimization, this study introduces a methodological innovation in drought research. The approach not only advances scientific understanding of spatiotemporal drought dynamics but also provides practical value for improving drought monitoring, informing adaptation planning, and strengthening early warning systems in Iran and other arid regions.
Introduction
Drought is a complex environmental hazard with profound ecological, agricultural, and social impacts. In Iran’s arid and semi-arid climates, its increasing frequency and severity have heightened risks to water and land resources. Conventional studies often rely on indices such as the Standardized Precipitation Index (SPI) and the Palmer Drought Severity Index (PDSI), but these approaches typically lack the spatiotemporal continuity needed to capture drought dynamics fully.
To address this limitation, we applied a three-dimensional (3D) clustering framework to monthly Self-Calibrating PDSI (scPDSI) data (1958–2022) from the high-resolution TerraClimate dataset. Using connected-component labeling and geometric threshold optimization, we identified and analyzed coherent spatiotemporal drought clusters across Iran. The objective was to evaluate structural patterns and regime shifts in drought behavior, offering a novel contribution to both drought monitoring and adaptation planning.
Material and Methods
We adopted a multi-step observational framework, adapted from Diaz et al. (2024), to analyze drought events in Iran (1958–2022). The process involved: (1) 3D connected-component labeling to identify continuous drought clusters; (2) cluster-size filtering to eliminate insignificant structures; and (3) threshold sensitivity analysis to determine optimal drought intensity levels.
Drought severity was measured using scPDSI derived from TerraClimate (~4 km resolution gridded climate variables). Data were processed in MATLAB by constructing a 3D data cube (latitude–longitude–time) and classifying drought conditions with five thresholds (–0.5 to –4.0). The resulting binary maps enabled the detection, filtering, and analysis of major drought events.
Results and Discussion
Over 7,400 drought clusters were identified across Iran between 1958 and 2022. Cluster number, size, and duration varied strongly with both scPDSI threshold and minimum-size filter. At more lenient thresholds (e.g., –0.5), clusters were abundant but small and short-lived; at more stringent thresholds (e.g., –4), clusters were fewer but spatially extensive and temporally persistent.
Geometric optimization (cos C method, based on curvature analysis) identified an optimal threshold of scPDSI = –2 with a minimum cluster size of 16 voxels. This balance retained hydrologically relevant events while reducing noise. Temporal analysis revealed a rising trend in the percentage of land under drought (PDA) since 2000, with increasingly frequent and spatially synchronized episodes. Moderate droughts (–2) peaked in the late 1960s, early 1980s, and after 2000, while extreme droughts (–4) intensified post-2000. Spatial mapping showed persistent droughts in central, western, and northeastern provinces, with newly emerging hotspots in Golestan, Ardabil, and Razavi Khorasan.
These results suggest a regime shift in Iran’s drought behavior over the past two decades, characterized by events that are more chronic, widespread, and synchronized. Such findings highlight the inadequacy of short-term or localized interventions and emphasize the need for integrated, long-term adaptation strategies.
Conclusion
This study presents a comprehensive structural analysis of drought events in Iran (1958–2022) using a novel 3D clustering framework applied to high-resolution scPDSI data. The approach captured droughts as continuous spatiotemporal entities, independent of administrative or station-based constraints.
Findings reveal a post-2000 regime shift toward longer, more frequent, and more extensive droughts, likely linked to climate change and large-scale ocean–atmosphere variability. Optimal detection was achieved with scPDSI = –2.0 and a cluster-size filter of 16 voxels, which balanced event frequency and reliability. Spatial analysis further identified newly emerging drought hotspots in central, western, and northern provinces, underscoring the need for updated risk maps and adaptive strategies.
The 3D clustering approach offers a transferable tool for advanced drought monitoring in arid regions globally. Future research should integrate climate change projections and link drought clusters to socioeconomic and ecological impacts to strengthen resilience at national and regional scales.
Keywords
Main Subjects
)
Send comment about this article