Development and Evaluation of a Principal Component-Based Composite Drought Index Considering Temporal Lag Dependencies Among Indices

This study introduces a composite drought index (CDI) that integrates multiple drought indices, including the Simplified Standardized Precipitation Index (SSPI), Simplified Standardized Precipitation-Evapotranspiration Index (SSPEI), soil moisture measured at depths of 0–10 cm (SM1) and 10–40 cm (SM2), Normalized Difference Vegetation Index (NDVI), and Vegetation Health Index (VHI), using principal component analysis (PCA). Data for Precipitation, temperature, SM1, SM2, NDVI, and VHI were re-gridded to a spatial resolution of 0.25° × 0.25° and used to compute SSPI and SSPEI over 3-, 6-, 9-, and 12-month timescales for grid points across Iran. SM1, SM2, NDVI, and VHI were similarly aggregated at these timescales and standardized using Box-Cox transformation. To facilitate PCA, the temporal lag dependency was adjusted to align all indices with SSPI as the primary reference, eliminating lag correlations. The analysis revealed a strong correlation between SSPI and SSPEI (r > 0.8) across most grids and timescales, alongside significant but weaker correlations with SM1 and SM2 (r > 0.5), VHI (r > 0.6), and NDVI (r > 0.4). The first principal component (PC1), representing the CDI, captured the majority of variance in the data matrix. Additional PCs explaining over 10% of the variance were combined to form a weighted version of the index (CDIw). While CDI showed the strongest correlation with SSPI and SSPEI, CDIw exhibited greater correlations with SM1, SM2, VHI, and NDVI, though with a slight reduction in its relationship with SSPI and SSPEI. Both CDI and CDIw demonstrated strong correlations with the Palmer Drought Severity Index, confirming their effectiveness in monitoring drought conditions in the study area.