Causality-based drought propagation analyses among meteorological drought, hydrologic drought, and water shortage

Droughts propagate through the hydrologic cycle, leading to water deficiencies in various hydro-climate variables, such as rainfall, streamflow, soil moisture, and/or groundwater. Understanding drought propagation characteristics is an essential issue in water resources planning and management. This study aims to detect the causal relationships from meteorological drought to hydrologic drought and how these natural phenomena cause water shortage using CCM (convergent cross mapping). The causal influences among the SPI (standardized precipitation index), SSI (standardized streamflow index), and SWHI (standardized water shortage index) of the Nanhua Reservoir-Jiaxian Weir system located in southern Taiwan are identified based on 1960–2019 records. Since water shortages are influenced by reservoir operation models, three different models, the SOP (standard operating policy), RC (rule-curve-based model), and OPT (optimal hedging model), are considered in this study. The results reveal that clear and strong causality is observed between SPI and SSI for both watersheds. The causality of SSI-SWHI is stronger than that of SPI-SWHI, but both causalities are weaker than that of SPI-SSI. Among the three operation models, the no hedging SOP leads to the weakest causal links of SPI/SSI-SWHI, and the strongest causality is noted for OPT since the optimally derived hedging policy uses future hydrologic information. The CCM-based causal network of drought propagation reveals that the Nanhua Reservoir and Jiaxian Weir are equally important for water supplies since nearly identical causal strengths are observed in both watersheds.