抄録
Climate change is exposing a fundamental limitation of conventional water-security thinking: hydrological risk can no longer be characterized by changes in water supply or hazard magnitude alone, because nonstationarity, compound extremes, system connectivity, and human responses jointly reshape how climatic forcing propagates into water-security outcomes. Yet evidence remains fragmented across hazard dynamics, infrastructure and nature-based adaptation, water-resource buffering, digital intelligence, and governance, leaving unclear how these pathways interact, where their effectiveness is conditional, and why adaptation outcomes diverge across scales and system states. This review addresses these gaps by organizing the evidence along a process-based chain from hydroclimatic forcing and system-state transformation to risk transmission, adaptive buffering, digital decision support, and governance feedbacks. Across this framework, the synthesis shows that climate change alters not only water availability but also the storage, timing, connectivity, and recoverability of water systems, while the resulting risks emerge from interactions among antecedent conditions, infrastructure capacity, ecological buffering, and human responses. Adaptation therefore shifts from isolated hazard protection toward context-dependent portfolios of gray, green, and blue infrastructure, managed aquifer recharge, source diversification, and ecosystem-based measures; however, their effectiveness is constrained by hazard intensity, environmental state, spatial scale, path dependence, and life-cycle trade-offs. Digital approaches can improve prediction and adaptive control through multisource observations, physics-informed learning, digital twins, and optimization, but their value is limited by nonstationarity, out-of-distribution conditions, uncertainty propagation, physical consistency, and operational safety. WEFE linkages and sociohydrological feedbacks further reveal that interventions optimized locally can redistribute risks across sectors, locations, and timescales. The resulting knowledge gap is therefore not a lack of predictive or technological capability, but the absence of a robust framework for linking changing system states to cross-scale risk transmission and adaptation performance under deep uncertainty. We argue that water-security research should move toward process-aware, multiscale, uncertainty-explicit decision frameworks that integrate hydrological processes, adaptive infrastructure, ecosystems, digital intelligence, and institutions, thereby reframing water security as the governance of coupled human–water dynamics rather than the management of individual hazards.
