Wildfire digital twins: a paradigm shift from monitoring systems to intelligent decision infrastructures

Wildfires are becoming more frequent, severe, and complex driven by global warming and climate change, posing escalating threats to ecosystems, infrastructure, public health, and community safety worldwide. Recent advances in Earth observation, artificial intelligence, physics-based modelling, and cyberinfrastructure have accelerated the development of wildfire digital twins (WFDTs), yet existing capabilities remain fragmented across monitoring, fusion, prediction, simulation, and decision support. Unlike prior reviews that primarily address individual aspect of wildfire, this review and vision paper provides an integrated synthesis of WFDTs. We propose a six-layer WFDT architecture, identify six core capabilities spanning situational awareness through adaptive learning, and organise applications across the wildfire lifecycle from prevention and detection to response, smoke management, and post-fire recovery. Building upon these advances, we outline a research roadmap toward autonomous WFDTs through advances in hybrid intelligence, trustworthy AI, uncertainty quantification, decision intelligence, and continual learning. We envision that the defining characteristic of future WFDTs is not increasingly realistic digital representation, but their evolution into a new paradigm that continuously integrates sensing, prediction, uncertainty quantification, optimisation, governance, and human expertise towards a paradigm of intelligent, trustworthy decision infrastructures that continuously support human-centred wildfire decision making.