Almost half of the U.S. population lives near a coast. Communities, businesses, and infrastructure in these areas are under increasing pressure from the impacts of flooding, climate change, and sea level rise. Hundreds of millions of people depend on coral reefs for food, livelihoods, cultural practices, and economic benefits. Reefs provide habitat for fish and other marine species and protection for valuable coastal infrastructure. Coral reefs are extremely valuable to the economy - annually contributing $10 trillion globally and over $3 billion domestically. Coral reefs are damaged due to changing water temperatures, ocean acidification, pollution, diseases, invasive species, changing weather patterns, and physical impacts from groundings and storms. The world has lost 30 to 50 percent of coral reefs already. Without significant intervention, tropical reef ecosystems could face global collapse by the end of the century. Advancements towards coastal and island community resilience are required for sustainable high-protein seafood security, supporting livelihoods, preserving cultural practices, and a variety of related economic benefits. Given coral reef degradation and species distribution (and biomass) shifts, information and decision support tools are needed for effective responses. Communities and managers (local, state, and federal) need information on historic and projected shifts and losses.

This project will support informed decision-making tasks through the design of an assessment and forecasting system to evaluate reef resilience across multiple species and spatio-temporal scales. Our target deliverable is a web-based tool which integrates Computer Vision AI and Machine Learning (ML), larval connectivity, and physical and biological assessment. The initial decision support includes measuring exposure, sensitivity and adaptive capacity towards quantifying and forecasting coral reef ecosystem resilience.

The proposed work revolves around (1) integration of curated observational and non-traditional data sources and (2) compositions of ML- and AI-based analytical, prediction, and forecasting models. It includes (i) data collation from multiple sources and types to reduce uncertainty, (ii) connectivity-modulated eco-resilience and stressors predictability assessment through ML, (iii) AI-powered computer vision and image classification of coral species and diseases, and (iv) a Bayesian forecasting integration node fed by the outcomes of (i)-(iii). Our decision support products will provide information and tools required for coastal and island communities and managers to make sound decisions and build resilience to climate change.