Operational ecoforecasting for coral reefs using artificial intelligence and integrated near real-time environmental data

Abstract

A synthesis of information products about environmental stressors provided in near real-time can serve environmental managers who seek to act decisively before stressors become unmanageable. We have created ecological forecasts, i.e., ecoforecasts, based on input from a variety of environmental sensors that report in near real-time, and we subsequently send those ecoforecasts to environmental managers. The application behind these ecoforecasts is Python-based software that uses an artificial intelligence (AI) inference engine called an expert system. The National Oceanic and Atmospheric Administration (NOAA) Environmental Information Synthesizer (NEIS), formerly the Environmental Information Synthesizer for Expert Systems (EISES), has been developed over two decades to meet the needs of environmental managers and scientists. NEIS integrates environmental data from multiple sources, including in situ and satellite sensors. The application produces ecoforecasts designed to identify environmental conditions conducive to mass coral bleaching and bleaching of specific coral species, as well as other marine environmental events such as algal blooms. This study evaluates the efficacy of coral bleaching ecoforecasts generated by NEIS for the Florida reef tract covering the years 2005–2017.