Treating coral bleaching as weather: a framework to validate and optimize prediction skill

Abstract

Data and code to analyze the statistical skill of capturing observed bleaching events globally using sea surface temperatures. The analysis here quantifies the hits, misses, false alarms, and correct negatives, and then applies these results to a variety of weather-forecasting metrics to quantify the skill of predicting bleaching. Finally, adjustments to sea surface temperature-based heat stress metrics are evaluated to test if prediction skill can be improved above typical approaches.