Detecting early warning signals of long-term water supply vulnerability using machine learning

摘要: Abstract Adapting water resources systems to climate change requires identifying hydroclimatic signals that reliably indicate long-term transitions vulnerable system states. While recent studies have classified the conditions under which vulnerability occurs (i.e., scenario discovery), there remains an opportunity extend such methods into a dynamic planning context design and assess early warning signals. This study contributes machine learning approach classifying occurrence of supply over lead times ranging from 0 20 years, using case northern California reservoir system. Results this predicts future vulnerabilities in validation significantly better than random classifier, given balanced set training data. Accuracy decreases at longer times, most influential predictors include monthly averages storage. Dynamic can be used inform monitoring detection changing climate.