A recent study published in GeoHealth by researchers at Pohang University of Science and Technology reveals how discrepancies between weather forecasts and actual disaster outcomes directly shape public emotional responses. Led by Professor Jonghun Kam and lead author Kiru Kim, the research team utilized artificial intelligence and natural language processing to analyze the psychological impact of forecast errors during Typhoon Khanun, which struck the Korean Peninsula in August 2023. The researchers evaluated the Korean Meteorological Administration predictions against observed rainfall data from 613 stations and processed more than 43,000 public posts from the NAVER Report Talk platform. Their analysis uncovered distinct spatial patterns in forecast accuracy and corresponding public sentiment. In western and metropolitan regions where rainfall was overestimated, online discourse was dominated by anxiety, worry, and fatigue. Conversely, eastern and southeastern areas experiencing heavier actual rainfall than predicted exhibited elevated levels of confusion, embarrassment, and sadness. Approximately fifty-five percent of all analyzed discourse reflected negative emotions, with anxiety and worry ranking as the most prevalent. The study also documented a clear temporal shift in public behavior. Information-seeking activity peaked before the typhoon made landfall, while real-time reporting and personal experience sharing surged during the event. This transition demonstrates a rapid adaptation by citizens from passive information consumers to active data providers as immediate threats materialize. Ultimately, the findings underscore that public emotional well-being is driven less by the disaster itself and more by the psychological gap between anticipated risk and lived experience. The research establishes that forecast accuracy extends beyond meteorological technicalities, functioning as a critical determinant of societal stress and perceived risk. The authors emphasize that transparently communicating forecast uncertainty can mitigate unnecessary panic and strengthen public trust during extreme weather events. By demonstrating how AI-driven discourse analysis can track emotional responses at scale, the study provides a scalable framework for improving risk communication protocols. As climate volatility increases, integrating real-time public sentiment monitoring with meteorological forecasting may prove essential for enhancing disaster preparedness and minimizing psychological distress across vulnerable populations.