This study investigates public sentiment during flash floods (Galodo) in West Sumatra by analyzing Twitter data using Natural Language Processing (NLP) via text2data.com. The research applies Latent Dirichlet Allocation (LDA) for topic modeling to identify dominant themes in public discussions. Findings indicate that 97.9% of sentiments expressed were positive, primarily centered on disaster impacts, situational updates, flood causes, and community reactions to government-led disaster management efforts. The study underscores social media’s influence in shaping public discourse during crises. A key contribution of this research is its integration of LDA-based topic modeling with sentiment analysis, specifically targeting Twitter discussions on flash floods in West Sumatra. This methodological approach offers valuable insights into how communities communicate and perceive natural disasters through digital platforms. The results suggest that social media fosters constructive dialogue during environmental emergencies, which can inform crisis communication strategies and enhance disaster response policies. By examining public sentiment and discussion trends, the study highlights the potential of social media analytics for improving disaster management frameworks. The predominance of positive sentiments reflects community resilience and engagement, providing policymakers with data-driven perspectives to optimize emergency responses. This research advances understanding of digital communication patterns during disasters, demonstrating the utility of NLP and topic modeling in crisis-related social media analysis.Ultimately, the findings emphasize the importance of leveraging social media data to gauge public sentiment, enabling more effective disaster communication and policy adaptations in vulnerable regions like West Sumatra.