This study introduces a new method to better understand extreme rainfall events by considering that heavy rain often occurs in clusters over time. Using rainfall records from northern Brazil, we developed and tested a statistical approach that improved prediction accuracy compared with existing methods. The results can support more reliable climate risk assessments and help improve planning for disaster prevention and climate change adaptation.