Urban flooding in a tropical coastal environment: Rainfall-impact patterns for early warning

Source abstract

Urban flooding disrupts mobility and municipal operations in large coastal cities, yet rainfall information and impact records are often analyzed separately. Fifteen-minute rainfall from 33 rain gauges is integrated with 4868 georeferenced, operationally classified urban flooding occurrences in Rio de Janeiro to characterize spatiotemporal impact patterns and to develop a rainfall-impact pairing framework with preliminary intensity-duration thresholds. Rainfall time series were segmented into independent events using a 6 h minimum dry interval; each event was described by duration and maximum 1 h intensity. Occurrence records were cleaned, assigned to three severity levels based on water depth and traffic disruption, and paired with rainfall using Thiessen polygon influence areas and timing overlap. Results show that moderate-severity road ponding dominates the database (86.6%), while high-severity street flooding accounts for 9.6%. Event persistence increases with severity: pooled low- and moderate-severity events have a median duration of 1.95 h (90th percentile 5.71 h), whereas high-severity events have a median of 2.99 h (90th percentile 8.36 h). Occurrences peak in summer (2330 events) and concentrate in late afternoon and evening during warm seasons, with persistent spatial hotspots and broader footprints for high-severity impacts. Network coverage is strong (median distance to nearest gauge 1.81 km; 90% within 3.59 km), supporting robust pairing at the city scale. The framework delivers evidence on when and where impacts cluster, and provides preliminary station-level intensity-duration thresholds, with antecedent soil moisture conditioning reducing false alarms, as a first step toward locally meaningful early warning criteria.