Vitória Novelino Bento Gonzaga

Intense rainfall events can lead to floods and related hazards, reinforcing the need for reliable design rainfall estimates. Intensity-Duration-Frequency (IDF) relationships quantify rainfall intensity across durations and return periods, yet systematic comparisons of IDF derivation methods that jointly consider multiple durations and return periods remain scarce. This study compares seven methods: two variants of the traditional procedure (TP.A; TP.B), the single-step procedure (SSP), the robust procedure (RP), the intensity transformation (IT), the third-order polynomial normal transform combined with least squares (TPNT), and the simple generalization procedure (SGP). A 40-year annual-maximum series from a pluviograph station in Porto Alegre, Brazil, was analyzed for 12 durations (5 min–24 h) and 13 return periods (2–100 years). Methods were assessed against observations using duration-specific RMSE and an overall RMSE across all duration-return-period pairs. Performance was strongly duration-dependent: fits were better at very short (5–15 min) and very long (≥ 14 h) durations, whereas short-to-intermediate durations (30 min–8 h) were consistently more challenging to represent within a single formulation. SSP showed the most consistent performance across durations, whereas TP.B showed the weakest performance. TPNT performed well at short durations but deteriorated as duration increased. TP.A showed moderate performance across durations, whereas RP, SGP, and IT had similar performance with minor duration-dependent variations. These results indicate that no single method dominates across all durations, highlighting that method selection should consider the duration range of interest, dataset characteristics, and the intended hydrological application to improve design rainfall estimates.