Rainfall is essential in hydrologic and hydraulic analyses, serving as critical parameter in water resource studies. Hydraulic structures are designed to manage flooding triggered by extreme rainfall events. One common approach to analyze these extreme occurrences is through probability distribution or frequency analysis. This study evaluates various methods of rainfall frequency analysis. Rainfall data was sourced from the Ethiopian Meteorological Agency (EMA), specifically the Addis Ababa Observatory. Before conducting frequency analyses, data quality was assessed for outliers, with findings within acceptable limits. The frequency analysis utilizes four different distribution methods: Gumbel Extreme Value I, Lognormal, Pearson II and Log-Pearson III. Moreover, these distribution methods were fitted using RMC BestFit software to select a method that fits best for the dataset. The fitted distribution methods were also calibrated with non-probability Intensity-Duration-Frequency (IDF) models. Results indicated that while all methods performed satisfactorily, the Gumbel EVI displayed the best balance between model fit and error reduction in this IDF analysis. The study underscores the importance of selecting appropriate statistical methods for accurate rainfall modeling, which is vital for the design and operation of hydraulic structures. Future research could investigate the applicability of these findings in other regions or integrate climate change variables into rainfall frequency analysis for enhanced flood risk management. Additionally, employing advanced techniques, like machine learning algorithms, may improve prediction accuracy and provide deeper understanding of rainfall variability and trends.
Mezen Agza (Tue,) studied this question.