Modeling intensity–duration–frequency curves for rainstorms in the cities of sokode and mango, northern Togo

Preventing flood-related natural disasters requires an understanding of climatic factors such as rainfall and, in particular, storm intensities. To address the shortage of flood-risk prevention data for the cities of Sokodé and Mango in northern Togo, an analysis of storm intensity characteristics was carried out. Pluviographic data from the Sokodé and Mango stations were collected over periods of 18 and 16 years, respectively. The various distribution fits were assessed using the chi-square (χ²) and Anderson–Darling tests. The Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) showed that the Gumbel distribution (GEV I) best describes short-duration rainfall of 5 and 10 minutes for Sokodé, and durations shorter than 55 minutes for Mango. Longer-duration rainfall—greater than 40 minutes for Sokodé and 60 minutes for Mango—follows the GEV III distribution. A comparison of the root mean square errors and correlation coefficients of four empirical equations (Montana, Talbot, Keifer & Chu, and Wanielista) showed that the Keifer & Chu formula best describes the maximum average rainfall in the cities of Sokodé and Mango. The Montana formula, commonly used in previous models, produced larger quadratic errors than the other formulas. Accordingly, intensity–duration–frequency (IDF) curves were established for the cities of Sokodé and Mango for return periods of 5, 10, 20, 50, and 100 years.