Evaluation of Various Probability Distributions for Deriving Design Flood Featuring Right-Tail Events in Pakistan

• Main Authors: Muhammad Rizwan, Shenglian Guo, Feng Xiong, Jiabo Yin
• Format: Article
• Language: English
• Published: MDPI AG 2018-11-01
• Series: Water
• Subjects: flood frequency analysis; probability distributions; L-moment; Monte Carlo simulation; right-tail behavior
• Online Access: https://www.mdpi.com/2073-4441/10/11/1603

Design flood estimation is very important for hydraulic structure design, reservoir operation, and water resources management. During the last few decades, severe flash floods have caused substantial human, agricultural, and economic damages in Pakistan during the Monsoon seasons. However, despite phenomenal losses, the flood characteristics are rarely investigated. In this paper, flood frequency analysis (FFA) on four major rivers over Pakistan is performed to probe probability distributions (PDs)at the right-tail flood events. For this purpose, (i) we employed ten different probability distributions associating with an L-moments method for constructing FFA models across Pakistan; (ii) we evaluated the best-fit distribution by using goodness-of-fit test and statistical criteria; and finally; (iii) we devised a Monte Carlo simulation to systematically evaluate the robustness of a selected distribution’s fitting performance by using a synthetic data series of different sizes. Our results indicated that generalized Pareto and Weibull emerged as the most viable options for quantifying hydrological quantiles for most of the river basins in Pakistan. Our main findings would provide rich information as references for flood risk assessment and water resource management in Pakistan.