Flood frequency analysis for the upper kalu river basin at Ratnapura, Sri Lanka

Abstract

Kalu River, sustains Sri Lanka's Western province and experiences an annual precipitation of 4000mm to 5000mm. Despite the region's susceptibility to frequent flooding, no previous studies had estimated flood return periods at Ratnapura. This research focused on Flood Frequency Analysis (FFA) in the Ratnapura region of Sri Lanka to estimate the return period of floods, which is crucial for flood management and infrastructure planning. Annual Peak Discharge (APD) data from the Ratnapura station over 44 years were fitted to five probability distributions: Generalized Extreme Value (GEV), Gumbel (GUM), Pearson Type 3 (PT3), Log Normal 3 (LN3), and Generalized Logistic (GLO). Stationarity was assessed using the Mann-Kendall and Levene's tests, indicating no trends in the series. Serial independence was analyzed using Turning Point and Anderson's autocorrelation tests, confirming randomness. L-moments, a parameter estimation method, were utilized, and the GEV distribution was the best fit from the Lmoment ratio diagram. The study estimated a major flood occurrence in Ratnapura approximately once every three years. However, accurate FFA is vital to minimize economic and social impact due to floods and to design resilient infrastructure. Further, this study shows that the GEV distribution is a potential model for regional FFA of Sri Lankan rivers.