Author(s): Swastika Devi; Ziyi Wu; Leonardo Alfonso; Biswa Bhattacharya

Linked Author(s):

Keywords: Flash floods; FFG; Frequentist; Forecasting; Fiji

Abstract: The current Fiji Flash Flood Guidance System utilized by the Fiji Meteorological Service provides binary flood alerts (flood/ no-flood), leading to uncertainties in decision-making. This study introduces a probabilistic Flash Flood Guidance (FFG) approach using a frequentist methodology to address this limitation. The focus is on the Ba Catchment in Fiji, aiming to establish risk classes under varying rainfall and antecedent soil moisture (AMC) conditions. The research pursues three main objectives: a) investigating historical rainfall and flash flood data under different AMC; b) evaluating flash flood occurrence probability, leading to the establishment of risk classes for effective decision-making; c) validating the approach with a case study in the Ba catchment. The study employs analysis of 43 flash flood events during January 2018 to March 2023, emphasizing the Frequentist FFG risk probabilities and their implications. Flood risk is grouped into four levels based on the likelihood of occurrence. Performance is evaluated for 17 events, and results indicate a higher Probability of Detection (93%) for the Frequentist FFG compared to existing FFG system (67%).

DOI: https://doi.org/10.3850/iahr-hic2483430201-136

Year: 2024