DEVELOPING A MULTIVARIATE FLOOD RISK INDEX FOR THE KELANI AND KALU RIVER BASINS IN SRI LANKA USING FUZZY LOGIC

Abstract

In Sri Lanka, rapid urbanization in recent years has intensified flood risks in vulnerable areas. Since assessments based solely on climate change factors may reduce the accuracy of evaluations, this study develops a Multivariate Flood Risk Index (MFRI) using fuzzy logic, integrating both climate-related and land-use factors. As a case study, flood risk is evaluated in three regions—Hanwella, Ratnapura Town, and Kalawellawa-Molkawa—covering the Kelani and Kalu river basins, which experience recurrent annual flooding. The MFRI integrates hazard risk factors, Rainfall Intensity (RI), Rainfall Season (RS), and River Level (RL), which are considered for all three regions. Additionally, Soil Moisture (SM) is included for Hanwella, and Reservoir Dam Status (RD) is included for Kalawellawa-Molkawa. Population Density (PD) and Dependency Ratio (DR) are considered as vulnerability risk factors. The Linguistic Ordered Weighted Averaging (LOWA) operator computes the hazard risk, while the Mamdani Fuzzy Inference System (MFIS) assesses vulnerability. Final flood risk combines both hazard and vulnerability risks using LOWA, with factors categorized as Low, Medium, or High. Three linguistic quantifiers were tested, with the 'mean' quantifier being the most effective. Analysis of five-year data revealed RI and RL as dominant hazard factors in Hanwella and Ratnapura, while RI, RL, and RD were most critical in Kalawellawa-Molkawa. Flood risk outputs indicated that high levels of RI and RL typically produced medium to high risk, while low levels led to consistently low risk. In the rainy season (on-season), even moderate levels of RI or RL were enough to trigger high risk, whereas in the off-season, both needed to exceed medium levels to yield comparable risk. Validation with the Irrigation Department's flood level data confirmed the model’s accuracy. The MFRI offers an adaptable framework for flood risk assessment, aiding targeted mitigation strategies in flood-prone regions.