Flood forecasting using adaptive network-based fuzzy inference system (ANFIS)

• Main Author: Lum, Marie Jia Ying.
• Other Authors: Chua Hock Chye Lloyd
• Format: Final Year Project
• Language: English
• Published: 2011
• Subjects: DRNTU::Engineering::Environmental engineering; DRNTU::Engineering::Civil engineering::Water resources
• Online Access: http://hdl.handle.net/10356/45023
• Institution: Nanyang Technological University

Results of a study investigating the applicability of Adaptive Networked-based Fuzzy Inference System (ANFIS) to forecast water levels up to a lead time of 5 days for the Lower Mekong River are reported. ANFIS is a black-box model which requires only a set of pre determined inputs and thus eliminates the need for complex hydrological assumptions. In this study, water level data from the mainstream and sub-catchments upstream of Thakhek, Lao People’s Democratic Republic are used as inputs to the ANFIS model to predict the water level at Thakhek for lead times of 1 to 5 days. Various ANFIS models were developed by varying the type and number of inputs. The ANFIS model that gave the best performance was compared to a statistical and a lumped parameter hydrological model, currently adopted for flood forecasting for the Lower the Mekong River. The models were compared on the basis of the Root Mean Square Error and Mean Absolute Error. Error analysis was also performed on different flow regimes as well as degrees of water level change.