Flood Forecasting via Time Lag Forward Network; Kelantan, Malaysia

International Conference on Advances in Renewable Energy and Technologies 2016

Jajarmizadeh, Milad^1 ; Mohd Sidek, Lariyah^1 ; Basri, Hidayah Bte^1 ; Jaffar, Aminah Shakira^1

Center for Sustainable Technology and Environment, College of Engineering, Universiti Tenaga Nasional, Jalan IKRAM-UNITEN, Kajang, Selangor

43000, Malaysia^1

关键词: Critical issues;    Flood forecasting;    Hidden layers;    Input and outputs;    Mean Square Error (MSE);    Optimum configurations;    Root mean square errors;    Water level forecasting;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/32/1/012043/pdf
DOI  :  10.1088/1755-1315/32/1/012043

来源: IOP

【 摘 要 】

Forecasting water level is one of the critical issues in Malaysia for Kelantan region. Based on the flood events in 2014, this study investigates the hourly-forecasting of water level in one station namely Kg Jenob in Kelantan. For this issue, Time Lag Forward Network (TLFN) is evaluated for forecasting the water level as dynamic model. Heuristic method in stepwise forward methodology is performed. Rainfall and water level are the input and output of the modelling respectively. For selected flood period 15/12/2014 to 30/12/2014, 8 scenarios are developed to obtain a minimum error in water level forecasting. By monitoring the error, it will show that the optimum configuration of network has 2 processors in hidden layer and 7 lags have enough contribution on the result of hourly forecasting. Transfer functions in hidden and output layers are is Tangent hyperbolic and bias. Observed and simulated data are compared with usual error criteria called Mean Square Error (MSE) and Root Mean Square Error (RMSE) which obtained 0.005 and 0.07 respectively. In conclusion, this study will be as a baseline for Kelantan to show that TLFN has promising result to forecast the flood events.