In-depth review on machine learning models for long-term flood forecasting

• Main Authors: Khairudin, Nazli Mohd, Mustapha, Norwati, Aris, Teh Noranis Mohd, Zolkepli, Maslina
• Format: Article
• Published: Little Lion Scientific R&D 2022
• Online Access: http://psasir.upm.edu.my/id/eprint/101869/; https://www.jatit.org/volumes/onehundred10.php

Flood is a natural disaster that can cause damage in human life, infrastructure, and socioeconomics. Forecasting the flood is essential to provide sustainable flood risk management for the people. Long-term flood forecasting is very important to provide early knowledge and information for decision maker in minimizing the impact of flood. Early warning can also be disseminated to the potential flood victim and area while proper action can be triggered such as mitigation and evacuation process. The development of long-term flood forecasting model has growing recently with the adoption of machine learning models. It has spark interest among researchers to explore the ability of machine learning characteristics in providing accurate forecasting. Nevertheless, the machine learning models has shown uncertainty and instability in their forecast. The goal of this paper is to provide an understanding and in-depth review of machine learning models in long-term flood forecasting. It includes investigating machine learning models used for long-term flood forecasting and performing comparative assessment in the type of parameters, pre-processing methods and performance measurements used by the models. This review indicates that machine learning models has widely been used involving single and hybrid models for long-term flood forecasting. Various parameters or flood variables have been used as the predictors. The performance of the forecast has been found to be improved through the hybridization of the model. Evaluation of the machine learning models can be done through various performance measurement that prove the models can provide acceptable forecast. The outcome of this study will help future researchers by providing insights of the current progress in the use of machine learning in long-term flood forecasting.