A Review on Machine Learning-Based Neural Network Techniques for Flood Prediction

Abstract

Floods are unexpected. A few subjective techniques exist in the literature for the prediction of the danger level of floods caused by natural events. In recent years, with the advancement of technologies and the machine learning (ML) field artificial intelligence (AI), artificial neural networks (ANN), we came across a completely new methodology which started to be used in the technology area and thus this problem was started to be solved by many other different approaches. GIS-based models and ANN have been extensively used in recent years. But there was no study which was comparing the different techniques and their accuracy. In this research, we present a novel SLR(Systematic Literature Review) on the (AI) artificial intelligence-based methodologies which have been published and compare their accuracy. The results of the research outline that hybrid techniques are more accurate in terms of error rates, the average error rate of RF, kNN, SVM, and LDA, as well as MSE of RF, kNN, SVM, and BNN for Level of water forecasting after comparing all of the approaches. This SLR is based on papers ranging from 2015 to 2021 and provides a combination of different algorithms and procedures based on artificial intelligence in the context of how these techniques assist in the early forecasting of floods

A review of flash‐floods management: From hydrological modeling to crisis management

“…Another (maybe the most) interesting feature of machine learning is the possibility of forecasting non-hydrological variables because machine learning models do not necessitate any hypothesis regarding physics (Dtissibe et al, 2020). When addressing crisis mangers, amongst the most popular tools used for flood prediction, are Artificial Neural Networks (ANN), in particular Multilayer Perceptron (MLP), and adaptive neuro-fuzzy inference system (ANFIS) (Dodangeh et al, 2020;Jones et al, 2023;Munawar, Hammad, & Waller, 2021;Munawar, Ullah, et al, 2021;Rasheed et al, 2022).…”