A Systematic Literature Review on Regression Machine Learning for Urban Flood Hazard Mapping

Abstract

Regression and classification serve as indispensable machine learning (ML) tasks in urban flood hazard mapping. While a machine learning classifier can effectively predict flood hazard, employing a regression model proves instrumental in estimating and predicting flood characteristics within urban areas. This approach reduces the computational time associated with the hydrodynamic models. This paper provides a systematic literature review (SLR) of the current state-of-the-art regression ML models applied in urban flood hazard mapping. A total of 30 articles published between 2018 and 2023 were collected from Five online libraries, i.e., Scopus, Web of Science, IEEE Xplore, Springer, MDPI, and Taylor and Francis. The SLR addresses the raised research questions and reveals that Convolutional Neural Network (CNN) emerged as both the most accurate and frequently employed model. Additionally, the SLR extends its scope to investigate aspects such as flood type, the input data, and the spatial scale of the study area.