ABSTRACT
Floods are one of the most hazardous natural calamities that put people’s lives and property in peril. All scientific and technological sectors now produce vast amounts of data since we are living in the “big data” era. The precision of forecasts for imminent floods is increased by a multidisciplinary analysis involving hydrology, meteorology, and remote sensing. Contrary to the conventional data processing approach, big data has the computational resources to analyze patterns, trends, and associations in data to help make better decisions and offer insightful information. Artificial intelligence/deep learning has the potential to capture hidden associations between climatological/meteorological forcing and hydrological responses. It modernizes not only scientific thinking but also predictive applications. This chapter discusses the potential of deep learning algorithms in the prediction of floods by leveraging the wealth of big data that is now available. Insight into the performance of state-of-the-art approaches like convolutional neural networks, recurrent neural networks, long-short term memory, gated recurrent unit, generative adversarial networks, autoencoders, transformers is also provided. The chapter concludes by highlighting the current limitations and future directions for deep learning-based flood prediction research with big data.
