Development in flood forecasting: A comprehensive review of complex and machine learning models

Abstract

The usefulness of Machine learning (ML) and Artificial intelligence (AI) models in flood prediction is an important field that has recently attracted much attention. ML measures can increase the correctness and timeliness of flood forecasts, which is essential for reducing the damage that floods cause to property and human life. In the last 20 years, ML models have been more helpful to creating more precise and cost-effective prediction systems. Assessment of the ML models' resilience, accuracy, efficacy, and speed: this study reviews the literature on benchmarking ML models for flood prediction. Therefore, it is compared and contrasted many ML algorithms to assess their performance in short- and long-term flood prediction. The study investigates the applicability of each algorithm for different flood prediction scenarios. Furthermore, the work finds ways to enhance the flood prediction model. These trends include hybridization, data decomposition, and model ensemble using satellite-based impact analysis. These models and methods can increase the precision and dependability of flood prediction models, leading to more efficient flood control plans. It provides an extensive overview of different models used for flood prediction, their strengths, limitations, and performance compared with other models. The potential of ML models to minimize the damage associated with floods and reduce the loss of human life focus. Overall, this review article highlights the importance of ML models and methods for flood forecasting and gives functional visions into the deepest learning models and techniques for the correctness of models performance on the flood forecasting.