<p>Flooding is a common yearly disaster that occurs when water covers dry land, leading to immediate damage to infrastructure, deaths, tainted water sources, and the spread of disease. Predictive models, which take into account the intricacies of hydrological systems, are essential for controlling the detrimental effects on human populations and the environment. Research on floods has been transformed by machine learning, which has shown amazing potential in forecasting, mapping, damage assessment, and prediction. Flood prediction may now be automated and streamlined with the help of An Automatic Machine Learning Tool, which represents a paradigm change. It makes it possible for a wider audience—including people without experience in statistics or machine learning—to create dependable and accurate prediction models for a variety of industries by lowering the need for significant human involvement. To further the ongoing conversation about improving the accuracy and timeliness of flood forecasts, this study investigates the relationship between An Automatic Machine Learning Tool and flood prediction. The goal of the research is to develop a practical and effective strategy for lessening the negative effects of flooding events on communities and ecosystems by examining earlier studies and methodologies. By utilizing an innovative and user-friendly interface specifically designed for flood research, An Automatic Machine Learning Tool not only signifies technological development but also has the potential to empower various stakeholders involved in disaster management. The tool is uniquely equipped with the ability to preprocess flood-related rainfall or weather datasets and employ data augmentation, utilizing Generative Adversarial Networks (GANs). This distinctive feature, absent in existing Automatic Machine Learning Tool tools, marks a significant contribution, particularly in the realm of flood prediction. The framework’s comprehensive capabilities extend beyond data augmentation, encompassing statistical analyses, flood forecasting with 9 machine learning algorithms, image segmentation through deep learning, flood image classification using transfer learning, and integration of Geographic Information System (GIS) functionality. In addition to our previous contributions, we have incorporated the emerging trend of machine learning, specifically Explainable AI, to enhance the interpretability of our results. This holistic approach positions the framework as an innovative solution within the Fourth Industrial Revolution, making substantial contributions to climate domain applications and addressing the challenges of climate change. The code is available here: <a href="https://github.com/Shakib1633/FloodWatch-Streamlit">https://github.com/Shakib1633/FloodWatch-Streamlit</a>. A detailed video presentation of the entire web app can be found at the following link: <a href="https://www.youtube.com/watch?v=iT2rulSI0LM">https://www.youtube.com/watch?v=iT2rulSI0LM</a>.</p>

FloodWatch: An Automatic Machine Learning Tool for Flood Forecasting and Segmentation using Weather Data and Images

期刊论文
发表日期：

2026年2月16日

查看全文

Tasnim Ullah Shakib,
Tariq Hasan Rizu,
Ellora Yasi,
Nusrat Sharmin,
Rubyeat Islam

摘要

Flooding is a common yearly disaster that occurs when water covers dry land, leading to immediate damage to infrastructure, deaths, tainted water sources, and the spread of disease. Predictive models, which take into account the intricacies of hydrological systems, are essential for controlling the detrimental effects on human populations and the environment. Research on floods has been transformed by machine learning, which has shown amazing potential in forecasting, mapping, damage assessment, and prediction. Flood prediction may now be automated and streamlined with the help of An Automatic Machine Learning Tool, which represents a paradigm change. It makes it possible for a wider audience—including people without experience in statistics or machine learning—to create dependable and accurate prediction models for a variety of industries by lowering the need for significant human involvement. To further the ongoing conversation about improving the accuracy and timeliness of flood forecasts, this study investigates the relationship between An Automatic Machine Learning Tool and flood prediction. The goal of the research is to develop a practical and effective strategy for lessening the negative effects of flooding events on communities and ecosystems by examining earlier studies and methodologies. By utilizing an innovative and user-friendly interface specifically designed for flood research, An Automatic Machine Learning Tool not only signifies technological development but also has the potential to empower various stakeholders involved in disaster management. The tool is uniquely equipped with the ability to preprocess flood-related rainfall or weather datasets and employ data augmentation, utilizing Generative Adversarial Networks (GANs). This distinctive feature, absent in existing Automatic Machine Learning Tool tools, marks a significant contribution, particularly in the realm of flood prediction. The framework’s comprehensive capabilities extend beyond data augmentation, encompassing statistical analyses, flood forecasting with 9 machine learning algorithms, image segmentation through deep learning, flood image classification using transfer learning, and integration of Geographic Information System (GIS) functionality. In addition to our previous contributions, we have incorporated the emerging trend of machine learning, specifically Explainable AI, to enhance the interpretability of our results. This holistic approach positions the framework as an innovative solution within the Fourth Industrial Revolution, making substantial contributions to climate domain applications and addressing the challenges of climate change. The code is available here: https://github.com/Shakib1633/FloodWatch-Streamlit. A detailed video presentation of the entire web app can be found at the following link: https://www.youtube.com/watch?v=iT2rulSI0LM.