Flood Forecasting in Critical Flood Prone Areas

Flood forecasting models use data from different categories to provide different types of information that contribute to understanding and predicting flood events. Different data categories contribute information on various parameters crucial for flood modeling as well as offering varying spatial and temporal resolutions. These categories include meteorological, hydrological, topographic, land cover and use, geological and geophysical, remote sensing, infrastructure and population, and historical floods data. By incorporating data from different categories, the model can identify critical factors and their interactions that contribute to a flooding event. The inclusion of diverse data enables better validation and calibration of the model. Within this topic, we are investigating different sub-topics of the project namely:

1. Analyzing Anthropogenic Time Series Data to Uncover Human Impacts on Flood Occurrences and Inform Mitigation Strategies

Flooding represents a prevalent and significant natural hazard in numerous regions globally, with projections indicating an anticipated rise in both its frequency and intensity due to climate warming. One of the challenges in flood risk management is to understand how human activities affect the occurrence and magnitude of floods. Anthropogenic factors, such as land use change, urbanization, and deforestation, can alter the natural hydrological cycle and increase the vulnerability of communities to flooding. Therefore, analyzing time series data related to these factors can provide valuable insights into the impact of human activities on flood occurrences and help design effective mitigation strategies. Research in this domain entails a multidisciplinary approach, combining hydrology, climate science, urban planning, and AI to develop comprehensive models that predict and mitigate the adverse effects of flooding on human settlements.

2. Harnessing AI and Data Augmentation to Address Data Scarcity in Flood Modelling

This research project seeks to address these obstacles by investigating innovative strategies that capitalize on a variety of data sources, harness advanced AI methodologies, and employ data augmentation techniques. Through the fusion of remote sensing data, Geographic Information Systems (GIS), and machine learning algorithms, our research seeks to bolster flood prediction capabilities, even in regions where hydrological and meteorological data are scant. AI-powered methodologies will be deployed to combine/enhance existing data, refine integration processes, and validate models to refine flood risk assessments. Furthermore, we will explore data augmentation methodologies to enrich the volume and quality of available datasets, thereby strengthening the robustness and accuracy of our models. Addressing these data-related challenges is pivotal in advancing flood management practices, furnishing data-driven insights that underpin effective disaster preparedness and response strategies. The findings from this research will hopefully showcase, the transformative potential of AI and big data in revolutionizing flood forecasting and management practices, ultimately fostering greater resilience and sustainable development in flood-prone areas.

3. Optimizing AI Algorithms for Flood Prediction in Resource-Constrained Environments

This research explores the enhancement of flood preparedness by integrating accuracy, computational efficiency, and data optimization. By harmonizing these aspects, we endeavor to develop sustainable solutions for effective flood prediction and mitigation. Employing advanced computational techniques, we assess and optimize flood prediction models, focusing on maximizing accuracy while minimizing computational resources and leveraging data optimization strategies. Our study aims to contribute to the development of resilient and sustainable flood preparedness approaches, capable of mitigating the adverse impacts of flooding events on communities and infrastructure.

4. Spatio-temporal analysis of flood risk, vulnerability and severity

Partner: Carnegie Mellon University Africa (opens in new tab)