Flooding has emerged as a critical environmental challenge, particularly in rapidly urbanizing regions like Chennai, where the dual pressures of urban expansion and climate change intensify the effects of extreme weather events. This study offers a comprehensive review of the current literature on flood risk assessment, focusing on extreme weather phenomena. We draw upon a variety of methodologies and findings to inform the development of a robust prototype aimed at estimating flood risk probabilities. Our approach involved the systematic collection of flood and non-flood imagery specific to Chennai, employing a web scraping technique that prioritized images categorized by year to maintain temporal relevance. Furthermore, we incorporated temporal sensor data from the ERA5 dataset, specifically tailored to the sub-regional context of Chennai, to enhance the predictive capabilities of our models. Our research investigates both early and late fusion techniques for the integration of these diverse data sources. While early fusion presented notable challenges regarding data compatibility and integration, the late fusion approach yielded superior performance, resulting in a prototype characterized by enhanced predictive accuracy. The findings underscore the efficacy of late fusion in leveraging the strengths of multimodal data, thereby providing a robust framework for flood risk estimation. This discusses the implications of these results for future research and emphasizes the potential of multimodal data fusion to advance methodologies in flood risk assessment, particularly in urban environments susceptible to flooding.