# Tunisian Integrated Water Resource Management and Urban Resilience System
A deep learning approach to flood forecasting has been explored in this project as a way to produce flood models that are scalable globally using **RNN** (Recurrent Neural Net) **LSTM** (Long Short-Term Memory) implemented in Python 3.7/Keras.
The project has been deployed on a website equipped with flood mapping visualization and a live user alert system. source
## Project Origin
Fighting climate change is one of the top priorities in our century. After finding out that floods are the most common and wide-reaching type of climate disaster, and realizing that there was a deep lack of innovation in this domain, I decided to undertake this as a science fair project.
## Abstract
> Flooding events not only cause significant damage to infrastructure and human populations but also result in the loss of vast amounts of potentially usable water. Between 1995 and 2015, over 2.3 billion people globally were affected by floods (Wallamacq et al., 2015), highlighting the urgent need for effective management strategies. In Tunisia, where water scarcity poses a persistent challenge, floods exacerbate the crisis by wasting valuable water that could be harnessed for future use.
The Tunisian Integrated Water Resource Management and Urban Resilience System (TIWRMURS) aims to turn the challenges posed by flooding into opportunities by identifying, storing, and optimizing the use of floodwaters. By leveraging satellite data from Sentinel-1 SAR for water body detection, Sentinel-2 for land classification, and SRTM DEM for topographic analysis, coupled with IoT sensors for real-time monitoring, the system identifies flood-prone zones and potential water storage sites. These insights enable the efficient manipulation of floodwaters into reservoirs, recharge zones, or sustainable agricultural systems.
The platform provides actionable data through an interactive dashboard, showcasing flood risk zones, infrastru …