Understanding the Inniscarra reservoir fluctuations to predict downstream flooding through hydrological modeling

Cork, Ireland is prone to flooding; the city has experienced more than 300 major floods over the past two centuries. To mitigate floods in the city, the Inniscarra dam was built between 1953-1957 about 14km upstream from the city. In the past decade, there have been 3 separate flood events costing 90 million, 40 million, and 30 million Euros in damage. To improve flood protection and reduce damage costs, it is necessary to understand the previous flooding scenarios. Water level data in the reservoir upstream has implications for Cork and surrounding areas.

The goal of this project is to develop Soil &Water Assessment Tool (SWAT) and Machin Learning (ML) techniques to understand and predict the downstream river discharge. This will provide insight into how the fluctuations in the upstream reservoir are predictive of downstream river discharge, and thus flooding in Cork City. Algorithms for this project are created by using climate variables as inputs and the algorithm predicts the fluctuations in terms of the discharge forecast as output. Multiple ML algorithms, such as random forest regression, support vector machine, and decision tree regressor, will be created to compare and determine the best predictive output. The discharge outputs from the ML models will be used to compare the results for the historical period with the SWAT model outputs. Overall, this project works to understand the fluctuations of the reservoir in order to predict and mitigate future flooding.