Flooding is a devastating natural disaster across the globe whose frequency and impacts have increased over the past few decades. In the United States of America (U.S.) flood is the most prevailing natural disaster, costing about 4.6 billion USD and claiming about 18 lives per event on average. The effects of floods are even worse when evacuation is not done in time due to flash floods that usually occur without conceivable warning. The degree of flood hazards have heightened the need for more accurate flood prediction and simulation models. This study seeks to develop a water level forecasting system for the Window Cliffs State Natural Area, TN. Based on this forecasting system, flood warnings can be issued ahead of time. Machine learning techniques shall be used with available datasets to generate an operational forecasting system. Datasets that can be used in building such a system include upstream water levels, precipitation, antecedent dry periods, temperature, wind direction, and characterized convective weather systems. While some data may not be readily available, the system shall be designed to utilize as minimal data as possible to generate a reliable prediction output. A successful implementation of this flood forecasting system at the Window Cliffs State Natural Area will provide grounds to upscale the design into larger domains.
Files coming soon.