Work

Prediction of Surface Water Supply Sources for the District of Columbia Using Neural Networks Methods

Year: 2025

Type: conference-paper

Abstract: Water availability for municipal and industrial use, irrigation, navigation support, hydropower, and environmental flows is a significant concern in regions throughout the United States.Hence, there is a need to develop analytical tools to assess and help in the long-term planning of the availability of water supply sources.In this paper, we studied the long-term prediction of surface water resources at the Potomac Watershed in the District of Columbia.A predictive model, based on recurrent neural networks, trained with the Levenberg-Marquardt backpropagation learning algorithm is proposed to forecast the runoff discharge using the past runoff discharge and gage height.Using this computational intelligence modeling tool, the impact of discharge and gage height to the longrun discharge forecast accuracy was studied.Our experimental results indicate that the proposed learning algorithm can successfully train the recurrent neural network for the runoff prediction.

DOI: 10.18260/1-2-1153-54285

Authors: Francisco Lourenco, Nian Zhang, Sasan Haghani

Institution: University of the District of Columbia

Language: English

Topic: Hydrological Forecasting Using AI

Subfield: Environmental Engineering

Field: Environmental Science

Domain: Physical Sciences

SDG: Clean water and sanitation

Open Access status: gold

Funders: National Science Foundation, U.S. Geological Survey, Water Resources Research Institute, North Carolina State University

Awards: Targeted Infusion Proposal: Course Development for a 21st Century Smart Grid Workforce, Research Initiation Award: An Intelligent Optimization, Clustering and Classification Framework for High Dimensional, Overlapped Classes, and Imbalanced Data, G8500555404