Water Resources Research Act Program

Details for Project ID 2019DC047B

Development of Streamflow Prediction Model and Software Package for Anacostia River at Non-gauged Locations based on Bayesian Approach

Institute: District of Columbia
Year Established: 2019 Start Date: 2019-03-01 End Date: 2020-02-28
Total Federal Funds: $10,000 Total Non-Federal Funds: $7,000

Principal Investigators: Dr. Amir Shahirinia

Abstract: Streamflow is an important component of watershed and hydrologic cycle. When a watershed drains upstream to downstream as streamflow, it defines the natural channel. Estimation of streamflow at the gauged location is very important from watershed management including water supply, flood control, irrigation, and hydro power generation etc. The federal agency, USGS, has installed gauges along the rivers and streams at strategic locations to measure the streamflow and current and historical data are available at the measured locations. Streamflow varies over time and space and estimation of streamflow is important from water management viewpoint. Generally a number of towns and cities have been developed along the rivers and streams which serves a primary source of water supply. Over the last few decades, urbanizations has been significant which resulted in significant variations in streamflows and estimation of streamflows at the non-gauged site is very difficult. Similar to other rivers, there are few gauged streamflow stations are located along the Potomac and Anacostia Rivers and Rock creek. To address this estimation of streamflow at the non-gauged locations, a number of researches have attempted streamflow forecasting using different methodologies such as computational intelligence techniques, particularly neural networks, swarm intelligence, chaos theory, artificial immune systems, and wavelets [1]-[31]. Be that as it may, a reality couldn't be disregarded that the greater part of such existing neural network based models have not yet fulfilled analysts and specialists in terms of accuracy until this point, and the speculation ability of these networks needs to be improved. For instance, most researches utilized the feedforward neural systems with Backpropagation calculations. In any case, a basic