Presenter: Eric Larsen
Authors: Eric Larsen, Som Dutta
Faculty Advisor: Som Dutta
Institution: Utah State University
Machine Learning methods were used to develop an artificial neural network (ANN) to approximate the flow between north and south arms of the Great Salt Lake (GSL). Knowledge of the flow between the arms of the lake is essential to understand the future of the ecology of the GSL. Current flow approximation means are no longer effective for the newly developed West Crack Breach(WCB) as of 2017. This study employed discharge measurement data provided by the United States Geological Survey (USGS) to train the ANN model for discharge between lake arms. This study first concluded the necessary input information for adequate training and fidelity of the discharge measurements. The flow approximation was first developed using a single discharge network and later developed with a dual discharge network. Overall the results of the training concluded that the flow could be approximated with a R squared value of 0.9440 for South to North flow, and a value of 0.8980 for North to South flow using the single discharge network. R squared value increased with the dual output network. The final results of the dual output discharge network using specific conductance as an input produced R squared values of 0.9339 for South to North flow, and 0.9030 for North to South flow.