Date of Award
January 2025
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Civil Engineering
First Advisor
YEO H. LIM
Abstract
Streamflow predictions are important in planning water resources and effective hydrological management. Physically based models such as the Hydrologic Engineering Center-Hydrologic Modeling System (HEC-HMS) have been used in the past for this purpose. Still, the use of machine learning models such as Long Short-Term Memory (LSTM) has been proven to complement the use of traditional models. This study compares the reliability performances of HEC-HMS and LSTM models to predict the streamflow of Goose River Watershed in North Dakota. The objective of the research is to perform a comprehensive comparative analysis between the performance of HEC-HMS and LSTM models to predict streamflow in the Goose River Watershed by evaluating the performance of both models in the fall, spring, summer, and winter seasons, and also computing the monthly NSE values for them. Twenty-five years of daily meteorological data from the National Weather Service (NWS) and North Dakota Agricultural Weather Network (NDAWN) and discharge data from the United States Geological Survey (USGS) station were used to conduct this study. The HEC-HMS model was calibrated using the Soil Moisture Accounting loss method and the Clark Unit Hydrograph transform method. Routing was done using the Muskingum method. The LSTM model was pre-processed using hyperparameter tuning, the train-test split ratio was 0.8:0.2, epoch=100, but there was an early-stop function which was added to avoid overfitting. The performance of both models was evaluated using NSE, RMSE, and PBias. The result shows a better performance by the LSTM model with NSE =0.97 and PBIAS = 5.45 for the testing set. The LSTM model could not capture the physically-based processes and unseen events that occur within the watershed, though it captured the peak flow better than the HEC-HMS. The HEC-HMS model provided a robust simulated discharge that is consistent across all seasons and months.
Recommended Citation
Ajao, Adewale, "Comparative Analysis Of LSTM-Based And Traditional Hydrologic Models For River Discharge Prediction: A Case Study Of The Goose River" (2025). Theses and Dissertations. 7089.
https://commons.und.edu/theses/7089