Masters Theses

Keywords and Phrases

discharge estimation; LSTM; radar velcoity; stage; streamflow

Abstract

Accurate river discharge estimation is essential for flood forecasting, water resources management, and hydraulic decision-making; however, continuous discharge records are unavailable at many river locations. Traditional stage-discharge rating curves are widely used but their reliability may decrease when channel conditions change or flow conditions vary rapidly. This study develops and evaluates Long Short-Term Memory (LSTM) models for discharge prediction using 15-minute time-series data from river monitoring stations in Missouri. Two model configurations, a baseline stage-only model and an enhanced stage-plus-velocity model, are developed and evaluated independently at two river sites to determine whether the inclusion of surface velocity improves discharge prediction and better captures dynamic flow behavior. Results show that the stage-only model reproduced the general discharge trend but produced larger errors during peak and rapidly changing flow conditions. In contrast, the stage-plus-velocity model more closely matched the observed discharge hydrograph and substantially reduced prediction error, achieving an R² of approximately 0.997 at the Meramec River and 0.976 at the Big Piney River, with much lower RMSE and MAE. A preliminary zero-shot transferability test was conducted by applying the Big Piney trained model directly to the Gasconade River near Vienna, where continuous discharge records are limited; full validation remains a direction for future work. These findings demonstrate that integrating radar-derived surface velocity with LSTM modeling provides a practical and cost-effective approach for improving continuous discharge estimation, particularly at ungagged or under-monitored river locations

Advisor(s)

Corns, Steven

Committee Member(s)

Lendasse, Amaury
Seo, BongChul

Department(s)

Engineering Management and Systems Engineering

Degree Name

M.S. in Engineering Management

Publisher

Missouri University of Science and Technology

Publication Date

Spring 2026

Journal article titles appearing in thesis/dissertation

Paper I, found on pages 5 to 49, to be submitted to Journal of Hydrology.

Pagination

x, 54 pages

Note about bibliography

Includes_bibliographical_references_(pages 51-53)

Rights

© 2026 Barkha Gautam , All Rights Reserved

Document Type

Thesis - Open Access

File Type

text

Language

English

Thesis Number

T 12592

Download

Share

 
COinS