Main content

Improved Streamflow and Flood Forecasting using Machine Learning

Show full item record

Title: Improved Streamflow and Flood Forecasting using Machine Learning
Author: Elkurdy, Mostafa
Department: School of Engineering
Program: Engineering
Advisor: Binns, Andrew
Abstract: As climate change leads to increasingly unpredictable flow rates associated with floods, droughts, and dam mismanagement, accurate streamflow forecasting is a critical endeavor within the field of hydrology. With advances in technology, machine learning (ML), and data collection, the future of hydrologic research will likely involve maximizing information extraction from complex observations and data collected throughout our environmental systems to improve forecasting ability of complex environmental variables. Forecasting these variables often requires extensive datasets and significant computation. This thesis aims to overcome these restrictions by applying two promising ML approaches to forecast streamflow using only previously recorded streamflow values. Variational mode decomposition, a data-driven time series decomposition technique, proves to drastically reduce imitation error, a common limitation in similar studies, when used alongside extreme gradient boosting, a newly promising ML model. This approach’s improvement to streamflow forecasting can consequently reduce economic losses associated with floods, droughts and improper dam management.
Date: 2020-12-14
Rights: Attribution 4.0 International
Terms of Use: All items in the Atrium are protected by copyright with all rights reserved unless otherwise indicated.

Files in this item

Files Size Format View
Elkurdy_Mostafa_202101_MASc.pdfuntranslated 2.442Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record

Attribution 4.0 International Except where otherwise noted, this item's license is described as Attribution 4.0 International