Date of Award

2020-01-01

Degree Name

Master of Science

Department

Civil Engineering

Advisor(s)

Ivonne Santiago

Second Advisor

Saurav Kumar

Abstract

As land use around bodies of water changes, the need to model the body of water increases. Models help to educate, understand, and predict the state of water. Process-based models are commonly used in modelling bodies of water, but there are challenges with these kinds of models. They require data which can be difficult for certain communities to obtain due to logistics or cost, are computationally intensive, technically complicated, and require calibration. In contrast, a data-driven model simply connect relationships from the data, are not as computationally intensive nor technically complicated, and do not require calibration. This research compared a data-driven model with a process-based model to verify if a data-driven model is a viable alternative to process-based model using the same sets of data. The research also attempted to find a relationship between water quality data, hydrological data, meteorological data, and remote sensing data in the form of electromagnetic radiation obtained by satellites Landsat 5 and Landsat 7. The study area for this research was in Occoquan Reservoir over a five-year period (2008-2012). A long short-term memory neural network model was developed and fed with data. The results of the model were then compared with the results from a CE-QUAL-W2 analysis. The comparison suggested that a data-driven model cannot be used as an alternative to a process-based model. Further research is required as the data used had multiple gaps which affected the performance of the data-driven model. Optimal data for future research should have high frequency of sampling, less censored data, and electromagnetic radiation readings obtained from an unmanned aerial vehicle as opposed to a satellite.

Language

en

Provenance

Received from ProQuest

File Size

72 pages

File Format

application/pdf

Rights Holder

Yohtaro Kobayashi

Download

Share

COinS