Forecasting Salinity in the Laguna Madre Using Deep Learning

Author

Martin J. Flores Jr., The University of Texas Rio Grande Valley

Date of Award

5-2022

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Civil Engineering

First Advisor

Dr. Jungseok Ho

Second Advisor

Dr. Dongchul Kim

Third Advisor

Dr. Fatemeh Nazari

Abstract

Salinity is an important metric in the Laguna Madre for establishing the long term health of the local ecological population. By utilizing Deep Learning (DL) techniques, the predicted and forecasted salinity in the Laguna Madre is generated from data provided by the Moderate Resolution Imaging Spectroradiometer (MODIS)-Aqua satellite.

Currently, only one other DL model has been used to forecast Sea Surface Salinity (SSS), being a Recurrent Neural Network (RNN). However, the RNN model requires the prediction of a full area of salinity to function.

As such, several model architectures were tested, with the best one, being a Multi-input MPNN, utilized to evaluate the feasibility of forecasting utilizing simpler DL models. The results show that a one-day forecast is plausible, while three and five-day forecasts would require a data-rich environment, unlike that of the Laguna Madre.

Comments

Copyright 2022 Martin J. Flores Jr. All Rights Reserved.

https://go.openathens.net/redirector/utrgv.edu?url=https://www.proquest.com/dissertations-theses/forecasting-salinity-laguna-madre-using-deep/docview/2707899424/se-2?accountid=7119

Recommended Citation

Flores, Martin J. Jr., "Forecasting Salinity in the Laguna Madre Using Deep Learning" (2022). Theses and Dissertations. 1039.
https://scholarworks.utrgv.edu/etd/1039