Using Machine Learning Models for Short-Term Prediction of Dissolved Oxygen in a Microtidal Estuary

Aug 5, 2026

Upload by Noah

This paper presents a comprehensive approach for predicting short-term (next two weeks) dissolved oxygen changes in a microtidal estuary using machine learning. The framework integrates historical water sampling, historical and upcoming two-week meteorological variables, and river discharge plus discharge metrics. Implemented for the Neuse River Estuary in North Carolina, it addresses long-term hypoxia-driven habitat degradation. After careful preprocessing and feature selection, the study compares a recurrent neural network with multilayer perceptron, LSTM, gradient boosting, and AutoKeras via sensitivity experiments, achieving R2 values up to 0.99.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages16

Reading Time40 min

AccessFree to download

Introduction

Dissolved oxygen as a key indicator

Hypoxia and forecasting importance

Existing approaches and challenges

Methods

Input predictors and data sources

Data preprocessing and feature selection

Model comparison and sensitivity experiments

Results and Discussion

Predictive performance and station-specific models

Hindcast vs operational forecast considerations

Implications for environmental monitoring and management

water Article Using Machine Learning Models for Short-Term Prediction of Dissolved Oxygen in a Microtidal Estuary Mina Gachloo 1, Qianqian Liu 2,3, *, Yang Song 1, Guozhi Wang 4, Shuhao Zhang 5 and Nathan Hall 6 Citation: Gachloo, M.; Liu, Q.; Song, Y.; Wang, G.; Zhang, S.; Hall, N. Using Machine Learning Models for Short-Term Prediction of Dissolved Oxygen in a Microtidal Estuary. Water 2024, 16, 1998. [https://doi.org/](https://doi.org/)[ ](https://doi.org/)[10.3390/w16141998](10.3390/w16141998) Academic Editor: Christos S. Akratos Received: 17 June 2024 Revised: 11 July 2024 Accepted: 12 July 2024 Published: 15 July 2024 Copyright: © 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license ([https://](https://)[ ](https://)[creativecommons.org/licenses/by/](creativecommons.org/licenses/by/)[ ](creativecommons.org/licenses/by/)[4.0/](4.0/)) . 1 Department of Computer Science, University of North Carolina Wilmington, Wilmington, NC 28403, USA; [mg4265@uncw.edu](mg4265@uncw.edu) (M.G.); [songy@uncw.edu](songy@uncw.edu) (Y.S.) 2 Department of Physics and Physical Oceanography, University of North Carolina Wilmington, Wilmington, NC 28403, USA 3 Center for Marine Science, University of North Carolina Wilmington, Wilmington, NC 28409, USA 4 Department of Information Systems and Management Engineering, Southern University of Science and Technology, Shenzhen 518055, China 5 School of Economics and Management, University of Science and Technology, Beijing 100083, China 6 Institue of Marine Sciences, University of North Carolina Chapel Hill, Morehead City, NC 28557, USA * Correspondence: [liuq@uncw.edu](liuq@uncw.edu) Abstract: This paper presents a comprehensive approach to predicting short-term (for the upcoming 2 weeks) changes in estuarine dissolved oxygen concentrations via machine learning models that integrate historical water sampling, historical and upcoming 2-week meteorological data, and river discharge and discharge metrics. Dissolved oxygen is a critical indicator of ecosystem health, and this approach is implemented for the Neuse River Estuary, North Carolina, U.S.A., which has along history of hypoxia-related habitat degradation. Through meticulous data preprocessing and feature selection, this research evaluates the predictions of dissolved oxygen concentrations by comparing a recurrent neural network with four other models, including a Multilayer Perceptron, Long Short-Term Memory, Gradient Boosting, and AutoKeras, through sensitivity experiments. The input predictors to our prediction models include water temperature, turbidity, chlorophyll-a, aggregated river discharge, and aggregated wind based on eight directions. By emphasizing the most impactful predictors, we streamlined the model-building processes and built a hindcast system from 2015 to 2019 . We found that the recurrent neural network model was most effective in predicting the dissolved oxygen concentrations, with an R2 value of 0 .99 at multiple stations. Different from our machine learning hindcast models that used observed upcoming meteorological and discharge data, an actual forecast system would use forecasted meteorological and discharge data. Therefore, an actual operational forecast may have lower accuracy than the hindcast, as determined by the accuracy of the predicted meteorological and discharge data. Nevertheless, our studies enhance our understanding of the factors influencing dissolved oxygen variability and set the basis for the implementation of a predictive tool for environmental monitoring and management. We also emphasized the importance of building station-specific models to improve the prediction results. Keywords: dissolve oxygen concentrations; Neuse River Estuary; prediction; machine learning models 1. Introduction The escalating threat to ocean water quality

FAQs

What time horizon does the prediction model target for dissolved oxygen?

The approach predicts dissolved oxygen concentration changes for the upcoming two weeks.

Which data sources are used as inputs to the machine learning models?

Inputs combine historical water sampling, historical and upcoming two-week meteorological data, and river discharge along with discharge-related metrics.

How do the models compare, and which one performs best?

The recurrent neural network shows the strongest performance, reaching an R2 value of 0.99 at multiple stations.

Aug 5, 2026

Upload by Noah