1. Title, Journal and Authors

Title: Improved Prediction of Harmful Algal Blooms in Four Major South Korea’s Rivers Using Deep Learning Models

Journal: Environmental Research and Public Health

Authors: Sangmok Lee and Donghyun Lee*

Department of Business Administration, Korea Polytechnic University

 

2. Summary

This paper is written to predict the chl-a concentration one-week in advance through some water quality models analyzing the time-series data in four major south Korea’s rivers. Contrary to ordinary models, deep learning models can predict future water quality without various kinds of water variables. One week intervals water quality data of reservoirs in four major south Korea’s rivers are utilized. OLS (Ordinary Least Square), MLP (Multilayer Perceptron), RNN (Recurrent Neural Network), LSTM (Long-short Term Memory) are used. Their performances are evaluated by RMSE.

Orders of the model performance are that: LSTM, RNN, MLP, OLS. Although other models have good performance in some points, overall LSTM shows the best performances. It means that LSTM is adaptable to treat the time-series data. For future research, it is needed to modulate the parameters more closely and to retain much more data without missing values.

 

3. Originality and Creativity

- To analyze the time series data, many neural network methodologies are utilized.

 

4. Application to research

- It is suitable to compare the peformance of many neural network when treating time-series data

 

5. Contact

Dae Seong Jeong / Integrated Ph.D. program

 

Environmental Systems Engineering Lab.

School of Earth Sciences and Environmental Engineering

Gwangju Institute of Science and Technology

1 Oryong-dong Buk-gu Gwangju, 500-712, Korea

 

Phone : +82-10-2003-7860

E-mail : jeongds92@gist.ac.kr