Abstract
The present study assessed the applicability of convolutional neural network (CNN), which showed superior performance for classification, segmentation, and natural language processing, to river water quality prediction. Monthly data was compiled from upstream and downstream water quality monitoring stations in the Hwang River over the period of January 2007 through December 2020, from which training and test sets were constructed in the ratio of 70:30. The performance of CNN consisting of single and multiple layers were evaluated separately using univariate data with single dependent variable (i.e., either chemical oxygen demand (COD) or chlorophyll-a (Chl-a) as well as multivariate data with dependent and 9 independent variables. The results showed that the prediction accuracy of the proposed CNN algorithm tested with univariate data was noticeably higher for COD than for Chl-a (in terms of target variable) as well as for multiple layers than for single layer (with respect to model architecture). In addition, the CNN algorithm evaluated with multivariate data achieved had better prediction performance than that of univariate data although its performance varied widely among data sets, and to a less extent, among stations and target variables. No measurable difference was also found in prediction performance of the CNN algorithm (for two target dependent variables) according to the number of (important) independent variables. All these results demonstrate that while the proposed CNN algorithm can be adopted to predict (monthly) water quality variables, its careful architecture design is yet required to achieve substantial performance improvement.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
More From: Journal of the Korean Society for Environmental Technology
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.