Environmental factor assisted chlorophyll-a prediction and water quality eutrophication grade classification: a comparative analysis of multiple hybrid models based on a SVM

In order to meet people's increasing demands for a better life and beautiful ecological environment, efforts to control environmental pollution have increased significantly. Eutrophication has become one of the greatest threats to aquatic environments and chlorophyll-a (Chl-a) has been proved to be a direct indicator for early warning of algal blooms. In this study, four SVM-based hybrid models integrating a support vector machine (SVM) with intelligent optimization algorithms are proposed and compared to understand the relationship between Chl-a prediction, water quality eutrophication grade classification and environmental factors. The intelligent optimization algorithms include the particle swarm optimization (PSO), artificial bee colony (ABC), cuckoo search (CS), and grey wolf optimizer (GWO). A dataset which contains eight environment factors over a 3.5-year period (from January 2013 to June 2016) was collected from the Pengxi River basin, a typical tributary of the Yangtze River, China. Multiple evaluation methods including the correlation coefficient (R), root mean square error (RMSE) and mean absolute percentage error (MAPE) were used to evaluate the performance of the SVM-based hybrid models. The results from low to high are PSO-SVM, ABC-SVM, CS-SVM, and GWO-SVM respectively. At the same time, the method of mean impact value (MIV) was used to investigate the relative importance of each input environmental factor to the SVM-based hybrid models and the results were pH (0.1974), WT (0.0366), EC (0.0212), CODMn (0.0130), NH3-N (0.0889), TN (0.1841), TP (0.1005), and SR (0.0184). This indicates that four environmental factors (pH, TN, TP, NH3-N) play a major role in the prediction models. This research aims to realize the early warning of algal blooms and water quality eutrophication grade classification, and finally provide a reliable solution for the improvement of water quality.

Automated summary

Efforts to control environmental pollution have increased significantly to meet people's demands for a better life and ecological environment. This study proposed and compared four SVM-based hybrid models with intelligent optimization algorithms to understand the relationship between Chl-a prediction and water quality eutrophication grade classification. Results showed that pH, TN, TP, and NH3-N played a major role in the prediction models.