Support vector machine―an alternative to artificial neuron network for water quality forecasting in an agricultural nonpoint source polluted river?

Abstract

Water quality forecasting in agricultural drainage river basins is difficult because of the complicated nonpoint source (NPS) pollution transport processes and river self-purification processes involved in highly nonlinear problems. Artificial neural network (ANN) and support vector model (SVM) were developed to predict total nitrogen (TN) and total phosphorus (TP) concentrations for any location of the river polluted by agricultural NPS pollution in eastern China. River flow, water temperature, flow travel time, rainfall, dissolved oxygen, and upstream TN or TP concentrations were selected as initial inputs of the two models. Monthly, bimonthly, and trimonthly datasets were selected to train the two models, respectively, and the same monthly dataset which had not been used for training was chosen to test the models in order to compare their generalization performance. Trial and error analysis and genetic algorisms (GA) were employed to optimize the parameters of ANN and SVM models, respectively. The results indicated that the proposed SVM models performed better generalization ability due to avoiding the occurrence of overtraining and optimizing fewer parameters based on structural risk minimization (SRM) principle. Furthermore, both TN and TP SVM models trained by trimonthly datasets achieved greater forecasting accuracy than corresponding ANN models. Thus, SVM models will be a powerful alternative method because it is an efficient and economic tool to accurately predict water quality with low risk. The sensitivity analyses of two models indicated that decreasing upstream input concentrations during the dry season and NPS emission along the reach during average or flood season should be an effective way to improve Changle River water quality. If the necessary water quality and hydrology data and even trimonthly data are available, the SVM methodology developed here can easily be applied to other NPS-polluted rivers.

Correcting Systematic Underprediction of Biochemical Oxygen Demand in Support Vector Regression

“…Each ðx i ; O i Þ whose corresponding α i and α Ã i are nonzero and contribute to the SVR function are known as support vectors (SVs). There are various kernel functions, but this paper adopts the radial basis function, which is the most commonly used kernel function and is the best at modeling nonlinear relations (Hsu et al 2003;Liu and Lu 2014). The radial basis function k RBF is defined as…”

“…The MAE measures the mean of the magnitudes of all residual errors between the modeled (y i ) and observed (O i ) values of the dependent variable (BOD). Bias represents the mean of all residual errors (Liu and Lu 2014), indicating the extent to which the modeled values overestimate or underestimate the dependent variable. The R 2 (of regression) measures the percentage of data variability explained by the modeled values and the goodness of fit, ranging 0 to 1 (zero correlation to a perfect match between the modeled and observed values).…”

“…A negative NSE value indicates that the model is predicting worse than than if the mean of observed values were used. The mathematical formulas for the four assessment metrics are not reproduced in this paper because they can be found in the modeling literature (e.g., Liu and Lu 2014).…”

“…Having more replicated extreme-high values above a certain threshold for the SVR algorithm to learn leads to improved training performance but poor generalization ability that subsequently worsens testing performance for the entire BOD range (Liu and Lu 2014).…”

Estimation of Water Quality Parameters With Data‐Driven Model

“…They observed that this method was effective in predicting water quality categories that are based on the US Salinity Laboratory diagram. Liu and Lu (2014) predicted water quality of rivers by estimating total nitrogen and total phosphorus using SVM and NN approaches and observed that SVMs provided higher accuracy results.…”