Spatial Regression and Prediction of Water Quality in a Watershed with Complex Pollution Sources

• Main Authors: Xiaoying Yang, Qun Liu, Xingzhang Luo, Zheng Zheng
• Format: Article
• Language: English
• Published: Nature Publishing Group 2017-08-01
• Series: Scientific Reports
• Online Access: https://doi.org/10.1038/s41598-017-08254-w

Abstract Fast economic development, burgeoning population growth, and rapid urbanization have led to complex pollution sources contributing to water quality deterioration simultaneously in many developing countries including China. This paper explored the use of spatial regression to evaluate the impacts of watershed characteristics on ambient total nitrogen (TN) concentration in a heavily polluted watershed and make predictions across the region. Regression results have confirmed the substantial impact on TN concentration by a variety of point and non-point pollution sources. In addition, spatial regression has yielded better performance than ordinary regression in predicting TN concentrations. Due to its best performance in cross-validation, the river distance based spatial regression model was used to predict TN concentrations across the watershed. The prediction results have revealed a distinct pattern in the spatial distribution of TN concentrations and identified three critical sub-regions in priority for reducing TN loads. Our study results have indicated that spatial regression could potentially serve as an effective tool to facilitate water pollution control in watersheds under diverse physical and socio-economical conditions.