Multivariate relations of river habitat to water–sediment indexes in the Yangtze Estuary

Abstract

River habitat varies with water–sediment. The relations of river habitat to water–sediment are significant in river ecosystems and water environment management but rarely studied. An index system regarding to river habitat and water–sediment was proposed, in which 14 parameters were monthly monitored at Xuliujing hydrological station of the Yangtze Estuary during 2015 to 2018. The relations of river habitat indexes to water temperature (T), discharge (Q) and suspended sediment concentration (SS) were investigated by using principal component analysis/factor analysis (PCA/FA), discriminant analysis (DA), artificial neural networks (ANNs) and multivariate regression (MR). The correlation coefficient matrix for 14 parameters indicated which indexes were well related to a specific index. Five principal components from PCA contributed to 70.2% of the total variance of all variables so that data dimension reduction was realized. Then five independent factors were obtained through FA, in which river habitat and water–sediment indexes were interrelated. The seasonal relations of river habitat to water–sediment were further developed after DA. Dissolved oxygen (DO), chloride (Cl), conductivity (COND), total nitrogen (TN), total phosphorus (TP), SS, T and Q were the most significant parameters for discriminating differences between the wet season and dry season, which could be applied to explain most of the expected temporal variations in river habitat and water–sediment. SS, T and Q was selected as input variables in the ANNs and MR models for predicting river habitat indexes. The predicted values of river habitat indexes DO, TN, TP, COND and Cl agreed well with the measured based on ANNs and MR models, which have high correlation coefficients ranged from 0.64 to 0.96. The results indicated that river habitat management policies should also consider the hydrological processes.