Modeling of Water Quality Response to Land-use Patterns in Taizi River Basin Based on Partial Least Squares

Abstract

Land use change in river basin is one of the most significant factors that influence river water quality. Based on remote sensing images and river water quality data of Taizi River Basin, partial least squares (PLS) model was applied to explore the relationship between water quality and land use patterns at sub-watershed scale. In this study, PLS model was established by using the percentage of area of land use patterns from seven different sub-watersheds as the variables and the concentrations of four water quality parameters as the observations. The established model was further tested with the data from other sub-watersheds in Taizi River Basin. The results indicated that total nitrogen (TN) and nitrate (NO3-) were significantly associated with land use types in Taizi River Basin with determination coefficient (R2) greater than 0.62. As for chloride (Cl-) and sulfate (SO42-), measured values and fitted values had a relatively low fitting precision (R2<0.5), which meant land use pattern was not the main factor affecting the concentrations of Cl-and SO42- in river water. Meanwhile, river water quality had a strong correlation with the percentage of area of unused land at the sub-basin scale (influence coefficient>0.24), although unused land area in the Taizi river basin was small.