Temporal-spatial estimation of the water quality variation in the middle and down streams of Han River using a modified ordinary-indicator kriging approach

Abstract

Accurately estimating the spatial variation of water quality is critical and difficult in complex river systems with various sources of pollutants. Water quality of Han River was influenced by land use variation, polluted tributaries, engineering projects, etc. This study detected the spatial variability of typical water quality variables and determined chlorophyll-a (Chl-a) in Han River using a modified indicator-ordinary kriging (IK-OK) approach with multiple thresholds. The 94 water sampling sites distributed from the middle and down streams of Han River were collected in November 2015 (Dry season) and May 2016 (Wet season). The results of sampling and analysis show that the pollution of the downstream Han River is more serious than that of the middle stream Han River in both dry and wet seasons. The nutrient variables were significantly influenced by tributary pollutants and land use variation and meanwhile lead to high Chl-a concentration which may finally lead to water bloom in Han River. The spatial variation of Chl-a concentration was firstly estimated using indicator kriging (IK) and ordinary kriging (OK). The results indicates that OK overestimates low values (Chl-a < 2.5 µg/L) and underestimates high values (Chl-a > 25 µg/L), and IK can be used as a more direct and reliable method for spatial analysis in the presence of extreme values. Therefore, the combination of IK and OK was adopted to probabilistically categorize water quality of Han River for reducing the underestimation of the extreme values. The results reveal that the extreme high extreme low concentrations of Chl-a were less frequently observed in Han River in both dry and wet seasons. The upper and lower limits of the most suitable categories determined using IK strongly influenced the spatial distributions of the trophic states determined using the combined IK-OK approach. The modified IK-OK approach with multiple thresholds can reduce the underestimation of high values and the saltation of water quality estimation, and finally obtain an accurate estimations of Han River water quality. The approach facilitates constructing a numerical model for more effectively evaluating water quality variation which can also provide useful site-specific managements to control water quality.