Predicting basin water quality using source-sink landscape distribution metrics in the Danjiangkou Reservoir of China

Abstract

Land use and land cover changes due to human development are a concern for water-resource managers as they increase the spread of non-point source pollution to rivers, lakes, and reservoirs, impacting the quality of freshwater resources. Danjiangkou Reservoir is the largest inter-regional water transfer project in China and provides water services to cities along the middle route of the South-to-North Water Transfer Project and Beijing. Hence, watershed management decisions that may affect the water quality of the reservoir are particularly important. This study investigated the land use and land cover of a typical agricultural basin near the Danjiangkou Reservoir. The basin was divided into 13 sub-basins and the water quality was sampled six times during 2018 to determine the water quality index (WQI) for each sub-basin. The land use type and digital elevation model of each sub-basin were derived from remote-sensing data and then used to determine source-sink landscape distribution metrics; including the flow accumulation index (FAI), location-weighted landscape contrast index (LWLI), and landscape capacity distribution index (LCDI). Correlation and regression analysis were then used to compare these indices with the WQI. The WQI had a significant negative correlation with both the LWLI and LCDI, and a weak correlation with the FAI. Results indicated that the land use classification, proportion of source and sink landscape types, landscape flow path length, and landscape nutrient inputs were the main explanatory factors for river water quality in the basin. The biggest challenge in the research process is to obtain the parameters of landscape pattern index and its tedious calculation formula. The most successful part of the study is that it takes into account the flow path, roughness coefficient of land, nutrient input and other factors, more truly simulates the rainfall runoff process under natural conditions, and obtains better regression results between each evaluation index and water quality index. These findings show that remote-sensing data can be used in lieu of direct sampling to evaluate basin-water quality. Furthermore, these data can be used to predict the impact of future land use changes on water quality and facilitate more effective watershed management.