Abstract: Exploring the relationship between landscape pattern and water quality at the spatial characteristic scale is of great significance to revealing the impacts of landscape pattern on water pollution migration. In this study, water quality data were collected from 26 monitoring sections in the Minjiang River basin in Fujian Province, which was divided into three spatial scales as sub-watershed, riparian buffer zone, and circular buffer zone. Landscape pattern characteristics were analyzed based on Landsat-8 images. Each spatial characteristic scale was identified to examine the impacts of landscape pattern on water quality under the spatial characteristic scale by means of semivariogram and redundancy analysis. The results showed that: (1) water quality of Minjiang River basin generally met the Level Ⅱ standard. However, there were still polluted waters that exceeded the Level Ⅲ standard, which was mainly caused by aquaculture and livestock breeding. (2) The characteristic scale of riparian zone was in 1000 m buffer zone, while the one of circular zone was in 2000 m. (3) Landscape pattern at sub-watershed had the strongest overall interpretation of water quality, up to 58.4%, followed by riparian buffer zone (45.2%) and circular buffer zone (38.4%). (4) Forest cohesion and urban largest patch index were the main landscape pattern variables affecting water quality. The upstream catchment is the key area for future water resources management. Reducing the discharge of pollution, increasing the advantage and aggregation of forest, and breaking the continuity of extensive built-up areas, could help improve water quality.

Key words: spatial characteristic scale, water quality, semivariogram, redundancy analysis.