RESUMO
Soil erosion associated with non-point source pollution is viewed as a process of land degradation in many terrestrial environments. Careful monitoring and assessment of land use variations with different temporal and spatial scales would reveal a fluctuating interface, punctuated by changes in rainfall and runoff, movement of people, perturbation from environmental disasters, and shifts in agricultural activities and cropping patterns. The use of multi-temporal remote sensing images in support of environmental modeling analysis in a geographic information system (GIS) environment leading to identification of a variety of long-term interactions between land, resources, and the built environment has been a highly promising approach in recent years. This paper started with a series of supervised land use classifications, using SPOT satellite imagery as a means, in the Kao-Ping River Basin, South Taiwan. Then, it was designed to differentiate the variations of eight land use patterns in the past decade, including orchard, farmland, sugarcane field, forest, grassland, barren, community, and water body. Final accuracy was confirmed based on interpretation of available aerial photographs and global positioning system (GPS) measurements. Finally, a numerical simulation model (General Watershed Loading Function, GWLF) was used to relate soil erosion to non-point source pollution impacts in the coupled land and river water systems. Research findings indicate that while the decadal increase in orchards poses a significant threat to water quality, the continual decrease in forested land exhibits a potential impact on water quality management. Non-point source pollution, contributing to part of the downstream water quality deterioration of the Kao-Ping River system in the last decade, has resulted in an irreversible impact on land integrity from a long-term perspective.
