APPLICATION OF THE ENVIRONMENTAL ZONING FACTORS IN DEEP-LONG RESERVOIR BY DATA MINING TECHNIQUES

摘要

The in-reservoir techniques for protection and restoration of water quality are different to the zones. In a rapidly flushed system, an algal bloom is less likely to depend on nutrient concentrations than on flushing rate. Unlike many natural lakes, where water may enter from several smaller tributaries draining comparatively small sub-watersheds, reservoirs have a distinct riverine (river-line) zone dominated by flow and mixing, in which nutrient supplies take place by advective transport, and algal cells settle down on the bottom of reservoir. This is followed by a transition zone where inflow velocity is slow, rapid sedimentation begins, and water clarity increases. Although almost of models simulating reservoir behaviors and water qualities provide the numerical values, they do not directly and comprehensibly explain which variable is most sensitive to the target of water quality in each zone or over any space, and relationships among variables related to the target. The objective of the research is to propose the environmental zoning factors for long-deep reservoir water quality management. We employed data mining techniques such as k-means and model tree to estimate the zoning factors and to predict the Chl-a concentration in each zone. Model tree is consisted of the tree-structured regression models that associate leaves with multiple linear regression functions. The Yongdam dam reservoir located upstream in the river Keumgang flowing through mid-west of Korean peninsular was selected for this study. The water quality parameters were measured such as DO, Ph, EC, water temperature, Secchi-depth, Turbidity, SS, BOD, COD_(Mn), COD_(Cr), Chlorophyll-a, TP, PO4-P, NH4-N, NO3-N, NO2-N, and DOC. The model tree analysis showed the dominant water quality parameters in each zone and showed fair performance on the prediction of the Chl-a concentration. These concepts of water quality management could be efficiently applied to the best management practices of the dam reservoir.