DEVELOPMENT OF A VEGETATION DISTRIBUTION MODEL FOR THE RESTORATION OF LATERAL CONNECTIVITY IN ISOLATED FORMER FLOODPLAIN

초록

Lateral connectivity between the stream channel and its floodplain is crucial to the structure and function of stream ecosystems. Whilst numerous assessment protocols for evaluation of physical, chemical and ecological states were developed in streams, the lateral connectivity between the channel and isolated former floodplains blocked by levees were not considered in most assessments. Additionally, ecological model provide prediction tools for lateral connectivity restoration. The aim of this study is to develop the lateral connectivity assessment and floodplain vegetation model between streams and isolated former floodplains in Korean streams. The assessment of lateral connectivity was made up of an overview assessment using GIS and field assessment in streams. The overview assessment included a method of mapping the isolated floodplain determining its connectivity to the channel and upland forest. The field assessment score of the connected floodplain was much higher than that of the isolated floodplain in the test stream. To develop floodplain vegetation model, vegetation and environmental characteristics were surveyed to understand ecosystem of isolated and connected floodplain in the Cheongmi Stream. The isolated floodplain showed lentic wetland environment where hydrophytes were dominant because of hydrological disconnection, while connected floodplain had lotic riparian ecosystem with hygrophytes. The vegetation were classified into 8 class according modified General Wetland Vegetation Classification System. Each vegetation class was modeled by statistical method with their environmental parameters. The results of generalized additive model (GAM ) showed high goodness value for each class. The model applied to Sagok area. We designed hydrological restoration of isolated floodplain and create virtual topography of restoration using GIS. After preparation hydrological and topographical parameters using numerical model and GIS, model predicted vegetation distri