This study investigated streamflow and water quality using the Soil and Water Assessment Tool (SWAT) and by applying two land use datasets from KOrea Multi-Purpose SATellite (KOMPSAT)-2 and Landsat TM (Thematic Mapper) images. For a 262.3 km2 agricultural watershed located in the northwest part of South Korea, the Landsat TM and KOMPSAT-2 could produce 8 and 25 land use classes from spatial resolutions of 30 m and 2 m, respectively. For each land use condition, the SWAT model was calibrated and validated using 8 years (2000-2007) of daily streamflow and monthly sediment, Total Nitrogen (T-N), and Total Phosphorous (T-P) data. The average Nash-Sutcliffe model Efficiencies (NSEs) of streamflow for KOMPSAT-2 and Landsat were 0.80 and 0.71, respectively. The average NSEs of sediment, T-N, and T-P were 0.52, 0.97, and 0.67 for KOMPSAT-2 and 0.30, 0.77, and 0.35 for Landsat, respectively. From the difference in the areal statistics and distribution between the two land uses, the key parameter for streamflow calibration was SCS-CN (Soil Conservation Service-CN). The watershed average SCS-CN value from KOMPSAT-2 was 1.05-fold greater than that from Landsat TM. The larger SCS-CN value increased the surface runoff and subsequently increased the sediment, T-N and T-P transport in the stream.

Keyword: scale effect, spatial resolution, remote sensing, streamflow, water quality, SWAT