미계측 유역에서의 수문 분석을 위한 글로벌 격자형 기후 자료의 활용성 평가

Abstract

We assessed the usability of satellite and reanalysis data for estimating streamflow amount in an ungauged watershed in consideration of the spatial distribution of precipitation. We used CHIRPS, PERSIANN-CDR, CFSR, ERA-Interim, and MERRA2 to obtain grid-based global data and daily GHCN as measured data for comparison. Evaluation results of daily GHCN data through comparison to the ASOS data from Gangneung station adjacent to Chungju Dam basin showed a greater than 60% missing data rate. The GHCN data underestimated extreme rainfall events caused by typhoons and overestimated rainfall events in certain years compared to the nromal range. In the comparison of grid-based global data, MERRA2 data showed the best reproducibility, with a 0.86 correlation coefficient, on the daily time scale. The SWAT model was applied to Chungju Dam basin for estimating streamflow and showed -16.1% total error when the MERRA2 data was used as model input without bias-correction. However, both total error and RMSE were decreased when the monthly biases of MERRA2 data at the weather stations were corrected by comparing to the 30-year average values. In addition, the biases were spatially interpolated for areas in the basin without a station. Total error was slightly improved to -13.1% when the bias was corrected using the GHCN data from Gangneung station, while the total error showed a satisfactory -6.3% when using ASOS data from Daegwallyeong station for streamflow estimation after bias-correction. In the absence of measured weather data, it is desirable to use the most appropriate grid-based data after bias-correction through comparison with neighboring weather stations to reflect the characteristics of the area.