강수진단모형을 이용한 미관측 지역 상세 강수량 자료 복원 방법에 관한 연구

Abstract

In various fields such as hydrology and agriculture, which is sensitive to precipitation and response to changes of rainfall patterns caused by global warming. Studies using numerical models that uses the meteorological data as the initial condition have been actively conducted. So, the demand high resolution precipitation grided data for model has increased. So in this study, we devised 1 km high-resolution synthetic precipitation data production method considering terrain effects uses Quantitative Precipitation Model (QPM) and Barnes method. We made synthetic precipitation data on rainfall interval at one-hour from 01. 01. 2010 to 01. 01. 2014 (KST). 48 observation station from Automatic Weather System (AWS), Automated Synoptic Observing System (ASOS) are removed and we carried out two procedure (Case 1, Case 2) by differencing the radius of influence for Barnes method with AWS, ASOS, Digital Elevation Model (DEM) and GFS analysis data of National Centers for Environmental Prediction (NCEP) to obtain the precipitation value at each grid point. Then the 48 removed observation and compared with the synthetic precipitation data on a validation. Stability verification results for the presence or absence of precipitation, is given by number of Hit, False, Miss, and Correct for Case 1 were 74738, 25778, 7544, 367981 respectively and for Case 2 were 74738, 24521, 6861, 398180 respectively. In Hit case, Bias is shown Case 1 and 2 were 1.22 and correlation coefficient for Case 1 and 2 were 0.74 and 0.76 respectively. Precipitation Ranges verification results, were shown for range of 0.5 ∼ 10 mm/hr, 10 ∼ 30 mm/hr, 30 ∼ 50 mm/hr and 50 ∼ 100 mm/hr; Mean Error (ME) for Case 1 was 0.3, -3.9, -14.4, -34.9 respectively and for Case 2 was 0.3, -3.9, -14.4, -34.9. Root Mean Square Error (RMSE); for Case 1 was 2.7, 8.3, 19.3, 39.6 respectively and for Case 2 was 2.4, 8.1, 19.9, 42.1 respectively. Also in each range, the 60% difference between the synthetic precipitation data and observation data of Case 1 was between -1.5 ∼ +1.5, -5.0 ∼ +5.0, -17.0 ∼ +17.0 and -33.0 ∼ +33.0 mm/hr respectively and Case 2 was between -1.2 ∼ +1.2, -4.5 ∼ +4.5, -17.0 ∼ +17.0 and -35.0 ∼ +35.0 respectively. Station verification result; 43 stations shown correlation coefficient of more than 0.7 for Case and 44 stations correlation coefficient of more than 0.7 for Case 2. Also 31 stations shown RMSE of less than 4.0 for Case 1 and 35 stations RMSE of less than 4.0 for Case 2. The significance of the correlation coefficient at the significance level of 0.05 for all points of Case 1 and 2 were significant.