전지구모형(General Circulation Model, GCM)이 모의한 수문기상 변수들은 시공간적 해상도가 매우 크고, GCM과 과거 관측 수문기상 변수들 사이에는 편차가 존재한다. 이를 극복하기 위해 다양한 편차보정기법을 활용하여 GCM 모의결과를 보정한다. 본 연구에서는 GCM 모의강수의 편차보정 개선 정도를 검증하기 위해 기본통계값 뿐만 아니라 entropy를 이용하였다. 우선 본 연구는 이수기와 홍수기에 각각 simple method와 Nested Bias Correction (NBC) method를 기반으로 하는 Composite Bias Correction (CBC) method를 제안하였으며, CBC method가 관측강수 대비 편차, RMSE, 연총강수량에서 각각 0.02 mm/month, 11.10 mm/month, 1325.99 mm/year(관측 연총강수량=1326.28 mm/year)을 나타내 가장 우수함을 나타냈다. 다음으로 entropy를 이용하여 CBC 편차보정방법 적용 전/후의 GCM 모의강수를 검증하였다. 편차보정 전 GCM 모의강수의 확률밀도함수는 발생빈도가 높은 적은 강수에 대해서는 과대모의, 홍수기에는 과소모의하는 것으로 나타났으나 편차보정된 GCM 모의강수의 확률밀도함수는 관측강수와 유사한 형태를 나타내었다. 편차보정된 GCM 모의강수의 entropy(=5.648)는 관측강수(=5.643)와 거의 비슷한 entropy를 나타내어, 관측강수와 통계특성뿐만 아니라 불확실성도 비슷하였다. 따라서 편차보정방법은 GCM 모의변수의 편차를 보정하고 관측변수의 불확실성을 잘 모의하는 것으로 나타났다.
전지구모형(General Circulation Model, GCM)이 모의한 수문기상 변수들은 시공간적 해상도가 매우 크고, GCM과 과거 관측 수문기상 변수들 사이에는 편차가 존재한다. 이를 극복하기 위해 다양한 편차보정기법을 활용하여 GCM 모의결과를 보정한다. 본 연구에서는 GCM 모의강수의 편차보정 개선 정도를 검증하기 위해 기본통계값 뿐만 아니라 entropy를 이용하였다. 우선 본 연구는 이수기와 홍수기에 각각 simple method와 Nested Bias Correction (NBC) method를 기반으로 하는 Composite Bias Correction (CBC) method를 제안하였으며, CBC method가 관측강수 대비 편차, RMSE, 연총강수량에서 각각 0.02 mm/month, 11.10 mm/month, 1325.99 mm/year(관측 연총강수량=1326.28 mm/year)을 나타내 가장 우수함을 나타냈다. 다음으로 entropy를 이용하여 CBC 편차보정방법 적용 전/후의 GCM 모의강수를 검증하였다. 편차보정 전 GCM 모의강수의 확률밀도함수는 발생빈도가 높은 적은 강수에 대해서는 과대모의, 홍수기에는 과소모의하는 것으로 나타났으나 편차보정된 GCM 모의강수의 확률밀도함수는 관측강수와 유사한 형태를 나타내었다. 편차보정된 GCM 모의강수의 entropy(=5.648)는 관측강수(=5.643)와 거의 비슷한 entropy를 나타내어, 관측강수와 통계특성뿐만 아니라 불확실성도 비슷하였다. 따라서 편차보정방법은 GCM 모의변수의 편차를 보정하고 관측변수의 불확실성을 잘 모의하는 것으로 나타났다.
Spatio-temporal resolutions of hydrometeorological variables simulated by General Circulation Models (GCMs) are very sparse and biases have occurred between GCM simulations and historical observations. To overcome these limitations, GCM simulations are corrected by utilizing various bias correction ...
Spatio-temporal resolutions of hydrometeorological variables simulated by General Circulation Models (GCMs) are very sparse and biases have occurred between GCM simulations and historical observations. To overcome these limitations, GCM simulations are corrected by utilizing various bias correction methods. This study has verified the improvement of GCMs simulated precipitation after applying a bias correction method using entropy as well as basic statistics. First and foremost, this study has proposed the Composite Bias Correction (CBC) method based on a simple method during the dry season and a nested bias correction method within a flood season. Bias, RMSE and annual total precipitation of the CBC method related to observations were at 0.02 mm/month, 11.10 mm/ month, and 1325.99 mm/year(observed annual total precipitation=1326.28 mm/year), respectively and the CBC method is superior to other methods. Furthermore, GCM simulated precipitation before/after applying the CBC method was verified using entropy. Although the probability density function (pdf) of GCMs simulated precipitation before the CBC method was overestimated in a low-frequency precipitation as well as being underestimated in the flood season, pdf of bias-corrected GCM simulated precipitation was similar to the observations. Bias corrected GCM simulated precipitation that had an entropy (=5.648) similar to observations (=5.643) had statistical characteristics and uncertainty analogous in comparison to observed precipitation. Therefore, the bias correction method does correct the biases of GCMs simulated variables and simulates the uncertainty of observation properly.
Spatio-temporal resolutions of hydrometeorological variables simulated by General Circulation Models (GCMs) are very sparse and biases have occurred between GCM simulations and historical observations. To overcome these limitations, GCM simulations are corrected by utilizing various bias correction methods. This study has verified the improvement of GCMs simulated precipitation after applying a bias correction method using entropy as well as basic statistics. First and foremost, this study has proposed the Composite Bias Correction (CBC) method based on a simple method during the dry season and a nested bias correction method within a flood season. Bias, RMSE and annual total precipitation of the CBC method related to observations were at 0.02 mm/month, 11.10 mm/ month, and 1325.99 mm/year(observed annual total precipitation=1326.28 mm/year), respectively and the CBC method is superior to other methods. Furthermore, GCM simulated precipitation before/after applying the CBC method was verified using entropy. Although the probability density function (pdf) of GCMs simulated precipitation before the CBC method was overestimated in a low-frequency precipitation as well as being underestimated in the flood season, pdf of bias-corrected GCM simulated precipitation was similar to the observations. Bias corrected GCM simulated precipitation that had an entropy (=5.648) similar to observations (=5.643) had statistical characteristics and uncertainty analogous in comparison to observed precipitation. Therefore, the bias correction method does correct the biases of GCMs simulated variables and simulates the uncertainty of observation properly.
AI-Helper
※ AI-Helper는 부적절한 답변을 할 수 있습니다.