한반도 서해안 안개예측을 위한 3차원 변분자료동화 및 분석넛징 효과 분석

Abstract

The effect of data assimilation (DA) on the prediction of the heavy fog event occurred at the west coast of the Korean Peninsula was evaluated using a high resolution numerical modeling approach during the spring of 2016 and summer of 2018. This analysis was performed by two sets of simulation experiments: (1) WRF run with DA (i.e., 3DVAR+NUD), including analysis nudging for meteorological variables (e.g., temperature, u and v wind components) within the planetary boundary layer, and (2) WRF run without DA (i.e., BASE). To reflect more effectively the surface conditions of the study area, the WRF simulation was performed using the 24-category land-use data (3s elevation data) from the United State Geological Survey (USGS) and the high-resolution topographic data (3s elevation data) from the Shuttle Radar Topography Mission (SRTM). For 3DVAR, OBSRPOC was generated using ADP Global Upper Air and Surface Weather Observations, and background error covariance was generated by NMC method. Overall, the results simulated by the 3DVAR+NUD for the fog event in the study area provided better agreement with those observed, compared to the BASE run. The results of model simulations (3DVAR+NUD, BASE) were compared with observed values of ASOS (Automated Synoptic Observing System) and AWS (Automatic Weather System) to confirm the effect of data assimilation. From the statistical analysis (e.g., Index Of Agreement (IOA) and Root Mean Square Error (RMSE)), the IOAs of air temperature, wind speed, relative humidity, visibility, dew point temperature, and T-Td for the fog event simulated by the 3DVAR+NUD were in the range of 0.80-0.88, 0.57-0.72, 0.76-0.91, 0.71-0.77, 0.72-0.86, and 0.66-0.98, respectively. 3DVAR+NUD showing significantly higher than those simulated by the BASE run (0.67-0.83, 0.60-0.73, 0.65-0.83, 0.59-0.70, 0.62-0.69, and 0.54-0.96, respectively), except wind speed during the spring of 2016. Wind speed may be affected by urban surface conditions and high buildings. During the summer of 2018, IOAs of meteorological variables simulated by 3DVAR+NUD (0.81-0.90, 0.71-0.84, 0.79-0.85, 0.61-0.65, 0.76-0.79, 0.79-0.86) show higher than those simulated by the BASE run (0.76-0.85, 0.68-0.82, 0.72-0.78, 0.51-0.62, 0.66-0.67, 0.75-0.78) in the inland station. In addition, the RMSEs of these meteorological variables simulated by the 3DVAR+NUD were somewhat lower than those simulated by the BASE run. The comparison between the two simulations indicated that the effect of DA with analysis nudging is crucial to enhance the modeling capability of fog simulation over the Korean Peninsula. In addition, it is estimated to improve the fog prediction ability by accurately describing the terrain and applying high-resolution sea surface temperature in the west coast of the Korean peninsula.