Application of a Bias Correction Method to Meteorological Forecast for the Pyeongchang Winter Olympic Games

Abstract

The 23rd Winter Olympics Games and the 13th Winter Paralympic Games are held in Pyeongchang, South Korea during 9-25 February 2018 and 8-18 March 2018. Supported by the WMO(World Meteorological Organization) WWRP(World Weather Research Project Group), ICE-POP 2018(International Collaborative Experiments for Pyeongchang 2018 Olympic and Paralympic winter games) is organized by KMA (Korea Meteorological Administration), which aims to improve the forecasting ability of convective scale numerical models for high-impact weather system under complex terrain and to support weather forecasting and meteorological services during the Winter Olympics through international cooperation. GRAPES_3 km, a high-resolution model independently developed by CMA (China Meteorological Administration), participated in this project and provided real-time forecasting products at the specific sites during the Winter Olympics Games. In order to improve the forecasting ability of GRAPES_3 km, a bias correction method named one-order adaptive Kalman filtering is applied, which compensate for the fact that the resolution of GRAPES_3 km is not sufficient to simulate the complex terrain of the Winter Olympic Games and GRAPES_3 km doesn't assimilate Korea's observations due to some objective reasons such as data transmission. 1-24-hour calibration products are provided twice a day, including 2 m temperature, 2 m relative humidity, 10 m wind speed, and 10 m wind direction on sixteen sites. The verification and evaluation are carried out in two aspects. First, it is to check whether the bias correction improves the accuracy of GRAPES_3 km, examined from the model's diurnal cycle, daily variation and the forecast ability of complex terrain. Second, to the performance of calibrated GRAPES_3 km during the Winter Olympic Games is and examined compared to LDAPS model from KMA and NU-WRF from NASA, which shows that the high-resolution GRAPES_3 km model has abilities to simulate the near-ground elements in this service and the bias correction technology makes model products better and effective. After bias correction, the root mean square error of model products is reduced to about 2℃ for the temperature, about 2 m·s-1 for the wind speed, and less than 20% for the relative humidity is reduced to. Compared with, NU-WRF and LDAPS models, the corrected GRAPES_3 km has the best wind speed forecast capability and root mean square error of the wind speed is significantly reduced. For the wind direction forecast of complex terrain, although all numerical models have limited capabilities, and their wind direction forecasting of all the stations are almost dominated by westerly during the Winter Games, the corrected GRAPES_3 km perform relatively well in some non-westerly dominated stations. However, GRAPES_3 km has no significant advantage in temperature and humidity compared with the other two models although they are improved more obviously than wind after the calibration. In addition, the bias correction can attenuate the diurnal variation characteristics of GRPAES_3 km, and to some extent, improve the simulation capability for complex terrain. In short, the application of the bias correction in the Pyeongchang Winter Olympic Games is an effective way for GRAPES_3 km.

Keywords

• bias correction