三种订正方法在航空气象要素预报中的误差分析与应用评估

Error Analysis and Application Evaluation of Three Correction Methods in Aviation Meteorological Element Forecasting

摘要: 为满足航空气象精细化、高准确性服务的需求，针对欧洲中期天气预报中心(ECMWF)数值模式的预报产品，采用平均偏差订正法(MBC)、线性回归订正法(LRC)和分位数映射法(QM)，基于2025年7月3日~8月10日的实况数据对10米风速、2米气温、2米露点温度及海平面气压4个关键要素进行逐小时订正。通过平均偏差(MB)、平均绝对误差(MAE)、均方根误差(RMSE)等指标评估，结果表明：线性回归订正法综合性能最优，在修正海平面气压(RMSE = 1.616 hPa)、10米风速(RMSE = 1.188 m/s)和2米露点温度(RMSE = 1.314˚C)上表现最佳；分位数映射法在2米气温预报中优势显著(RMSE = 2.095˚C)，较线性回归法降低35.6%；平均偏差订正法计算简便但综合误差最大。基于此，提出“气温采用分位数映射、其余要素采用线性回归”的混合订正策略，为航空气象精细化预报提供技术支撑。

Abstract: To meet the demand for refined and high-accuracy aviation meteorological services, this study focuses on the forecast products of the European Centre for Medium-Range Weather Forecasts (ECMWF) numerical model. Three correction methods—Mean Bias Correction (MBC), Linear Regression Correction (LRC), and Quantile Mapping (QM)—are adopted to perform hourly corrections on four key meteorological elements, namely 10-meter wind speed, 2-meter air temperature, 2-meter dew point temperature, and sea level pressure, based on the observed data from July 3 to August 10, 2025. Evaluation using indicators such as Mean Bias (MB), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE) shows the following results: The Linear Regression Correction (LRC) method exhibits the best comprehensive performance, with optimal corrections for sea level pressure (RMSE = 1.616 hPa), 10-meter wind speed (RMSE = 1.188 m/s), and 2-meter dew point temperature (RMSE = 1.314˚C). The Quantile Mapping (QM) method has a significant advantage in 2-meter air temperature forecasting (RMSE = 2.095˚C), which is a 35.6% reduction compared with the LRC method. The Mean Bias Correction (MBC) method is simple in calculation but has the largest comprehensive error. Based on these findings, a hybrid correction strategy is proposed: using QM for air temperature and LRC for other elements, which provides technical support for refined aviation meteorological forecasting.