利用2014~2018年的ERA-Interim数据建立适用于日本区域的加权平均温度模型，分析其误差并进行季节性改正。结果表明：1）建立的线性加权平均温度模型精度比Bevis模型提高约16%；2）线性模型的残差时间序列存在季节性变化，因此对模型进行季节性改正，改正后的模型精度比Bevis模型和线性模型提高约37%和25%；3）将3种模型与探空站积分Tm进行比较，进一步说明季节性改正后模型的优越性。"/>
A weighted mean temperature model is established using the ERA-Interim data from 2014 to 2018 for the Japanese region. We analyze the errors of the Tm model and carry out seasonal corrections. The results show that: 1) A linear weighted mean temperature model suitable for the Japanese region is established, and accuracy is improved by about 16% compared with the Bevis model; 2) It is found that the residual time series of the linear model shows seasonal variations, so seasonal correction of the linear model is implemented. The accuracy of the corrected model is improved by about 37% and 25% compared with the Bevis model and the linear model; 3) Comparing the three models with the integral Tm of sounding stations, the superiority of the seasonal corrected model is further illustrated."/>
Weighted Mean Temperature Modeling with Seasonal Variations in Japan
Abstract A weighted mean temperature model is established using the ERA-Interim data from 2014 to 2018 for the Japanese region. We analyze the errors of the Tm model and carry out seasonal corrections. The results show that: 1) A linear weighted mean temperature model suitable for the Japanese region is established, and accuracy is improved by about 16% compared with the Bevis model; 2) It is found that the residual time series of the linear model shows seasonal variations, so seasonal correction of the linear model is implemented. The accuracy of the corrected model is improved by about 37% and 25% compared with the Bevis model and the linear model; 3) Comparing the three models with the integral Tm of sounding stations, the superiority of the seasonal corrected model is further illustrated.
