通过分析广西4个探空站资料，结合GGOS Atmosphere格网Tm数据，建立随高程增大的温度递减率模型。根据温度递减率模型分别采用反距离加权法、双线性插值法、新反距离加权法和新双线性插值法计算探空站Tm，通过分析插值误差建立广西非气象参数Tm模型，并与Bevis模型、中国东部模型、广西模型进行比较。结果表明，温度递减率模型的Tm插值精度相对其他3种模型有比较明显的提升，4种方法的平均绝对误差（MAE）和均方根误差（RMSE）在1~2 K之间；广西非气象参数Tm模型的插值精度得到进一步提高，百色站的MAE约为2 K，其余站点的MAE和RMSE均在1 K左右，能满足可降水量反演的精度要求。"/>
Based on the data analysis of 4 radiosonde stations in Guangxi and GGOS Atmosphere grid Tm, we establish the temperature decline rate model with the increase of altitude. According to the temperature decline rate model, the inverse distance weighted, bilinear interpolation, new inverse distance weighted and new bilinear interpolation are used to calculate the Tm of the sounding stations. Based on the analysis of interpolation error, we establish the Tm model of non-meteorological parameters in Guangxi and compare it with Bevis model, eastern China model, and Guangxi model. The results show that the interpolation accuracy of the temperature decline rate model is obviously higher than those of the other 3 models. The average absolute error (MAE) and root mean square error (RMSE) of the 4 methods are between 1 and 2 K. We further improve the interpolation accuracy of Tm model of non-meteorological parameters in Guangxi. The MAE of Baise station is about 2 K, and the MAE and RMSE of other stations are about 1 K, which can meet the accuracy requirements of precipitation water vapor inversion."/>
广西非气象参数Tm模型研究
Abstract：Based on the data analysis of 4 radiosonde stations in Guangxi and GGOS Atmosphere grid Tm, we establish the temperature decline rate model with the increase of altitude. According to the temperature decline rate model, the inverse distance weighted, bilinear interpolation, new inverse distance weighted and new bilinear interpolation are used to calculate the Tm of the sounding stations. Based on the analysis of interpolation error, we establish the Tm model of non-meteorological parameters in Guangxi and compare it with Bevis model, eastern China model, and Guangxi model. The results show that the interpolation accuracy of the temperature decline rate model is obviously higher than those of the other 3 models. The average absolute error (MAE) and root mean square error (RMSE) of the 4 methods are between 1 and 2 K. We further improve the interpolation accuracy of Tm model of non-meteorological parameters in Guangxi. The MAE of Baise station is about 2 K, and the MAE and RMSE of other stations are about 1 K, which can meet the accuracy requirements of precipitation water vapor inversion.
