青藏高原地区大气加权平均温度模型的构建

加权平均温度（T_m）是将天顶湿延迟转换为大气可降水量的关键参数，针对青藏高原地区海拔高、地形起伏大、水汽高度分布复杂的特点，本文利用2010—2014年GGOS Atmosphere T_m格网数据和地表高程数据建立T_m垂直递减率函数，进而建立一种顾及T_m垂直递减率变化的适合青藏高原地区的新模型（QTm模型）。此外，利用2015年青藏高原地区14个探空站和GGOS Atmosphere T_m格网数据评估模型精度和适用性。试验结果表明，与GGOS Atmosphere T_m相比，QTm模型的年均Bias和RMSE分别为0.29和2.49 K，相对于GPT2w-1和GPT2w-5模型，RMSE分别提升了38.97%、67.06%；与探空数据相比，QTm模型的年均Bias和RMSE分别为0.16和2.90 K，相对于GPT2w-1和GPT2w-5模型分别提升了31.12%、39.46%。新模型的构建为青藏高原地区提供了可靠的T_m值，进而提供实时、高精度GNSS水汽信息。

Construction on atmospheric weighted mean temperature model over the Tibetan Plateau

The weighted average temperature (T_m) is a key parameter that converts the zenith wet delay into atmospheric precipitation. According to the characteristics of high altitude, high terrain fluctuation, and complicated distribution of water vapor in the Tibetan Plateau, this paper uses the GGOS Atmosphere T_m grid from 2010 to 2014 net data and surface elevation data to establish a T_m vertical decline rate function, and then establishes a new model (QTm model) suitable for the Tibet Plateau region that changes in T_m vertical decline rate is taked into account. In addition, the accuracy and applicability of the model are evaluated using 14 radiosonde stations and GGOS Atmosphere T_m grid data from the Tibet Plateau in 2015. The results show that compared with GGOS Atmosphere T_m, the average annual Bias and RMSE of the QTm model are 0.29 and 2.49 K. Compared with the GPT2w-1 and GPT2w-5 models, the RMSE increases by 38.97% and 67.06%. In comparison, the annual average Bias and RMSE of the QTm model are 0.16 and 2.90 K, which are 31.12% and 39.46% higher than GPT2w-1 and GPT2w-5 respectively. The construction of the new model provides a reliable T_m value for the Tibet Plateau region, and thus provides real-time, high-precision GNSS water vapor information.