A Study of Near Real-time Water Vapor Analysis Using a Nationwide Dense GPS Network of Japan
A Study of Near Real-time Water Vapor Analysis Using a Nationwide Dense GPS Network of Japan
复制标题
DOI：
10.2151/jmsj.87.1
复制
发表时间：
2009-02-01
影响因子：
3.1
通讯作者：
Shoji, Yoshinori
中科院分区：
文献类型：
--
作者：
Shoji, Yoshinori
为了利用日本全球定位系统（GPS）的全国密集接收器网络来促进水汽监测和数值天气预报，在气象研究所进行了近实时（NRT）分析试验。本文沿着介绍了NRT分析过程和一些验证结果。考虑到分析日本1，300多个GPS站所涉及的计算负荷，采用了精密单点定位（PPP）过程。PPP过程需要GPS卫星的位置和时钟偏移的准确信息。国际GNSS服务（IGS）已经常规地提供超快速星历（IGU），其包括具有约3小时的延迟的卫星轨道和时钟偏移。我们发现IGU卫星钟差的精度不足以通过PPP程序反演可降水量（PWV）。因此，我们使用IGS站“USUD”处的预测时钟偏移对IGU时钟应用校正。USUD的氢脉泽原子钟与GPS时间也有一定的差异。然而，我们发现它可以用一个线性方程来拟合和预测，得到的卫星钟差对IGS最终星历有一定的偏差，但卫星钟差的时间常数偏差对PWV反演没有任何影响。反演的PWV数据与无线电探空仪观测结果吻合良好。夏季和冬季的均方根差分别约为3.4 mm和1.6 mm。其结果与先前使用最终星历的研究结果相当。一次暴雨过程中反演的PWV时空变化表明了NRT PWV反演方法在天气监测中的有效性。我们可以捕捉到暴雨前的水汽增加。
In order to use the nationwide dense receiver network of the Global Positioning System (GPS) in Japan to contribute to water vapor monitoring and numerical weather prediction, a near real-time (NRT) analysis trial was performed at the Meteorological Research Institute. In this paper, the NRT analysis procedure is described along with some validation results.In view of the Computational load involved in analyzing more than 1,300 GPS stations in Japan, the precise point positioning (PPP) procedure was adopted. The PPP procedure requires accurate information of GPS satellites' positions and clock offsets. International GNSS Service (IGS) has been routinely providing ultra rapid ephemeris (IGU) that includes satellite orbits and clock offsets with latency of about 3 hours. We found the accuracy of satellite clock offsets in IGU was insufficient for the retrieval of precipitable water vapor (PWV) through the PPP procedure. Therefore, we applied correction to the IGU clock using the predicted clock offset at an IGS station "USUD". The hydrogen maser atomic clock at USUD also had some differences with GPS time. However, we found it could be fitted and predicted by a linear equation for a period of several days.The resulting satellite clock offsets exhibited some biases toward the IGS final ephemeris, but the time constant biases of satellite clock offsets did not affect the PWV retrieval at all. The retrieved PWV data agreed well with those obtained from radio-sonde observations. The root mean square differences in summer and in winter were around 3.4 mm and 1.6 mm, respectively. The results were comparable with those obtained by preceding studies using the final ephemeris. The Retrieved spatial and temporal variation of PWV in a heavy rainfall case demonstrated the usefulness of the NRT PWV retrieval for weather monitoring. We could capture the water vapor increase that preceded torrential rain.