利用地基GPS技术反演武汉地区大气可降水分

利用武汉地区的探空资料和GPS实测数据，对对流层干分量延迟、对流层加权平均温度进行了检验分析．结果表明，对于武汉地区而言，常用的大气干分量延迟模型(SAAS Hopfield and Black)存在着1-2cm的系统误差，这在利用GPS资料估算大气可降水分(PWV)时会引入2-3mm的误差；对流层加权平均温度与常用的Bevis公式也存在着一定的差异，但这种差异对PWV结果影响很小．为此，提出了校正对流层干分量延迟的方法，并利用实测数据对该方法进行了检验．实践证明，这种校正方法基本上可以消除常用干分量模型的系统误差。

Remote Sensing of PWV Using Ground-Based GPS Data in Wuhan Region

Since early nineties last century, ground based GPS meteorology has been getting great progress in the world. Many countries including China have been constructing continuous GPS network for the purpose of research and application of this new technology. The results from these networks are encouraging and then near real-time estimation of Precipitable Water Vapor (PWV) are undergoing to build a perfect system in some countries too. Despite of these, one fundamental element is usually neglected, which was found by analyzing the radiosonde data collected in Wuhan Observatory to prepare for the incoming ground-based GPS meteorology application in Wuhan. That is the accurate model of Dry air Zenith Delay (DZD) in certain region. Usually, three most popular Dry air Zenith Delay models --Saastamoinen (SAAS), Hopfield and Black are believed to be accurate in several millimeters and then are thought to be almost no influence on the estimation of PWV. In practice, due to the relationship between DZD and the temperature and pressure profile above site, DZD models in different regions may be a little different from each other. To reduce the influence of DZD model errors on the estimation of PWV, DZD models accuracy should be investigated so that a precise model suitable to this region is obtained. This paper is trying to do some work on this aspect using radiosonde data in Wuhan region. The following results indicate that there is a systematic error which can be up to more than 16 mm in the three popular DZD models, which will introduce more than 2 mm into the result of PWV. Therefore, a bias and some scale parameters are introduced in these models to eliminate the systematic errors. The parameters are estimated using the Least Squares method and are shown in Table 3 below. To validate these models after calibration, a set of data was analyzed. The results show that the influence of DZD on PWV is less than 1 mm after calibration in Wuhan. On the other hand, the same case occurs on another fundamental element--the weighted mean tropospheric temperature. With the same way as Bevis used, more than 40 days' radiosonde data were processed and it was found that there was a good linear relationship between the mean tropospheric temperature and the surface temperature in Wuhan area while linear regression was involved in analysis. Moreover, the linear equation resulted from those radiosonde data is so close to Bevis' formula that the equation owns the influence of 1 mm on the estimation of PWV in Wuhan compared with Bevis'.