无气象参数的对流层延迟改正模型研究

发布时间：2018-04-27 02:03

本文选题：对流层延迟改正 + BP神经网络　；参考：《东南大学》2017年硕士论文

【摘要】：对流层延迟是影响GNSS高精度定位的主要原因之一。GNSS卫星信号经过大气层时,受到中性大气折射的影响会产生时延和路径弯曲的现象,从而造成GNSS信号的传播延迟。目前常用的改正对流层延迟的方法是模型改正法,而模型改正法又分为实测气象模型与无气象参数模型两种。实测气象模型需要获取测站处实测的气象参数,在实际工程应用中会受到一定的限制。而无气象参数模型计算较为方便,但其对流层延迟改正的精度要低于实测气象模型的改正精度。本文提出了一种基于BP神经网络技术改进无气象参数模型的新方法,最终建立起适用于中国地区的高精度天顶对流层延迟改正模型。本文的主要内容和结论如下:(1)利用IGS站提供的高精度对流层延迟数据,对中国地区对流层延迟的时空变化规律进行了详细分析。分析发现,中国地区对流层延迟随纬度的增加而减少,东部沿海地区的对流层延迟要高于西部内陆地区,且青藏高原等高海拔地区的对流层延迟较小。(2)在分析无气象参数的EGNOS模型和余弦函数模型的基础上,利用BP神经网络技术对EGNOS模型进行误差补偿,提出了一种新的IEGNOS融合模型。在中国地区5个IGS站上,IEGNOS模型的偏差绝对值的平均值和平均中误差都要优于EGNOS模型和余弦函数模型。其中EGNOS模型在5个IGS站上的平均中误差为±5.5cm,而IEGNOS模型的平均中误差为±2.9cm,相对于传统的EGNOS模型,IEGNOS模型的精度提高了 47%。(3)对天顶湿延迟反演大气可降水量PWV的过程和误差进行了分析。研究了利用探空气象资料计算加权平均温度和水汽转换系数的方法。以探空资料计算出的水汽转换系数当作真值,利用最小二乘法来建立适用于云南地区无需气象参数的Emardson水汽转换系数计算模型,即IEmardson模型。Emardson模型在云南地区4个站的平均中误差为±0.00495,IEmardson模型的平均中误差为士0.00112,相比于Emardson模型,其精度提高了约77%。因此,该IEmardson模型更适用于云南地区反演大气可降水量PWV。(4)利用IEGNOS模型来反演昆明地区的大气可降水量。通过IEGNOS模型得到测站天顶对流层湿延迟,结合改正后的Emardson水汽转换系数计算模型获取昆明地区一年的大气可降水量PWV,并与探空资料获取的大气可降水量PWV进行了对比。其12个月的月平均可降水量的平均偏差为1.38mm,平均中误差为±3.58mm,与探空资料获取的大气可降水量的变化趋势基本一致,因此在无实测气象参数时其反演的大气可降水量PWV具有较高的可信度。
[Abstract]:Tropospheric delay is one of the main reasons that affect GNSS high precision positioning. When GNSS satellite signals pass through the atmosphere, the phenomenon of delay and path bending will occur due to the influence of neutral atmospheric refraction, resulting in the propagation delay of GNSS signal. At present, the commonly used method to correct tropospheric delay is the model correction method, and the model correction method is divided into two kinds: the measured meteorological model and the non-meteorological parameter model. The measured meteorological model needs to obtain the measured meteorological parameters at the station, which will be limited in practical engineering application. However, the accuracy of tropospheric delay correction is lower than that of measured meteorological model. In this paper, a new method based on BP neural network to improve the model without meteorological parameters is proposed. Finally, a high-precision zenith tropospheric delay correction model suitable for China is established. The main contents and conclusions of this paper are as follows: (1) based on the high-precision tropospheric delay data provided by IGS station, the temporal and spatial variation of tropospheric delay in China is analyzed in detail. It is found that the tropospheric delay decreases with the increase of latitude in China, and the tropospheric delay in the eastern coastal area is higher than that in the western inland area. The tropospheric delay of Qinghai-Xizang Plateau is small. Based on the analysis of EGNOS model and cosine function model without meteorological parameters, a new IEGNOS fusion model is proposed by using BP neural network technology to compensate the error of EGNOS model. In five IGS stations in China, the mean and mean median errors of the absolute deviation of the IEGNOS model are better than those of the EGNOS model and the cosine function model. The mean median error of EGNOS model on 5 IGS stations is 卤5.5 cm, while that of IEGNOS model is 卤2.9 cm. Compared with the traditional EGNOS model, the accuracy of IEGNOS model is improved 47%. The error is analyzed. The method of calculating weighted mean temperature and water vapor conversion coefficient using sounding meteorological data is studied. Taking the water vapor conversion coefficient calculated from the radiosonde data as the true value, a calculation model of Emardson water vapor conversion coefficient suitable for Yunnan region without meteorological parameters is established by using the least square method. That is to say, the mean median error of IEmardson model. Emardson model in four stations in Yunnan area is 卤0.00495. The average median error of IEmardson model is 卤0.00112, which is about 77% higher than that of Emardson model. Therefore, the IEmardson model is more suitable for retrieving the atmospheric precipitable water in Yunnan area. The IEGNOS model is used to retrieve the atmospheric precipitable water in Kunming area. The wet delay of tropospheric troposphere at the zenith of the station is obtained by IEGNOS model, and the annual precipitable water PWV of Kunming area is obtained by combining with the corrected Emardson water vapor conversion coefficient calculation model, and compared with the atmospheric precipitable water PWV obtained from the sounding data. The average deviation of monthly average precipitable water is 1.38 mm and the mean median error is 卤3.58 mm, which is basically consistent with the variation trend of atmospheric precipitable water obtained from sounding data. Therefore, the inversion of atmospheric precipitable water PWV without measured meteorological parameters has a high reliability.
【学位授予单位】：东南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P228.4

【相似文献】