GNSS的出现给空间定位技术带来了革命性的进展，其优异的定位性能给人们创造了精准的空间大地坐标。随着GNSS技术的发展和完善，定位精度要求越来越高，平面精度已经达到毫米级，但是在高程方面还存在需要改进的问题。利用GNSS定位技术测定的GNSS高程是基于WGS-84参考椭球的大地高，而工程中所采用的高程是基于似大地水准面的正常高，由于两者的基准面不一致,导致了GNSS大地高不能直接应用于实际。如何将GNSS测量所得的大地高转换为工程实际可用的正常高是一项有着现实意义的工作，有非常广阔的应用前景。基于此本文对上述问题进行了研究，主要研究工作以及最终成果如下：

(1)本文基于差分GNSS技术以及GPS重力位水准理论，给出一种计算两点间正常高高差的方法，即改进的GNSS重力位水准方法。该方法通过差分GNSS技术能够提供空间两点高精度的大地高高差，也不需要考虑国家高程起算点水准面的重力位，是一种快速的获取两点间正常高高差的方法。本文从Molodensky定义的正常高出发，根据多个已知点的高程异常数据，分别测试了EGM96、EGM2008、EIGEN-6C4等重力场模型，发现EIGEN-6C4模型为最优模型并作为本文实验选用的重力场模型。在西安科技大学骊山校园测区，我们分别基于GPS高程拟合法、GPS重力位水准法、单点GNSS重力水准法、改进的GNSS重力位水准法等四种方法计算不同点间的高差，高差的中误差分别±0.0371m；±0.0434m；±0.1641m；±0.0368m。改进的GNSS重力位水准的精度高于其他三种方法，单点GNSS重力水准方法的精度最低（需要说明的是单点GNSS观测的时间为2小时），而GPS高程拟合方法与GPS重力位水准方法的精度介于前两者之间。改进的GNSS重力位水准法比GPS重力位水准法的精度提高±0.0066m。在太白山测区，我们分别基于GPS重力位水准法、改进的GNSS重力位水准法计算不同点间的高差，中误差分别为±0.049m、±0.039m。改进的GNSS重力位水准法比GPS重力位水准法的精度提高±0.010m。不论是在平原小区域，还是在山区，改进的GNSS重力位水准法是一种有效并快速地确定正常高差的方法。

(2)在使用GPS重力位水准法计算正常高高差时，需要考虑国家高程起算点水准面的重力位，其误差会进一步影响高差。而改进的GNSS重力位水准方法的高差计算与高程基准选择无关。此外，影响改进的GNSS重力位水准精度的因素有大地高高差、重力场模型、平均正常重力等，其中大地高高差影响最大、平均正常重力次之、重力场模型最小。

(3)尽管采用精密单点定位的方法获得大地高高差精度也可达毫米级，但是需要耗费的时间相当长。经过实验验证，发现2小时后大地高高差的精度±0.031m，随着观测时长的增加，大地高高差的精度越来越高，24小时后在大地高高差的精度为±0.006m。而采用差分GNSS进行2小时的观测，获得大地高高差的精度为±0.005m。因此，想要短时间获得高精度的大地高高差，首选差分GNSS观测方法。

The emergence of GNSS has brought revolutionary progress to the spatial positioning technology, and its excellent positioning performance has created accurate spatial geodetic coordinates for people. With the development and improvement of GNSS technology, the requirement of positioning accuracy is getting higher and higher, and the plane accuracy has reached the millimeter level, but there are still some problems that need to be improved in elevation. The GNSS height measured by GNSS positioning technology is based on the geodetic height of the WGS-84 reference ellipsoid, while the elevation used in the project is based on the normal height of the quasi-geoid. Because the datum of the two are inconsistent, the GNSS geodetic height can not be directly applied to practice. How to convert the geodetic height obtained by GNSS survey into the normal height that can be used in engineering is a work of practical significance and has a very broad application prospect. Based on this, this paper studies the above problems, the main research work and the final results are as follows:  

(1) based on the difference GNSS technique and GPS gravity potential leveling theory, this paper presents a method to calculate the normal height difference between two points, that is, the improved GNSS gravity potential leveling method. This method can provide high-precision geodetic height difference between two points in space through differential GNSS technology, and does not need to consider the gravity potential of the leveling surface of the national elevation starting point. it is a fast method to obtain the normal height between the two points. In this paper, starting from the normal height defined by Molodensky, according to the elevation anomaly data of several known points, the gravity field models such as EGM96, EGM2008 and EIGEN-6C4 are tested respectively. It is found that the EIGEN-6C4 model is the optimal model and used as the gravity field model selected in this experiment. In the Lishan campus survey area of Xi'an University of Science and Technology, we calculate the height differences between different points based on GPS height fitting method, GPS gravity potential leveling method, single point GNSS gravity leveling method and improved GNSS gravity potential leveling method, and the median errors are ±0.0371m, ±0.0434m, ±0.1641m and ±0.0368m, respectively. The accuracy of improved GNSS gravity potential leveling is much higher than that of other methods, and the accuracy of single point GNSS gravity leveling method is the lowest, while the accuracy of GPS height fitting method and GPS gravity potential leveling method is between the former two. The accuracy of the improved GNSS gravity potential leveling is ±0.007m higher than that of the GPS gravity potential leveling. In Taibai Mountain survey area, we calculate the height differences between different points based on GPS gravity potential leveling method and improved GNSS gravity potential water method, and the median errors are ±0.049m and ±0.039m, respectively. The accuracy of the improved GNSS gravity potential leveling is ±0.010m higher than that of the GPS gravity potential leveling. Whether in the plain area or in the mountain area, the improved GNSS gravity potential leveling is an effective and rapid method to determine the normal height difference.

 (2) when using the GPS gravity level method to calculate the normal height difference, it is necessary to consider the gravity potential of the national height starting point, and its error will further affect the height difference. The height difference calculation of the improved GNSS gravity potential leveling method has nothing to do with the selection of elevation datum. In addition, the factors that affect the accuracy of the improved GNSS gravity leveling are geodetic height difference, gravity field model, average normal gravity and so on, in which the geodetic height difference is the biggest, the average normal gravity is the second, and the gravity field model is the smallest.

(3)  although the accuracy of geodetic height difference can reach millimeter level by using the method of precision point positioning, it takes a long time. After experimental verification, it is found that the accuracy of geodetic height difference is ±0.031m after 2 hours. With the increase of observation time, the accuracy of geodetic height difference becomes higher and higher, and the accuracy of geodetic height difference after 24 hours is ±0.006m. Using differential GNSS to observe for 2 hours, the accuracy of geodetic height difference is ±0.005m. Therefore, in order to obtain high-precision geodetic height difference in a short time, differential GNSS observation method is preferred.

[1] 孔祥元、郭际明. 大地测量学基础[M]. 武汉:武汉大学出版社,2005.

[2] 周忠谟, 易杰军. GPS卫星测量原理与应用[M]. 北京:测绘出版社,1992.

[3] 刘大杰, 施一民, 过静珺. 全球定位系统(GPS)的原理与数据处理[M]. 上海:同济大学出版社,1996.

[4] 邓罡. GPS高程拟合代替水准测量研究[D]. 长沙:中南大学,2012.

[5] 简程航. GPS高程拟合方法研究及其工程应用[D]. 北京: 中国地质大学,2014.

[6] 贺园园, 孟鲁闽, 王梅.GPS高程拟合在工程测量中的应用研究[J].测绘与空间地理信息,2010, 33 (2): 127-129.

[7] 刘杨, 申文斌, 夏敏, 等.利用GPS共视法确定重力位差及海拔高差的实验研究[J]. 武汉大学学报：信息科学版,2011, 36 (6): 640-643.

[8] 戴吾蛟, 陈招华, 梁铭.高差对GPS大地高测量精度的影响[J].大地测量与地球动力学,2009, 29 (3): 80-83,87.

[9] Nahavandchi H.Two different methods of geoidal height determinations using a spherical harmonic representation of the geopotential, topographic corrections and the height anomaly–geoidal height difference[J]. Journal of Geodesy,2003, 76 (11-12): 714-714.

[10] Wang J, Xie D.GPS elevation fitting study based on ternary polynomial regression[J]. Earth Sciences Research Journal,2020, 24 (2): 201-205.

[11] Trojanowicz M.Use of the UNB_TopoDens model for local modelling of chosen gravity field parameters in the Western Carpathians[J].Acta Geodynamica et Geomaterialia,2020, 17 (1): 113-118.

[12] 郭春喜, 聂建亮, 田婕, 等.GNSS水准高程变化自适应融合的垂直形变分析[J]. 武汉大学学报,2020,45(1):7-12.

[13] 章传银, 郭春喜, 陈俊勇, 等.EGM 2008地球重力场模型在中国大陆适用性分析[J].测绘学报,2009, 38 (4): 283-289.

[14] Zhang Y, Wang B.A new method of triangular elevation by total station and accuracy estimation[J].Engineering of Surveying and Mapping,2007, 16 (6): 50-52.

[15] Yang D, Zou J.Precise levelling in crossing river over 5 km using total station and GNSS[J].Scientific Reports,2021, 11 (1): 1-12.

[16] Wei Z, Wang G.Determination of Quasi-geoid in Mainland China Using Geopotential Model and GPS/Leveling Data[J].Acta Geodaetica et Cartographica Sinica,2003, 32 (1): 1-5.

[17] Li J.The Recent Chinese Terrestrial Digital Height Datum Model: Gravimetric Quasi-Geoid CNGG2011[J].Acta Geodaetica et Cartographica Sinica,2012, 41 (5): 651-660.

[18] Foroughi I, VanícEk P, Kingdon R W, et al.Sub-centimetre geoid[J].Journal of Geodesy,2019, 93 (6): 849-868.

[19] Wei Z.GPS Gravity-potential Leveling[J].Springer Berlin Heidelberg,2009, 134: 279-283.

[20] Liu Y, Shen W, Xia M, et al.Determination of Geopotential Difference and Orthometric Height Difference Using GPS Common View[J].Geomatics Information Science of Wuhan University,2011, 36 (6): 640-643.

[21] Chang C, Wang S Z, Wang Q B, et al.Analysis of truncation error based on degree variance of the gravity anomaly of high-order earth gravity model[J].Science of Surveying and Mapping,2015, 40 (12): 31-33.

[22] Qingwu M, Jianguang Q, Xiuhai L, et al.Compensating Method of the Height Anomalies for Ultra-High Earth’s Gravity Field Model[J].Wireless Personal Communications,2017, 97 (1): 75-94.

[23] Betti B, Biagi L, Crespi M, et al.GPS sensitivity analysis applied to non-permanent deformation control networks[J].Journal of Geodesy,1999, 73 (3): 158-167.

[24] Teng L, Yuan Y, Zhang B, et al.Multi-GNSS precise point positioning (MGPPP) using raw observations[J].Journal of Geodesy,2016, 91 (3): 1-16.

[25] Klokoník J, Kosteleck J, Bezděk A, et al.Support for two subglacial impact craters in northwest Greenland from Earth gravity model EIGEN6C4 and other data[J].Tectonophysics,2020, 780: 228-396.

[26] 孔祥元, 郭际明, 刘宗泉. 大地测量学基础.[M]. 武汉:武汉大学出版社,2010.

[27] Santos M C, Vanicek P, Featherstone W E, et al.The relation between rigorous and Helmert's definitions of orthometric heights[J].Journal of Geodesy,2006, 80 (12): 691-704.

[28] Trojanowicz M.Use of the UNB_TopoDens model for local modelling of chosen gravity field parameters in the Western Carpathians[J].Acta Geodynamica et Geomaterialia,2020, 17: 113-118.

[29] Filmer M, Featherstone W, Kuhn M.The effect of EGM2008-based normal, normal-orthometric and Helmert orthometric height systems on the Australian levelling network[J].Journal of Geodesy,2010, 84 (8): 501-513.

[30] Heiskanen W A, Moritz H.Physical geodesy[J].Bulletin Gæodésique,1967, 86 (1): 491-492.

[31] 李克昭. GNSS定位原理[M]. 北京: 煤炭工业出版社,2014.

[32] 王伟伟. BDS/GPS组合差分定位技术研究[D]. 桂林:桂林电子科技大学,2017.

[33] 王广运.载波相位差分GPS定位技术[J]. 测绘工程,1999, 8 (1): 12-17.

[34] 孙晋生, 孙中行, 代丽霞.GPS RTK技术在地形测量中的应用[J].吉林地质,2011, 30 (4): 113-116.

[35] 吴雨航, 陈秀万, 吴才聪, 等.电离层延迟修正方法评述[J].全球定位系统,2008, 33 (2): 1-5.

[36] 何怡, 李扬继.浅析差分GPS的算法及数据格式[J].电讯技术,2004.

[37] 崔建勇. 多模区域差分增强系统相关技术研究[D]. 郑州:解放军信息工程大学,2012.

[38] 朱留军, 徐爱功.GPS测量中多路径误差分析[J].交通科技与经济,2009.

[39] 陈水航.GPS高程拟合代替四等水准测量精度分析与探讨[J].经纬天地,2018 (2): 51-54.

[40] 刘帅, 王礼江, 朱建军, 等.GPS高程拟合模型的优选[J]. 测绘工程,2006, 15 (4): 17-19.

[41] 张兴福, 沈云中, 周全基.GPS高程异常拟合精度的估算方法[J].测绘通报,2003 (8): 21-22.

[42] 章传银, 晁定波, 丁剑, 等.厘米级高程异常地形影响的算法及特征分析[J].测绘学报,2006, 35 (4): 308-314.

[43] 陈俊勇.高程异常的推估及精度[J].测绘通报,1993 (4): 3-8.

[44] 许厚泽.全球高程系统的统一问题[J].测绘学报,2017, 46 (8): 939-944.

[45] 魏子卿.GPS重力位水准[J].大地测量与地球动力学,2007, 27 (4): 1-7.

[46] Nagy D.Theoretical gravity formulae and a mean gravity value[J].Acta Geodaetica Et Geophysica Hungarica,1999, 34 (1): 23-31.

[47] Pick M.On the normal gravity formulae[J].Studia Geophysica et Geodaetica,1990, 34 (4): 289-312.

[48] Wei Z.Normal Gravity Formulae[J].Acta Geodaetica Et Cartographic Sinica,2003, 32: 95-102.

[49] 党亚民, 章传银, 张鹏, 薛树强. 现代大地测量基准 [M]. 北京:测绘出版社,2015.

[50] 莫志明, 王建华, 罗志才, 等.中国大陆和香港高程基准面重力位差的测定[J].大地测量与地球动力学,2004, 24 (1): 36-40.

[51] 郭海荣. 建立高程基准的理论与方法的研究[D]. 郑州:解放军信息工程大学,2003.

[52] 韩建成. 基于地球重力场模型和地表浅层重力位确定大地水准面[D]. 武汉:武汉大学,2012.

[53] 宁津生, 邱卫根, 陶本藻. 地球重力场模型理论[M]. 武汉:武汉测绘科技大学出版社,1990.

[54] 常岑, 王世忠, 王庆宾, 等.重力异常阶方差模型精度及其截断误差分析[J]. 测绘工程,2016, 25 (10): 38-41,48.

[55] 常岑, 王世忠, 王庆宾, 等. 基于超高阶重力场模型阶方差的截断误差分析[C].测绘科学,2015: 31-33,49.

[56] 梁建青, 沈云中, 张兴福.基于EIGEN6C2模型的Kaula规则精化[J]. 测绘工程,2016, 25 (12): 21-24.

[57] Kaula, W. M.Tests and combination of satellite determinations of the gravity field with gravimetry[J].J Geophys Res,1966, 71 (22): 5303-5314.

[58] 胡明城.高精度和高分辨率的全球重力场模型EGM96[J].测绘科学,1998, 000 (003): 2-5.

[59] 张精明, 闫建强, 王福民.EGM2008地球重力场模型精度分析与评价[J].石油地球物理勘探,2010, 45 (z1): 230-233.

[60] Kosteleck J, Klokoník J, Bucha B, et al.Evaluation of gravity field model EIGEN-6C4 by means of various functions of gravity potential, and by GNSS/levelling[J].Geoinformatics FCE CTU,2015, 14 (1): 7-7.

[61] 李立功. 基于旋转模型的青藏高原重力变化研究[D]. 西安:长安大学,2011.

[62] Zhang C, Jiang T, Ke B, et al.The Analysis of Height System Definition and the High Precision GNSS Replacing Leveling Method[J].Acta Geodaetica et Cartographica Sinica,2017, 16: 915-951.

[63] Zhao D, Yabo L I.Precision Single Point Positioning in RTKLIB Precision Analysis[J].Geomatics Spatial Information Technology,2019, 42 (8): 52-54.

[64] Cetin S, Aydin C, Dogan U.Comparing GPS positioning errors derived from GAMIT/GLOBK and Bernese GNSS software packages: A case study in CORS-TR in Turkey[J].Survey Review,2019, 51 (369): 533-543.

[65] Sun T Y.The error in the monitoring of mine subsidence area is studied by using the second-class level measurement[J].World Nonferrous Metals,2017 (24): 275-277.