NING Jinsheng, LI Jiancheng, LUO Zhicai, CHAO Dingbo and JIANG Weiping
School of Geodesy and Geomatics, Wuhan University, 129 Luoyu Road, Wuhan 430079, China
I. ESTABLISHMENT OF NFGN 2000
As same as the establishment of other geodetic control network, the gravity network is also set up by the way of step control. The national fundamental gravity network (NFGN) can provide the gravity datum and the highest order control of gravimetry for different purposes. The first NFGN in P.R.China, named NFGN 57, was established in 1957, its mean accuracy is ± . Afterwards, NFGN 85 was constructed in 1985 for the requirements of the national economic construction and the development of science and technology. The systematic error of Potsdam absolute gravimetric system was corrected for NFGN 85, more gravity fiducial points are available, and the density of absolute gravimetric point improved. So the mean accuracy of NFGN 85 is better than that of NFGN 57 by one order of magnitude, and can arrive at ± . Now the accuracy and the reliability of NFGN 85 cannot meet with the requirements of resource exploration, national defense construction, surveying and mapping, spaceflight technology, geoscience, etc., due to the lower observation accuracy and inhomogeneous distribution of absolute gravimetric points, not completely reasonable figure structure of NFGN 85, some gravity basic points destroyed, etc. For these purposes, NFGN 2000 covering the whole territory of China except for Taiwan was designed and measured since 1999. This network consists of 133 gravity points, and there are 17 fiducial points (absolute gravimetric points) and 116 basic points (relative gravimetric points) among these gravity points. Furthermore, one derived point was laid out for every point of 106 basic points, which were used as the spare point of basic point. NFGN 2000 has the following characteristics and improvements by comparing to NFGN 85:（1）17 hi gh accuracy absolute gravimetric points were laid out reasonably and homogenously over the whole territory of China, which further improved the gravity datum of national fundamental gravity network.
（2）The figure structure of NFGN 2000 was designed optimally, and the observational figure structure and the figure strength of west China especially improved and enhanced, so the overall accuracy of NFGN 2000 was improved.（3）The calibration line of gravimeter (long base line) and the national high accuracy calibration site of gravimeter constant (short base line), were reconstructed and improved for the unification of gravimetric scale and the precise calibration of various gravimeter constants.
（4）NFGN 2000 was connected to NFGN 85, and joined together with the gravity basic network of national important scientific engineering “Monitoring Network of Crustal Movement in China”.The mobile absolute gravimeter FG-5 and LCR-G gravimeter were used in NFGN 2000. The effects of tide, pressure, polar movement, vertical gradient, etc., were corrected for the absolute gravimetric measurements of NFGN 2000. So the observation accuracy of gravity fiducial points is better than ± . For the relative gravimetric measurements, the effects of tide, pressure, height of instrument, zero drift, etc., were also corrected. The relative observation accuracy of gravity basic points is better than ± . Therefore, the standard error of NFGN 2000 is not larger than ± after adjustment. The heights of the gravity fiducial point, basic point and derived point are relative to China's 1985 Yellow Sea Height Datum, and their planar coordinates with respect to Xi'an 80 coordinate system.
II．REFINEMENT OF LOCAL GEOID
The local or regional geoid with high resolution and high accuracy can provide the fundamental geo-spatial information not only for surveying and mapping, geophysics oceanography and geodynamics, but also for the construction of “digital China”, and now especially for applying GPS technique to determine orthometric or normal height in geodesy and surveying engineering. For these purposes, the new quasi-geoid model CQG2000 with the accuracy of decimeter level has been constructed, which covers the whole territory of China (Chen et al., 2001). The local quasi-geoid with the accuracy of centimeter level and the resolutions of 2¢.5 and 1km has also been determined respectively for Jiangsu Province and Shenzhen City, P.R.China. In future the local quasi-geoid with high accuracy for other provinces and developing regions in China will be determined continually. It is hopeful that the high accuracy quasi-geoid can be employed to the substitution of traditional third and fourth order spirit leveling, the large-scale digital mapping, etc. This will accelerate the construction of “digital city” and “digital China”.
With remove-restore technique Wuhan University calculated the quasi-geoid of Jiangsu Province using the following data: (1) 8756 discrete gravity data on land; (2) marine gravity anomalies on 422475 cross points derived from the multi-satellite altimetric data of version 3 T/P (cycle 1 to cycle 249), Geosat/GM/ERM, ERS-1, ERS-1/168 days, and ERS-2 (cycle 0 to cycle 52); (3) digital terrain model (DTM) with resolution of 18².75´28².125 covering the whole territory of Jiangsu Province, 2¢´2¢ global DTM2000 provided by NASA/NIMA; (4) high quality GPS/leveling data; and (5) WDM94 and EGM96 geo-potential models. Moreover, all integral computations such as terrain correction, first term correction of Molodensky solution, etc., were carried out by rigorous 1-D FFT technique. The resolution of the quasi-geoid is 2¢.5´2¢.5, and its accuracy is better than ±0.078 m.
With the fast development of science and technology, and the requirements of economic construction in Shenzhen City, the land planning department of Shenzhen decided to construct the Shenzhen quasi-geoid with the resolution of 1km and the accuracy of centimeter level in 2000. For
this purpose, the GPS/leveling network consisting of 73 control points was laid out in 2001, and 4870 discrete gravity points were also collected using Lacoste & Romberg model ‘G' and “D” land gravimeter and model ‘S' marine gravimeter. And then with almost the same method as that computing the quasi-geoid of Jiangsu Province, Wuhan University calculated the Shenzhen quasi-geoid with the resolution of 1km, using 65 high accuracy GPS/leveling data, 5213 discrete gravity data, digital topographic model with the resolution of 100 m which covers the whole territory of Shenzhen and its neighboring region, and WDM94 geo-potential model. The geoid covers the area of 8 km to 60 km along south-north direction and 79 km to 179 km along west-east direction in Shenzhen local grid coordinate system. Finally, the modeled geoid heights and geoid height differences were compared to those from 29 high precision GPS/leveling data that not used to the calculation of the geoid. And the check results show that the accuracies (standard deviations) of the geoid height and geoid height difference are ±0.014 m and ±0.019 m, respectively, and its overall relative accuracy is better than 1ppm.
Chinese geodesists have always investigated the methods for transferring the height within long distance across sea. Up to now there are four methods used to the height transference, such as static leveling, dynamic leveling, GPS/leveling and conventional geodetic method. The conventional height transferring methods are not practical due to the time-consuming and labour-intensive. Wuhan University studied the method of the height transference within long distance across sea by combining the ellipsoidal height from GPS and the precise gravimetric geoidal height. Using this method the China's 1985 Yellow Sea Height Datum is transferred to Yangshan Island, which is 30 km far away from Shanghai. The transferred heights are then compared with those determined from two independent sets of gauge records, and the differences are 1.0 cm and 5.0 cm respectively. Moreover, the transferred height differences on the island are also compared with those derived from third order precise spirit leveling, which indicates that the differences are 0.2 cm and 0.7 cm respectively. These test results show that the method is inexpensive, very effective and reliable for the height transference within long distance across sea.
III. SATELLITE ALTIMETRY
Since 1970s, the marine gravity field derived from multi-satellite altimetry missions has rapidly been developed. Combining processing the altimeter data of TOPEX/Poseidon (9-249 cycle), ERS2 (0-44 cycle), Geosat/GM(1-25 cycle) and Geosat/ERM (1-66 cycle), it was solved out for crossover points of respective satellite altimetric mission and the deflection of the vertical along altimetric track profiles, and then the marine gravity anomalies gridded in 2.5¢×2.5¢ size over the South China Sea were determined using inverse Vening-Meinesz formula. Comparing the gravity anomalies derived from the altimeter data with 600 000 gravity anomalies measured by marine gravimeter, it shows that RMS and STD of the difference between them are ± 9.4 mGal and ±9.3 mGal respectively. A new gravity recovery method with along-track vertical deflections is developed independently on the basis of the Laplace's equation in Cartesian system. The method can be easily approached with 1D Hilbert transform and the along-track gravity anomalies can be derived directly from the along-track slopes of sea surface height (SSH). In other words, only along-track slope of SSH needs to be calculated other than two slopes of along-track and cross track at crossover point（Li et al., 2001; Wang et al., 2001; Huang et al., 2001）.
Four-year altimeter data from T/P and one-year altimeter data from ERS-1 in the China sea and its vicinity are used to determine the mean sea surface (MSS) with stacking method along the satellite repeated collinear tracks. After the contributions of sea surface topography are reduced from the MSS, the 30'´30' geoid undulations are obtained, and the accuracy of geoid is 8.5 cm (RMS) (Xu et al., 1999). The marine deflection of the vertical is computed from the altimeter data of Topex/Poseidon, ERS-2 and Geosat/GM ERM, in which 2.5'´2.5' grids for calculation are used, and then 5.0¢´5.0¢ geoid determination of China Sea Area is carried out by Molodensky method. In order to check the computed results, the geoid which is directly solved out by Molodensky formula mentioned above is compared with that computed by Stokes formula using the deflection-inversed gravity anomaly data as an inner examination, and the standard deviation is ±0.025 m. Considering the reality that there is lack of gravity data in China coast area, an extending method is advanced for piecing the two types of geoid determined by different principle and data set. Finally, the continent-marine gravity geoid after piecing together is then fitted with quasi-geoid determined by National GPS leveling network, and the corrected gravity geoid called CQG2000 is obtained (Chen et al., 2001).
The method for determining mean sea surface (MSS) by using multi-altimetric data is developed. The data used to compute WHU2000 MSS include 7 years of Topex/Poseidon data (cycle 11 to 249), 2 years of Geosat ERM data (cycle 1 to 44), 5 years of ERS2 data (cycle 1 to 52) and all ERS-1 168-day data. The WHU2000 MSS is determined with resolution of 2¢´2¢ within the ±82° latitude and its precision is better than 0.05 m. Comparing WHU 2000MSS with 3.75¢×3.75¢ CLS-SHOM98.2 MSS, 3¢×3¢ GFZ MSS95A and 3.75¢×3.75¢ OSU MSS95, as an external check, the corresponding STDs of their differences are 0.090 m, 0.211 m and 0.079m respectively (Jiang et al., 2002； Li et al.，2001).
The collinear method is used to determine MSS heights and their variations in the regions of China seas including Yellow Sea, East China Sea and South China Sea with T/P and ERS-1 altimeter data during the period from October of 1992 to June of 1998. After having done the corrections of T/P altimeter instruments bias, and geophysical environment corrections of tide, ionosphere, troposphere, sea-state bias and inverse barometry, we find that rising rates vary in different regions. Compared with global annual sea level ring rate (+2.1±1.3) mm/a, the annual rising rate in Yellow Sea, East China Sea and South China Sea is (+3.44±0.61) mm/a, (+3.12±0.47) mm/a and (-1.41±0.48) mm/a, respectively. From sea level anomalies, it can be seen clearly that the influences of El Nino in 1993, 1994, and 1997-1998 are greatest in South China Sea, less in East China Sea and least in Yellow Sea (Hu et al., 2001; Wang et al., 2000).
Based on the characteristic of the perfect spatial distribution of the T/P altimeter data, a spatial analysis method is performed, which transfers the constituents harmonic constant H and g into a pair of orthogonal parameters U and V, and then expresses each of them with a polynomial function. As parameters, the polynomial coefficients are derived with altimeter data on the least squares criteria. Thus the models of the main tidal wave in south sea are established. 72 weeks T/P data through weeks 11 to 82 are included in the calculation. The models are evaluated with different approaches and data set. The conclusion is that the tide models can provide partial tide amplitudes with 3cm accuracy, phase lags deviation of those amplitudes which are larger over 10 cm are within ±10°, and the tide models derived are better than those of Schwiderski and SR95.1 (Bao et al., 1999; Bao et al., 2000).