Keywords: Satellite Altimeter, Sea Level Anomaly, China Seas

Abstract
In this paper, the collinear method is used to determine mean sea surface heights and its variations in the region of China Seas with Topex/Poseidon and ERS-1 altimeter data during the period from October of 1992 to June of 1997. After having done the corrections of T/P altimeter instrument bias and the geophysical environment corrections of tide, ionosphere, wet and dry troposphere, sea-state bias and inverse barometry, we find that the rising rates vary in different regions. Compared with the global annual sea level rising rate (+2.1 ± 1.3mm/yr), the annual rising rate in Yellow Sea, East Sea and South China Sea is +3.44 ± 0.61 mm/yr, +3.12 ± 0.47 mm/yr, and –1.41 ± 0.48 mm/yr, respectively. From the sea level anomalies, it can be seen clearly that the influence of El Nino in 1993, 1994, and 1997-1998 is greatest in South China Sea, less in East Sea and least in Yellow Sea. In addition to normal FFT spectrum analysis, a new wavelet analysis of WAPS was also used to the sea level anomalies. The results show that the annual period is stationary in all interest regions, but the semi-annual and seasonal periods have some degree of period drift. In addition, we also found there exists a clear two months period in all three regions whose mechanism is still to be explored.

The tidal gauge measurements are usually used in the traditional analysis of sea level variations but ones often have to face the problem of even locations of tidal gauges and the deformation of crust where tidal gauges locate. Thanks to the new remote sensor of satellite altimeter which was actually started with the launch of Skylab in 1978 by NASA, the world-wide sea surface heights in all-weathers can be measured in short time, such as 10 days, 17 days or 35 days. The altimeter missions of Geos-3 (NASA, 1975), Seasat (NASA, 1978) and Geosat (US Navy, 1985) prove the unique role of satellite altimeter in aspects of the determination of ocean circulation, sea surface topography and geoid undulations, and the improvement of gravity geopotential model. Because of one order improvement in radial orbit error, better corrections of atmospheric medium (ionosphere and troposphere), and the better oceanic tidal model, the unprecedented accuracy of ERS-1 (ESA, 1991), Topex/ Poseidon (NASA/CNES, 1992), ERS-2 (ESA, 1995), or GFO (US Navy, 1998) satellite altimeter is achieved. The altimeter data, which directly measure satellite-to-sea distances or indirectly measure sea levels, are very important to research in geodesy, geophysics as well as oceanography. As a result, a new generation of satellite altimeter, such as Envisat (ESA, 2005), Jason (NASA, 2005) will be launched in the first decade of coming 21^st century.

In this paper, Topex/Poseidon and ERS-1 altimeter data during the period from October of 1992 to June of 1997 are used to determine mean sea surface heights and its variations in the region of China Seas with the collinear method. After having done the corrections of T/P altimeter instrument bias and geophysical environment corrections of tide, ionosphere, wet and dry troposphere, sea-state bias and inverse barometry, we get the variations of sea surface heights in the China Seas and vicinity. Besides common harmonic analysis and FFT analysis, a new wavelet analysis of WAPS is also used to the variations of mean sea level. The results in the region of Yellow Sea, East Sea and South China Sea will be discussed in detail.

Altimeter Data Processing
Topex/Poseidon (T/P) and ERS-1 satellite altimeter data used in this paper are from the AVISO CD ROMs of corrected Sea Surface Heights (SSH) (AVISO, 1996). The time span of altimeter data covers from October 3^rd, 1992 to June 12, 1998. The interest regions are China Yellow Sea (119° - 126° E, 33° - 37° N), East Sea (120° - 127° E, 26° - 33° N) and South China Sea (110° - 119° E, 14° - 23° N). The period of each T/P cycle is 10 days and the period of each ERS-1 cycle is 35 days. The length of each 1-second corrected SSH is 32 bytes which contains 6 parameters, i.e., latitude, longitude, time, the corrected SSH, a reference mean sea surface (OSUMSS95) and inverse barometer correction. The corrected SSHs have been corrected by the EM bias, tidal corrections, wet and dry troposphere corrections, ionosphere correction, radial orbit corrections, gravity center correction of instrument, and inverse barometer effect. The correction models used for corrected SSH are shown in Table 1. For T/P altimeter data, the accuracy of the 1- second corrected SSH is estimated about 5 cm (Fu, et al., 1994), while the precision of T/P radial orbit is about 3-4 cm due to the higher precision tracking system of SLR and DORIS, the higher altitude of satellite orbit, the improvement of non-conservative force model and Earth gravity model, and the satellite-to-satellite tracking technique of GPS (Tapley, et al., 1994). Therefore, no orbit corrections are calculated for T/P, i.e., T/P orbits are taken as the reference level and the radial orbit error of ERS-1 is reduced by fitting the ERS-1 orbits into the more precise T/P orbits with dual-satellite crossover adjustment. This provides the accuracy of ERS-1 orbit similar to that of T/P orbit: 2 cm (Le Traon, et al., 1995). Moreover, the ERS-1 bias and any long-wavelength errors are simultaneously estimated with orbit error. This provides homogeneous precision between ERS-1 and T/P altimeter data. Because the repeating periods of T/P and ERS-1 cycles are different, ERS-1 altimeter data are also assimilated to T/P altimeter data, i.e., the repeating period of the assimilated altimeter data is 10 days. The assimilated altimeter data contain whole T/P data in a 10-day cycle and a part of ERS-1 data of a 35-day cycle. In addition, the Topex altimeter range calibration corrections (http://obs.wff.nasa.gov) derived from on-broad internal instrument measurements [Hayne et al., 1994] are also used. There is a shortening of the range over 1992-1998 of 0.52 ± 0.9 mm/yr. However, much of the error due to random noise still exists. So if an approximate smoothing is made, the accuracy can be increased to about 2-3 cm (Cheney, et al., 1994; Mitchum, 1994). In this paper, an average of measurements is interpolated every 10 seconds along each T/P and ERS-1 track with a criterion of three times of standard deviation. This smoothing scheme reduces the altimeter noise by approximately a factor of 2 while preserving sufficient horizontal resolution.
Table 1. The environmental corrections used in the corrected sea surface heights

corrections		Topex/Poseidon	ERS-1
Orbit		NASA JGM3 orbits	D-PAF orbits
Geophysical Corrections	Dry troposphere Wet troposphere Ionosphere Inverse barometer	from ECMWF from TMR radiometer TOPEX:from dual frequency altimeter range POSEIDON:from DORIS from ECMWF	from ECMWF from ATSR-M radiometer BENT model from ECMWF
Sea State Bias (EM Bias)		BM4 formula	-5.5% of significant wave height
Tides	Ocean tide & loading tide Solid tide Pole tide	CSR 3.0 Cartwright and Taylor model (1971) applied	CSR 3.0 Cartwright & Edden model with Wahr’s radial correction not applied