Space application data promoting service platform for China Manned Space Engineering

Current location:

A Preliminary Analysis of Sea Surface Height Measurement Accuracy of Tiangong-2 Three-dimensional Imaging microwave altimeter

2018-06-05 09:54

Accurate measurement of sea surface height has important implications for studying global environmental changes such as sea level changes and ocean flow fields. At present, sea surface altitude measurement is still based on satellite altimeter, but it can only measure one-dimensional sea surface height information in the direction of the flight trajectory. As shown in Figure 1, if we want to measure the height of a two-dimensional sea surface, we can only use the satellite's adjacent flight orbit to perform two-dimensional spatial sampling along the distance. Since the distance between adjacent satellite orbits is usually between 200-300 km, traditional satellite altimeters are mainly used to study large-scale ocean phenomena above 300 km. In recent years, there has been an increasing demand for research on marine phenomena such as medium and small-scale vortices and ocean currents of the order of 15-300 km, while InSAR combines the advantages of high-precision height measurement and wide swaths. Therefore, InSARs designed specifically for sea level measurement have been designed. It has become the most promising solution for studying this type of oceanic phenomenon.

Fig. 1 Schematic survey of sea surface height by satellite altimeter

The Interferometric imaging radar altimeter (InIRA) is the world's first wide-shunt ocean altimeter. Its two Ku-band antennas have a baseline length of 2.3m and an incident angle of 2.5-7.5 degrees. At the right side of the sub-satellite point, sea surface height measurement data exceeding 30 km in width is obtained. Since the launch of Tiangong-2, the platform has been operating stably, InIRA is working properly, and a large amount of high-quality detection data has been acquired. In 2021, the SWOT satellite will carry the same type of altimeter jointly developed by the United States and France.
Figure 2 is the amplitude map of part of the South China Sea acquired by InIRA. The South China Sea is famous for its inner waves. The ocean internal wave can be highlighted on the SAR image by modulating the roughness of the sea surface and influencing the radar echo intensity. This oceanic phenomenon is exactly recorded in this data. Under medium incident angle, the sea surface microwave signal received by InIRA is mainly based on Bragg scattering energy. Under the near-nadir incident angle, the received sea surface microwave signal is mainly specular scattering energy. Therefore, the sea surface echo in Figure 2 is significantly stronger than land, which is in contrast to the general SAR image.

Fig. 2 InIRA acquires the amplitude of the complex image

Since the interferometric phase accuracy of InIRA is mainly affected by factors such as decoherence of signal-to-noise ratio, scattering of wave body, decoherence of spatial waves, and decoherence of spatial baseline, the precision of InIRA is simulated in accordance with the system parameters of InIRA in accuracy analysis. The accuracy of the data estimation is compared and the accuracy is evaluated.
Figure 3 shows the phase diagram of the interferometric interferometric phase showing the significant phase fringes on the sea surface and flat land. This is a typical “flat” phase phenomenon in InSAR elevations, ie no change in the height of the sea surface or terrain, but there will be regularly varying interferometric phase fringes in the distance.

Fig. 3 Interferometric phase diagram after registration

Figure 4 shows the coherence coefficient between complex images. It can be seen that most of the sea surface coherence coefficients are above 0.9, but the far-end coherence coefficient is reduced, which is mainly due to the decrease of the echo signal-to-noise ratio; while the mountain area coherence coefficient is overall low, about 0.3-0.7. This is mainly due to the fact that lush mountain vegetation produces significant bulk scattering of Ku-band short-wavelength signals, which in turn causes decoherence.

Fig. 4 Coherence coefficient diagram

Figure 5 is an interference phase diagram after removing the ground phase. On the left side of the map, visible interference fringes can be seen in the mountainous area, which is caused by the height change in the mountains. The height of the sea surface on the right side of the graph is much smaller than that of the land, and the corresponding elevation phase is very small, so the overall sea surface interference phase tends to zero.

Fig. 5 Interferometric phase diagram after removing the ground phase

Due to the short InIRA baseline, the interference phase corresponding to a 10 cm sea level change is only 0.01-0.001 rad from the proximal to the far end of the swath and the phase noise is significant. Even if there are some fluctuations in the sea surface, the elevation phase is difficult to be reflected in the interference phase diagram at this spatial resolution. Therefore, first select the sea surface interference phase in the black box in Figure 5, and perform a large-scale spatial multi-view processing at the sampling intervals of 1 km and 5 km, respectively, to reduce the phase noise, and finally compare and analyze with the theoretically measured high accuracy. As shown in Figure 6, it can be seen that in the entire mapping zone, the high precision of the 5km sampling interval is generally higher than the 1km sampling interval, which is mainly due to the larger spatial multi-view number and the phase of the 5km sampling interval. The noise level is lower; at different sampling intervals, the overall accuracy of the actual data estimation is lower than the theoretically high accuracy.

Fig. 6 Comparison of theoretical and practical estimation accuracy at different pixel sampling intervals

In order to better evaluate the accuracy, this time in the range of 1-7° incidence angle, the difference between the actual estimation accuracy and theoretical accuracy of the 1km and 5km sampling intervals was analyzed. As shown in Figure 7, it can be seen that the accuracy of the 1km and 5km sampling intervals is very close within the range of 4-5° incidence angle, and they are all around 4-6cm. This result also shows that there are biases introduced by the ideal sea-level hypothesis when estimating high-precision data using actual data, as well as residual calibration errors and data processing deviations.

Fig. 7 Difference between actual estimation accuracy and theoretical accuracy

The preliminary analysis of the high-precision data accuracy of the three-dimensional imaging altimeter of Tiangong-2 in some sea areas of the South China Sea shows that the high-precision measurement at a sampling interval of 1 km is low, but if the sampling interval of the pixel is increased to 5 km, a better measurement accuracy can be obtained. . In addition, further reducing the impact of system error will make the Tiangong-2 three-dimensional imaging microwave height measurement and high precision to better meet the research needs of various types of marine phenomena.

Author:Kong Weiya, Zhong Jinsong. Institute Of Electronics, Chinese Academy Of Scences.


Weiya Kong, Jinsong Chong*, Hong Tan*. Performance Analysis of OceanSurface Topography Altimetry by Ku-Band Near-Nadir Interferometric SAR[J],Remote Sensing, 9(9), 2017, doi:10.3390/rs9090933.