Skip navigation


You are in:  Home » Publications » Reports » Report 23

SCAR Report 23

Application of SAR Interferometry in Grove Mountains, East Antarctica

E Dongchen(1), Zhou Chunxia(1), and Liao Mingsheng(2)
1 Chinese Antarctic Center of Surveying and Mapping, Wuhan University, Wuhan, 430079, P.R.China
2National Key Lab. For Information Eng. in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan, 430079, P.R.China,

Email: edc1939@public.wh.hb.cn

Abstract

Synthetic aperture radar interferometry has been proposed as a potential technique for digital elevation model (DEM) generation, topographic mapping, and surface motion detection especially in the inaccessible areas. Grove Mountains Area locates to the southwest of Princess Elizabeth Land, inland areas of east Antarctica. The topographical map of the core area (11°¡10 KM2) was printed after the field surveying with GPS and total station was finished under the atrocious weather conditions during the 16th CHINARE (Chinese National Antarctic Research Expedition) 1999/2000. This paper will present an experimental investigation of the ERS-1/2 SAR tandem data in 1996 on DEM generation of the Grove Mountains Core Area, analyze the data processing, and compare the DEM with the actual topographic form. It is confirmed that InSAR is a very useful technique to be utilized in Antarctica, and can be used to produce more products instead of dangerous field surveying.
Key words: Synthetic aperture radar interferometry, ERS-1/2 Tandem, GPS, DEM, Antarctica.

1. Introduction

Maps of Antarctica’s interior remained mostly white blanks into the mid-1980s. Satellites using visible light had produced detailed surface images, but their angles of view excluded more than 1.2 million square miles poleward of about 82° south latitude. Then in 1997, the Canadian RADARSAT-1 satellite was rotated in orbit. With its synthetic aperture radar (SAR) antenna looking south towards Antarctica, it permitted the first high-resolution mapping of the entire continent of Antarctica. In other areas of Antarctica, DEM and topographic mapping have been obtained by means of different methods or their integration such as synthetic aperture radar, radar altimeter, laser altimeter, radar echo sounding, GPS surveys, aerial photographs, and geodetic maps.
In 1998, China planned to take the first time expedition to Grove Mountains Area (see also Figure 1), which locates to the southwest of Princess Elizabeth Land, inland areas of east Antarctica. Adopting Landsat4 TM images, Chinese Antarctic Center of Surveying and Mapping (CACSM) had completed the colorful satellite image map of the Grove Mountains in the scale of the 1:100 000 in August 1998 to ensure the expedition route and navigation to Grove Mountains Area during the 15th CHINARE 1998/1999. Then the topographical map of the core area (see also Figure 2) was printed after the field surveying with GPS and total station was finished under the atrocious weather conditions during the 16th CHINARE 1999/2000.

Figure 1: Grove Mountains, East Antarctica.

Figure 2: Grove Mountains Core Area.

However, in Antarctica, traditional mapping method is no longer a most efficient means to obtain the topographical maps or DEM in large areas especially in abominable environment. Synthetic aperture radar interferometry has been proposed as a potential technique for digital elevation model (DEM) generation, topographic mapping, and surface motion detection especially in the inaccessible areas. So we bought radar image data from ESA and a case study for DEM generation was done in Grove Mountains. In this paper, the primary experiment result based on ERS-1/2 tandem data is presented. Moreover, with the hard-won field surveying data, comparing and analyzing DEM generated by using tandem radar image data with DEM generated with the topographic points is carried out.

2. Methodology

The tandem operation of ERS-1 and ERS-2 satellites, with a short temporal baseline, put forward a better time correlation for DEM generation. It utilizes the two single look complex image of the same area to form interfere and further obtain the three dimension information. The principle to get height h is illuminated in the following geometry figure.

Figure 3: Geometry of SAR interferometry.

In this study, the interferometric SAR data processing mainly includes: (1) coregistration of the complex image data; (2) formation of the interferogram; (3) Phase unwrapping; (4) DEM generation. Indeed, baseline refinement, removal of flat earth, noise filtering, etc, are always indispensable to obtain a high-precision DEM.
DEM of Grove Mountains was generated by the ERS tandem data. Since we have DEM generated with the topographic points obtained during the field surveying, then we can perform a comparison.

3. Field Surveying

Grove Mountains Area, with bare peaks at inland areas of east Antarctica, is located to the south of the Zhongshan Station about 400 km. Its geographical extension is 72°„40°‰– 73°„10°‰S, 74°„00°‰– 75°„45°‰E, and the area is about 3200 km2; meanwhile, the core area extension is 72°„50°‰54°Â– 72°„56°‰20°ÂS, 74°„54°‰07°Â– 75°„14°‰09°ÂE, and its area is about 110 km2. Grove Mountains is of typical in-land character and also an ideal midway station place for expedition teams extending to the South Pole.
In the core area, there are two exposed mountains, many rock peaks, and detritus strips on the surface of the ice sheet with the altitude of 2000 meters, which has great topographical undulation and is densely covered by ice crack. The weather there is atrocious for it has blustery or milky weather half of a year and the average temperature is about thirty degrees below zero centigrade, which brings great difficulties for field surveying and operations.
TM color satellite image map and ERS-1 radar image (1996/02/10) of the core area are illustrated in Figure 4 and Figure 5 respectively. In TM image map, ice face is in light green, blue-ice face is in blue green, horn and bare rocks are in brown. The Mount Harding, Zhakroff Ridge, Jingyu Peak, Tianhe Range, Zhonghua Peak, and Lianhua Peak can be easily interpreted from both of them, and the general position is coherent. Meanwhile some difference occurs inevitably because radar image and TM image are in different characters and the error exists.

Figure 4: TM color satellite image map.

Figure 5: ERS-1 radar image.

In order to provide geologists with the topography of Grove Mountains, our geodetic surveyors has conducted the field surveying with GPS and total station and completed the mapping experiment in the Grove Mountains Core Area during the 16th CHINARE 1999/2000 summer expedition. The topographical map of the core area at the scale of 1:25 000 (see also Figure 6) was printed after the field surveying was finished in 31 days by two surveyors of CACSM under the atrocious weather conditions, and 14,300 topographical points were obtained through post-processing differential GPS (DGPS) technique and forward intersection method with total station. WGS-84 coordinate system and Transverse Mercator map projection are adopted. The center meridian is 75_E, and the vertical contour interval is 10 meters.

Figure 6: Topographical map of Grove Mountains Core Area.

Figure 7. Interferogram of the core area

4. DEM Generation with INSAR and data interpretation

The information of the tandem single-look complex image pair is listed in Table 1. The perpendicular baseline is about 164m, and parallel baseline is about 94m. Baseline parameter plays a very important role in flat earth effect removal and geometric transformation from phase to elevation, so it is very important to adopting precise baseline parameter. In practice, baseline accuracy at centimeter level is the basic requirement for producing high-precision DEM. ESA attached five sets of orbit vector data at intervals of 4.2 seconds within the head file while distributing SAR image data. In this study, we adopted ERS-1/2 high precision orbits calculated and provided by Delft Institute for Earth-Oriented Space Research (DEOS).

Table 1. ERS-1/2 tandem data information

Interferogram, which is defined as the product of the complex SAR values of the second image with the complex conjugate of the reference image, includes both the amplitude and phase information of SAR image pair. And it is the basis for DEM generation. Figure 7 shows the interferogram of Grove Mountains Core Area. After the fieldwork, we get to know that disordered blocks of rocks and snow cover on the tops of the mountains and peaks, which causes the discontinuous interferometric phase fringe of these areas.
Once the phase is unwrapped, an absolute phase is required to obtain the absolute pixel height. A point with known elevation in the scene can be used to provide an absolute elevation reference. In Grove Mountains, two geodetic control points and one are set on the top of Mount Harding and Zhakroff Ridge respectively, while it is impossible to find their location in radar image because the field GPS surveying was done in 2000 and the SAR images were acquired in 1996. And even if they are in the radar image, it is also difficult to find them. For the particular environment in this location, it’s trouble to find feature points too. On the other hand, ice-sheet flowed and changed a lot in such a long period. So we could only find a relatively flat and stable area to appoint a point as the reference point. Area west to Mount Harding is relatively stable because of the blocking effect of Mount Harding. Meanwhile, with reference to the coherence of the core area (see also Fig.8), which depends on the terrain conditions, the brighter means the better coherence. Generally speaking, the areas covered mostly by ice and snow to the west of the Mount Harding and Zhakroff Ridge have relatively better coherence. Other areas covered by ice and snows such as northwest terrace take second place. And coherence of the top of the mountains and peaks are the worst. A point to the west of Mount Harding with rough elevation of 1867 meters is selected as the reference point.

Figure 8: Coherence of the core area

Figure 9: DEM generated by using tandem radar image data

After phase unwrapping, the DEM was generated. In order to show intuitionistic vision, color perspective model of the grid DEM is formed as Figure 9. In those areas of high topographic deviationism, the deviation is fairly large for the layover and shadow effect caused by side-glanced radar observing mode. Compared with the perspective model from the DEM generated with the topographical points obtained during the field surveying (see also Figure 10), the terrain tendency and the main terrain character are coincident. Red stands for the higher, and blue the lower. Mount Harding and Zhakroff Ridge are obvious in Figure 9 and Figure 10, while Jingyu Peak, Zhonghua Peak and Lianhua Peak are not exactly presented in Fig.9. Meanwhile, Fig.9 doesn’t show the valley and lower terrain that can be clearly given in Fig.10. The details of the topography information in the two following figures exist differences and need more study and analysis.

Fig.10. DEM generated with the topographic points obtained during the field surveying

Table 2. Results of the comparison of the two DEM

From 1996 to 2000, changes in Grove Mountains may be caused by snow accumulation, ice-snow melting and ice sheet flowing, and ice sheet flowing is the main factor. However, according to the topographic feature, ice sheet flow and the coherence of the core area, relatively flat and stable areas can be found. It is more reasonable to assess the DEM difference in three different terrains listed in Table 2 than only to compare the whole area.
With the measured results listed in Table 2, the quality of DEM difference of the relatively stable areas is preferable, which confirms that InSAR is valid to be utilized in this area. For the flowing ice surface, the corresponding DEM difference gets larger, which is mainly caused by the changes in ice surface. The result is unsatisfied in the hilly terrains because of the surface of the mountains covered by snow and blocks of rocks, which brings noises and difficulties to obtain accurate height.

5. Conclusions and Future Activities

From the primary study and other researchers’ study on DEM generation and topographic mapping, it could be shown that InSAR will be very useful to be utilized in Antarctica as a new mean for producing topographic products more effective instead of field surveying.
In order to obtain high-precision DEM, formation of the interferogram, phase unwrapping, and other crucial steps must be further studied to reduce the error. Band C can penetrate to the ice surface, but here this penetration effect to DEM generation is neglected. Moreover, we can analyze the properties of snow cover on InSAR phase and coherence, and the effect on DEM generation and mapping.
Scientists have been very interested in Antarctic ice sheet flow, ice sheet kinematic characteristics, and mass balance. InSAR can also be an effective tool in these research fields by adopting more pairs of radar image data.
During the 19th CHINARE 2002/2003, our researchers has set eight ground control points covering all Grove Mountains area, which will help us to utilize interferometric SAR image data or others to precisely produce the topographical map of the whole Grove Mountains area and extract the topographic information of Antarctic inland ice sheet.

References

  1. P.A. Rosen, I.R. Joughin, F.K. Li, S.N. Madsen, E. Rodriguez and R.M. Goldstein, “Synthetic Aperture Radar Interferometry”, Proceedings of the IEEE, vol.88, no. 3, pp. 333-382, March 2000.
  2. G. Rufino, A. Moccia and S. Esposito, “DEM Generation by Means of ERS Tandem Data”, IEEE Trans. Geosci. Remote Sensing, vol. 36, no. 6, pp. 1905-1912, Nov. 1998.
  3. M. Coltelli, G. Fornaro, G. Franceschetti, R. Lanari, G. Puglisi, E. Sansosti and M. Tesauro, “ERS-1ERS-2 tandem data for digital elevation model generation”, IEEE Proceedings of IGARSS ’98, vol. 2, pp. 1088 –1090, 1998.
  4. H.A. Zebker and R.M. Goldstein, “Topographic mapping from interferometric SAR observations ”, J.Geophys.Res., vol. 91, pp. 4993-4999,1986.
  5. H.A. Zebker, C.L. Werner, P. Rosen, and S. Hensley, “Accuracy of Topographic Maps Derived from ERS-1 Interferometric Radar”, IEEE Trans. Geosci. Remote Sensing, vol. 32, no. 4, pp. 823-836, July 1994.
  6. M. Schwäbisch, M. Matschke, W. Knöpfle, and A. Roth, “Quality Assessment of InSAR-derived DEMs Generated with ERS Tandem Data”, IEEE Proceedings of IGARSS ’96, vol. 2, pp. 802-804, 27-31 May 1996.
  7. R.M. Goldstein, H.A. Zebbker, and C.L. Werner, “Satellite Radar Interferometry: Two-Dimensional Phase Unwrapping”, Radio Science, vol. 23, no. 4, July-August 1988.
  8. M. Schwäbisch, “Large Scale Interferometric DEM Generation using ERS Tandem Data: Example of the Czech Republic”, IEEE Proceedings of IGARSS’98, vol. 3, pp. 1641-1643, 6-10 July 1996.
  9. K.E. Mattar, A.L. Gray, Marco W.A. van der Kooij and P.J. Farris-Manning, “Airborne Interferometric SAR Results from Mountainous and Glacial Terrain”, IEEE Proceedings of IGARSS ’94, vol. 4, pp. 2388-2390, 8-12 Aug 1994.
  10. R. Abdelfattah, J.M. Nicolas, M.R. Boussema, “Impact of Orbital Parameters on DEM Production by SAR Interferometry”, IEEE Proceedings of IGARSS’2000, vol. 2, pp. 739-741, 2000.
  11. R. Goldstein, “Atmospheric Limitations to Repeat-Track Radar Interferometry”, Geophys. Res. Letters, vol. 22, no. 18, pp. 2517-2520, 1995.
  12. K. A. Hogda, T. Guneriussen and I. Lauknes, “Synthetic aperture radar for DEM generation in snow-covered mountain terrain”, IEEE Proceedings of IGARSS ’2002, vol. 4, pp. 2193-2195, 2002.
  13. Shi Shiping, “DEM Generation using ERS-1/2 Interferometric SAR Data”, IEEE Proceedings of IGARSS’2000, vol. 2, pp. 788 –790, 2000.
  14. RADARSAT-1 Antarctic Mapping Project, http://www.asf.alaska.edu/4_4.html.