Nazemeh Ashrafianfar, Hans-Peter Hebel and Wolfgang Busch

Size: px

Start display at page:

Download "Nazemeh Ashrafianfar, Hans-Peter Hebel and Wolfgang Busch"

Elisabeth Wilkinson
5 years ago
Views:

MONITORING OF MINING INDUCED LAND SUBSIDENCE - DIFFERENTIAL SAR INTERFEROMETRY AND PERSISTENT SCATTERER INTERFEROMETRY USING TERRASAR-X DATA IN COMPARISON WITH ENVISAT DATA ABSTRACT Nazemeh

1 MONITORING OF MINING INDUCED LAND SUBSIDENCE - DIFFERENTIAL SAR INTERFEROMETRY AND PERSISTENT SCATTERER INTERFEROMETRY USING TERRASAR-X DATA IN COMPARISON WITH ENVISAT DATA ABSTRACT Nazemeh Ashrafianfar, Hans-Peter Hebel and Wolfgang Busch Institute of Geotechnical Engineering and Mine Surveying, Clausthal University of Technology, Erzstr.18, Clausthal-Zellerfeld, Germany nazemeh.ashrafianfar@tu-clausthal.de; hans-peter.hebel@tu-clausthal.de; wolfgang.busch@tu-clausthal.de In October 2006 an unexpected flooding occurred in the first potash mining area of Uralkali Company (BKRU-1) in Berezniki city, western Ural region, Russia. This phenomenon caused a huge sinkhole in the area in July Regards to the geological and mining situation of the area, it has been expected to occur more disaster and land subsidence in parts of the city. The subsidence in the area has been monitored since May 2006 by using ENVISAT data and since May 2008 by TerraSAR-X data. All TerraSAR-X data (StripMap mode) have been ordered by DLR through the TerraSAR-X project GEO0348. TerraSAR-X data due to high spatial (ground resolution better than three meters) and temporal (every 11 days) characteristics offer the best possibilities to monitor vertical displacement in this area. We present the results of this project in time periods of 2008 to TerraSAR-X data have been analyzed by Differential SAR Interferometry (DInSAR) and Persistent Scatter Interferometry (PSI) methods and also have been compared with results using ENVISAT data. Results of DInSAR and PSI methods using TerraSAR-X data show high agreement, only in some parts of the area with high subsidence in a short period of time, PSI results cannot be more reliable. As the results of comparison TerraSAR-X and ENVISAT analysis by DInSAR method, it is considerable that TerraSAR-X data could detect all subsidence areas resulted from ENVISAT data, in addition to some other subsidence areas, which were detected only by using TerraSAR-X data. It is due to higher spatio-temporal resolution of the TerraSAR-X data. The DInSAR analyses of both data sets show comparable results. 1. INTRODUCTION This paper presents the results of monitoring progressive subsidence due to mining activities in Berezniki city, Russia (Fig. 1). Berezniki is a mining city in the perm region, which is famous due to producing around 10% of the world potash. In October 2006 an unexpected water flooding occurred in the first potash mine of Uralkali Company. In summer 2007 the first sinkhole in Berezniki occurred (Fig. 2). In October 2008 this sinkhole extended about 423 meters long, 310 meters wide and 100 meters deep [3] and swallowed the main railroad line between Perm and Cherdyn. The actual distance of this sinkhole to the bypass railroad line via Berezniki city was in November 2010 only about 60 meters. Due to importance of monitoring this critical condition in the Berezniki area, in this research progressive subsidence in different parts of the city has been monitored via Differential SAR interferometry (DInSAR) and Persistent Scatter Interferrometry (PSI) [4] techniques. To this aim, ENVISAT data (since May 2006) and TerraSAR-X data (since May 2008) has been used. According to previous studies [3], several subsidence areas (Fig. 3) have been detected in Berezniki. In this paper, we present more details about the subsidence condition from 2008 to In the following sections, we explain briefly existing data and compare results of DInSAR and PSI methods using ENVISAT and TerraSAR-X data. Figure 1: Berezniki city in Perm region, Russia 1

2 Figure 2: Sinkhole in June 2008 [3] Figure 3: Subsidence areas in Berezniki city, Russia. Red boxes show subsidence areas detected with TerraSAR-X and ENVISAT data. Purple boxes show subsidence areas detected only with TerraSAR-X data. 2

2. TERRASAR-X DATA TerraSAR-X data for this project have been programmed and ordered by German Aerospace Center (DLR) through the TerraSAR-X project GEO0348.

59 meters in slant range and 1.90 meters in azimuth. Until November 2010, about 74 TerraSAR-X StripMap scenes have been delivered by DLR (Fig. 4).

Blue boxes show the winter time with snow-cover in the area. Figure 4: Time-Baseline diagram of TerraSAR-X data 3. METHODS AND RESULTS 3.

3 2. TERRASAR-X DATA TerraSAR-X data for this project have been programmed and ordered by German Aerospace Center (DLR) through the TerraSAR-X project GEO0348. They are strip_009r data with incident angle between 34 and 37 degrees. DLR has delivered all data as single look slant range complex. StripMap data have a pixel spacing of 1.59 meters in slant range and 1.90 meters in azimuth. Until November 2010, about 74 TerraSAR-X StripMap scenes have been delivered by DLR (Fig. 4). However, about 17 scenes of these TerraSAR-X scenes due to snow cover couldn t be used in DInSAR analysis. Fig. 4 shows time to baseline diagram of TerraSAR-X data. Blue boxes show the winter time with snow-cover in the area. Figure 4: Time-Baseline diagram of TerraSAR-X data 3. METHODS AND RESULTS 3.1 Differential SAR Interferometry (DInSAR) To use DInSAR method all scenes have been co-registered respect to the scene of , and an average intensity image has been calculated. The SRTM DEM (by DLR) also has been co-registered respect to the average intensity image. Then, all differential interferograms have been calculated, and linear phase trends have been subtracted (Fig. 5). Afterwards interferograms have been spectrally filtered and unwrapped. In the last step, vertical subsidence has been calculated. Figure 5: Original interferogram (left), linear phase trend of the original interferogram (middle) and interferogram after subtraction of linear trend (right). Time interval of the interferogram: to

To calculate subsidence over a longer time than 33 days we added results of existing 11, 22 or 33 days successive interferograms.

4 To reach high coherency in the procedures of TerraSAR-X interferogram analysis, spatio-temporal baseline has the most important effect. Interferograms for longer time intervals than 33 days show low coherency due to change of vegetation cover in the area. To calculate subsidence over a longer time than 33 days we added results of existing 11, 22 or 33 days successive interferograms. It can be explained when there are two successive interferograms AB and BC, we can calculate subsidence of AC. In this case, the most amount of atmospheric effect of the scene B is eliminated. Fig. 6 and Fig.7 show the results of such additions in the study area. Fig. 6 shows subsidence contour lines in a part of the area for the time interval between and (left) and for the time interval between and (right). Figure 6: Subsidence contour lines in millimeter (addition of all successive 11-day-interferograms) in a part of the area, time intervals between and (left), and (right) Fig. 7 shows vertical displacements during different time intervals along a profile line in the time span between and The displacement data are the average values of all PS-points within 20 meters from both perpendicular sides. We compared results of the DInSAR method using TerraSAR-X and ENVISAT data. Fig. 8 shows results of this comparison. In spite of different spatio-temporal properties of TerraSAR-X and ENVISAT data, the results of DInSAR using both data shows high comparability. This diagram also has been 20 meters from both sides smoothed. 4

5 Figure 7: Vertical displacements along a profile line over time intervals between and (addition of all successive 11/22/33-day-interferograms) Figure 8: Diagram of time to vertical displacement along a profile for comparison results of DInSAR method by TerraSAR-X and ENVISAT Data 3.2 Persistent Scatter Interferometry (PSI) PSI is a powerful technique to measure and monitor vertical subsidence. There are two main differences between DInSAR and PSI techniques [2]: first, PSI uses large series of RADAR images (typically more than 20) and second, implements suitable data modeling and analysis procedures to achieve time series of the deformation. This method is sensible to small deformation rates until a few millimeters per year [2]. For PSI analysis with TerraSAR-X data in this project the new method of GAMMA Remote Sensing AG [4] has been used. In this method, to optimize the processing in the case of the high deformation gradients and non-linear movements, a multi reference stack has been used, which includes pairs with shorter time intervals [4]. An overview of PSI-points diversion in Berezniki city is shown in the Fig. 9 (left). The average point density over the entire study area 5

is about 2050 points per km². The point-wise results of PSI have been converted to a surface-model by Kriging interpolation method (Fig.

and 02.11.2010. The results of the Kriging interpolation from PSI points (right) Fig.

However, in areas with high subsidence in a short period of time (here between 10.05.2010 and 02.11.2010) these results show high difference to each other (Fig.

6 is about 2050 points per km². The point-wise results of PSI have been converted to a surface-model by Kriging interpolation method (Fig. 9, right) Figure 9: Point-wise results of PSI method (left) for the time interval between and The results of the Kriging interpolation from PSI points (right) Fig. 10 shows the results of comparing the PSI and DInSAR methods using TerraSAR-X data. These results show high comparability. However, in areas with high subsidence in a short period of time (here between and ) these results show high difference to each other (Fig. 9, red-box). It is because of the existing only a few numbers of the PSI points with suitable neighborhood distance in the procedures of the PSI method in this area. Figure 10: Comparison the results of DInSAR, PSI-points and PSI-Kriging, red box shows the area with incorrect result from PSI method, due to high rate subsidence in a short period of time in this area. 6

7 4. CONCLUSIONS The results of the project confirm that TerraSAR-X data can be used to detect land surface displacement in a long period of time in Berezniki. It is due to high spatio-temporal resolution of TerraSAR-X data for monitoring subsidence. The new PSI method of GAMMA Remote Sensing AG can be used to calculate vertical subsidence in this area by TerraSAR-X data, in spite of non-linear behavior of the subsidence. It is first due to 11 days repeat of TerraSAR-X data and second, high spatial resolution of the data. Results of DInSAR analysis using TerraSAR-X and ENVISAT show high comparability. Results of the DInSAR and PSI method using TerraSAR-X data show high agreements, only in some parts of the area with high subsidence in a short period of time there is difference between these results. Results of this project confirm that TerraSAR-X data with high spatio-temporal resolution using PSI and DInSAR methods makes a powerful combination for monitoring land subsidence in Berezniki. ACKNOWLEDGMENT TerraSAR-X data used in this project has been provided by DLR through the project GEO0348 for Institute of Geotechnical Engineering and Mine Surveying (IGMC), TU Clausthal, Germany (PI Hans-Peter Hebel). REFERENCES 1. Busch, W., (2008): Radarinterferometrische Erfassung von Bodenbewegunggen im Gebiet Berezniki (Perm, Russland). 8th Altbergbaukolloquium, 6-8 November 2008, TU Clausthal, VGE Verlag GmbH, Essen, pp Crosetto, M., Montserrat, O., Iglesias, R., Crippa, B., (2010): Persistent Scatter Interferometry: potential, limits and initial C- and X-band comparison. Photogrametric Engineering& Remote Sensing, Vol. 76, No. 9, September 2010, pp Hebel, H.-P., Busch, W., Schäfer, M., Walter, D., (2008): RadarMon 2008: Subsidence monitoring over a collapsed mine in Berezniki, Russia. 3rd TerraSAR-X Science Team Meeting, November 2008, DLR, Oberpfaffenhofen, Germany. 4. Wegmüller, U., Walter, D., Spreckels, V., Werner, C., (2008): Evaluation of TerraSAR-X DINSAR and IPTA for ground motion monitoring. 3rd TerraSAR-X Science Team Meeting, November 2008, DLR, Oberpfaffenhofen, Germany. 7

Fringe 2015 Workshop

Fringe 2015 Workshop On the Estimation and Interpretation of Sentinel-1 TOPS InSAR Coherence Urs Wegmüller, Maurizio Santoro, Charles Werner and Oliver Cartus Gamma Remote Sensing AG - S1 IWS InSAR and