Ground subsidence monitoring based on synthetic aperture radar differential interferometry

(1) Algorithm selection and data processing flow

The data processing flow of differential interference is: first obtain the DEM and SAR interference image data of the experimental area, check whether the data meets the algorithm requirements, and then Perform image registration, calculate the coherence coefficient and generate an interference pattern, perform 5-view processing in the azimuth direction; remove the flat ground phase and terrain phase, filter the differential interference pattern, and obtain deformation information along the slant range according to the imaging geometric relationship, and Projected into the vertical direction, the desired settlement map is generated.

From a physical point of view, the interference phase can be decomposed as the following formula:

Research on remote sensing information extraction of degraded wasteland

Where: φflat is the flat land effect The phase caused by the imaging geometry is used to eliminate the flat earth effect; φtopo is the phase caused by terrain; φdef is the last remaining deformation signal; φorb is the phase caused by orbit error, and precision orbits can be used to reduce the error; φatm is caused by tropospheric and ionospheric delays. The phase can be ignored when the weather is clear; φnoi is the phase caused by noise, which can smooth and denoise the interference pattern.

According to the elimination method of terrain phase φtopo, differential interference is divided into two-rail method, three-rail method and four-rail method.

The two-track method uses two SAR images and external DEM data (such as SRTMDEM). The external DEM data is used to eliminate the terrain phase. The elimination process is differential processing. The advantage of the two-track method is that it does not require phase unwrapping of the DEM data, so it does not introduce errors related to it. The disadvantage is that the resolution of the deformation map obtained is affected by the spatial resolution of the DEM data.

The three-track method uses three SAR images, one primary and two secondary. The time interval between Image 1 and Image 2 is generally short to ensure that there is almost no change in the surface during the two imaging periods. The first interference pattern formed can be approximately considered to only contain the interference phase generated by the terrain, which is used to eliminate terrain information. The three-track method Can be used in areas without DEM. Then, image 1 and image 3 are subjected to interference processing to generate a second interference pattern containing the terrain phase and deformation signal. The difference between the latter and the former is the displacement between image 1 and image 3.

The four-track method uses four SAR images, two primary and two secondary. The first interference pattern is the same as the three-track method and is generated from image 1 and image 2. The difference is that the second interference pattern is generated from image 3 and image 4. Subtracting the first interference pattern is the sum of image 3 and image 4. deformation in between. The four-track method is similar to the three-track method. The difference is that the terrain interference pattern and the deformation interference pattern are independent of each other. The selection space is larger and the application is more flexible. It is often used in situations where there are Tandem image pairs.

Considering data cost and result accuracy, this study uses the two-track method to monitor mining area subsidence.

The data processing flow of the two-track method is shown in Figure 6-15.

Figure 6-15 Two-track method data processing flow chart

(2) Analysis of influencing factors

Decoherence factors in interference processing include time decoherence, Spatial decoherence, data processing decoherence, tropospheric and ionospheric effects. Therefore, the total coherence can be expressed as:

Research on remote sensing information extraction of degraded abandoned lands

1. Time loss of coherence

In many cases, spaceborne SAR Orbital interference images are acquired at different times, ranging from one day to several months or even years. During this period, the ground may change, and any change may change the phase of the radar signal and its statistical distribution. The resulting coherence weakening or even disappearing is called temporal decoherence. The main factors causing time incoherence are: plant growth or changes in vegetation caused by harvesting, farming, strong winds, etc.; constant movement of liquid surfaces, such as oceans, lakes, ponds, etc., mixed with swamps or unstable areas; ground landslides, earthquakes and other emergencies; other changes caused by human activities, such as the spatial development of commercial center parking lots, construction projects, deforestation, etc.; environmental changes such as precipitation, ice and snow cover, and melting. In short, surface displacement and environmental factors are the main factors causing temporal decoherence.

Assuming that the surface displacement is Gaussian distribution, then the coherence can be approximately replaced by the RMS displacement of the scatterer (Zebker, 1994):

Research on remote sensing information extraction of degraded abandoned lands

In the formula: σy and σz are the displacements along the intersection rail and the vertical direction respectively. For ERS-1/2C band satellites, take λ=5.7cm, reference incident angle θ=23°, JERS-1L band satellite, λ=23.5cm, reference incident angle θ=35°, Figure 6-16 and Figure 6- 17 reflects the relationship between horizontal and vertical RMS displacement changes of the ERS satellite and JERS-1 and time decoherence.

Figure 6-16 The relationship between time decoherence and scatterer RMS displacement (ERS-1/2)

Figure 6-17 The relationship between time decoherence and scatterer RMS displacement Figure (JERS-1)

It can be found from Figure 6-16 and Figure 6-17 that the RMS displacement of about 3cm is enough to completely decoher the ERS-1/2 C band data. The JERS-1 satellite requires an RMS displacement of about 10cm to cause complete temporal decoherence. The maximum RMS displacement allowed by its coherence is much higher than that of the ERS satellite. This fully demonstrates that the C-band radar wave is smaller than the L-band radar wave. It is more sensitive to ground changes. It can also be said that for the same RMS displacement, L-band radar waves can maintain higher coherence than C-band radar waves.

2. Spatial decoherence

The vector sum of the echoes of each scatterer in the ground resolution unit constitutes the echo amplitude and phase. If the geometric conditions are the same when acquiring ground images twice, and the position of the scatterer does not change, then the amplitude and phase of the two imaging will be the same; if the geometric conditions change, such as the incident angle of the antenna, then the echo phase will change. changes, this phenomenon is called spatial decoherence. Any interferometer will inevitably encounter such problems.

For ENVISAT ASAR, the nominal critical baseline distance is 1. 1km. For ALOS PALSAR, the nominal critical baseline distance is 12. 6km. Assuming that the effective baseline length is known, the spatial decoherence caused by the baseline can be calculated by Equation (6-14):

Research on remote sensing information extraction of degraded wasteland

3. Data processing decoherence

Data processing decoherence includes many aspects, such as registration decoherence, interpolation decoherence, interferogram filtering, phase unwrapping, etc. Generally, registration decoherence has the most significant impact, and other aspects can be solved through the corresponding method to suppress, if the method is improper, it will lead to registration failure or excessive error. The error introduced by the registration process will reduce the coherence of the interference pattern, thereby introducing phase noise. When the registration error reaches 1 pixel, the two images will be completely incoherent. Just and Bamler (1994) gave the registration incoherence formula in the range and azimuth direction:

Research on remote sensing information extraction of degraded wasteland

Where: μr is the registration error , between 0 and 1.

(3) Research data and methods

This study uses the two-track method to process image data to obtain land subsidence information. First, using the geometric relationship between the two SAR images, the DEM is inverted into an interference pattern containing only terrain information and projected into the SAR image coordinate system. Then, the interference pattern containing the surface deformation information obtained from the two SAR images is differentiated to obtain the deformation information. From the above analysis, it can be seen that obtaining DEM data with accuracy that meets the requirements is the key to the two-track method. In February 2000, the United States conducted the Space Shuttle Mapping Mission (SRTM), which conducted interferometric measurements over a vast area between 60° north latitude and 54° south latitude, providing coverage with a resolution of 30m and an elevation accuracy better than 16m. DEM data of 80% of the land surface (Figure 6-18, Figure 6-19).

The radar data used in the study is the two-scene image data of the Xuzhou area obtained by the ASAR (Advanced Synthetic Aperture Radar) synthetic aperture radar sensor of the European Space Agency's ENVISAT-1 satellite. The release of SRTMDEM data provides data guarantee for the wide application of two-pass differential interferometry. Therefore, the data used this time are the level 0 raw data of the imaging mode in the ASAR data product, the product code is ASA_IM_0C, and the elevation data provided by SRTMDEM data.

Figure 6-18 DEM plane diagram of Xuzhou City

Figure 6-19 DEM three-dimensional diagram of the experimental area near Peixian County, Xuzhou City

The data names are as follows:

p>

ASA_IM__0CNPDE20090120_022105_000000642075_00404_36029_9461.N1

ASA_IM__0CNPDE20070327_022114_000000792056_00404_26510_1403.N 1

N34E116.hgt

N34E117.hgt

File from ASAR data It can be seen from the name that the data of these two scenes are both N1 format files. The images of the two scenes were obtained on January 20, 2009 and March 27, 2007 respectively. The Track numbers are both 404, and the first track number is 36029, the second orbit number is 26510. Through GAMMA software processing, the level 0 raw data on January 20, 2009 and March 27, 2007 were processed into single-view complex images (SLC). The vertical baseline of the two scene images obtained was 271.95m and the time baseline was 665d. Intercept the required research area range on the SLC and perform data processing to obtain the regional deformation amount.

(4) Land subsidence monitoring in Xuzhou urban area

The image of Xuzhou urban area was intercepted, ranging from 34°11'7.58″ to 34°24'0.34″ north latitude, 117 °23'1.19″～117°17'48.62″ (Figure 6-20), the image on January 20, 2009 is the main image (Figure 6-21 is the intensity image), and the image on March 27, 2007 is the secondary image Image, two-track analysis of external DEM.

Figure 6-20 Geomorphic map of Xuzhou urban area

From the coherence coefficient map in Figure 6-22, the overall coherence is relatively good, and the coherence coefficients in most areas are greater than 0.5. It can be seen from the settlement map in Figure 6-23 that there is ground subsidence in Xuzhou urban area, with the settlement amount reaching about 10mm, and the settlement amount in some areas reaching about 38mm. Judging from the distribution of settlement, the settlement in the urban center is relatively small and is mainly distributed in the peripheral areas of the urban center. This is also in line with the geographical distribution of Xuzhou City, a large coal mining city. Generally, the coal mines in Xuzhou City are far away from the urban center. Distributed around the city center. The settlement at ① in Figure 6-24 is relatively large, reaching 38mm. Based on the analysis of the local geographical environment, there is a large modern mine nearby - Pangzhuang Coal Mine. The coal mine consists of three pairs of wellheads, Pangzhuang and Zhangxiaolou. The mine field area is 18.3km2 and the industrial square area is 1.36km2. After the successful reconstruction and expansion, the depth of Zhangxiaolou's new well reached -1025m, making it the deepest well in East China. The annual coal mining volume reaches 2.6 million tons. Possibly due to the annual coal mining and the continuous exploitation of groundwater, the site and surrounding areas have experienced ground subsidence, and it has also shown a trend of subsidence to the northeast. It can also be seen from Figure 6-24 that the settlement in the Pangzhuang Coal Mine area is much more obvious than the settlement in the urban center, but the average settlement in the entire Xuzhou City is still relatively small.

Figure 6-21 Xuzhou urban intensity map (left and right inverted)

Figure 6-22 Xuzhou urban coherence coefficient map (left and right inverted)

Figure 6 -23 Xuzhou urban subsidence map (left and right inverted)

Figure 6-24 Subsidence funnel (left and right inverted)

(5) Datun Town ground subsidence monitoring

Datun Town is one of the “Top Ten Towns” in Xuzhou City. It has proven coal reserves of 2.4 billion tons and can be mined evenly for 100 years. Its annual output of raw coal is 12 million tons. Datun Coal and Electricity Group Company is located in the hinterland of the town. Home to Longdong Coal Mine, Yaoqiao Coal Mine, Xuzhuang Coal Mine and Kongzhuang Coal Mine, Datun Central District is a typical area in the coal mining city of Xuzhou. Figure 6-25 is an image of the central area of ??Datun, ranging from 34°45'56.78″ to 34°53’58.23″ north latitude and 116°51’23.46″ to 117°0’3.27″ east longitude. Due to the Longdong Coal Mine It is not on SLC, so the interception range only includes the other three coal mines. In Figure 6-26, we can see the obvious coal mining area, and its coherence coefficient is very high, generally greater than 0.6 (Figure 6-27). The deformation map of Datun center obtained using the two-track method is shown in Figure 6-28.

Figure 6-25 Geomorphic map of Datun central area

Figure 6-26 Intensity map of Datun central area (left and right inverted)

Figure 6-27 The coherence coefficient map of Datun central area (left and right inversion)

Figure 6-28 The deformation map of Datun central area (left and right inversion)

After inverting the deformation map obtained by the difference left and right, we can Obtain the settlement map of the central area of ??Datun (Figure 6-29). It can be seen from the settlement map that in the time span of 665 days from March 27, 2007 to January 20, 2009, the central area of ??Datun There is an obvious subsidence trend in most areas of the district, and the subsidence distribution is basically consistent with the distribution of mining areas. Yaoqiao, Xuzhuang, and Kongzhuang coal mines have all experienced ground subsidence, and the settlement amount in more than 70% of the areas is greater than 10mm. The triangle marked area in Figure 6-29 is the central area of ??Datun, where the maximum settlement reaches 61mm, the average settlement is 3mm, and the annual average maximum accumulated settlement reaches 33.5mm.

Figure 6-29 Settlement map of Datun central area

According to the above level measurement results of Datun central area, it can be seen that the maximum cumulative settlement is expected to reach 753mm by 2010 , the annual average maximum settlement accumulation during these five years reached 30. 6mm. Comparing the results of leveling measurement and D-InSAR two-track method monitoring, the difference between the two is only 2. 9mm. It can be seen that using D-InSAR two-track method The annual average maximum settlement accumulation in the central area of ??Datun, Xuzhou City monitored by the method is consistent with the results obtained from leveling measurements (Table 6-15).

Table 6-15 Comparative unit of the results of the two monitoring methods: mm

With the economic development of Datun Central District, the population is gradually increasing, the coal resources are continuously exploited, and the land subsidence The trend will inevitably intensify, and the harm caused by land subsidence, such as the reduction of urban flood control capabilities and the destruction of underground infrastructure, will definitely affect the production and life in the area and cause huge economic losses. Therefore, the central area of ??Datun should rationally utilize groundwater resources, rationally mine coal resources, improve the dynamic monitoring system of land subsidence, and take measures to slow down the subsidence trend as soon as possible.