Deformation identification using DInSAR multi temporal analysis method in supporting infrastructure development (case study of the area around the nation’s new capital)

The relocation of nation’s capital is one of the government’s flagship programs at this time. The lack of information on soil structures that are prone to deformation can bring an impact on the failure of infrastructure development in the areas of New State Capital (known as Ibu Kota Negara/IKN). Therefore, a significant study is important to find out the deformations that may occur in the region as an initial information in determining the location for safer infrastructure development based on the soil structure conditions in the research area. This research used the Differential Interferometry Synthetic Aperture Radar (DInSAR) method with a multi-temporal approach to identify the deformations in the research area. The data included SAR images of Sentinel 1A type SLC with C band (5.405 GHz) for the period 2015-2019. The results confirmed that deformation had been identified in several areas, both those experiencing subsidence and uplifting. The maximum subsidence was 12.97 cm at Sepaku district on period 2018-2019 and the maximum uplift was 10.01 cm on period 2017-2018. The identified areas with deformation generally take place in areas with a high density of buildings, construction areas, road infrastructure, and river alluvial deposits.


Introduction
One of the considerations for choosing the North Penajam Paser and Kutai Kertanegara, East Kalimantan Province as candidates for the New State Capital (known as Ibu Kota Negara/IKN) is that they are the safest areas from disasters, especially geological disasters such as earthquakes and tsunamis [11]. However, based on its geological conditions, the IKN areas do not escape from potential geological disasters caused by ground deformation in the area. Based on the lithology, the research areas are dominated by a fairly thick sedimentary structure, which is part of the Kutai basin. The thick sedimentation is very prone to deformation since it has a relatively weak soil structure.
Based on the geological map, the North Penajam Paser area, which is the site of point 0 of the new IKN development, takes place is included in Balikpapan area. Almost all the rocks in this area had experienced deformation which forms anticlines, synclines, and faults [13]. In the southern area, there International Conference on Geological Engineering and Geosciences IOP Conf. Series: Earth and Environmental Science 851 (2021) 012017 IOP Publishing doi: 10.1088/1755-1315/851/1/012017 2 is a Holocene age (quarter layer) alluvial deposits with unconformity structure. Based on these geological conditions, the new IKN area has the potential to experience deformation that can trigger geological disasters, such as landslides, subsidence, earthquakes, sink holes, and soil liquefaction. Therefore, it may have a direct impact on the conditions of the infrastructure. Based on the data obtained from the Cent of Volcanology and Geological Hazard Mitigation in the research area, a landslide occurred in the Sepaku district in April 2018, which caused damage to dozens of houses [5].
Therefore, this research used a method to identify and to map the areas that are prone to deformation. This step is one of the attempts to mitigate geological disasters, as well as to provide initial information for planning infrastructure development in the research area. The method used in this research is the Differential Interferometry Synthetic Aperture Radar (DInSAR) multitemporal analysis using SAR Sentinel 1A radar image type Single Look Complex (SLC) with C band to identify the deformation [1][2] [4][6][7] [10]. The output parameters of the processing results using the DInSAR method was mapped using QGIS software to produce a thematic map of land deformation in the new IKN area.

Data
The data used in this study are secondary data obtained from the following sources: • Sentinel satellite image data set 1-A with C band (working at a frequency of 5.405GHz and a wavelength of 5.6 cm), Single Look Complex (SLC) type 1st level, period 2015-2019. The data can be downloaded from ASF Alaska via https://vertex.daac.asf.alaska.edu (see Table 1). • Digital Elevation Model Nasional (DEMNAS) data used to validate the DEM model generated from DInSAR processing. The DEMNAS data has a spatial resolution of 0.27-arcsecond. It can be downloaded via http://tides.big.go.id/DEMNAS/. The software used in this research includes SNAP (Sentinel Application Platform) version 7.0, and QGIS version 3 software.

Processing Method
The processing method used in this research includes the acquisition stage, pre-processing stage, data processing stage, and postprocessing stage. The acquisition process was carried out by downloading secondary data in the form of Sentinel 1A satellite images and DEMNAS data in the research area with the links listed previously. The further details are presented in figure 2, showing the research data processing method. The data processing and analysis techniques of this research were divided into three stages: the data processing technique using the DInSAR method using Snaphu software and the GIS processing using QGIS software.  Data processing using DInSAR method in the initial stages, the TopSar Split process is carried out to divide one satellite image scene into three subwadths (IW) and burst mode (1-9) which covers the research area. This research, the researchers used subwadth IW3 with burst 6 to 9. After the TopSar split process is carried out, the researchers then input the satellite orbit value data. The next process is the formation of an interferogram and input the DEM value, i.e., the SRTM 30 meters 3 seconds [6]. To reduce the noise and to sharpen the fringer on the interferogram, a filtering process was performed, i.e., enhanced spectral diversity. The next step was carried out using TopSar Deburst followed by the process of eliminating the topography phase (Thopo phase removal) and reducing the error phase with the multilooking process. The next process is Goldstain phase filtering to remove noise and increase the coherence value. The final stage is to perform terrain correction. The post processing stage was performed with GIS using QGIS version 3 software to display deformation maps in the research area. Figure 3 shows the phase coherence value resulting from DInSAR processing using SNAP software after the Goldstein filter was carried out. Most of the pixels still have a low coherence value <0.4. The low coherence value indicates that the research area still has a fairly high vegetation density and high noise even though it has gone through the filtering stage. It indicates that the frequency C band (wavelength of 5.6 cm) is not capable enough to penetrate the dense vegetation canopy, so there is considerable noise in the backscatter phase received by the satellite. Therefore, this research only analysed areas that have a coherence value > 0.5 to get the interferogram phase with the best accuracy [4] [8].   Table 2. On the deformation map ( figure 5, 6, 7), the deformation area of the Sentinel 1A image data processing used the DInSAR method with SNAP software. The identified areas getting vertically deformed are indicated by red blocks, showing that the subsidence and the deformation phase is negative, while blue blocks are areas that had experienced an uplift and showed positive phase values. The white areas are generally areas that are relatively stable and do not experience change or deformation. The results of the DInSAR multitemporal processing for the 2015-2019 period showed a maximum subsidence of 12.97 cm and a maximum land increase (uplift) of 10.01 cm (figure 4). The deformation process that occurred in the research area was quite complex. The area where deformation was identified changes every year, whether in the deformation area, the area, or the amount of deformation (cm).

Discussion
The location of point 13 (figure 5) is the terminal area of PT. Pertamina's oil refinery in the Penajam district. In the 2015-2019 period, the location of this point experienced an average land subsidence of 6.41 cm per year. The subsidence occurred at point 13 due to the continuous load from buildings and oil refineries. Meanwhile, the soil structure at the location of that point was not strong enough to withstand the loads of the building above it. The point 26 was a densely populated port city with a fairly high building density. This area was identified as having deformations in the form of subsidence and uplift. The subsidence that occurred at point 26 had a speed of 6.51 cm per year. The subsidence occurred in this area due to the load of buildings with high density and human activities such as groundwater extraction. Based on geomorphology, the uplift that occurred in this area is closely related to alluvial material deposits, because this area is located at the mouth of rivers and the sea.   [5] as shown in figure 7. This area is a hilly area with steep slopes with a height of 47 meters. The DInSAR multitemporal processing results confirms that this area had been identified as experiencing subsidence in the time span before the landslide disaster of 9.15 cm [4] [9]. The average land subsidence value in this area is 8.84 cm per year. Hence, the underlying cause for landslides at point 27 based on this study is deformation in the form of subsidence due to the loading of buildings with a high enough density in the area around the landslide, the topographical conditions, and the presence of high rainfall.  The observational areas generating a deformation process can be used to monitor the developments in an area. Areas under development can generally be identified from the deformation contrast values that have changed significantly, compared to areas that are not under development [6] [12]. This can be seen from areas that experience deformation in the form of subsidence with high building density, and this is closely related to the loading of big building volumes on the soil structure. Land subsidence may also occur due to the extraction of big amount of groundwater, particularly in densely populated urban areas [2].

Conclusion
The deformations that occur in the research area can be identified properly using the DInSAR method. Multitemporal analysis using DInSAR on Sentinel 1A images in the research area is able to detect changes of ground deformation, both subsidence and uplift from time to time. This method can identify areas that experience landslides caused by subsidence that have occurred in the research area. Based on the results of data processing, the area with the greatest land subsidence rate is the location of point 4 at Sepaku Distrct, with an average annual subsidence of 10,75 cm per year. This location also experienced a maximum uplift of 10 cm which occurred in the 2017-2018 period. Meanwhile, the area with the lowest land subsidence is the location of point 0 IKN (point 3) at Sepaku subdistrict with a land subsidence rate of 4,02 cm per year. The identified areas with deformation generally take place in areas with a high density of buildings, construction areas, road infrastructure, and river alluvial deposits.