Measuring Land Subsidence with PS-InSAR in Bangkok using Sentinel-1 Time-Series Techniques

Abstract

Bangkok, the capital city of Thailand, has been affected by land subsidence since early 1970. Subsidence related factors are analyzed, including groundwater over-exploitation, aquifer structure, crustal deformation, urban construction, land use types, and clay thickness. These different factors have effects on the urban area and can lead to hazardous consequences. Nowadays, the space-borne Synthetic Aperture Radar (SAR) can detect structural deformation due to several phenomena of the Earth’s surface. SAR images offer high resolution, high observation density, high accuracy, resource-saving and are cost-effective for subsidence estimation. The Sentinel-1 radar satellite open data acquires data systematically and frequently over the Earth.

The questions analyzed in this study are: what PS-InSAR techniques measure and how can the land subsidence rate (mm/year) in Bangkok be derived from Sentinel-1 satellite image? What are the main factors of subsidence that influence/induce on Bangkok subsidence? Therefore, this study aims to apply the PS-InSAR time-series technique to measure the relative velocity (mm/year) and vertical movement (mm/year) of land subsidence in Bangkok and its surrounding area during 2016 until 2019. The study area consists of 3,450 sq.km in 80 ascending images and 3,450 sq.km in 93 descending images, and both areas include Bangkok city. The total study area covers Bangkok and also some part of Pathum Thani, Nonthaburi, Samut Sakhon, Samut Prakan, Nakhon Nayok, Chachoengsao, Ratchaburi, as well as Nakhon Pathom provinces in Thailand.

The land subsidence was measured by using PS-InSAR time-series techniques. Furthermore, it compares the subsidence from the descending orbit in the Western Greater Bangkok and validates the relative movement between descending PS-InSAR results and the leveling survey technique in the study area. Lastly, the objective is to evaluate the performance of Sentinel-1 PS-InSAR time-series techniques from the intersection area of ascending and descending orbit, whereby the vertical velocity is decomposed from deformation along with the line-of-sight directions. The vertical velocity can correlate analysis of the geological subsidence with a spatial pattern of the PS-InSAR time-series techniques in Bangkok city.

The advantage of the PS-InSAR technique is suitable for detecting the point-like stable object and strong reflectors for estimating ground surface movement. So, the PS-InSAR time-series result was validated by using the depth of leveling benchmark that is more than or equals to 20 meters below the surface in ascending and descending orbits of research areas because the ground surface movement affects the construction in an urban area that is built of foundation engineering through the sedimentary bed in Quaternary basin.

The PS-InSAR Sentinel-1 time-series processing was administered in the SARPROZ software. The R-squared (RSQ) and standard deviation (σ) are used to validate the relative velocity between the PS-InSAR and leveling techniques. Additionally, the correlation pattern between subsidence and geology-geomorphology at the hexagonal area of interest of ascending orbit and descending orbit also relates to the vertical movement in this area.

The PS-InSAR result shows that the study area has been subsiding at the relative rate of 8-25 mm/year in the line-of-sight direction. 80 points/km2 of 276,774 PS points, a stable point in the ascending area are detected. 232,814 PS points, with a density of 67 points/km2 are detected as a stable point in the descending area. Nevertheless, limitation of PS subsidence point estimating can be related to the stable material, building, construction or less vegetation. The validate analysis between PS-InSAR and leveling benchmarks as a result of the analysis revealed four points benchmarks satisfyingly agree with the result from the PS-InSAR point. While the trendline of the BKK6062 benchmark point and PSI95832 PS-InSAR time-series point that both trendlines conform to the ninety-five-confidence interval, but the 95-confidence interval trendline is rather wide owing to the dispersion of leveling survey points. It could be an error from leveling survey because it showed a low R-squared equal to 0.1, or phase unwrapping, microwave double-bounce, et cetera from PS-InSAR time-series technique.

The decomposed vertical velocity analysis revealed an area approximately 1700 sq.km within the interesting hexagonal area showing subsidence of averaging 8.93 mm/year and exceeding 20.01 mm/year in 2016-2019. Extreme subsidence occur at Khok Krabue, Bang Nam Chuet, and Phanthai Norasing in Mueang Samut Sakhon, Samut Sakhon province, as significant subsidence in red risk zone area. Because these areas are underlaid by soft Bangkok clay (Qa, Qc) and are accommodated by agriculture land use, aquaculture, industrial factories, and residential zone. Furthermore, it is for this reason that these areas have used groundwater overexploitation in substantial land subsidence. The groundwater level is a significant part of the factor for the observed subsidence, but the stratigraphic and the morphostructural setting in Bangkok’s Quaternary basin is an important factor as well.

-