Insar time series analysis for land subsidence monitoring in Bangkok and its vicinity area

Abstract

Since early 1970s, leveling data has revealed land subsidence occurring in Bangkok and surrounding provinces. Though leveling is reliable and accurate, serious drawbacks of the technique are that 1) it measures only places that accessible by roads and 2) it is time-consuming and labor-intensive. Consequently, the density of measuring points is less than one point/km2 which is far too low to capture short wavelength phenomena like land subsidence. This research applies InSAR, a space-based radar technique which can accurately measure distance or change of distance from radar antenna to objects of interest on the ground, to measure land subsidence. A main objective is to apply a combination of two specific methods of InSAR namely Persistent Scatterer (PS) and Small Baseline (SB) to measure occurring subsidence from radar data. PS and SB were employed to analyze 19 images of Radarsat-1, the footprint of which covers most of Bangkok and certain areas of Nothaburi, Pathumthani and Samutprakarn, for the period October 2005 until March 2010. More than 300,000 points, yielding the density of 120 points/km2 are detected as stable which can be treated like survey benchmarks. Thus, the number of points provided by InSAR and its density has largely overcome the under-sampling and bias problem of leveling data. InSAR reveals that Muang, Samutprakarn and Lam luk ka, Pathumthani have been subsiding rapidly at the rate of 20-30 mm/yr. Central Bangkok on the east of Chao Phraya river appears to be subsiding slowly at rates around 10 mm/yr or slower while in northern and western suburban areas are found to subside at rates between 10-20 mm/yr. InSAR also shows fast subsiding areas in Saphansung and Ladkrabang district, Bangkok which have been missed by leveling due to its under sampling problem. At 71 locations where leveling and InSAR data of the same period are available, statistical test (t-test) of subsidence rates obtained from both methods found that 52 of 65 locations where InSAR results available within radius 100 m around leveling benchmarks are identical. Most of the 10 places where InSAR rates are larger are found to be in suburban areas where building density are low. Larger opening between buildings make it more likely for double-bounce of radar echo from building to ground slab to occur and so faster subsiding ground slab contaminated into that of building. The 3 areas where InSAR rates are slower can be explained by phase unwrapping problem. The research result leads to a conclusion that InSAR can be utilized as a geodetic tool to monitor land subsidence but a certain number of leveling points are still required as check points and for certain places where limitations of InSAR may provide unreliable rates.