InSAR Decomposition Reveals Varied Deformation Patterns in Coastal Subsidence Regions

The impact of climate change was reported to cause more frequently extreme rainfall or drought events in decades. The disaster like typhoon coming up with excessive rainfall would flood low-lying coastal areas, resulting in devastating damage to local industry. Moreover, the coastal area may also be threatened by relative sea level rise due the lowering land surface height by land subsidence hazards. The Pingtung Plain, one of the major aquaculture-rich counties in Taiwan, suffered from up to 3.5-meter cumulative subsidence in the coastal area from 1972 to 2019. Featuring tropical monsoon climate properties, the land displacement in Pingtung area is highly variated with interannually seasonal changes between dry and rainy seasons. In this study, we apply Persistent Scatterer InSAR (PSI) technique into the high variability coastal area to make up for the lack of either spatial or temporal resolution of in-situ grounded measurements like continuous GNSS station or precision leveling. To comprehensively analyze the transient land deformation responded within a short period, we further employ an unsupervised decomposition method called Principal Component Analysis (PCA) into InSAR observation matrix to distinguish individually spatiotemporal patterns. First two principal components (PCs) reveal that the coastal and inland area in the Pingtung Plain characterize different amplitude of seasonal variations and long-term trends. In addition, the third principal component indicates the heterogeneous patterns due to different types of industrial groundwater usage and hydrogeological environment. The unsupervised method is capable to retrieve different spatial deformation patterns in the high variability coastal area from the bulk InSAR time-series matrix, which can contribute to comprehensive understandings of the relation between land deformation and aquifer system.