Mapping the Yellow River Delta land subsidence with multitemporal SAR interferometry by exploiting both persistent and distributed scatterers

Due to highly compressible soil and a large amount of human activity, the costal deltas are more prone to ground subsidence. Many major costal deltas in the world are facing subsidence and are consequently more susceptible to flooding, salinization and seawater infusion or even permanent submergence. Therefore, ground subsidence has been a significant problem in coastal delta areas worldwide. The Yellow River Delta (YRD) is the second largest river delta in China. On the one hand, the YRD contains a large area of wetlands rich in biodiversity, and on the other hand, industrial activities and urbanization are extensive due to abundant underground resources such as oil, gas and brine. Excessive land use has caused different degrees of ground subsidence in this area. However, a detailed and comprehensive description of the ground subsidence pattern over the YRD has not been provided. Also, widespread non-urban area in Yellow River Delta region, such as wetlands, farmland and coastal tidal areas, hinders the application of persistent scatterer interferometry method (PSI) for comprehensive subsidence measurement over the whole area. In this paper, we developed a multitemporal InSAR method to map ground subsidence over the YRD area by exploiting both persistent scatterers (PS) and distributed scatterers (DS). This method is characterized by employing the coherence-weighted phase-linking algorithm for fast and reliable optimal phase reconstruction of each DS point and a two-tier network of PS and DS for the robust analysis of land subsidence. To extract the detailed and comprehensive ground subsidence over the whole YRD, we apply our method to 30 ENVISAT ASAR images (2007-2010) and 49 Sentinel-1A (S-1A) images (2015-2018) and obtain measurements of the ground subsidence during these two periods. Forty-one Sentinel-18 (S-1B) images (2016-2018) are also exploited for cross-sensor consistency validation with the result derived from the S-1A dataset. Our method shows a great advantage over the PSI method, providing much higher measuring point (MP) density in mapping land subsidence over the YRD, including 15-fold higher density for the ASAR dataset, 5.1-fold for the S-1A dataset and 5.3-fold for the S-1B dataset, which enables a very detailed description of local ground deformation patterns. Cross-track consistency in the derived measurements from the S-1A and S-1B datasets shows a standard deviation of 9.6 mm/yr for the vertical subsidence rate. A quantitative validation of the derived subsidence results compared with leveling measurements suggests an accuracy of 4.58 mm/yr for the standard deviation term. By comparing the ground deformation over the YRD during the periods of 2015-2018 and 2007-2010, we find that the subsidence in this region shows an overall intensification trend and many new and severe subsidence depressions appear along the coastline, with a maximum vertical subsidence rate of 432 mm/yr. Subsequently, the overextraction of underground brine for salt production is identified as the primary factor causing the ground subsidence near the YRD coastal area.