Monitoring riverbank deformation and influencing factors in the Inner Mongolia section of the Yellow River using InSAR technology

Riverbank collapse has seriously threatened the river channels under the combined interactions of freeze-thaw cycles, water flow erosion and groundwater level fluctuations. Particularly, a typical seasonal permafrost is located in the Inner Mongolia section of the Yellow River. However, the precision and continuity of conventional monitoring can be confined to the low coherence and vegetation-covered soil surfaces. This study aims to monitor the riverbank deformation for the prevention and mitigation of riverbank collapse. A systematic investigation was implemented to determine its patterns and influencing factors. 153 Sentinel-1A images were acquired from 2018 to 2024. SBAS-InSAR (Small Baseline Subsets-Interferometric Synthetic Aperture Radar) was applied to derive the spatiotemporal distribution of riverbank deformation in two typical meander bends, Shisifenzi and Wenbuhao. A 120-day temporal baseline was adopted to effectively capture the seasonal deformation during freeze-thaw cycles and flood seasons. Goldstein filtering and minimum cost flow (MCF) phase unwrapping were also employed to enhance the monitoring accuracy. Sentinel-2 optical imagery was then integrated with the InSAR monitoring. The location and causes of deformation were then determined, according to the land use types, topographical features and normalized difference vegetation index (NDVI). The NDVI values were calculated from the images of the vegetated flood season and sparsely vegetated winter. The InSAR coherence maps were also utilized to assess the impact of vegetation coverage on the monitoring accuracy. The results reveal that the significant linear correlation was observed between the SBAS-InSAR monitoring and leveling measurement, with a root mean square error (RMSE) of 1.11 mm and a coefficient of determination (R²) of 0.754. The high consistency and reliability of the SBAS-InSAR were then achieved to monitor the riverbank deformation under freeze-thaw cycles and complex geomorphology. There was the seasonal and spatial heterogeneity of deformation. Specifically, the annual average deformation rate was ranged from -20.0 to 24.8 mm/a in the Shisifenzi area. Two significant uplift regions were observed with the tendency to merge into a contiguous area. The most pronounced subsidence occurred in farmland areas of Shisifenzi Village, with a rate of 15.2 mm/a. Two subsidence zones were identified along the embankments at the apex of the Shisifenzi meander and the right riverbank near Guanniuju Village, where the seasonal deformation exceeded 6 mm. In the Wenbuhao area, the annual average deformation rate was ranged from -18.7 to 19.5 mm/a. Uplift regions were sporadically distributed, with an average deformation rate of 16.7 mm/year. While the subsidence zones were concentrated along the eastern riverbank, with the maximum rate of 14.9 mm/a and seasonal deformation between 8 and 14 mm. Overall, the riverbank deformation exhibited a pattern of slight general uplift interspersed with localized subsidence. Significant deformation areas were dominated by the influencing factors, such as the soil structure, agricultural activities, freeze-thaw cycles, and water flow erosion during flood seasons. Meander bends were particularly affected by the water flow erosion, leading to the reduced embankment and notable subsidence. Furthermore, the strong correlation of riverbank deformation was also found in the groundwater levels, river water levels, soil temperature and moisture. The weaker correlation was found in the precipitation, due to the arid and low-precipitation climate in the study area. Therefore, the SBAS-InSAR technology can be expected to effectively monitor the riverbank deformation under complex environments. The finding can also provide the reliable technical support to the disaster prevention and mitigation in seasonal permafrost regions.