Feasibility study on monitoring soil erosion using persistent scatterer synthetic aperture radar interferometry technology

Synthetic aperture radar time series difference interferometry (PS-InSAR) has widely been expected to provide an opportunity for the measurement of large-scale surface deformation in recent years. But it is still lacking for the PS-InSAR in soil erosion monitoring in China. Taking Linger County, Inner Mongolia of western China as the research area, an attempt was made to monitor the soil erosion using persistent scatterer synthetic aperture radar interferometry (PS-InSAR). Firstly, the accumulated point cloud of surface deformation was captured using the time-series differential interferometry of sentinel-1A satellite data. Secondly, the multi-scale filtering was constructed in time and spatial domain using the phenological characteristics of soil erosion. The points of soil erosion were then selected from the accumulated point cloud of surface deformation. Finally, the points of soil erosion and multi-source data were superimposed to evaluate the spatial and temporal characteristics of regional distribution in soil erosion. A total of 1 007 958 deformation points were extracted from the operation of PS-InSAR. It was found that surface deformation was the common response of human activities, vegetation growth, soil erosion, landslide, debris flow, frozen soil expansion and contraction, rock weathering and transportation, litter accumulation, and geological movement. The influence of interference factors was then removed using the filtering of vegetation, stable surface, human activity in the spatial domain, linear, and distortion in the time domain. Correspondingly, the monitoring object was anchored to the surface deformation caused by soil erosion. Eventually, 10 596 points of soil erosion were separated from 1 007 958 accumulated points of surface deformation, indicating the average erosion rate was 15 mm/a. A field test was then verified that the accuracy of monitoring and actual erosion point was 86.4%, and the accuracy of monitoring elevation was centimeter level. Superimposed multi-source data showed that the soil erosion in the study area was mainly distributed in the sparse vegetation at the junction of plain and mountain areas, with the slope less than 30° and annual rainfall less than 300 mm. There was a strong correlation between the amount of soil erosion and the rainfall in time. But, the amount of soil erosion was still increasing in the months without rainfall, indicating that the erosion exogenous force was the mixture of wind and water. Furthermore, the soil erosion was mainly concentrated in the transition zone of plain and mountainous areas, where human activities were relatively intensive. More importantly, soil erosion can be serious, due to excessive cultivation and deforestation in extensive management and large-scale farming in recent years. It was also found that 1 101 soil erosion points were distributed in the scope of farmland, particularly in the extensive management of hillside farmland or abandoned land when combining with the land use data. Consequently, it is highly demanding to strengthen the protection of farmland and ecological security in eroded areas, thereby increasing the intensity of returning farmland to forest and grassland. The main measure can also be made to avoid the steep slope reclamation. Anyway, the improved PS-InSAR technology presented macroscopicity, accuracy, long-term, and economy, suitable for monitoring soil erosion, compared with conventional or optical remote sensing. © 2021, Editorial Department of the Transactions of the Chinese Society of Agricultural Engineering. All right reserved.