Early Identification of Geological Disasters in Loess Hilly Area of Ningnan Based on InSAR Technology: A case study of Yuanzhou District, China

Using small baseline subset InSAR (SBAS-InSAR) technology and multi-temporal synthetic aperture radar data, multitemporal phase and long-term sequential deformation monitoring and early identification of geological hazards were carried out in the loess hilly area of southern Ningnan-Yuanzhou area. Based on the analysis of radar visibility in the loess hilly area of southern Ningxia. The Sentinel-1A historical archived data of 36 periods from January 2019–December 2021 and descending orbit data from January 2019–May 2021 were used to carry out monitoring of surface deformation in the river basin, and a total of 61 hidden danger. 49 new concealed high risk points were identified by comprehensive optical remote sensing interpretation. The annual average deformation rate of the radar line-of-sight direction of the ascent data was -47.6746–71.6472mm/a, and the cumulative deformation variable was -81.20–121.61mm. There were 23 new concealed high risk points identified. The annual average deformation rate was -72.1574–51.6028mm/a, and the cumulative deformation variable was -165.6–112.4 mm. During the descending orbit, there were 21 obvious deformations and 16 newly discovered geological hazards. Taking a group of ascending orbit deformation zones in Yuanzhou District as an example, which proved that the application of SBAS-InSAR technology in the loess hilly area of Ningnan was feasible.


Introduction
The middle and upper reaches of the Yellow River in the Ningxia Loess Hills are located in the southcentral region of Ningxia, China.This includes parts of Yanchi County, Hongsibao District, Zhongning County, and Zhongweicheng District and most of Pengyang County, Haiyuan County, Yuanzhou District, Xiji County, and Longde County.Owing to highly undulating terrain, loose soil, and abundant precipitation, these areas are highly prone to geological disasters such as landslides and collapses.The tributaries of the Yellow River in this region are well developed, and geological disasters often occur along the Qingshui River, a first-order tributary of the Yellow River, and the Hulu, Jing, and Ru Rivers, which are second-order tributaries.Furthermore, the loess covered hilly area of Ningxia in the middle and upper reaches of the Yellow River is affected by the southwest Hua Mountain fault and several faults in the southern part of Liupan Mountain.This has led to a landslide group with a discernable correlation with eological structure.Also, road cut slopes or loading and mining activities in these areas sometimes cause slope instability or geological disasters such as landslides and collapses.
Yuanzhou District, located in Guyuan City, a mountainous area in southern Ningxia, is prone to geological disasters in China.The prevention and control of geological disasters has always attracted the attention of governments, academia, and enterprises.The main types of geological disasters in the study area are landslides and collapses, followed by debris flows, ground subsidence, ground fissures, and land subsidence.Among them, landslides, collapses, and debris flow disaster points are multifaceted, sudden, and serious, which are the main disasters that cause loss of life and property and restrict social and economic development.This paper primarily used the "sky-air-ground" integrated method [1] to combine InSAR (Interferometric Synthetic Aperture Radar) deformation interpretation (which combines ascending and descending orbits), high-precision remote sensing surveys, ground surveys, and other means [2,3] to identify geological disasters early in Ningxia.From this we established a technical method for identifying hidden dangers relating to geological disasters in the loess covered hilly areas of Ningxia, in order to provide a solid and feasible basis for disaster prevention and mitigation management.

Overview of the Research Area
Yuanzhou District is in the middle of the Ningxia loess covered hilly area, northeast of Liupan Mountain, the source of the Jing River.The geographical location is between latitudes and longitudes of 35° 34′ N-36° 38′ N and 105° 28′ E-106° 30′ E, respectively.It has a total area of 2,739.01km2,and includes 3 subdistricts, 7 towns and 4 townships.Yuanzhou District is located inland, and most of the area has a typical continental semi-arid climate.In 2020, the average annual temperature in Yuanzhou District was 7.8°C, with the lowest average temperature in January and the highest average temperature in July, with the lowest annual temperature of -14.7°C and the highest annual temperature of 30.5°C.The water resources in the region are poor, with good water quality in the south and poor water quality in the north.Surface water primarily originates from atmospheric precipitation.Owing to the elevated terrain of Liupan Mountain, there are three major water systems: the Qingshui, Ru, and Zhangyi rivers.The geographical location of the study area is shown in Figure 1.

Data Sources
In this paper, Sentinel-1 data in the C-band of the ascending and descending orbit were used to detect the distribution of geological disasters in the Yuanzhou District.The data are presented in Table 1.According to the monitoring requirements, one image of the target monitoring area could cover the main areas of Yuanzhou District-analyzing the characteristics of disasters and query data-selecting 36 ascending orbit data from January 2019-December 2021 combined with descending orbit data from January 2019-May 29, 2021.This was used for SBAS (Small Baseline Subsets)-InSAR processing and analysis.

The Principle of Early Identification of Geological Hazards
Synthetic aperture radar expands radar resolution in via pulse compression technology in the range direction.It utilizes synthetic aperture technology to simulate a large antenna system to improve azimuth resolution.This greatly improves the resolution of real aperture radar images and generates radar images with relatively accurate amplitude and phase.The geometric relationship of the spaceborne radar wave imaging scan is shown in Figure 2.

Figure 2. Radar imaging geometric relationships
Considering the geomorphological topography, vegetation coverage, and geological hazard types in Yuanzhou District, the InSAR treatment method suitable for Yuanzhou District is SBAS-InSAR [4][5][6][7].The SBAS method was initially proposed to solve large-scale deformation phenomena at low resolution.The unwrapping phase map based on the multi disparity interference atlas for subsequent phase unwrapping operations is used to analyze surface deformation.This reduces the influence of temporal and spatial decoherence factors and obtains the average deformation velocity of the monitoring area [8,9].The small baseline technique performs differential interference processing on the short-time baseline image pair and the spatial baseline image pair in the baseline combination and then effectively solves the problem of the overall method equation rank loss through the SVD (Single Value Decomposition) method after multiple detangled differential interference maps-fuses the data such that

Technical Methods for Early Identification of Geological Hazards
The monitoring of surface deformation adopts InSAR technology and SBAS to extract the slope surface deformation information of different scales and deformation features.Based on the preliminary results of regional geological hazard identification using high-resolution optical satellite images, collected and archived radar satellite data for key areas with potential hazards and risks.The InSAR technology uncovered surface deformation in concealed high risk areas and dynamically monitored it (Figure 3).

Figure 3. Geological hazard identification and dynamic monitoring process of InSAR
First, the preliminary identification results of geological hazards guided the collection of radar data in the work area.Based on the preliminary identification results of high-resolution optical remote sensing geological hazards, key areas with concealed dangers and risks were selected and areas with lower risks were excluded.We then collected and archived Sentinel-1 radar satellite data for the at-risk areas.Thus, the area to be studied was significantly narrowed, saving a lot of manpower, material resources, and time.Second, regional-scale hazard monitoring included time-series analysis techniques-such as SBAS-InSAR-to carry out dynamic monitoring, extract the deformation rate and cumulative deformation variable information, and analyze the change trend.InSAR monitoring and deformation information was extracted during orbit ascent and descent when conditions permitted, forming multiangle observations of landslides, effectively monitoring deformation.Finally, field validation was performed.After obtaining the deformation result map of specific geological hazard areas using the SBAS method, one step can be added to the actual area for inspection and verification to verify whether the accuracy meets the requirements.

InSAR Processing Results
The master images selected for this study were dated 28 July 2020 and were registered to the same coordinate system as a measure of complex coherence coefficients.All points with coherence greater than 0.

Comprehensive Remote Sensing Identification Results of Geological Hazards and Concealed Risks in Yuanzhou District
Based on the deformation rate and cumulative variable amount of the ascending orbit data, the distribution pattern of deformation anomalies within the area were analyzed, and a total of 26 deformation constant areas were identified.Using the deformation rate and cumulative variable amount of the descending orbit data, the distribution pattern of deformation anomalies within the area were analyzed, with 15 deformation constant areas identified.Based on the deformation rate in the abnormal deformation area, combined with previous research and practical work experience, a deformation area with a deformation rate value less than -10 mm/a in Yuanzhou District was designated as a high deformation rate area.According to the "Code for Risk Assessment of Geological Disasters" (GB/T40112-2021), the minimum scale of landslides is 0.0001 km2.Further screening of slopes within high deformation areas, that is removing areas with slopes less than 10°; Take the disaster bearing body within from the edge of the deformation zone as the threat object of the geological disaster, and ultimately determine it as the identified hidden danger area.In combination with the above factors, spatial superposition analysis was carried out on the study area.Combined with optical images, this identified 61 geological disaster hazards, 46 landslide hazards, five collapse hazards, eight debris flow hazards, two land subsidence hazards, and threats to roads, houses, and water bodies.Among them, 49 were newly identified, and 12 were registered.The results are listed in Table 2, and their distributions are shown in Figure 8.

Time Sequence Monitoring and Verification of Typical Concealed Risk Points
To verify the results of geological hazard identification using the InSAR temporal monitoring method, 44 geological hazards were verified in the field during September 2022.Thirty-five concealed risks were consistent with the early identification results of the InSAR temporal monitoring method with an accuracy rate of 79.5%.
The SAR dataset was used to conduct long-term monitoring of landslide deformation using a group of ascending orbit deformation areas in Mingchuan Village, Hechuan Township, Yuanzhou District, Guyuan City (Figure 9).The data was used to determine the historical deformation characteristics of disaster bodies, confirm the acceleration process of deformation of disaster bodies, and evaluate the reliability of InSAR technology for early identification of geological hazards.

Figure 9. Time series deformation polyline
The deformation area is located in the first group of Mingchuan village, Hechuan Township, Yuanzhou District, Guyuan City.The residential area is located below the deformation area, with large height differences, steep slopes, sparse vegetation on the surface, and loose deposits at the bottom of the slope.High deformation rate, large deformation range, and thick loess cover, which mainly threatens the houses below.After field verification, the concealed risk at this location is a landslide hazard, which is consistent with the interpretation point, and mainly threatens the houses and roads below.The hill is 31 m high, 76.2 meters long, with a slope of 22.1 °.Multiple cracks were observed on the walls of the houses (Figure 10).Threatens the winding mountain road there are many parallel cracks on the road surface, which are severe, continuous, and dense.After field verification, the deformation time series is consistent with the actual situation.

Discussion
(1) Owing to the dense vegetation coverage in the Liupan Mountains, some areas have significant incoherence, resulting in ineffective monitoring results in some areas.In future work, on the one hand, it is suggested to plan data acquisition in advance, especially radar satellite data in the L-band, adopting satellite data with shorter time sampling intervals, and combining multiple data to solve simultaneously, so as to make monitoring more comprehensive and effective.On the other hand, it is recommended to layout fixed corner reflectors to enhance the signal of vegetation coverage areas and obtain more effective radar data.
(2) It is recommended to carry out long-term comprehensive remote sensing technology monitoring planning for areas similar to loess areas, where geological hazards are widely distributed and have advantages in obtaining data for identifying geological hazards.
Comprehensive remote sensing methods in geological hazard wide area identification and time series monitoring should be fully utilized, and optimization combined with professional instrument monitoring should be achieved to achieve efficient and accurate geological hazard monitoring technology.

Conclusions
(1) Conduct INSAR monitoring research on geological disasters of Loess Hills in Yuanzhou region.The annual average deformation rate of radar line-of-sight direction of the ascent data in the study area was -47.6746-71.6472mm/a, and the cumulative deformation variable was -81.20-121.61mm.There were 29 deformations in the results of the ascending orbit hidden danger, of which 6 were registered and 23 were not.
(2) The annual average deformation rate in the line-of-sight direction of the demotion data radar in the study area was -72.1574-51.6028mm/a, and the cumulative deformation variable was -165.6-112.4mm.The deformation of 21 places in the descending orbit is evident, of which 5 are registered and 16 are not.
(3) The study obtained and processed 65 ascending and descending orbit Sentinel-1 data points from January 2019 to December 2021 and obtained deformation rate plots and cumulative variable charts.Combined with optical images, a total of 61 geological hazard hazards, 46 landslide hazards, 5 collapse hazards, 8 debris flow hazards, and 2 ground subsidence hazards were identified, posing a threat to roads, houses, and water bodies.Of these, 49 new geological hazards were identified.Combined with the field verification of 44 locations, 35 were consistent with the interpretation results, with an accuracy rate of 79.5%.Prove the feasibility of using SBAS-InSAR technology for the early identification of geological hazards in loess hilly areas.

Figure 1 .
Figure 1.Location map of the study area 3 in the working area, were selected to establish the observation equation, and the SBAS time series analysis model was used to solve the deformation rate and cumulative deformation variable.Using data combination and surface deformation inversion, deformation rate plots and cumulative variable charts-along the radar line of sight of the study area-were obtained using the Sentinel-1 ascending and descending orbit data from 23 January 2019 until 26 December 2021 (figures 4-7).The annual average deformation rate along the radar line of sight was 47.6746-71.6472mm/a, and the cumulative deformation variable was -81.20-121.61mm during the ascending orbit (figures 4 and 5).Over a period of more than two years, the annual average deformation rate in the radar's line of sight was -72.1574-51.6028mm/a,and the cumulative deformation variable was -165.6-112.4mm, during the descending orbit (Figure6 and 7).

Figure 10 .
Figure 10.Signs of deformation at the foot of the slope

Table 2 .
Geological hazards with newly identified concealed risks