Keywords

1 Introduction

Taiwan is located in the place where the Banyan Tree is squeezed. Earthquakes are frequent, the geology is fragmented, the terrain is mountainous, the rivers are short and steep, and rainfall is unevenly distributed in time and space. Slope collapse disasters such as rockfalls, landslides, and debris flows seriously endanger the development of regional transportation and the industrial economy. Mountain roads are often the only external access in the region, and the stability of road slopes is critical to the safety of life and property of residents. It is therefore important to introduce automatic monitoring systems and emergency notification using various advanced technologies in steep-sloped areas with disaster potentials.

Disaster prevention and response systems have evolved from the traditional disaster prevention operations into disaster management, becoming an important direction for slope disaster prevention and control in recent years. The Global Positioning System (GPS) is a satellite-based radio positioning system. Utilizing positioning satellites in the earth’s orbit, it can provide users, operating receiving equipment around the world, with highly accurate three-dimensional positioning, velocity, and time data without restrictions. The system is unaffected by weather, and can provide globally unified longitude and latitude reference data. It has an extensive range of applications and is already a mature technology. Due to the mass production of GPS modules, the low price further increases the accessibility of the technology and their intrinsic value to slope disaster monitoring applications.

In addition to accumulating sufficient historical monitoring data to provide follow-up scientific research and development, the automated monitoring system also hopes to capture the precursors of slope disasters and their dynamic behaviour when they occur, decreasing the response time of residents, and providing a reference for disaster prevention and responsible decision-making. Since these benefits significantly reduce the damage caused by disasters, real-time and effective communication transmission will be one of the critical links in the overall monitoring system.

This paper is an in-depth discussion on the application of a GPS and automatic monitoring system in road slope and landslide disaster reduction. It proposes, for the first time in Taiwan, that after a rainfall event, the maximum moving ground speed velocity of a section of the Alishan Highway section is 2.5 cm every six days, while the monthly total displacement is up to 10 cm. This information can be used as a reference for mitigation and basis for the regulation of this section of the Alishan Highway.

2 Literature Review

The landslide in Lishan area (Fig. 1) covers a vast area (Su et al. 2009), mainly including the Lishan Village at the intersection of Zhongheng Highway (Taiwan Line 8) and Yilan Branch Line (Taiwan Line 7A).

Fig. 1
figure 1

Lishan area. Source Open street map

Full size image

Large-scale stratigraphic sliding occurred in this area in April 1990. To prevent the continuous expansion of landslide disasters and maintain the safety of residents and smooth roads, the Bureau of Soil and Water Conservation of the Agriculture Committee of the Executive Yuan implemented the “Slope Disaster Remediation” plan in the Lishan area in 1995. Soil moisture detection (TDR) and GPS observation systems were established.

The Lishan Skid Warning Center was established to provide real-time notifications. A static GPS generated baseline measurements of high accuracy within 3 h of receiving satellite data that is suitable for ground slip monitoring. The system uses GPS to measure the baseline length change for 3 h on different days, then compares the displacement and amount of displacement using a ground extensometer in the Lishan slip area. These displacement values, along with a wide range of environmental baseline measurements, are used to better understand the trend of large-area soil slides. To assess the feasibility of long-term GPS monitoring of landslides, Zeng (2003) and Shen (2004) undertook observations in the Lishan landslide area. Both researchers found that the most suitable measurement method for long-term landslide monitoring was static GPS baseline measurements, receiving satellite data at a rate of one measurement every second for a 3-h static baseline solution, averaging the movement over a 1-h interval. This method was most suitable for landslide monitoring with millimeter-level accuracy. A long-term GPS monitoring landslide reference value was determined using the GPS accuracy and displacement velocity. Preliminary monitoring results confirms the feasibility of this management reference value.

Jian (2002) studied the influence of various factors on the sliding mechanism of the slope using in situ measurement data. The study area was on the hillside along the highway near Wuwanzai, Gongtian Village, Fanlu Township, Chiayi County (Jian 2002). This area belongs to the old landslide area and has suffered from strata slip since the road was opened. Beginning in 2000, several monitoring instruments were set up in this area to observe the local strata, surface deformation, groundwater level, and rainfall for a continuous period of about two years, concurrently with theoretical analysis and laboratory test results. Observations show that this area is in a volatile state, with the sliding behavior closely related to the rainfall. According to the monitoring data in 2000, the accumulated rainfall required for accelerated sliding of the slope was about 80–270 mm, which is equivalent to the accumulated rainfall of 3–5 days.

Qiu (2004) and Jian (2002) used in situ measurement and theoretical analyses to explore the influence of related factors on the slope sliding mechanism. Long-term monitoring of surface and strata displacement, rainfall, groundwater level, groundwater flow, and other projects began in 2000, near Wuwanzai on the 18th line of Taiwan. On June 26, 2003, severe damage occurred, causing the loss of 150 m of road subgrade and interrupting traffic flow for about 1.5 months. The cause of edge damage was analyzed, discussed and compared with on-site monitoring results and theoretical analyses. The stability of the residual slope after the injury was evaluated, and a possible remediation plan was proposed.

The Linbeikeng area of Pingding Village, the upper reaches of Pingding River, Linnei Township, Yunlin County, known as the Little Yellow Mountain (Fig. 2), experienced a significant collapse after 921 earthquakes in 1988. Slope collapses continued to occur during Typhoon Lee and continuous heavy rain in 2006. The Bureau of Soil and Water Conservation conducted on-site slope hydrology and geology data collection to monitor collapse disasters in these areas. The ground collapse was monitored with surface extensometers and traditional GPS slope ground deformation-monitoring systems. Completing the real-time measurement of slope collapse during the typhoon established monitoring and management values, and strengthened the software and hardware of the local disaster prevention and response center in the Linbeikeng area of Pingding Village and the upper reaches of the Pingding River (Min 2020).

Fig. 2
figure 2

Collapse areas of Pingding River upstream, Linbeiken area, and Pingding Village during different periods of time [24]. Note The yellow parts refer to the collapse condition on June 2002 (caused by Jiji Earthquake and Typhoon Toraji); the black parts are the collapses on 9 September 2004, while the parts in red are the collapses on 11 January 2005. The arrows indicate the sliding directions of each collapse

Full size image

Lumb (1975), Brand (1982), and Slosson and Larson (1995) all show that slope failure often occurs after heavy rain. In addition to torrential rain, triggering slope damage also requires certain immediate hydrological conditions. Early on, rainfall must saturate the slope’s surface, making it easier for water to flow downslope. Then subsequent heavy rain can start the mechanism of slope damage. The impact of previous rainfall on slope stability has been studied for many years. Lumb (1975) found that previous rainfall damaged slopes, especially if the current rainfall was high, and will be accompanied by more slope damage events. Four types of slope damage events were caused by rainfall:

  1. 1.

    The worst incident resulted in more than 50 slope damages in one day,

  2. 2.

    A severe incident caused 10–50 slope damages in a day,

  3. 3.

    Minor incidents caused less than ten slope damages in a day,

  4. 4.

    An independent incident caused only one slope damage in a day.

Based on the rainfall data obtained, Lumb (1975) defined the range of events of different magnitudes, illustrated by 15 days of precipitation and 24 h of heavy rain. The most severe event occurred with more than 100 mm of rain in 24 h and 15 days of upfront rainfall of more than 350 mm. Severe incidents occurred within 24 h of heavy rain of more than 100 mm when earlier rainfall reached 200 mm. With the correlation between rainfall and slope damage confirmed, it is possible to determine when slope damage has reached the threshold for research.

3 Research Methodology

3.1 Study Site

One of the limiting factors affecting the large-scale use of GPS in landslide monitoring is the high cost of hardware. For continuous monitoring of deformable bodies, each monitored point needs to be equipped with high-quality geodetic GPS receivers. This makes the equipment cost too expensive. To this end, we introduced the concept of using multiple GPS antennas connected to one receiver. This one-machine multi-antenna GPS is especially suitable for structural deformation and displacement monitoring, using an external off-the-shelf, standard commercial GPS receiver and a specially designed GPS multi-antenna switch (GMS: GPS multi-antenna switch). This allows multiple antennas to be connected to one receiver, reducing the cost of hardware at each monitoring point exponentially, and making GPS more practical in many applications.

The one-machine multi-antenna GPS adopts the principle of time-sharing to allow each antenna to be connected to the receiver, and to collect GPS observation data according to the set time sequence. This principle is implemented under the control of an industrial personal computer (PC) using a specially designed GPS multi-antenna switch (GMS). Data collected by each antenna is used to calculate the position of the responding antenna, or the displacement of each corresponding antenna using the differential positioning mode. The architecture of a complete multi-antenna GPS is shown in Fig. 2. The system only needs two GPS receivers. The Reference Point refers to a fixed point and the Antenna Array is a moving point, while displacement is obtained through the Data Processing Center.

The Alishan Public Works Section commissioned the Feng Chia University team to complete the installation of various instruments, and share the project implementation results of the fourth bend of Taiwan 18 Line 44K and Xiaogongtian Village. Ground surface displacement measuring equipment included three CCD cameras with infrared projection, two lamps, a displacement meter, a surface extensometer, and image data-transmission equipment. The instrument deployment is shown in Fig. 3 (after Zheng et al. 2008).

Fig. 3
figure 3

Instrument installation location map

Full size image

3.2 Introduction to Global Satellite Positioning Monitoring System

In the past two decades, GPS has been widely used in many disciplines and fields, proving to be a potent tool for monitoring deformation and displacement of artificial and natural structures. GPS has many advantages over traditional measurement technology: it is more accurate and efficient, is high suited for automation, and installation requires low labor intensity. The dual-channel GPS satellite-positioning receiver has many advantages, including ultra-high precision, ultra-high sampling frequency, and all-weather observations. GPS is often used as a continuous tracking station for high-precision geodetic surveying and crustal change measurement. Since the On-The-Fly (OTF) method can solve in real-time, the undetermined value of the cycle, it is also often used for fast dynamic (kinematic) measurements (i.e. RTK), with aerial photogrammetry, or in fast-running vehicles. Since hardware is relatively expensive, single-channel satellite positioning receivers are used since they are relatively cheap to manufacture when compared with dual-channel GPS satellite positioning receivers. However, the number of receiving channels is small, so the ability to support dynamic measurement is inferior when compared to the dynamic capabilities of dual-channel GPS receivers. After years of changes in software and hardware, and with new technologies for solving undetermined cycles, single-channel GPS satellite positioning receivers are still inferior to dual-channel GPS receivers, but can be used for monitoring and measurement with smaller dynamic amplitudes. Performance is also excellent, with GPS successfully solving the undetermined value of the cycle and providing the accuracy of long-term monitoring. In addition, with wireless communication equipment and sophisticated analysis software, the single-channel GPS satellite-positioning receiver is sufficient to support the monitoring of displaced and deformed buildings and slopes (Fig. 4).

Fig. 4
figure 4

Differential positioning structure of a new generation of multi-antenna systems

Full size image

3.3 Global Satellite Positioning Monitoring System Test

To ensure the accuracy and stability of the one-machine multi-antenna monitoring system, laboratory tests were completed before onsite installation for real-time monitoring. This test was conducted at the Business School of Feng Chia University.

3.3.1 Introduction to the Experimental Site

The test was conducted on the top floor of the Business School of Feng Chia University to ensure a convenient place where the experiment had an unobstructed view of the horizon and sky. The storage room used as the instrument control room contained a multi-antenna GPS switch (Switch), two GPS receivers (Receiver), and a PC host. An existing GPS fixed station at the Ministry of the Interior was used as the multi-antenna GPS measurement base station. The layout and location of the entire one-machine multi-antenna GPS are shown in Fig. 5, comprising one base station and four mobile stations.

Fig. 5 
figure 5

Multi-antenna GPS test configuration diagram (4 mobile stations)

Full size image

3.3.2 Static Initial Value Measurement

To confirm the accuracy of the GPS measurement, a laser theodolite was used to measure the level and elevation of the fixed and mobile stations (GPS0 to GPS7). Since the leveling measurement use a particular direction as fixed, other points were measured relative to this direction. With the connection direction between Base and GPS1 taken as fixed, the results obtained were compared with the multi-antenna GPS measurements (Fig. 6). Coordinates at the end of the test are shown in Table 1. Differences in the vertical direction are large, while a large error in the horizontal direction of the GPS8 is also noted. The poor accuracy of the multi-antenna GPS8 resulted from positioning the GPS antenna too low to the ground so that the signal was reflected. Antennas should be placed more than 100 cm from the ground to avoid signal interference. In addition, a multi-path effect was generated because the position of the roof (instrument house) on the top floor blocks about half of the line of sight to the air.

Fig. 6
figure 6

Eight multi-antenna GPS test location diagram (Unit: cm)

Full size image
Table 1 Comparison of multi-antenna GPS measurement results
Full size table

4 Field Test of Global Satellite Positioning Monitoring System

The fixed station with instruments installed by early October 2007, and included one set of GPS base stations and four sets of mobile monitoring GPS antennas, as shown in Fig. 7.

Fig. 7
figure 7

Schematic diagram of local multi-antenna instrument configuration

Full size image

Static measurements and reception data were collected for two days (i.e. 48 h) from 2007/10/17 to 2007/10/19 to calculate the static coordinates (Table 2). Dynamic real-time monitoring began on 2007/10/19. Results for three days from 2007/10/19 to 2007/10/21 are shown in Table 3. Except for GMS3 and GMS4, which have a continuous displacement in the Y direction (about 2.4 cm), the rest oscillate back and forth, and are in a stable state. The dynamic real-time monitoring results for 20 days from 2007/10/19 to 2007/11/12 show that, except for the continuous displacement in the Y direction of GMS3 and GMS4 (about 9.76 cm), the remaining antennas continued to oscillate back and forth in a stable state. On-site observations on 2007/10/25 confirmed that the behaviours recorded were the result of slope displacement and ground slip, and not caused by abnormal instruments or shaking of the fixed GPS rod. Monitoring results over 100 days from 2007/10/19 to 2008/1/25 show that there was a continuous displacement (about 17.19 cm) in the Y direction of GMS1 to GMS4. A site visit in March 2008 confirmed the displacement of the slope and landslide. Figures 8 and 9 show graphs of the dynamic coordinates of continuous time and displacement changes of GPS3 and GPS4, with displacement direction and displacement (X, Y, H) corresponding to (N, E, H) in Table 3.

Table 2 GPS static coordinates
Full size table
Table 3 Multi-antenna GPS dynamic coordinates
Full size table
Fig. 8
figure 8

Road slope automatic monitoring system (Antenna No. 3 2007/10/19–2008/1/25)

Full size image
Fig. 9
figure 9

Road slope automatic monitoring system (No. 4 Antenna 2007/10/19–2008/1/25)

Full size image

Finally, the results measured by the above four sets of GPS were compared with the dynamic coordinates of the last stroke and the initial static coordinates and a spatial relationship diagram is drawn, as shown in Fig. 10. The direction of the arrow in the figure represents the direction of GPS displacement, and the length of the arrow represents the amount of displacement. It shows that GPS1–GPS4 move in the same direction and are the same moving the N2 block, as shown in Fig. 10.

Fig. 10
figure 10

Schematic diagram of GPS sliding displacement direction and relative displacement

Full size image

GPS data received by the base station antennas were collected for seven days a month from October 2007 to February 2008 to confirm the GPS fixed station was not moving (Fig. 11). The Zhuzaki County e-GPS fixed station was used as the benchmark for baseline calculation (this station was about 13.4 km from the experimental area). The five-month data from the base station recorded a horizontal coordinate difference of about 1 cm and is regarded to be within the error range. The average value is calculated as the coordinates of the base station, and is shown in Table 4.

Fig. 11
figure 11

GPS antenna position and moving block diagram

Full size image
Table 4 Calculated average differences in horizontal coordinate
Full size table

Slope landslides often occur after heavy or torrential rains, hinting at a specific relationship between slope damage and rainfall, in addition to certain hydrological conditions. Early rainfall can saturate surface soils with moisture, making it easier for water to flow over the slope and initiate slope failure. Rainfall data was collected in Alishan during the typhoon period from 2007 to 2008, as shown in Table 5. Figure 8 shows the relationship between the ground slip velocity and rainfall from October 2007 to February 2008. To represent the significance of ground slip, velocities are presented as displacement amounts over six days (cm/6 days). GPS1, GPS4 display severe displacement in the east–west direction, with the maximum westward sliding velocities equivalent to 10 cm per 30 days. Elevation changes and displacements in north–south direction were not large and values are not presented.

Table 5 Statistical table of rainfall brought by typhoons in Alishan from 2007 to 2008
Full size table

Figures 12 and 13 compare sliding velocity (cm/6 days) and accumulated rainfall for 6 days (1093 mm) from 2007/10/19 to 2008/02/14. Smaller rainfalls before the end of the year had a relatively insignificant impact. The figure shows that sliding velocity is obviously the fastest when the typhoon rainfall hits. When the rainfall stops, sliding gradually slows. The sliding velocity presents a parabolic trend when each point is fitted with a binary linear equation (Figs. 14 and 15). The R-square values of the fitting trend lines of Antenna 3 and Antenna 4 are both as high as 0.9, indicating a trend in sliding velocities during this period. The mathematical relationship of a slow descent of the quadratic equation of one variable, and the similar fitting equations of Antenna 3 and Antenna 4, indicate their sliding velocity and direction are the same. Following the rains of Typhoon Kerosa, the maximum displacement velocities of Antenna 3 and Antenna 4 were 2.4 cm/6 days and 2.5 cm/6 days, respectively. Displacement of GPS3 and GPS4 slowed at the same time, 74 days after the rainfall event.

Fig. 12
figure 12

Analysis of E-coordinate sliding speed and rainfall of Antenna 3

Full size image
Fig. 13
figure 13

Analysis of E-coordinate sliding speed and rainfall of Antenna 4

Full size image
Fig. 14
figure 14

E-coordinate sliding velocity fitting of Antenna 3

Full size image
Fig. 15
figure 15

E-coordinate sliding velocity fitting of Antenna 4

Full size image

5 Conclusion

This study proposes a specific solution for disaster prevention with the application of a one-machine multi-antenna GPS. This innovative monitoring system is tested on a section of the Alishan highway traversing steep sloping land (the fourth bend area), and forms the basis for a disaster prevention and response approach and subsequent research.

The fourth bend area experienced severe weather with sustained, heavy rainfalls during typhoons Pabuk, Uti, Sanpa and Kelosa (Table 5). A site investigation of this area in early October 2007 found many surface sliding phenomena triggered by typhoon rainfall, including a large (20 cm) fissure. Such large-scale sliding conditions in the Wuwanzai area can be understood through continuous measurement of absolute coordinates with the satellite tracking station data at the Land Survey Bureau and the GPS base station.

After the installation of the one-machine multi-antenna GPS was completed, monitoring was undertaken. After 100 days of monitoring, there was a maximum displacement of 17.19 cm. It is speculated that due to the influence of Typhoon Kerosa upon Taiwan at the time, the accumulated rainfall in the area around the Alishan Highway was greater than 600 mm. It was due to this large rainfall amount and surface runoff, that the slope sliding phenomenon in the fourth bend area was aggravated.

The one-machine multi-antenna GPS proven to monitor ground slip of the colluvium along the Alishan Highway. After the torrential rain, there was obvious ground slip in the sensitive areas of this road section, with the displacement rate reaching 10 cm/month (30 days). During sunny conditions, ground slip velocity slows in the form of a mathematical quadratic equation with time as a variable. The results of measurements by four sets of GPS are compared with the dynamic coordinates of the last stroke and the initial static coordinates. A spatial relationship diagram is shown in Fig. 10, with the direction of arrows depicting the 100-day dynamic coordinates, while the length of arrows represent the amount of displacement. This relationship shows that GPS1 to GPS4 move in the same direction, and are part of the same NE moving block (Fig. 11).