LANDSLIDE STUDY USING TERRESTRIAL LASER SCANNER (LiDAR) ANALYSIS

Commission III

KEY WORDS: Remote sensing, Landslide, terrestrial laser scanner, LiDAR, CloudCompare

ABSTRACT:

In Reina del Cisne (Cuenca-Ecuador) a dynamic sliding process was created due to a cut that was applied at the beginning of the
year 2018 to the hillside without technical considerations for the construction of an access road to a house in the sector. From May
2018 to January 2019, period analyzed in this work, the landslide has caused total structural damage (dwellings near the hillside) or
partial (houses away from the hillside) and the total collapse of the path that caused the landslide. The field visits from the month of
May 2018 and the comparison with CloudCompare of the clouds of points obtained with terrestrial laser scanner (LiDAR) between
the months of May 2018 and January 2019 (house) and December 2018 and January 2019 (profiles), have highlighted the high
activity of this gliding. It has been analyzed three-dimensionally several profiles along the landslide, in addition to an affected house,
where they have experienced phenomena such as: sinking and tilting to slope down with values from 3 cm to 30 cm (profiles) in 30
days to 2.28 m (house) in 230 days.

1. INTRODUTION

Landslides are one of the most destructive geological processes

affecting humans, causing deaths and property loses, worth tens
of trillions of dollars each year (Brabb & Harrod, 1989). In the
city of Cuenca-Ecuador, it is a common problem that several of
its inhabitants aim to build their homes in steep slopes, due to
the low prices of these lands and the accelerated growth of the
urban area. These slopes experience landslides, which causes
damage to homes and constructions in general, as is the case of
Reina del Cisne sector. It is essential to carry out monitoring of
slides which are of vital importance for the mapping of the
same, even for the correct urban planning and prevention of the
inhabitants at this risk. This monitoring can be carried out by
means of classical techniques of topography in situ such as
Figure 1. Study area (Cuenca – Ecuador) with information
differential GPS or total station (González-Zúñiga, 2010)
combined using active sensors such as RADAR (Bardi et al., about the landslide and distribution of the profiles studied.
2014; Martie et al., 2016; Ventisette et al., 2014), or the laser
scanner (Chen et al., 2014; Eeckhaut et al., 2011; Conner & 2.1 LiDAR
Olsen, 2014; Hernandez et al., 2012; Travelletti et al., 2014;
Wang et al., 2011). The terrestrial LiDAR used has a range of 130 m, an accuracy
The objective of this work is to monitor and record sliding of ± 2 mm, fires an infrared laser in waveform and calculates
movements located on a steep hillside in Reina del Cisne sector the distance of the objects depending on the phase of the wave
(SE Cuenca) that is affecting a group of homes, conclusions and during its departure and return to the sensor. For this type of
future work. lifts at short distances, this type of laser scanner based on the
wave phase is more productive than the laser pulse-based
2. METHODS scanners, which calculate the distance to the objects according
to the flight time of the laser. This implies that they are slower,
The study area was delimited with PPGIS. Figure 1 shows the but reach greater distances. For this reason, the use of terrestrial
study area. This zone is located at UTM coordinates 17s area, laser scanner is an appropriate method for the study of
726.775 m E and 9.679.327 m N with an elevation of 2.600 landslides (Revuelto et al., 2013).
m. This area is characterized by slopes of medium to high.
2.1.1 LiDAR processing: The 7 external individual house 2.2 CloudCompare
scans (May 2018) and two sets of 10 individual scans
throughout the study area (December 2018 and January 2019), To align and compare in CloudCompare the point clouds of
have joined each other through the FARO Scene© software both dates (December 2018 and January 2019) were used 3
using references. static reference points, located outside the landslide area
For the house scans, 140 mm spheres and checkerboard-type indicated in Figure 1; in addition, a comparison has been made
tables and plans of the house were used. of the deformations that the dwelling has suffered using the
For the two scans throughout the study area, 246 mm spheres same 3 reference points located outside the main escarpment
were used. The distance between scans of the house has been (May 2018 and January 2019).
less than 15 m, so all the references used to join the scans have
received the sufficient number of impacts of the laser to be used 2.2.1 Error obtained in the point clouds alignment: The 3
for that purpose. On the other hand, the distances of the two sets reference points were precisely located and selected in each of
of scans along the area of study had a distance equal to or the point clouds to align them. The points (A0, A1, and A2)
greater than 15 m, so they used larger spheres to be able to correspond to the December point cloud and the points (R1, R2,
achieve that these spheres receive sufficient number of impacts and R3) correspond to the January point cloud, for the first
to be able to correspond it individually without problem. alignment, which is indicated in Figure 4. The same process was
used in the alignment for May 2018 and January 2019 clouds.
To join 2 continuous scans, at least 3 correctly matched
references are required (Barbarella & Fiani, 2013). The errors achieved in both alignments were low, because all
While joining scans, it is important to check that the selected errors have a value less than 1 mm, as shown in Figure 5 and
references are matched, which can be viewed in the workspace. Figure 6.

Figure 2. Two continuous scans of LiDAR joined in FARO

Scene© software.
Figure 4. Location of point cloud alignment points.
2.1.2 Errors obtained in the scans join: The join of scans to
form each of the 3 point clouds was satisfactory. Figure 3 shows
the highest average error of the 3 dates was only 1.5 mm.

Figure 5. Error values in the alignment between December 2018

and January 2019 clouds.

Figure 3. Average errors in mm when joining the May (above), Figure 6. Error values in the alignment between May 2018 and
December (half) and January (below) scans. January 2019 clouds.
2.2.2 Extraction of analysis areas: After the alignment of
the point clouds (May 2018 and January 2019), the area
corresponding to the house analyzed was delimited using
CloudCompare segment tool as shown in Figure 7.

Figure 10. Profile 2 exported to AutoCAD® Civil 3D.

Figure 11. Profile 3 exported to AutoCAD® Civil 3D.

3. DETECTION OF LANDSLIDE ACTIVITY USING

Figure 7. Segment extraction process. EXTRACTED POINT CLOUDS

For a specific structure, it is measured how much this structure

2.2.3 Profiling in aligned point clouds: At the end of the
has moved over the time of the study. This process was done
alignment of the point clouds (December 2018 and January
using CloudCompare with the 2 clouds of extracted points
2019), 3 profiles are extracted for each point cloud (6 in total)
corresponding to the house analyzed (May 2018 and January
to perform the deformation analysis of the house located in the
2019).
landslide shown in Figure 1.
The segmentation method for scanning profile was used for
Obtaining the profiles is done with the CloudCompare extract
comparing profiles, consisting segmenting the point cloud
cloud section tool as shown in Figure 8, the new point clouds
following lines. This technique consists of freely drawing a
corresponding to the extracted profiles are saved in the .las
profile, looking to cross perpendicularly the points. The tracing
format for use in AutoCAD® Civil 3D.
of this line can be done where it is most convenient, in this case,
it was where the LiDAR scanner was fixed for scanning
(Gonzalez, Woods, & Eddins, 2004).
Then, the profile representation was generated using
AutoCAD® Civil 3D with the profiles extracted from the 2
point clouds (December 2018 and January 2019). This allows
obtaining profiles with the respective relevant landslide
movements information.

Figure 8. Profile extraction process. 3.1 Measurement process in a specific structure

The method for comparing 2 point clouds (May 2018 and

January 2019) in CloudCompare consists of 1) segment the the
2.3 AutoCAD® Civil 3D
structure to be analyzed; 2) locate and select the point at which
The profiles obtained in CloudCompare are exported to the movement will be measured within the 2 point clouds; 3)
represent, locate and quantify the movements that the landslide read the difference distance between the 2 selected points.
has suffered as shown in Figures 9, 10 and 11. Figure 12 shows the process.

Figure 9. Profile 1 exported to AutoCAD® Civil 3D.

Figure 12. Measurement process in CloudCompare.

3.2 Procedure with extracted profiles

The method for comparing profiles (December 2018 and

January 2019) extracted from CloudCompare consists of 1)
export point clouds to AutoCAD® Civil 3D; 2) create a surface
from the point cloud; 3) draw a line at the top of the point
cloud; 4) create a profile of the surface to which the line
corresponds; 5) extract and display the abscissas and heights
information corresponding to the profile. Figures 13, 14 and 15
show the 3 January 2019 point cloud profiles.

Figure 16. The slab piece has shifted 2.28 m between May 2018
(blue) and January 2019 (RGB).

Figure 13. Profile 1 in AutoCAD® Civil 3D.

Figure 14. Profile 2 in AutoCAD® Civil 3D.

Figure 17. The slab piece has shifted 2.18 m between May 2018
(blue) and January 2019 (RGB).

Figure 15. Profile 3 in AutoCAD® Civil 3D.

4. RESULTS

To measure the movement of the house, different 3 parts of a

slab affected by the landslide were selected (May 2018 and
January 2019).
For the comparison of profiles, the dimensions of each of the
point clouds and their differences were correctly analyzed. For
profile comparison, a 5 m division was used for abscissas and a
division of 1 m for heights. Each profile has been superimposed Figure 18. The slab piece has shifted 2.23 m between May 2018
both clouds points to illustrate and quantify the deformations (blue) and January 2019 (RGB).
that has experienced in the area of study in the period of time
analyzed (December 2018 and January 2019). 4.2 Profiles comparison

4.1 Measurement of movement in slab In Profile 1, 2 and 3, (Figure 1, 19, 20 and 21), it is observed
that the main escarpment has increased its height (abscissa 0 +
Figure 16 shows a sector of the slab of the house scanned, near 015); the growth of corn (abscissa 0 + 020 to 0 + 035) is
a secondary landslide escarpment, has shifted 2.28 m downhill. observed; in general, it is seen as the ground sinks (abscissa 0 +
Figure 17 shows a portion of the slab of the scanned house that 040 to 0 + 090) and as slowly it is rising (abscissa 0 + 090 to 0
has shifted 2.18 m down the hillside. Figure 18 shows another + 105).
piece of slab, which has been displaced 2.23 m downhill and In Profile 1, (Figure 19) and Profile 3, (Figure 21) are displayed
has been slightly tipped. All these structures located in full along the same as the recorded terrain movements, although
landslide produced by a secondary escarpment. The result of they are removed from the central axis of the landslide these are
this alignment has revealed that the landslide has moved produced by mini-slides produced by secondary escarpments.
significantly in 230 days (May 2018 and January 2019).
In Profile 2, (Figure 20), it can be noticed that the recorded it is possible to see that the primary escarpment (abscissa
movements are greater, this is because the profile is located in 0+015) continues to increase its height, by an average of 6 cm
the central axis of the landslide, besides being influenced by the in this study time: December 2018 and January 2019.
movements of the mini-landslide produced by a secondary As well, the use of terrestrial laser scanner (LiDAR) for this
escarpment. research has been fundamental, since with this technique has
been able to reliably detect the movement generated by a
landslide; which could help identify areas vulnerable to this
type of problem. But, remote scanning techniques must be
complemented by traditional topography techniques to achieve
more accurate results.
It is planned to continue monitoring and structural analysis of
buildings in this sector with the fundamental support of aerial
images of a drone. In this way, it will be possible to monitor the
entire landslide and its surroundings.

Figure 19. Comparison profile 1, December 2018 (yellow) and REFERENCES

January 2019 (RGB).
+: Growth or uplift. Barbarella, M., & Fiani, M., 2013. Monitoring of large
-: Sinking. landslides by Terrestrial Laser Scanning techniques: field
data collection and processing. European Journal of remote
sensing, 46(1), 126–151.
Bardi, F., Frodella, W., Ciampalini, A., Bianchini, S.,
Ventisette, C. Del, Gigli, G., … Casagli, N., 2014.
Integration between ground based and satellite {SAR} data
in landslide mapping: The San Fratello case study.
Geomorphology, 223, 45–60.
Brabb, E. E., & Harrod, B. L., 1989. Landslides: extent and
economic significance. Proceedings of the 28th
International Geological Congress: Symposium on
Figure 20. Comparison profile 2, December 2018 (yellow) and Landslides.
January 2019 (RGB). Chen, W., Li, X., Wang, Y., Chen, G., & Liu, S., 2014.
+: Growth or uplift. Forested landslide detection using LiDAR data and the
-: Sinking. random forest algorithm: A case study of the Three Gorges,
China.
Conner, J. C., & Olsen, M. J., 2014. Automated quantification
of distributed landslide movement using circular tree trunks
extracted from terrestrial laser scan data. Computers &
Geosciences, 67, 31–39.
Eeckhaut, M. Van Den, Poesen, J., Gullentops, F.,
Vandekerckhove, L., & Hervás, J., 2011. Regional mapping
and characterisation of old landslides in hilly regions using
LiDAR-based imagery in Southern Flanders. Quaternary
Research, 75(3), 721–733.
González-Zúñiga, J. C., 2010. Monitorización de
Figure 21. Comparison profile 3, December 2018 (yellow) and
deslizamientos de ladera mediante estación total y GPS
January 2019 (RGB).
diferencial. Aplicación al deslizamiento del kilómetro
+: Growth or uplift.
35+000 de la vía Loja-Cuenca (Ecuador), 71.
-: Sinking.
Gonzalez, R. C., Woods, R. E., & Eddins, S. L. (2004). Digital
image processing using MATLAB. (Vol. 624). Pearson-
5. CONCLUSIONS
Prentice-Hall Upper Saddle River.
According to interviews with the residents, the landslide began Hernández, M. a., Pérez-García, J. L., Fernández, T., Cardenal,
in the winter of 2015 as a latent, slow slippage that reactivated F. J., Mata, E., López, A., … Mozas, A., 2012.
only in times of heavy rainfall. Cutting into the hillside, Methodology for landslide monitoring in a road cut by
transformed this latent and slow slippage into a very active and means of terrestrial laser-scanning techniques.
fast slide. From May 2018 to January 2019, movements of up to International Archives of the Photogrammetry, Remote
2 m have been recorded. Also, in just one month, this new Sensing and Spatial Information Sciences - ISPRS Archives,
version of the landslide has forced a house to be evacuated 39(September), 21–26.
because of the important deformations that its structure has Martire, D. Di, Tessitore, S., Brancato, D., Ciminelli, M. G.,
experienced. Costabile, S., Costantini, M., … Calcaterra, D., 2016.
The results of the profiles confirm that this is a major and Landslide detection integrated system (LaDIS) based on in-
rotational landslide, since from abscissa 0+015 to 0+085 it is situ and satellite {SAR} interferometry measurements.
observed that the terrain has been sinking; to the extent that {CATENA}, 137, 406–421.
happens from abscissa 0+085 to 0+105 where the ground Revuelto, J., López-Moreno, J. I., Azorín-Molina, C., Arguedas,
begins slightly to rise. By means of the three profiles analyzed, G., Vicente Serrano, S. M., & Serreta Oliván, A., 2013.
Utilización de técnicas de láser escáner terrestre en la
 monitorización de procesos geomorfológicos dinámicos: el
manto de nieve y heleros en áreas de montaña.
Travelletti, J., Malet, J.-P., & Delacourt, C., 2014. Image-based
correlation of Laser Scanning point cloud time series for
landslide monitoring. International Journal of Applied
Earth Observation and Geoinformation, 32(0), 1–18.
Ventisette, C., Righini, G., Moretti, S., & Casagli, N., 2014.
Multitemporal landslides inventory map updating using
spaceborne {SAR} analysis. International Journal of
Applied Earth Observation and Geoinformation, 30, 238–
246.
Wang, G., Philips, D., Joyce, J., & Rivera, F., 2011. The
integration of TLS and continuous GPS to study landslide
deformation: a case study in Puerto Rico. Journal of
Geodetic Science, 1(1), 25–34.