International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
  
scanners are able to survey objects and scenes with greater 
distances and so were adopted in the frame of this work. 
3.1 First terrestrial laser scanning campaigns 
Two 3D Terrestrial Imaging Laser Scanners were used to 
realisc in different epochs the landslide survey: a Riegl LMS- 
Z210 system (May 2001) and a Riegl LMS-Z420i system (April 
2004). These scanners, born to cover wide areas, collect four 
measurements for each impulse: 2 angles, the distances and the 
intensity of the received echo impulse. The polar coordinates 
are immediately transformed in a local 3D Cartesian system 
(sensor svstem). Automatic recognition of targets by intensity 
images matching is supported. 
Riegl LMS-Z210 system can perform data acquisition at a 
distance from 2 to 350 metres with a nominal accuracy in the 
distance of about + 2.5 centimetres. The system is able to 
acquire intensity range and also RGB images for an angular 
Field Of View (FOV) of 370 gon (horizontal; angular resolution- 
+ 20 mgon) x 88 gon (vertical; angular resolution + 40 mgon), 
with a minimum angle step resolution of 80 mgon and a capture 
rate of 6000 points per second. The new Riegl LMS-Z420i 
system (Riegl, 2004) has better performances: concerning the 
rangefinder, the measurement range is from 2 to 800 metres 
with a nominal accuracy of + | centimetre; the system is able to 
acquire intensity range for a scanning range of 360 gon 
(horizontal; angular resolution + 2 mgon) x 80 gon (vertical; 
angular resolution + 4 mgon); the minimum angle step width in 
vertical scans is 9 mgon, 11 mgon in horizontal scans and an 
acquisition rate of 12000 points per second is achieved. 
Three scans were carried out in 2001 survey, with 80 mgon 
resolution, starting from different positions on the landslide in 
order to minimise the absence of data in the clouds of points 
due to the perspective view of the scans. In order to registry 
together the multiple scans, 6 reflective 6x6 centimetres targets 
tapes and 4 retroreflector prisms were placed inside the 
landslide. 
  
^ 
Figure 3. Instruments adopted for the Laser Seanning surveys: 
A) Riegl LMS-Z210: B) Riegl LMS-Z420i; C) Cylindrical 
Retroreflector (height 100 mm x diameter 100 mm); D) 
Retrorcflector prism used in the first campaign. 
With the purpose to define a topographic local reference system 
and to validate the accuracy of the laser scanning measurement 
over the control points, a survey was realised by a Leica 
TC2000 Total Station from two stations located inside and in 
the proximity of the landslide, with high precision results. 
The 3D-RISCAN software package was used to locate in 
automatic mode the targets, based on their intensity values. The 
coordinates of these points were used to stitch together the 
different scans, by means of 3D similarity transformation. The 
residuals coming from these transformations produced standard 
deviation values less than 4 em. 
À similar survey was performed in 2004 with the LMS-Z4201 
instrument. Four scans were realised with 80 mgon resolution 
and, in order to registry the multiple scans, 13 dedicated 
cylindrical retro reflectors were placed inside the landslide 
(figure 3C). 
The acquired points clouds data have to be connected together 
in order to reconstruct the continuous surface of the landslide, 
so they were automatically aligned during the survey and 
merged by the RISCAN PRO, the new software package for 
RIEGL 3D laser imaging sensor of the LMS-Z series, that 
permits even to filter, to create polygonal surfaces and to map 
the surface with textures using digital photos obtained during 
the scanning by a digital semi-metric camera placed over the 
laser instrument. 
3.2 Procedures for DTMs extraction 
The first products resulting from clouds alignment arc DSMs 
(Digital Surface Models) of the arca in correspondence of each 
survey; the second step of data analysis is to obtain the DTMs 
(Digital Terrain Models) that reproduce the natural surface of 
the ground, without vegetation and buildings. The software 
used for segmenting purpose has been the Microstation module 
TerraScan, by TerraSolid Inc. 
The instruments used are able to estimate, during data 
acquisition, two different responses of the same pulse but it's 
necessary to decide for the first echo or the second echo 
response before starting data recording. One of the own 
technical characteristics of these instruments is the ability to 
distinguish the two different echoes only if their difference is 
greater than 2 meters. Considering the purpose of the survey, 
the scanners were set-up to record the last echo in case of 
discrimination between the two responses. 
The most important phase is being able to discriminate terrain 
points for DTM construction; opportune editing routines (data 
segmentation) are necessary for this purpose. Data 
segmentation requires for a explicit parametrization of the 
classification algorithms provided by the software, following 
some criteria related to the object and surface. 
TerraScan provide different segmentation procedures depending 
on the kind of classification requested. In this case, two routines 
were adopted in sequence, named "Low point" and “Ground”. 
The first one permits to identify the points subject to errors on 
the basis of an analysis of outliers and clustering in the original 
data. The second corresponds to an iterative process: starting 
from a TIN surface derived from a subset of points, identified 
as terrain because of their smaller elevation in respect of the 
neighbouring, the surface incorporate other points on the basis 
of a geometrical analysis of slopes and distances, based on 
some user-assigned parameters, finally. generating a surface 
defined as “ground”. 
The use for terrestrial applications of a software normally 
adopted in airborne laserscanning impose a different operating 
approach, firstly at all because of the very high density of the 
points (a value of 1000 points per square meter can be 
achieved) and geometrical condition of the scanning, and also 
for the unavailability of multiple echoes. 
The definition of a set of optimal parameters for data 
segmentation needs a careful examination of sample areas and 
some testing on different morphological situations, different 
materials and different distances; normally, this choice is 
instead more general and reproducible for airborne scanning. In 
the case study, morphological situation can vary abruptly from 
      
  
     
     
   
   
   
   
   
   
   
   
   
    
   
  
  
    
   
    
   
    
   
  
  
  
    
    
   
  
  
  
  
  
  
  
  
  
  
  
     
   
   
   
    
   
   
     
    
         
Internat 
an area 
presence 
uniform 
In any : 
successi 
assisted 
with las 
  
  
  
Figure - 
section 
after rei 
2001 su 
3.3 AL 
To inve 
results 
system 
manual 
simultai