International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
The capture is not nadiral but it presents a rather high off-nadir 
angle equal to 15.04°. However, this feature is not too sensible 
for the aim of the study, but it could be important in case of 
study of urban areas with high buildings and artefacts (problems 
of shadows, perspective views, etc.) for which the nadiral view 
is fundamental. 
The area of interest is the mountainous part of the QuickBird 
scene. Unfortunately, because of the season and the hour of the 
day during which the image has been collected (mid morning in 
January), it is affected by strong shadows, especially in the most 
interesting areas, along valleys and gullies and, in some cases, 
over the landslide detachment areas. On the contrary, capture 
during the winter season, allows the recognition of some 
features usually covered by tree canopies. Furthermore, even if 
the visual quality is good on the whole, the presence of clouds 
(percentage equal around to 16%) right over landslides areas 
prevents a clear visibility of study features. 
1.2 Airborne Laser Data 
Two different areas, included in the QuickBird scene limits, 
have been scanned through the Airborne Laser Scanning 
mounted aboard the helicopter. Both LIDAR surveys have been 
performed by the TopEye System, that allows to record two 
echoes and the intensity for each points in the near infrared 
wavelengths. 
Over the largest area, around 21 km? wide, the eight of flight, 
equal to 400 m has brought to a density of 1 point to 1 m? 
instead, over the other area, around 2 kn wide, the lower 
height of flight, equal to 200 m, has led to a higher density of 
points, equal to 4 points to | m”. In the whole, thirty million 
points have been collected. 
  
Figure 2. LIDAR DTM over two areas 
As the results of the present work have to be framed in a 
national contest and have to be used from different users in a 
local contest, all data have to be expressed in the Italian 
Cartographic System Gauss-Boaga. For this reason, from the 
original reference system, WGS84, laser data have been 
transformed in Gauss-Boaga system through a Molodensky 
transformation with a precise estimation of parameters through 
ground points of a local reference network. 
Data have been processed thanks to TerraSolid package. First of 
all, they have been checked in order to eliminate outlier points, 
then, through a TIN densification algorithm (Vosselman, 2000) 
they have been filtered and, finally, classified in three classes: 
vegetation, ground and buildings. The ground class points have 
been thinned up to three million points and, finally, they have 
been used in order to create the LIDAR DTM. 
Moreover, because the LIDAR survey is composed by several 
strips, it has been necessary to join them together (Barbarella, 
2003). The result of the strips adjustment is shown 
in figure 2, where it's also evident that strips don't cover the 
area with continuity but several holes among different strips are 
visible. For this reason the necessity of integrate data where 
they lack has risen. 
As regards the intensity data, the infrared image derived from 
reflectivity of laser points has been utilised for the integration 
with the satellite image. 
  
Figure 3. LIDAR radiometric response over Bracigliano 
landslide. In green the digitalisation of the landslide boundaries 
1.3 IGM DTM 
A digital terrain model provided by the Italian Military 
Geographical Institute (IGM) was also available over the whole 
area corresponding to the QuickBird scene. This DTM has been 
derived from the digitalisation of contour lines of 1:25000 maps 
in the Italian Cartographic Reference System Gauss-Boaga. 
Maps obviously date back the landslide event, so the DTM 
doesn’t correctly represent the real and present morphology of 
the terrain along the mountainsides, from which a large amount 
of soil and mud teared away. 
The IGM DTM over the scene is composed by four files for a 
total amount of 1 million points regularly gridded with a 
spacing of 20 meters. 
3. INTEGRATION OF TERRAIN MODELS 
The availability of the two digital terrain models with different 
resolution, allows to exploit advantages and quality of both of 
them, that is the greater detail of the laser DTM and the 
extension of the IGM DTM over the whole scene, needed for a 
correct orthorectification of the entire QuickBird image. 
Therefore, the two available DTMs have been integrated 
together. 
For this aim, in order to integrate the raw data, and not 
interpolated data, an automated procedure has been created on 
purpose with Matlab package. Regular gridded data of the IGM 
DTM, corresponding to the area in common with the laser 
strips, have been removed from the file and replaced with laser 
points belonging to the relative gridding cells. The final result 
of the operation is a file composed by around 4680000 of sparse 
points, with a gridding space variable from 20 to 0.3 meters, 
denser over landslides and impervious areas where it’s clearly 
necessary to have a great amount of details to describe the 
earth-bare, and a less density in the plain area and over the 
mountainous area not hit by landslides. 
946 
Inter! 
  
F 
Befor 
been 
of ele 
GPS- 
deteri 
has e 
strips 
momé 
and a 
coher 
Even 
to 7- 
surve 
has b: 
The 
accur; 
morpl 
corre 
that o 
The f 
mode 
à prof 
Ortho 
using 
adopt: 
param 
are th 
It sho 
entire 
previc 
at lea 
whole 
the ri 
delive