B3. Istanbul 2004 
  
    
TERRAIN MODELING IN AN EXTREMELY STEEP MOUNTAIN: A COMBINATION 
OF AIRBORNE AND TERRESTRIAL LIDAR 
A.Ruiz, W.Kornus, J.Talaya, J.L.Colomer 
Institut Cartografic de Catalunya (ICC), Parc de Montjuic, E-08038 Barcelona 
toni(Dice.es, wkornus(dice.es, talaya(dice.es, colomer(dice.es 
Commission II, WG III/3 
KEY WORDS: Integration, DEM/DTM, Aerial, Terrestrial, Laser scanning, LIDAR 
ABSTRACT: 
A combination of airborne and terrestrial LIDAR data has been used to model extremely steep mountains that are crossed by the 
Nüria cog railway. This cog train is the only terrestrial transportation resort to reach the Nüria Valley in the Spanish Pyrenees. The 
purpose of this Digital Elevations Model (DEM) is the modeling of rocks that fall over the railway track in order to implement 
protection measures to mitigate this risk. 
The airborne LIDAR system was an Optech ALTM 3025. Special parameter settings were selected to improve the coverage of the 
area but as the mountains contain many overhangs and vertical walls some occlusions appeared in the airborne LIDAR data. A 
terrestrial survey was also carried out in order to improve the terrain modeling. The terrestrial campaign consisted of 5 scenes 
observed with a Riegl LMS-Z210 mounted on a tripod in 5 static positions in front of the problematic vertical areas. Terrestrial laser 
scenes were oriented identifying previously surveyed reflectors. 
The poster presents the methodology applied to integrate data from both LIDAR sensors and shows the obtained results. 
1. INTRODUCTION 
In this paper it is described the procedure that was used to build 
a 3D terrain model of an extremely steep terrain. The surveyed 
area comprises 71 Ha in the mountains crossed by a railway 
track in its path to the Nüria Valley in the Spanish Pyrenees. 
The generated terrain model was required to analyze the risk 
and to implement protection measures against the hazard of 
rock falling over the railway track. 
2. DATA AND METHODOLOGY 
The data was captured with two different instruments owned by 
the Institut Cartographic de Catalunya (ICC), an Optech ALTM 
3025 airborne lidar and a Riegl LMS-Z210 terrestrial lidar. The 
second instrument has been used in static positions and in 
dynamic mode (Talaya et al, 2004), however, the 
measurements done in dynamic mode had not been used to 
generate the final terrain model. 
The data processing and terrain model computation has been 
done using different programs, some of them commercial and 
some of them developed at the ICC, with successive 
approximations in a rather tricky way that is explained in the 
paper. 
2.1 Airborne lidar data 
The airborne lidar flight was done on July 28", 2003 and 
consisted of seven parallel strips with 20% overlap that fully 
covered the area of interest. These strips had a half scan angle 
of 7° (setting A in table 1). The almost vertical pointing of view 
reduced the probability of occlusions due to the mountains at 
the bottom of the canyon. Two additional strips were flown one 
over each side of the canyon with the purpose of getting more 
points distributed on the vertical walls of the mountains. These 
additional two strips had a half scan angle of 20°, the maximum 
allowed by the instrument (setting B). 
  
  
  
Setting 
A B 
Velocity (knots) 120 120 
Half Scan angle (degrees) 7 20 
Scan frequency (Hz) 33 20 
Pulse repetition (Hz) 25,000 25,000 
Height above ground (m) 1300 1300 
Strip overlap (96) 20 - 
Ray divergence (mrad) 0.2 0.2 
Point distance along (m) 0.88 1.54 
Point distance across (m) 0.89 E51 
Footprint (m) 0.260 0.260 
  
  
  
Table 1. Flight parameter settings. 
Finally, a cross strip was flown over the rest of the strips and 
also over a control field in a flat arca. A set of 48 points was 
measured with GPS-RTK on the control field with an estimated 
accuracy of 3 cm (1 sigma) to be used as ground control. 
Systematic errors in elevation for each strip were reduced using 
the strip adjustment procedure that is routinely applied to 
airborne lidar data at the ICC (Kornus and Ruiz, 2003). 
Corrections between —1.3 and 13.6 cm were applied to the 
elevations in each strip. Applying this approach accuracies in 
the order of 10-15 cm in elevation are usually obtained for lidar 
points in flat areas measured from 2300 m altitude above 
ground. 
Last echo airborne lidar points were classified into ground and 
non-ground points with the help of TerraScan software