In the recent years an increasing demand on precision digital elevation models (DEM) has to be noted – high-precision in this sense means that the accuracy of the DEM is at least ± 0.5 m in x/y and better than 0.2 m in z. Typical applications of such DEM are monitoring of coastal erosions, simulation of floods in river basins, and simulation of antennae sites or line-of-sight calculations for mobile communication networks. Many of those applications not only request for high precision, but also for short delivery times.
The TopoSys laserscanner system FALCON is a LIDAR system which is based on glass fiber tech-nology and especially designed to produce precision elevation models. Recently, FALCON has been complemented by an four channel image scanner. This configuration offers various advan-tages: First of all, elevation and image data is taken at exactly the same time. So, these data sets do not suffer from e.g. such problems that there are new buildings in one data set which in the older one do not exist. Another pro is that the LIDAR digital surface model (DSM) may be taken as data set to rectify image data to a true-ortho projection.
2. Toposys Lidar and line Scanner Camera
Since 1996 TopoSys produces precision LIDAR DEM, which mainly are used in large scale appli-cations. Both the LIDAR systems operated by TopoSys GmbH for services in Europe are supple-mented by line scanner cameras for about three years.
2.1. Precision LIDAR
The TopoSys LIDAR is a fiber optical line scanner system, which generates 83,000 measurements per second. Due to the glass fiber technology, the distance of measurement spots on ground is regu-lar in across- and along-track direction. In flight direction the scan lines slightly swing, which in-creases the overall distribution of measurement spots (see also Figure 1).
TopoSys LIDAR DEM are raster DEM typically available at raster widths between 0.5m and 2 m. At an aircraft speed of 70 m/sec and a survey altitude of e.g. 900 m above ground the LIDAR pro-vides an average density of 4 to 5 measurements per m². The 1 m standard product is calculated from these 4 to 5 measurements per m². The processing of the data will be adapted to the final use of the elevation data. So, the application determines, which of the measurements will be the best representatives within a raster cell, i.e. whether highest or lowest measurements are the best or whether measurements will be averaged.
Figure 1: Advanced scan pattern of the FALCON LIDAR
Table 1: Some parameters of TopoSys LIDAR.
|Measurement rate||83000 measurements per second|
|Laser wave length||1.5 µm (eye-safe)|
|Field of view||14°|
|range||< 1600 m|
|Swath width (at max. range)||390 m|
|Average measurement density (at max. range)||3 measurements / m²|
|Registration of first and last echo|
2.2. Line scanner camera
Table 2: Parameters of TopoSys Line Scanner Camera.
|Pixel size (at survey height: 900 m above ground)||0.5 m|
|Field of view||21°|
|Pixel per line||682|
|Spectral channels||(1) 440 – 490 nm
(2) 500 – 580 nm
(3) 580 – 660 nm
(4) 770 – 890 nm
The TopoSys line scanner camera gathers image data in four spectral channels. The design require-ment was that image data must have double the resolution of the raster elevation models. So, from e.g. a survey altitude of about 900 m above ground, image pixel size is 0.5 m, while DEM raster width is 1.0 m.
Gathered strips of image are mosaiced and compensated for radiometric differences. Images are available in true-ortho projection, either as four separate channels or as true-color (R,G,B) and color-infrared composites (NIR, R,G) in Geo-TIF format.
2.3. Amount of data
Operating line scanner camera and LIDAR system simultaneously nearly doubles the amount of data to be stored during a survey flight. Table 3 gives shows the amount of data collected during the survey flight assuming a scan area of 60 km² and a survey height of 900 m above ground.
Table 3: Amount of LIDAR and image data after survey flight and during processing
|Amount of data after survey flight|
|Navigation data||0.1 GByte|
|LIDAR data||3 GByte|
|Image data||4 Gbyte|
|Amount of data during processing|
|Navigation data||0.5… 1 GByte|
|LIDAR data||1.5…. 6 GByte|
|Image data||11….86 GByte|
Really significant is the storage capacity needed during data processing, when merging image strips to a complete image and applying radiometric correction.
3. DTM, DSM and TRUE ORTHO Images
One input for rectification of aerial images is the elevation model. The traditional way to correct image geometry is by using a bare ground model (digital terrain model DTM). The result of this kind of rectification is acceptable for small scale application; however, large scale images suffer from the fact that 3D objects have a lean. 3 D objects are not represented geometrically correct, because their height is not taken into account. In case of buildings, displacement of footprints and roofs is extremely unpleasant, when e.g. merging such image data with other, geographically cor-rect information.
In the past, 3D information of objects was frequently only extracted for “relevant” objects like building, bridges etc. Objects like trees, cars and other irrelevant objects were not considered, be-cause it was too much work to determine their heights.
Airborne Laserscanning is a means to generate a complete 3D model of the surface. A LIDAR DSM forms the ideal basis for a true-ortho rectification supposed that both scales (resolutions) of image data and LIDAR elevation data are harmonized. As result of such a work flow, Figure 2 shows a true-color and a color-infrared image of the TopoSys line scanner camera.
Figure 2: True-ortho images rectified with help of the LIDAR DSM.
4. Large Scale Applications
Precise LIDAR elevation models are needed in a large number of applications. In this sense “pre-cise” means that the elevation model is available at an accuracy of at least ± 0.5 m in x/y and better than 0.2 m in z (in the local coordinate system). The following shows elevation models of such quality used for
- simulation of floods
- monitoring of coastal erosion
- 3d city models
4.1. Simulation of floods
Traditionally, simulation of floods needs very precise elevation models. Aim of such simulations is to determine areas to be prevented, to determine area in which water shall run without causing big damages (retention areas), and to define respective engineering works.
Already for some time precise LIDAR DEM serve as input data for simulation. Due to the LIDAR’s ability to penetrate a forest canopy, bare ground models (DTM) can be produced even in vegetated areas. Figure 3 shows the LIDAR DSM as well as the DTM. The depicted area is about 2 * 2 km².
Simultaneously taken images are frequently also requested, in order to derive further parameters (like land use), which are needed for an accurate simulation.
Figure 3: 1 m raster LIDAR DSM (left) and DTM (right) of a river basin
© Landesamt f. Wasserwirtschaft, Munich, Germany
4.2. Monitoring of costal erosion
Figure 4 shows a part of the island Sylt in Germany. The erosion at the western part of the island amounts to about 1 million m³ per year. The total cost for coastal prevention of the western part of the island is more than 10 million Euros per year.
Figure 4: 1 m raster LIDAR DSM (left) of the island Sylt
© Amt f. Ländliche Räume, Husum, Germany
Precise LIDAR elevation models of the beach area are gathered regularly after the winter storms. LIDAR DTM in combination with bathymetric measurements, taken at the same time of the LIDAR survey, allow to determine the erosion volume as well as the locations, which have to be filled up.
As the water surface is clearly visible at FALCON’s laser wave length of 1.5 µm, it is easy to de-termine the land/water boundary. In Figure 4 the different textures of water surface (waves) and beach can be recognized.
4.3. 3 D city models
Layout of network for mobile communication, urban and environmental planning, simulation of floods, and simulation of noise distribution are some tasks, a precise 3 D city model may be used for. In case of the city of Venice, Italy (Figure 5) the building structure is extremely dense and streets and channels are very small. So, a grid width of a raster DEM must be 1 m, better 0.5 m. Therefore, Venice was scanned with a scan angle of ± 7° in order to avoid shadowing.
3D city models also allow for accurate, precise and up-to-date mapping of a road network. The LIDAR DSM, combined with complementary information (such as street names and house num-bers) in a GIS, provide an up-to-date coverage for vehicle navigation and positioning systems (see Figure 6). Of course, building blocks and road network may be vectorized to produce the conven-tional road map.
Figure 5: 1 m raster DSM of Venice, Italy
© CGR, Parma, Italy
Figure 6: 1 m raster LIDAR DSM of Athens, Greece
© Nama Geoinformatics, Athens, Greece
Precise LIDAR DSM like that of Venice may serve as basic data for a virtual city model. The first step to come to an virtual 3D model is to convert the buildings of the LIDAR DSM to vector for-mat. ATOP is a software package especially developed for this purpose. It was applied on the DSM of the city of Parma, Italy to produce CAD conform buildings.
After vectorisation the 3D building model may be draped with the true-ortho images in order to get texture on roofs and pavements (see alos Figure 7, left). Additionally taken images from ground may be put on the walls of the buildings to get a model, which looks close to reality (Figure 7, right).
Figure 7: 3 D virtual city model of Parma, Italy produced from LIDAR DSM
(© CGR, Parma, Italy; Real.IT, Schwäbisch Gmünd, Germany)
Beside small scale applications for LIDAR DEM (like topographic mapping), there is an increasing demand on precise LIDAR data for large scale applications. It could be observed that the market even asks for data of higher resolution than 1 m raster DEM.
The shown examples of applications have been selected to illustrate the quality of precise LIDAR DEM. The spectrum of application for precise elevation data is much bigger as shown here and in-cludes such applications like power line mapping, precision forest management, and opens pit monitoring.
During the relatively short time in which the complementary line scanner camera is operational, it turned out that more and more customers asked for both, elevation models and images. As all data is digital from the beginning, data processing is done relatively quick and highly automated – mainly quality control needs operator support.
- Lohr, Uwe; Digital Elevation Models by Laserscanning; CoastGIS ‘97, Second International Symposium on GIS and Computer Mapping for Coastal Zone Management; Aberdeen, UK; August 29-31, 1997
- Lohr, Uwe; The TopoSys Laserscanning system; Photogrammetric Record of the Royal Photogrammetric Society; London, UK; October 13, 1997
- Lohr, Uwe; Laserscan DEM for Various Applications; ISPRS Commission IV Symposium „GIS – Between Visions and Applications“, Stuttgart, Germany; September 7-10, 1998
- Wehr, Aloysius; Lohr, Uwe; Airborne Laserscanning – an introduction and overview ISPRS Journal of Photogrammetry and Remote Sensing, Volume 54 NOS. 2-3; July 1999
- Lohr, Uwe; High Resolution Laserscanning, not only for 3D-City Models; in Photogrammet-ric Week ’99; Dieter Fritsch, Rudolf Spiller (Eds.); Wichmann; 1999
- Schaller, Jörg; Digital Elevation Models by Laserscanning and Their Application in GIS Pro-jects; Proceedings of 20nd Annual ESRI International User Conference; San Diego; June 26-30, 2000
- Schiewe, J,; 3D city modelling using ultra high resolution and multi-sensorial remote sensing. In: Geo-Informationsystems GIS 6 / 2003, Hüting Verlag, Germany