- Email: [email protected]
Some examples of European activities in airborne laser techniques and an application in glaciology
Journal of Geodynamics 34 (2002) 347–355 www.elsevier.com/locate/jog
Some examples of European activities in airborne laser techniques and an application in glaciology E. Faveya,*, A. Wehrb, A. Geigera, H.-G. Kahlea a
Geodesy and Geodynamics Lab, Swiss Federal Institute of Technology, ETH-Ho¨nggerberg, CH-8093 Zu¨rich, Switzerland b Institute for Navigation, University of Stuttgart, Geschwister-Scholl-Str. 24D, D-70174 Stuttgart, Germany
Abstract Airborne Laser Altimetry (ALA) has experienced a rapid increase in popularity as a method serving a wide range of applications in Remote Sensing, Geodesy, Geophysics and Geodynamics. Besides the ‘traditional’ approach of using laser scanning solely as a supplement for photogrammetry in acquiring digital terrain models, ALA has also been applied to various geoscience research problems. After a short overview of airborne laser altimetry activities in Europe, an application of airborne laser scanning dedicated to Alpine Glaciology is presented. # 2002 Elsevier Science Ltd. All rights reserved.
1. European activities in airborne laser altimetry During the last decade Airborne Laser Altimetry (ALA) has matured so that very precise extended digital surface models were obtained operationally with height accuracies better than 20 cm. At present automatic algorithms are under development and partly operational for deriving digital elevation models (DEM) on the basis of ALA-data. Typical laser surveying results reveal that DEMs derived from ALA-data have at least the same and in many cases higher accuracy than photogrammetric surveys. Even in forest areas DEMs can be generated (Lindenberger, 1993; Kraus and Pfeifer, 1998), because the number of laser returns from ground are sufficient if the survey is carried out during the winter season. Also, laser altimeters which record the backscattered energy allow for correction to the lowest return, imaging ‘bald earth’ topography. This is impossible by photogrammetric means. Recognizing these advantages the group of customers
* Corresponding author at current address: u-blox AG, Zu¨rcherstr. 68, CH-8800 Thalwil, Switzerland. Fax: +41-1722-7447. 0264-3707/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved. PII: S0264-3707(02)00039-X
348
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
for ALA has been growing rapidly. Various service providers and airborne laser scanning system manufacturers have been established on the market. Manufacturers are e.g. Optech, Toposys and SAAB. In Europe service providers are e.g. Toposys, TopScan and Terra Surveys. In the middle of the last decade, national surveying agencies in the Netherlands and Germany already applied ALA in extensive areas. Today the prices for ALA surveys are competitively low compared to other methods, so that entire federal states in Germany are surveyed by ALA. In order to make quotations comparable and to validate the surveys, standards must be defined. Therefore, a user group of the Surveying and Mapping Agencies (SMA) of the Federal States of Germany proposes a standard procedure for verification and handling of laser scanner data. Petzold et al. (1999) discuss their experiences validating the products delivered by their laser scanning data providers. They compare and contrast the quality of the results with those of photogrammetric techniques. For this purpose, the ‘Organisation Europe´enne d’Etudes Photogrammetriques Experimentales’ (OEEPE) defined a test area with different terrain features, e.g. small towns, villages, forests and open country. They concluded that laser data was highly efficient in recovering the relief information needed for river management with the needed quality. The Survey Department of Rijkswaterstaat in the Netherlands used laser scanning for topographic measurements of the entire Dutch coast. They inventory sources of error and stress the importance of minimizing them by determining best conditions for the survey, using high point density, understanding all components of laser scanning processing, and improving it by formulation of models to compensate for errors, while developing adequate techniques for data reduction. It is also used for studying flooding scenarios and monitoring coastal morphology (Huising and Gomes Pereira, 1998). Another very important branch of ALA are bathymetry and topography in coastal regions. In Europe, the Swedish Hydrographic Dept. and the Swedish Defense Research Agency (FOA) are carrying out such surveys with the FOA Laser Airborne Sounder for Hydrography. This system has been operational since 1996 (Baltsavias, 1999, p. 197). ALA-data may be used for the determination of tree heights. After the DEM is derived, the height of the trees is extracted from the data set by subtracting the DEM from the digital surface model (DSM). The DSM describes the ground surface including surface features such as trees, vegetation and all manmade objects (Kraus and Pfeifer, 1998). Airborne laser profiling has been intensively used for Geoid determination in Greece and Switzerland. The principle with a self-calibration technique is described by Geiger et al. (1994). For geoid determination, the laser distance to the sea surface is measured. Since the sea surface height is close to the equipotential surface of the gravity field (discarding currents and wind effects) it is possible to obtain the relief of the gravity equipotential surface by sampling the sea surface height. The self-calibration procedure assumes, that the measured surface is flat, which is almost true in the case of lakes. Cocard et al. (1997) applied this method to map the surface of the Ionian Sea. Favey and Schlatter (1998) determined the geoid of the Lake of Geneva by both airborne laser altimetry and by conventional methods to serve as reference. Thus a qualitative and quantitative comparison could be made. The geoid determined by laser profiling matches the reference geoid by 3 cm with a systematic offset of 10 cm. An interesting application in Glaciology is measuring glacier surface elevations by remote sensing as an alternative to the traditional time-consuming stake methods. Kennett and Eiken (1997) made investigations on Hardangerjo¨kulen, Norway. They state a good match of traditional mass
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
349
balance measurements to the volume change measured by ground-based GPS elevations in spring, and airborne laser scanning measurements 6 months later. Yet they did not make any accuracy analysis using independent elevation mesurements of the glacier surface for the same time. On Unteraargletscher, Bernese Alps, Switzerland, Favey (2001) has conducted several laser scanning flights over a designated glacier part at the end of summer in 1997, 1998 and 1999 (Fig. 1). This example is discussed below.
2. Feature extraction and DEM generation With regard to the requirements of the national surveying agencies the first processing objective of laser scanner data is the derivation of a DEM. This means the application of very reliable filter algorithms which extract automatically e.g. buildings, trees and vegetation. The task of classifying ground points, vegetation and buildings solely from laser scanner data without the input of additional data is addressed by Axelsson (1999). He presents an algorithm based on the ‘minimum description length’ criterion. Buildings, breakpoints and vegetation are successfully classified apart from ordinary ground points.
Fig. 1. Map of Unteraargletscher and its main contributaries Lauteraargletscher, Strahlegggletscher and Finsteraargletscher, showing the location of the test site. The area to be mapped by airborne laser scanning was the upper part of the glacier to the west of the 658 km line. The glacier boundary is drawn as a bold line. Contour interval 100 m.
350
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
The determination of the topology of the building and the distinction between manmade objects and vegetation is of most interest for Geoinformation Systems (GIS). Haala and Brenner (1999) attempt to extract buildings and trees with the use of additional information helping the classification of the laser scanner data. In a first approach, multispectral imagery is combined with laser scanner data to be processed in an integrated classification algorithm. In a second method, 2-D ground plan information is used to help 3-D reconstruction of buildings. Hug (1996) uses a laser scanner which measures not only distances but also ground reflectance. This additional information is exploited when classifying laser scanner data in urban areas based on the knowledge of the spectral reflectance properties of the different surface types. With respect to automatic feature extraction Maas and Vosselman (1999) focus on two methods for generating 3-D city models and extraction of buildings from laser scanner data. In the first method they try to model buildings based on invariant moments of point clouds, allowing asymmetries such as dorms on a gable roof. The second approach is based on the intersection of planar roof faces derived from 2-D triangulation. They apply these techniques to original laser scanner data points acquired with a high point density of more than 1 point/m2 without previous interpolation on a regular grid, and find these data very valuable for the determination of 3-D models.
3. Example of an application in glaciology The following example forms part of a project funded by the Swiss National Science Foundation on ‘Mass Balance Determination of Glaciers with the Use of State-of-the-art Remote Sensing Methods and a Numerical Flow Model’ involving the Geodesy, Photogrammetry, Hydraulics, Hydrology and Glaciology labs of ETH Zurich, with the cooperation of the Institute of Navigation, University of Stuttgart. To achieve the aims of the project, observations from remote sensing and ground-based instruments are combined in a numerical model described by GuDmundsson and Bauder (1999). The geodetic part of the project involves airborne laser scanning to produce a surface model of the upper parts of the glacier with an accuracy of about 0.5–1 m in time periods of 1–5 years. Up to now, these products were derived manually by analytical photogrammetry. A Twin Otter of the Swiss Directorate of Surveying (Fig. 2) was used for the laser scanning flights. GPS measurements for positioning were acquired at a rate of 2 Hz using two double frequency Trimble SSI receivers, one on board the aircraft, the other one on ground at a fixed, known location about 15 km away from the glacier. In post-processing, the coordinates of the aircraft were estimated using ‘on the fly’ ambiguity resolution techniques for differential carrier phase measurements (Cocard, 1995). In addition to the double frequency receiver for positioning, the aircraft was equipped with four single frequency GPS antennae on fuselage, wings and tail, each connected to a NovAtel receiver. In conjunction with an INS, the GPS antenna array was used for attitude determination using a loosely-coupled implementation of merging the two methods. Fig. 3 shows the data flow of processing laser data. The determination of GPS attitude and its merging with INS is described in detail by Favey (2001). The laser scanning system is a circular scanner using the continuous wave ranging method described by Wehr and Lohr (1999). The wavelength of the laser light is 810 nm, and distances
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
351
Fig. 2. Laser scanning principle. Aircraft: Twinotter of Swiss Directorate of Surveying used for the campaign. The laser scanner works as a circular scanner ‘drawing’ ellipses on the ground. All instruments are drawn at their actual location during measurement flights. The angle ’ depicts the swath angle of the laser scanner.
are sampled at a rate of 7.69 kHz. The footprint of a single measurement is 1.13 m at a height of 700 m above ground. Reflections are obtained up to 1300 m above snow and ice. The swath angle is 13.6 which corresponds to a swath width of 190–470 m at a flying height of 400–1000 m above ground respectively. The scanning system measured groundpoints approximately every 2 m at a flight speed of about 70 m s 1. At the end of summer 1998 and 1999, laser scanning campaigns were carried out. Laser scanning measurements of two consecutive years were evaluated in order to determine the surface elevation change of Lauteraargletscher. The results shown in Fig. 4 indicate a difference of up to 4 m in the firn area of the glacier. This is mainly due to the fact, that winter 98/99 was a season with extraordinary precipitation which fell mostly as snow above 2500 m (about the double of a winter’s mean precipitation). The cliff parts in the northwestern corner of Fig. 4, which were not snow-covered in both years show a difference of 0 m (Fig. 5). The measurements on the lower part (southeastern corner in Fig. 4) are much noisier due to the higher flying height. Large local differences throughout the glacier are caused by crevasses that opened or closed or merely moved between 1998 and 1999. To verify the accuracy of the laser scanning system, two flight lines in opposite direction were flown over a runway (Favey et al., 1999). This runway was also measured terrestrially using GPS to provide ground truth. Table 1 shows, that the laser DEM is offset about 0.2 m. The comparison of the laser DEM with known ground heights shows an RMS of 0.3 m. This absolute offset,
352
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
Fig. 3. Data flow of processing laser data.
Table 1 Statistics of differences between ground based GPS heights and airborne laser heights Version
No. of pts
RMS (m)
Ground GPS—laser line east Ground GPS—laser line west
2066 2467
0.26 0.30
Mean (m) 0.24 0.22
Max abs (m) 2.50 6.23
probably due to a bias in antenna height determination, does not affect the quality of relative evaluations like calculating temporal differences of the glacier, as long as the system setup is not changed. The generated DEM on the glacier itself was compared with a DEM, generated independently by photogrammetric means (Favey et al., 2000; Favey, 2001). The temporal difference data for the years 1999 1998 was determined by airborne laser scanning and by photogrammetric means. The two entirely independent datasets, compared in Table 2, match with an RMS of about 0.7 m on the glacier area. The a priori accuracy for a single DEM determined by either method is
353
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
Fig. 4. Lauteraargletscher: Difference of laser DEM 1999 and laser DEM 1998 in m. This figure shows a difference of up to 4 m at higher altitude (northwestern corner) and a gradual decrase towards the lower glacier area. The white line denotes the location of the profile in Fig. 5. Contour lines show the glacier topography as derived from the laser data 1999. Contour interval of 100 m.
Table 2 Comparison of surface elevation change distribution determined by laser scanning and independently by photogrammetry for the region of Lauteraargletschera Dataset
No. of pts
RMS (m)
Mean (m)
Max abs. (m)
All points Glacier points
25 461 24 896
0.99 0.74
0.10 0.07
33.58 15.89
a
The lower line contains only the measurements, that were within a defined glacier boundary polygon.
valued at about 0.4 m one sigma. Since glacier surface roughnesses are on the same order, a better accuracy DEM will hardly result in better estimates of changes in surface altitude. The comparison in Table 2 shows, that the surface elevation change distribution was achieved by airborne laser scanning with an accuracy of about 0.5 m. A detailed analysis of these data is given by Favey (2001). To improve the accuracy of the laser scanning, further research on modeling of the tropospheric influences onto DGPS trajectory determination needs to be done. During this campaign, the data suffered most from the sensitivity of the scanning mirror to flight dynamics. For Alpine areas, a scanning mechanism without a rotating part would improve the data quality significantly.
354
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
Fig. 5. Profile through laser data at Lauteraarsattel (white line in Fig. 4). Over the cliff part (350–540) the profiles match each other, whereas on the glacier firn (540–700) the profile is systematically higher in 1999 by about 4 m.
4. Conclusion Airborne Laser Altimetry is a powerful tool serving a multitude of applications. In Europe, many well established companies offer ALA services. The ALA researchers focus mainly on algorithm development for GIS integration of laser data by automatically detecting buildings, vegetation, and generating DEMs. On the other hand, some research is done in the field of laser hardware and ALA system development. ALA offers many interesting applications for geosciences such as geoid determination, or volume change of changing topography like glaciers, coastal zones, and land slides. In the presented example, we were able to detect a volume increase of the firn area by up to 3–4 m. This shows the potential of ALA especially for regions, where conventional photogrammetric methods encounter serious problems because of e.g. lack of sufficient texture or fast deforming terrain, where the establishment of reliable ground control points is almost impossible. In numerous examples, ALA has proven to be a suitable and operational method for geoscience and surveying with increasing popularity. References Axelsson, P., 1999. Processing of laser scanner data—algorithms and applications. ISPRS Journal of Photogrammetry & Remote Sensing 54 (2–3), 138–147. Baltsavias, E.P., 1999. Airborne laser scanning: existing systems and firms and other resources. ISPRS Journal of Photogrammetry & Remote Sensing 54 (2–3), 164–198.
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
355
Cocard, M., 1995. High Precision GPS Processing in Kinematic Mode, Vol. 52. Schweizerische Geoda¨tische Kommission (geoda¨tisch-geophysikalische arbeiten in der schweiz edition). Cocard, M., Geiger, A., Kahle, H.-G., Veis, G., Milas, P., 1997. Regional sea surface determination by using Airborne Laser techniques. In: Annales Geophysicae. XXII General Assembly of EGS, Vol. 5 of Part I Suppl. I, p. C191. Favey, E., 2001. Investigation and Improvement of Airborne Laser Scanning Technique for Monitoring Surface Elevation Changes of Glaciers. PhD Thesis, Swiss Federal Institute of Technology, Zurich, No. 14045. Favey, E., Cerniar, M., Cocard, M., Geiger, A., 1999. Sensor attitude determination using GPS antenna array and INS. In: ISPRS WG III/1 Workshop: Direct Versus Indirect Methods of Sensor Orientation, Barcelona, 25–26 November, pp. 28–38. Favey, E., Pateraki, M., Baltsavias, E.P., Bauder, A., Bo¨sch, H., 2000. Surface Modelling for Alpine Glacier Monitoring by Airborne Laser Scanning and Digital Photogrammetry. In: Int’l Archives of Photogrammetry and Remote Sensing, XIXth ISPRS Congress, Amsterdam, Vol. XXXIII (B4), pp. 269–277. Favey, E., Schlatter, A., 1998. Geoidbestimmung am Lac Le´man. Technical Report, Institut fu¨r Geoda¨sie und Photogrammetrie, ETH Zu¨rich, Bericht Nr. 282. Geiger, A., Cocard, M., Peter, Y., Kahle, H.-G., 1994. Precise DGPS and self-calibrating Laser altimetry for airborne marine geoid determination. In: Proceedings of 7th Int. Techn. Meeting of Institute of Navigation (ION) GPS, pp. 471–476. GuDmundsson, G.H., Bauder, A., 1999. Towards an indirect determination of the mass-balance distribution of glaciers using the kinematic boundary condition. Geogr. Ann. 81A (4), 575–584. Haala, N., Brenner, C., 1999. Extraction of buildings and trees in urban environments. ISPRS Journal of Photogrammetry & Remote Sensing 54 (2–3), 130–137. Hug, C., 1996. Combined use of laser scanner geometry and reflectance data to identify surface objects. In: OEEPEWorkshop ‘3-D city models’, Bonn, 9–11 October. Huising, E.J., Gomes Pereira, L.M., 1998. Errors and accuracy estimates of laser data acquired by various laser scanning systems for topographic applications. ISPRS Journal of Photogrammetry & Remote Sensing 53 (5), 245–261. Kennett, M., Eiken, T., 1997. Airborne measurement of glacier surface elevation by scanning laser altimeter. Ann. Glaciol. 24, 293–296. Kraus, K., Pfeifer, N., 1998. Determination of terrain models in wooded areas with airborne laser scanner data. ISPRS J. Photogram. Rem. Sensing 53 (4), 193–203. Lindenberger, J., 1993. Laser-Profilmessungen zur topographischen Gela¨ndeaufnahme. PhD Thesis, Universita¨t Stuttgart, Deutsche Geoda¨tische Kommission, Mu¨nchen. Reihe C, Heft Nr. 400. Maas, H.-G., Vosselman, G., 1999. Two algorithms for extracting building models from raw laser altimetry data. ISPRS Journal of Photogrammetry & Remote Sensing 54 (2–3), 153–163. Petzold, B., Reiss, P., Sto¨ssel, W., 1999. Laser scanning—surveying and mapping agencies are using a new technique for the derivation of digital terrain models. ISPRS Journal of Photogrammetry & Remote Sensing 54 (2–3), 95–104. Wehr, A., Lohr, U., 1999. Airborne laser scanning—an introduction and overview. ISPRS Journal of Photogrammetry & Remote Sensing 54 (2–3), 68–82.
Some examples of European activities in airborne laser techniques and an application in glaciology E. Faveya,*, A. Wehrb, A. Geigera, H.-G. Kahlea a
Geodesy and Geodynamics Lab, Swiss Federal Institute of Technology, ETH-Ho¨nggerberg, CH-8093 Zu¨rich, Switzerland b Institute for Navigation, University of Stuttgart, Geschwister-Scholl-Str. 24D, D-70174 Stuttgart, Germany
Abstract Airborne Laser Altimetry (ALA) has experienced a rapid increase in popularity as a method serving a wide range of applications in Remote Sensing, Geodesy, Geophysics and Geodynamics. Besides the ‘traditional’ approach of using laser scanning solely as a supplement for photogrammetry in acquiring digital terrain models, ALA has also been applied to various geoscience research problems. After a short overview of airborne laser altimetry activities in Europe, an application of airborne laser scanning dedicated to Alpine Glaciology is presented. # 2002 Elsevier Science Ltd. All rights reserved.
1. European activities in airborne laser altimetry During the last decade Airborne Laser Altimetry (ALA) has matured so that very precise extended digital surface models were obtained operationally with height accuracies better than 20 cm. At present automatic algorithms are under development and partly operational for deriving digital elevation models (DEM) on the basis of ALA-data. Typical laser surveying results reveal that DEMs derived from ALA-data have at least the same and in many cases higher accuracy than photogrammetric surveys. Even in forest areas DEMs can be generated (Lindenberger, 1993; Kraus and Pfeifer, 1998), because the number of laser returns from ground are sufficient if the survey is carried out during the winter season. Also, laser altimeters which record the backscattered energy allow for correction to the lowest return, imaging ‘bald earth’ topography. This is impossible by photogrammetric means. Recognizing these advantages the group of customers
* Corresponding author at current address: u-blox AG, Zu¨rcherstr. 68, CH-8800 Thalwil, Switzerland. Fax: +41-1722-7447. 0264-3707/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved. PII: S0264-3707(02)00039-X
348
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
for ALA has been growing rapidly. Various service providers and airborne laser scanning system manufacturers have been established on the market. Manufacturers are e.g. Optech, Toposys and SAAB. In Europe service providers are e.g. Toposys, TopScan and Terra Surveys. In the middle of the last decade, national surveying agencies in the Netherlands and Germany already applied ALA in extensive areas. Today the prices for ALA surveys are competitively low compared to other methods, so that entire federal states in Germany are surveyed by ALA. In order to make quotations comparable and to validate the surveys, standards must be defined. Therefore, a user group of the Surveying and Mapping Agencies (SMA) of the Federal States of Germany proposes a standard procedure for verification and handling of laser scanner data. Petzold et al. (1999) discuss their experiences validating the products delivered by their laser scanning data providers. They compare and contrast the quality of the results with those of photogrammetric techniques. For this purpose, the ‘Organisation Europe´enne d’Etudes Photogrammetriques Experimentales’ (OEEPE) defined a test area with different terrain features, e.g. small towns, villages, forests and open country. They concluded that laser data was highly efficient in recovering the relief information needed for river management with the needed quality. The Survey Department of Rijkswaterstaat in the Netherlands used laser scanning for topographic measurements of the entire Dutch coast. They inventory sources of error and stress the importance of minimizing them by determining best conditions for the survey, using high point density, understanding all components of laser scanning processing, and improving it by formulation of models to compensate for errors, while developing adequate techniques for data reduction. It is also used for studying flooding scenarios and monitoring coastal morphology (Huising and Gomes Pereira, 1998). Another very important branch of ALA are bathymetry and topography in coastal regions. In Europe, the Swedish Hydrographic Dept. and the Swedish Defense Research Agency (FOA) are carrying out such surveys with the FOA Laser Airborne Sounder for Hydrography. This system has been operational since 1996 (Baltsavias, 1999, p. 197). ALA-data may be used for the determination of tree heights. After the DEM is derived, the height of the trees is extracted from the data set by subtracting the DEM from the digital surface model (DSM). The DSM describes the ground surface including surface features such as trees, vegetation and all manmade objects (Kraus and Pfeifer, 1998). Airborne laser profiling has been intensively used for Geoid determination in Greece and Switzerland. The principle with a self-calibration technique is described by Geiger et al. (1994). For geoid determination, the laser distance to the sea surface is measured. Since the sea surface height is close to the equipotential surface of the gravity field (discarding currents and wind effects) it is possible to obtain the relief of the gravity equipotential surface by sampling the sea surface height. The self-calibration procedure assumes, that the measured surface is flat, which is almost true in the case of lakes. Cocard et al. (1997) applied this method to map the surface of the Ionian Sea. Favey and Schlatter (1998) determined the geoid of the Lake of Geneva by both airborne laser altimetry and by conventional methods to serve as reference. Thus a qualitative and quantitative comparison could be made. The geoid determined by laser profiling matches the reference geoid by 3 cm with a systematic offset of 10 cm. An interesting application in Glaciology is measuring glacier surface elevations by remote sensing as an alternative to the traditional time-consuming stake methods. Kennett and Eiken (1997) made investigations on Hardangerjo¨kulen, Norway. They state a good match of traditional mass
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
349
balance measurements to the volume change measured by ground-based GPS elevations in spring, and airborne laser scanning measurements 6 months later. Yet they did not make any accuracy analysis using independent elevation mesurements of the glacier surface for the same time. On Unteraargletscher, Bernese Alps, Switzerland, Favey (2001) has conducted several laser scanning flights over a designated glacier part at the end of summer in 1997, 1998 and 1999 (Fig. 1). This example is discussed below.
2. Feature extraction and DEM generation With regard to the requirements of the national surveying agencies the first processing objective of laser scanner data is the derivation of a DEM. This means the application of very reliable filter algorithms which extract automatically e.g. buildings, trees and vegetation. The task of classifying ground points, vegetation and buildings solely from laser scanner data without the input of additional data is addressed by Axelsson (1999). He presents an algorithm based on the ‘minimum description length’ criterion. Buildings, breakpoints and vegetation are successfully classified apart from ordinary ground points.
Fig. 1. Map of Unteraargletscher and its main contributaries Lauteraargletscher, Strahlegggletscher and Finsteraargletscher, showing the location of the test site. The area to be mapped by airborne laser scanning was the upper part of the glacier to the west of the 658 km line. The glacier boundary is drawn as a bold line. Contour interval 100 m.
350
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
The determination of the topology of the building and the distinction between manmade objects and vegetation is of most interest for Geoinformation Systems (GIS). Haala and Brenner (1999) attempt to extract buildings and trees with the use of additional information helping the classification of the laser scanner data. In a first approach, multispectral imagery is combined with laser scanner data to be processed in an integrated classification algorithm. In a second method, 2-D ground plan information is used to help 3-D reconstruction of buildings. Hug (1996) uses a laser scanner which measures not only distances but also ground reflectance. This additional information is exploited when classifying laser scanner data in urban areas based on the knowledge of the spectral reflectance properties of the different surface types. With respect to automatic feature extraction Maas and Vosselman (1999) focus on two methods for generating 3-D city models and extraction of buildings from laser scanner data. In the first method they try to model buildings based on invariant moments of point clouds, allowing asymmetries such as dorms on a gable roof. The second approach is based on the intersection of planar roof faces derived from 2-D triangulation. They apply these techniques to original laser scanner data points acquired with a high point density of more than 1 point/m2 without previous interpolation on a regular grid, and find these data very valuable for the determination of 3-D models.
3. Example of an application in glaciology The following example forms part of a project funded by the Swiss National Science Foundation on ‘Mass Balance Determination of Glaciers with the Use of State-of-the-art Remote Sensing Methods and a Numerical Flow Model’ involving the Geodesy, Photogrammetry, Hydraulics, Hydrology and Glaciology labs of ETH Zurich, with the cooperation of the Institute of Navigation, University of Stuttgart. To achieve the aims of the project, observations from remote sensing and ground-based instruments are combined in a numerical model described by GuDmundsson and Bauder (1999). The geodetic part of the project involves airborne laser scanning to produce a surface model of the upper parts of the glacier with an accuracy of about 0.5–1 m in time periods of 1–5 years. Up to now, these products were derived manually by analytical photogrammetry. A Twin Otter of the Swiss Directorate of Surveying (Fig. 2) was used for the laser scanning flights. GPS measurements for positioning were acquired at a rate of 2 Hz using two double frequency Trimble SSI receivers, one on board the aircraft, the other one on ground at a fixed, known location about 15 km away from the glacier. In post-processing, the coordinates of the aircraft were estimated using ‘on the fly’ ambiguity resolution techniques for differential carrier phase measurements (Cocard, 1995). In addition to the double frequency receiver for positioning, the aircraft was equipped with four single frequency GPS antennae on fuselage, wings and tail, each connected to a NovAtel receiver. In conjunction with an INS, the GPS antenna array was used for attitude determination using a loosely-coupled implementation of merging the two methods. Fig. 3 shows the data flow of processing laser data. The determination of GPS attitude and its merging with INS is described in detail by Favey (2001). The laser scanning system is a circular scanner using the continuous wave ranging method described by Wehr and Lohr (1999). The wavelength of the laser light is 810 nm, and distances
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
351
Fig. 2. Laser scanning principle. Aircraft: Twinotter of Swiss Directorate of Surveying used for the campaign. The laser scanner works as a circular scanner ‘drawing’ ellipses on the ground. All instruments are drawn at their actual location during measurement flights. The angle ’ depicts the swath angle of the laser scanner.
are sampled at a rate of 7.69 kHz. The footprint of a single measurement is 1.13 m at a height of 700 m above ground. Reflections are obtained up to 1300 m above snow and ice. The swath angle is 13.6 which corresponds to a swath width of 190–470 m at a flying height of 400–1000 m above ground respectively. The scanning system measured groundpoints approximately every 2 m at a flight speed of about 70 m s 1. At the end of summer 1998 and 1999, laser scanning campaigns were carried out. Laser scanning measurements of two consecutive years were evaluated in order to determine the surface elevation change of Lauteraargletscher. The results shown in Fig. 4 indicate a difference of up to 4 m in the firn area of the glacier. This is mainly due to the fact, that winter 98/99 was a season with extraordinary precipitation which fell mostly as snow above 2500 m (about the double of a winter’s mean precipitation). The cliff parts in the northwestern corner of Fig. 4, which were not snow-covered in both years show a difference of 0 m (Fig. 5). The measurements on the lower part (southeastern corner in Fig. 4) are much noisier due to the higher flying height. Large local differences throughout the glacier are caused by crevasses that opened or closed or merely moved between 1998 and 1999. To verify the accuracy of the laser scanning system, two flight lines in opposite direction were flown over a runway (Favey et al., 1999). This runway was also measured terrestrially using GPS to provide ground truth. Table 1 shows, that the laser DEM is offset about 0.2 m. The comparison of the laser DEM with known ground heights shows an RMS of 0.3 m. This absolute offset,
352
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
Fig. 3. Data flow of processing laser data.
Table 1 Statistics of differences between ground based GPS heights and airborne laser heights Version
No. of pts
RMS (m)
Ground GPS—laser line east Ground GPS—laser line west
2066 2467
0.26 0.30
Mean (m) 0.24 0.22
Max abs (m) 2.50 6.23
probably due to a bias in antenna height determination, does not affect the quality of relative evaluations like calculating temporal differences of the glacier, as long as the system setup is not changed. The generated DEM on the glacier itself was compared with a DEM, generated independently by photogrammetric means (Favey et al., 2000; Favey, 2001). The temporal difference data for the years 1999 1998 was determined by airborne laser scanning and by photogrammetric means. The two entirely independent datasets, compared in Table 2, match with an RMS of about 0.7 m on the glacier area. The a priori accuracy for a single DEM determined by either method is
353
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
Fig. 4. Lauteraargletscher: Difference of laser DEM 1999 and laser DEM 1998 in m. This figure shows a difference of up to 4 m at higher altitude (northwestern corner) and a gradual decrase towards the lower glacier area. The white line denotes the location of the profile in Fig. 5. Contour lines show the glacier topography as derived from the laser data 1999. Contour interval of 100 m.
Table 2 Comparison of surface elevation change distribution determined by laser scanning and independently by photogrammetry for the region of Lauteraargletschera Dataset
No. of pts
RMS (m)
Mean (m)
Max abs. (m)
All points Glacier points
25 461 24 896
0.99 0.74
0.10 0.07
33.58 15.89
a
The lower line contains only the measurements, that were within a defined glacier boundary polygon.
valued at about 0.4 m one sigma. Since glacier surface roughnesses are on the same order, a better accuracy DEM will hardly result in better estimates of changes in surface altitude. The comparison in Table 2 shows, that the surface elevation change distribution was achieved by airborne laser scanning with an accuracy of about 0.5 m. A detailed analysis of these data is given by Favey (2001). To improve the accuracy of the laser scanning, further research on modeling of the tropospheric influences onto DGPS trajectory determination needs to be done. During this campaign, the data suffered most from the sensitivity of the scanning mirror to flight dynamics. For Alpine areas, a scanning mechanism without a rotating part would improve the data quality significantly.
354
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
Fig. 5. Profile through laser data at Lauteraarsattel (white line in Fig. 4). Over the cliff part (350–540) the profiles match each other, whereas on the glacier firn (540–700) the profile is systematically higher in 1999 by about 4 m.
4. Conclusion Airborne Laser Altimetry is a powerful tool serving a multitude of applications. In Europe, many well established companies offer ALA services. The ALA researchers focus mainly on algorithm development for GIS integration of laser data by automatically detecting buildings, vegetation, and generating DEMs. On the other hand, some research is done in the field of laser hardware and ALA system development. ALA offers many interesting applications for geosciences such as geoid determination, or volume change of changing topography like glaciers, coastal zones, and land slides. In the presented example, we were able to detect a volume increase of the firn area by up to 3–4 m. This shows the potential of ALA especially for regions, where conventional photogrammetric methods encounter serious problems because of e.g. lack of sufficient texture or fast deforming terrain, where the establishment of reliable ground control points is almost impossible. In numerous examples, ALA has proven to be a suitable and operational method for geoscience and surveying with increasing popularity. References Axelsson, P., 1999. Processing of laser scanner data—algorithms and applications. ISPRS Journal of Photogrammetry & Remote Sensing 54 (2–3), 138–147. Baltsavias, E.P., 1999. Airborne laser scanning: existing systems and firms and other resources. ISPRS Journal of Photogrammetry & Remote Sensing 54 (2–3), 164–198.
E. Favey et al. / Journal of Geodynamics 34 (2002) 347–355
355
Cocard, M., 1995. High Precision GPS Processing in Kinematic Mode, Vol. 52. Schweizerische Geoda¨tische Kommission (geoda¨tisch-geophysikalische arbeiten in der schweiz edition). Cocard, M., Geiger, A., Kahle, H.-G., Veis, G., Milas, P., 1997. Regional sea surface determination by using Airborne Laser techniques. In: Annales Geophysicae. XXII General Assembly of EGS, Vol. 5 of Part I Suppl. I, p. C191. Favey, E., 2001. Investigation and Improvement of Airborne Laser Scanning Technique for Monitoring Surface Elevation Changes of Glaciers. PhD Thesis, Swiss Federal Institute of Technology, Zurich, No. 14045. Favey, E., Cerniar, M., Cocard, M., Geiger, A., 1999. Sensor attitude determination using GPS antenna array and INS. In: ISPRS WG III/1 Workshop: Direct Versus Indirect Methods of Sensor Orientation, Barcelona, 25–26 November, pp. 28–38. Favey, E., Pateraki, M., Baltsavias, E.P., Bauder, A., Bo¨sch, H., 2000. Surface Modelling for Alpine Glacier Monitoring by Airborne Laser Scanning and Digital Photogrammetry. In: Int’l Archives of Photogrammetry and Remote Sensing, XIXth ISPRS Congress, Amsterdam, Vol. XXXIII (B4), pp. 269–277. Favey, E., Schlatter, A., 1998. Geoidbestimmung am Lac Le´man. Technical Report, Institut fu¨r Geoda¨sie und Photogrammetrie, ETH Zu¨rich, Bericht Nr. 282. Geiger, A., Cocard, M., Peter, Y., Kahle, H.-G., 1994. Precise DGPS and self-calibrating Laser altimetry for airborne marine geoid determination. In: Proceedings of 7th Int. Techn. Meeting of Institute of Navigation (ION) GPS, pp. 471–476. GuDmundsson, G.H., Bauder, A., 1999. Towards an indirect determination of the mass-balance distribution of glaciers using the kinematic boundary condition. Geogr. Ann. 81A (4), 575–584. Haala, N., Brenner, C., 1999. Extraction of buildings and trees in urban environments. ISPRS Journal of Photogrammetry & Remote Sensing 54 (2–3), 130–137. Hug, C., 1996. Combined use of laser scanner geometry and reflectance data to identify surface objects. In: OEEPEWorkshop ‘3-D city models’, Bonn, 9–11 October. Huising, E.J., Gomes Pereira, L.M., 1998. Errors and accuracy estimates of laser data acquired by various laser scanning systems for topographic applications. ISPRS Journal of Photogrammetry & Remote Sensing 53 (5), 245–261. Kennett, M., Eiken, T., 1997. Airborne measurement of glacier surface elevation by scanning laser altimeter. Ann. Glaciol. 24, 293–296. Kraus, K., Pfeifer, N., 1998. Determination of terrain models in wooded areas with airborne laser scanner data. ISPRS J. Photogram. Rem. Sensing 53 (4), 193–203. Lindenberger, J., 1993. Laser-Profilmessungen zur topographischen Gela¨ndeaufnahme. PhD Thesis, Universita¨t Stuttgart, Deutsche Geoda¨tische Kommission, Mu¨nchen. Reihe C, Heft Nr. 400. Maas, H.-G., Vosselman, G., 1999. Two algorithms for extracting building models from raw laser altimetry data. ISPRS Journal of Photogrammetry & Remote Sensing 54 (2–3), 153–163. Petzold, B., Reiss, P., Sto¨ssel, W., 1999. Laser scanning—surveying and mapping agencies are using a new technique for the derivation of digital terrain models. ISPRS Journal of Photogrammetry & Remote Sensing 54 (2–3), 95–104. Wehr, A., Lohr, U., 1999. Airborne laser scanning—an introduction and overview. ISPRS Journal of Photogrammetry & Remote Sensing 54 (2–3), 68–82.