The Geoglyphs of San Ignacio - New Results from the NASCA Project
Armin Grün, Sabine Beutner and ETH Zurich
Institute of Geodesy and Photogrammetry
E-Mail: agruen@geod.baug.ethz.ch, beutner@geod.baug.ethz.ch
Web-Site: http://www.photogrammetry.ethz.ch
Armin Grün, Sabine Beutner and ETH Zurich
Institute of Geodesy and Photogrammetry
E-Mail: agruen@geod.baug.ethz.ch, beutner@geod.baug.ethz.ch
Web-Site: http://www.photogrammetry.ethz.ch
Abstract
We report about the current state of our Nasca project. The block Sacramento (called Palpa in earlier publications) is fully processed and the block San Ignacio is almost completed. We show the results of San Ignacio, which were obtained through the processes of photo triangulation, DTM measurement, orthoimage and orthomosaic generation and 3-D vector extraction. High quality photorealistic visualizations of the terrain and the geoglyphs are already available. The next task will be the integration of the data on a GIS platform.
Introduction
The geoglyphs of Nasca, Peru, are considered one of the world's major mysteries. Despite the fact that many efforts have been made in the past 70 years to record and analyze the geoglyphs no convincing explanation has emerged yet which would explain the purpose of these drawings in the sand beyond any doubt. Many hypotheses have been put forward and each author will find a certain number of lines and figures in support of his/her opinion. The huge number of lines, trapezoids, star centers, biomorphic figures, etc. has up to now defied any comprehensive recording and analysis.
We are currently conducting an ambitious project which aims at the complete 3-D recording of the geoglyphs of the areas of Nasca, Sacramento and San Ignacio (a total of about 300 km2) with photogrammetric means. This is done in cooperation with a group of archaeologists who will try to relate their excavation results to the geoglyphs in order to get a more complete picture of the overall cultural and social background of the Nasca people (200 B.C. - 600 A.D.).
We are using the latest digital photogrammetric technology for the recording of the geoglyphs and some of the most advanced visualization software. Finally, all data will be integrated on a GIS platform, which will require the efficient management of both vector- and extensive raster image data.
We have aquired over 1000 aerial photographs at scales 1:5 000 and 1:10 000 both in colour and B/W. In addition we have collected existing, older photographs which can give us an indication of the damage that has been done to the geoglyphs in recent years.
The block Sacramento has been fully processed and visualization results have been shown to the public on a SGI Onyx2 in an interactive mode at the Nasca exhibition in the Rietberg Museum, Zurich and in the VisDome of ETH Zurich as stereo projection on a large 4 x 12 m2 screen and on other occasions. For details on the complete project (motivation, goals, data acquisition, processing and visualization of the Sacramento block) we refer to Grün, et al., 2000.
In this paper we report about the ongoing work on the San Ignacio block. Here we have only B/W images available for processing. The phototriangulation has been performed on an Analytical Plotter, since automated triangulation does not deliver the desired results. The DTM has been measured completely in manual mode, because of the high quality requirements. Digital orthoimages have been produced over the largest part of the area and a mosaic has been generated. This is the basis for photorealistic visualization. A number of geoglyphs have been measured in the stereomodels in vector mode (so far we have included about 6 700 vectors in the dataset), but this work is not finished yet.
Photoriangulation
Figure 1 shows the location of the three blocks Sacramento (in earlier publications called "Palpa"), San Ignacio and Nasca, which are also processed in this sequence. A detailed layout of the triangulated part of the block San Ignacio is given in Figure 2.
Figure 1: Location of the blocks Sacramento, San Ignacio, Nasca in southern Peru
The block San Ignacio was flown in May 1998 with a Zeiss RMK A15 camera at a nominal image scale of 1:5 000 in B/W with 60 % forward overlap and 60 % sidelap. Out of the total number of 189 aerial images 169 images, arranged in 11 strips of varying length, were selected for triangulation. The 9 control points are natural features and were measured during a GPS field campaign in June 1999.
Earlier tests have shown that automated triangulation does not deliver the required results, therefore the triangulation was measured on the S9 Analytical Plotter. Since all points except the control points are natural features with low point definition quality the estimated s0 = 9.5 mm is a realistic value. However, this noise level does not allow us to apply self-calibration.
Figure 2: San Ignacio block layout (only triangulated part) with 9 control points
DTM Generation
Especially in the geoglyph areas we do have very high accuracy requirements for the DTM. Geoglyphs are usually layed out in fairly smooth terrain. Therefore, even small deviations between them and the surrounding DTM will produce "pumpy" effects in visualization, although the geoglyphs are integrated as 3-D vectors into the DTM a priori.
Figure 3: DTM of San Ignacio in shaded relief representation
We have performed tests with automated image matchers, but they did not deliver acceptable results. Therefore, similar as in triangulation, we had to measure the DTM manually on the S9 Analytical Plotter. 94 models were measured in profile mode with 20 m profile spacing, resulting in 296 731 points. In addition 834 breaklines were measured and (up to now) 6 706 vectors belonging to geoglyphs, which were also integrated as breaklines into the DTM. As the other geoglyphs are becoming available they will be integrated as well. Figure 3 represents the complete DTM of San Ignacio with a gross extension of 11 x 6 km2.
Orthoimage Derivation and Mosaicking
Currently this process is not completed. At the time of the writing of this paper we have generated orthoimages from 21 aerials, which cover about 70 % of the required area. The aerial images were scanned on a Zeiss SKAI with 21 mm pixelsize at the Federal Office of Topography, Wabern, Switzerland.
Figure 4: Current status of orthomosaic
The orthoimages and the mosaics were produced with Socet Set. We have orthoimages available at 2 m and 1 m resolution for the whole area, and at 0.5 m and 0.25 m resolution for subareas of special interest. Figure 4 shows the current coverage of the orthomosaic in overview representation.
Geolyph Extraction and Visualization
The geoglyphs are extracted as 3-D vectors from the individual stereomodels on an Analytical Plotter. Since they appear in a great variety of sizes and forms, partly overlaying each other and partly only barely visible as very faint structures in the images it is important that the interpretation and attributation this is done by an expert, in our case by a project archaeologist. Up to now we have generated 6 706 vectors, but this is an ongoing process which will only be finished in a couple of months (see Figure 5).
For visualization and interactive roaming we have the following products available:
Figure 5: Portion of the San Ignacio geoglyphs (yet uncompleted)
Currently we use as visualization platforms ERDAS VirtualGIS on an NT machine and SGI SceneViewer. We will also prepare canned image sequences as flyovers and walktroughs, which can be presented by laptops. For real-time visualization we have access to the VisDome facilities of the ETH Zurich, where a large screen stereo display with polarization glasses is available.
Figure 6: Part of the orthomosaic (left) and textured 3-D model (right) of the same area
Conclusions
We have reported about the status of the second processing phase of our Nasca project. After completion of the Sacramento dataset (Grün, et al., 2000) we have now almost completed the block San Ignacio. The very high data fidelity specifications required the manual measurements of triangulation, DTM and 3-D geoglyphs on Analytical Plotters. The combination of modern analytical and digital (Socet Set by LH Systems for orthoimage and orthomosaic generation) equipment together with up-to-date visualization software resulted in accurate and stunning results. After the implementation of the data on a GIS platform scientists from different disciplines will have a large dataset and tools for analysis at their disposal, which have never been available before. We believe that this will give archaeologists and other experts new and valuable perspectives for their work. It is also interesting to see how much attention such representations find in the general public.
Acknowledgements
We are grateful for the project funding provided be the AVINA foundation and ETH Zurich, and for the support in scanning the aerial images by Landestopographie (Federal Office of Topography), Wabern, Switzerland. We also acknowledge the contributions of the following people to the San Ignacio part of the project: Nicola d'Apuzzo, Simon Bär, Xioayun Fu, Gillian Grün, Karsten Lambers and Markus Reindel.
References
We report about the current state of our Nasca project. The block Sacramento (called Palpa in earlier publications) is fully processed and the block San Ignacio is almost completed. We show the results of San Ignacio, which were obtained through the processes of photo triangulation, DTM measurement, orthoimage and orthomosaic generation and 3-D vector extraction. High quality photorealistic visualizations of the terrain and the geoglyphs are already available. The next task will be the integration of the data on a GIS platform.
Introduction
The geoglyphs of Nasca, Peru, are considered one of the world's major mysteries. Despite the fact that many efforts have been made in the past 70 years to record and analyze the geoglyphs no convincing explanation has emerged yet which would explain the purpose of these drawings in the sand beyond any doubt. Many hypotheses have been put forward and each author will find a certain number of lines and figures in support of his/her opinion. The huge number of lines, trapezoids, star centers, biomorphic figures, etc. has up to now defied any comprehensive recording and analysis.
We are currently conducting an ambitious project which aims at the complete 3-D recording of the geoglyphs of the areas of Nasca, Sacramento and San Ignacio (a total of about 300 km2) with photogrammetric means. This is done in cooperation with a group of archaeologists who will try to relate their excavation results to the geoglyphs in order to get a more complete picture of the overall cultural and social background of the Nasca people (200 B.C. - 600 A.D.).
We are using the latest digital photogrammetric technology for the recording of the geoglyphs and some of the most advanced visualization software. Finally, all data will be integrated on a GIS platform, which will require the efficient management of both vector- and extensive raster image data.
We have aquired over 1000 aerial photographs at scales 1:5 000 and 1:10 000 both in colour and B/W. In addition we have collected existing, older photographs which can give us an indication of the damage that has been done to the geoglyphs in recent years.
The block Sacramento has been fully processed and visualization results have been shown to the public on a SGI Onyx2 in an interactive mode at the Nasca exhibition in the Rietberg Museum, Zurich and in the VisDome of ETH Zurich as stereo projection on a large 4 x 12 m2 screen and on other occasions. For details on the complete project (motivation, goals, data acquisition, processing and visualization of the Sacramento block) we refer to Grün, et al., 2000.
In this paper we report about the ongoing work on the San Ignacio block. Here we have only B/W images available for processing. The phototriangulation has been performed on an Analytical Plotter, since automated triangulation does not deliver the desired results. The DTM has been measured completely in manual mode, because of the high quality requirements. Digital orthoimages have been produced over the largest part of the area and a mosaic has been generated. This is the basis for photorealistic visualization. A number of geoglyphs have been measured in the stereomodels in vector mode (so far we have included about 6 700 vectors in the dataset), but this work is not finished yet.
Photoriangulation
Figure 1 shows the location of the three blocks Sacramento (in earlier publications called "Palpa"), San Ignacio and Nasca, which are also processed in this sequence. A detailed layout of the triangulated part of the block San Ignacio is given in Figure 2.
Figure 1: Location of the blocks Sacramento, San Ignacio, Nasca in southern Peru
The block San Ignacio was flown in May 1998 with a Zeiss RMK A15 camera at a nominal image scale of 1:5 000 in B/W with 60 % forward overlap and 60 % sidelap. Out of the total number of 189 aerial images 169 images, arranged in 11 strips of varying length, were selected for triangulation. The 9 control points are natural features and were measured during a GPS field campaign in June 1999.
Earlier tests have shown that automated triangulation does not deliver the required results, therefore the triangulation was measured on the S9 Analytical Plotter. Since all points except the control points are natural features with low point definition quality the estimated s0 = 9.5 mm is a realistic value. However, this noise level does not allow us to apply self-calibration.
Figure 2: San Ignacio block layout (only triangulated part) with 9 control points
DTM Generation
Especially in the geoglyph areas we do have very high accuracy requirements for the DTM. Geoglyphs are usually layed out in fairly smooth terrain. Therefore, even small deviations between them and the surrounding DTM will produce "pumpy" effects in visualization, although the geoglyphs are integrated as 3-D vectors into the DTM a priori.
Figure 3: DTM of San Ignacio in shaded relief representation
We have performed tests with automated image matchers, but they did not deliver acceptable results. Therefore, similar as in triangulation, we had to measure the DTM manually on the S9 Analytical Plotter. 94 models were measured in profile mode with 20 m profile spacing, resulting in 296 731 points. In addition 834 breaklines were measured and (up to now) 6 706 vectors belonging to geoglyphs, which were also integrated as breaklines into the DTM. As the other geoglyphs are becoming available they will be integrated as well. Figure 3 represents the complete DTM of San Ignacio with a gross extension of 11 x 6 km2.
Orthoimage Derivation and Mosaicking
Currently this process is not completed. At the time of the writing of this paper we have generated orthoimages from 21 aerials, which cover about 70 % of the required area. The aerial images were scanned on a Zeiss SKAI with 21 mm pixelsize at the Federal Office of Topography, Wabern, Switzerland.
Figure 4: Current status of orthomosaic
The orthoimages and the mosaics were produced with Socet Set. We have orthoimages available at 2 m and 1 m resolution for the whole area, and at 0.5 m and 0.25 m resolution for subareas of special interest. Figure 4 shows the current coverage of the orthomosaic in overview representation.
Geolyph Extraction and Visualization
The geoglyphs are extracted as 3-D vectors from the individual stereomodels on an Analytical Plotter. Since they appear in a great variety of sizes and forms, partly overlaying each other and partly only barely visible as very faint structures in the images it is important that the interpretation and attributation this is done by an expert, in our case by a project archaeologist. Up to now we have generated 6 706 vectors, but this is an ongoing process which will only be finished in a couple of months (see Figure 5).
For visualization and interactive roaming we have the following products available:
- Contour representation of the DTM (Fig. 2)
- Wireframe and shaded relief representation of the DTM (Fig. 3)
- Orthomosaic (Fig. 6)
- Texture map at various resolutions (Fig. 6)
- Combination of 3-D vectors and texture map
Figure 5: Portion of the San Ignacio geoglyphs (yet uncompleted)
Currently we use as visualization platforms ERDAS VirtualGIS on an NT machine and SGI SceneViewer. We will also prepare canned image sequences as flyovers and walktroughs, which can be presented by laptops. For real-time visualization we have access to the VisDome facilities of the ETH Zurich, where a large screen stereo display with polarization glasses is available.
Figure 6: Part of the orthomosaic (left) and textured 3-D model (right) of the same area
Conclusions
We have reported about the status of the second processing phase of our Nasca project. After completion of the Sacramento dataset (Grün, et al., 2000) we have now almost completed the block San Ignacio. The very high data fidelity specifications required the manual measurements of triangulation, DTM and 3-D geoglyphs on Analytical Plotters. The combination of modern analytical and digital (Socet Set by LH Systems for orthoimage and orthomosaic generation) equipment together with up-to-date visualization software resulted in accurate and stunning results. After the implementation of the data on a GIS platform scientists from different disciplines will have a large dataset and tools for analysis at their disposal, which have never been available before. We believe that this will give archaeologists and other experts new and valuable perspectives for their work. It is also interesting to see how much attention such representations find in the general public.
Acknowledgements
We are grateful for the project funding provided be the AVINA foundation and ETH Zurich, and for the support in scanning the aerial images by Landestopographie (Federal Office of Topography), Wabern, Switzerland. We also acknowledge the contributions of the following people to the San Ignacio part of the project: Nicola d'Apuzzo, Simon Bär, Xioayun Fu, Gillian Grün, Karsten Lambers and Markus Reindel.
References
- Grün, A., Bär, S., Beutner, S., 2000. Signals in the Sand - 3-D Recording and Visualization of the Nasca Geoglyphs. PFG (Photogrammetrie, Fernerkundung, Geoinformation), No. 6/2000. pp. 385-398.