PosterPDF Available

ANALYSIS OF THE DIGITAL TERRAIN MODEL ON THE ARCHAEOLOGICAL SITE LOCATED IN THE CHODEL BASIN

Authors:
XXI Conference of PhD Students and Young Scientists
ANALYSIS OF THE DIGITAL
TERRAIN MODEL ON THE
ARCHAEOLOGICAL SITE LOCATED
IN THE CHODEL BASIN
M.Sc. eng. Bartosz Naumowicz
University of Warmia and Mazury in Olsztyn, Faculty of Geoengineering, Institute of Geodesy and Civil Engineering,
Department of Geoinformation and Cartography
INTRODUCTION
One of the methods of massive geodetic data collection is laser scanning, which is becoming more and more popular. Three
types of scanning can be distinguished: terrestrial, aerial and satellite. Aerial laser scanning (ALS) is often used in
archaeological work to determine the location and identification of potential archaeological sites. It gives a detailed picture of
the terrain, allowing to detect and identify its unnatural changes. On the other hand, terrestrial laser scanning (TLS) is
particularly useful for an inventory of excavation sites, mainly due to the possibility of taking measurements during
archaeological work and creating models of subsequent layers. This work aims to compare digital terrain models of selected
archaeological sites made based on ALS and TLS. The study consisted of an early settlement area in Chodlik, which is located
in the Lublin province, district Opole, village Karczmiska. A single archaeological site was selected for the analysis (Fig. 1). In
the selected place, there are 3 earth embankments - the remains of fortifications built in this place in the Early Middle Ages.
MATERIALS & METHODS
For this work, about 15 GB of data was obtained from measurements with a TLS with Leica ScanStation C10 scanner.
The fieldwork took two days. Almost 80 acres of land were scanned. The office works and data processing lasted about 2
months. 9 measurement stations were established in the area of the early medieval stronghold of Chodlik (Fig. 2). The least
wooded and bushy area was selected to obtain more accurate scans. It was also chosen because of the distinctly distinguished
3 earth embankments - remnants of an early medieval stronghold. Unfortunately, the scanning area had taller grass, low
shrubbery and a few trees, which contributed to cloud contamination. The area coincided with the LiDAR measurements. A
high resolution (scanner preset) was selected for scans, giving X and Y spacing values of 5 mm. Pictures were also taken so
that the points of the cloud had RGB colours. Most scans were taken by the full scan method to fill in "gaps" in previous
scans. Additionally, each point where 4 tie points (HDS targets) were set was stabilized with a stake and measured with a
GNSS receiver Leica CS15. The site of the stronghold in Chodlik is located in zone 7 of the PL-2000 system. The final
number of points in the cloud obtained from the TLS method is: 17,575,516. The ".ptx" file takes up: 569 MB. The final
appearance of the cleaned and prepared belt is shown in Figure 3. Later it was compared with mesh created from ALS cloud.
Software used for this study is Leica Cyclone 9 and CloudCompare.
Figure 1. Localization of analysed terrain (source: own study)
CONCLUSION
As part of the work, LIDAR measurements were obtained and ground measurements were made. Based on the above data, numerical models of archaeological sites were built and their
mutual comparison was made. The results obtained show that the differences between models developed with two techniques are at the level of several centimeters. The acquisition time of
TLS data was aproximetly 2 days. During the acquisition of ALS data, the measurement time was even shorter. An area of approximately 1 km2 was scanned in one day. This shows how fast
the data acquisition from the ALS is.As shown in this work, the measurements obtained by ALS and TLS can be easily compared with each other and they differ from each other by several
centimeters. However, this is not a glaring difference, taking into account the accuracy obtained from the ALS during these tests (about 50 cm horizontally and 15 cm vertically, according to
flight performers). Finally, the compared model assumed to be the least contaminated and with no outliers has ranges from 0 cm to +/- 50 cm between the two scans. Creating a mesh from
ALS data allowed for almost two times better matching with the point cloud from the TLS. The average value of the vertical distance between the mesh made from ALS and the cloud acquired
from TLS is -0.001 m with a standard deviation of 0.173. In conclusion, laser scanning in archeology is very useful for the identification, inventory and documentation of potential or already
discovered archaeological sites. Depending on the conditions, area or research object, its different methods should be used: terrestrial, aerial or satellite scanning.
RESULTS
After a few attempts, it was decided to create a so-called mesh from the ALS data (a plane that is calculated based on
a triangle mesh with points closest to each other). A total of 20 trials were performed using various parameters and
methods. The following are selected as the best fit using 1,000,000 points by Cloud to Cloud (coregister RMS: 0.301)
and Cloud to Mesh using only 50,000 points (coregister RMS: 0.182). Then, using the C2M Distance tool, it was
calculated how far apart the points obtained from the measurements are. The results are presented for the different
distance ranges in Table 1. Observations that were larger than 80 cm, both positive and negative, were removed from
the cloud so as not to distort the actual picture of the distribution of the mesh-to-cloud distance values. The use of
different ranges, developed by trial and error, 3 cloud spans were determined that showed the greatest changes and
differences between the mesh and the cloud: +/- 80cm, +/- 50cm and +/- 25cm.We can see the visualization in Figure
4a-d. Note that most of the high difference values are due to residual scanned grass, bushes, and noise, most of which
was removed in earlier stages. Trying to limit the outliers, we remove more and more points, which also limits the
possibility of a real measurement value of the scanned area that interests us. The biggest difference is between the range
of +/- 50 cm and +/- 25 cm (Fig. 4cd). Therefore, the range of +/- 50cm was found to be a good compromise between
the accuracy of fit and the amount of surface covered (Fig. 4c).
Figure 2. Raw, unprocessed point cloud with marked
measurement stations (yellow triangles) (source: own study)
Figure 3. Cleaned cloud of points from terrestrial laser scanning prepared for comparisons with aerial laser scanning (source: own study)
Range
[m]
Number of
points
Min. Value
Max. Value
Mean value
Standard
deviation
Full
17 575 516
-0.866 1.030 0.002 0.182
+/- 0,80
17 574 575
-0.800 0.800 0.002 0.182
+/- 0,50
17 377 418
-0.500 0.500 -0.001 0.173
+/- 0,25
14 687 952
-0.250 0.250 -0.016 0.119
Table 1. The distances between the TLS point cloud and the mesh created
from the ALS cloud (source: own study)
a) b) c) d)
Figure 4. Visualization of the distance between the cloud from TLS and the mesh from ALS. a) Full range; b) Range +/- 80cm; c) Range +/- 50cm; d) Range +/- 25cm. (source: own study)
ResearchGate has not been able to resolve any citations for this publication.
ResearchGate has not been able to resolve any references for this publication.