Accurate runoff estimation on outdoor heritage sites from point clouds: A way to prevent water erosion

Jesús Balado, Lucía Díaz-Vilariño and Ernesto Frías, University of Vigo
Keywords: LiDAR; terrestrial laser scanning; archaeological site; erosion; D8 algorithm
The preservation of heritage sites is essential for researchers to study and understand the past. Preservation can respond to historical, social and economic objectives (Sardón de Taboada 2015). Different objectives often conflict, promoting knowledge and visits lead to deterioration of the elements to be preserved (Comer 2012). Outdoor heritage sites are particularly sensitive to atmospheric and climatological phenomena, such as rain (Wischmeier and Smith 1978; Malavieille 2010), wind (Erkal, D’Ayala, and Sequeira 2012), extreme thermal contrasts (Cassar and Pender 2005) and fires (Robichaud et al. 2007). These produce a slow and constant erosion of the heritage elements that can be increased by adverse meteorological phenomena.
Castros are fortified villages normally located on hills. This location gave them defensive advantages and allowed them to have good visibility and control of the nearby territory. Their location in these high areas makes the buildings especially unprotected against rain and wind.
LiDAR technology makes it possible to acquire the built environment quickly, accurately and without the need for contact. The result of a LiDAR acquisition is a point cloud, sets of 3D coordinates with information on material reflectivity, number of returns, scanning angle, etc. Point clouds allow to obtain direct measurements of the environment and generate digital models (Balado et al. 2017; Rubinowicz and Czyńska 2015; Y.F. Tang and S.C. Pun-Cheng 2004). However, there are some limitations that must be taken into account when generating methodologies to study the environment: information contained in point cloud is disordered and redundant, density of the points is not uniform and point clouds contain areas without data.
The objective of this work is to evaluate the behaviour of water in a Castro and determine those areas where there is a greater risk of erosion, landslides and the creation of water ponds. The methodology input is a point cloud of the Castro. The point cloud is rasterized for conversion into an image with information on the maximum height of the points belonging to each pixel. The D8 runoff calculation algorithm is applied to the image to obtain the direction in which water is evacuated from the Castro. In this direction the difference between height of the target pixel and the source pixel is calculated, which is related to the force of the water in the fall.Case study
The Castro of Santa Trega has been chosen as a case study. It is located in A Guarda, in the autonomous region of Galicia (Spain), coordinates WGS84 (-8.869156,41.891454). The Castro is situated on a hill 341 metres high at the estuary of River Miño with Atlantic Ocean. This site is prone to heavy rains and winds during the winter months. The Castro of Santa Trega was declared a National Historic-Artistic Monument in 1931. The Castro had a continuous occupation between the 1st century B.C. and the 1st century A.D. Partial archaeological excavations have been carried out during the 20th century.
The study area corresponds to an excavated area of approximately 2300 square metres located in the lower part of the Castro near the outer wall and the north gate. The area contains 35 buildings of which the lower wall has been preserved (30cm high), except for one which has been entirely recreated. The buildings are distributed without following a square mesh along a slight slope to the west. The northern area is delimited by a reconstructed wall of 2 meters at its highest point and rocky ground. All the buildings are made of small irregular stones.
Twelve scans have been regularly distributed to acquire the case study area with a terrestrial laser scanner Faro Focus X330 (Roca et al. 2014). From each scan a point cloud of 18 million points was obtained. Acquisitions have been recorded with respect to the first scan position and coloured with photographs of the environment. Due to the magnitude of the resulting cloud, a downsampling has been applied to obtain a maximum density of 10 thousand points per square meter. The resulting cloud contains 20 million points.Methodology
The methodology is based on two phases: firstly, the rasterization, and secondly, the application of the D8 algorithm for the runoff calculation. The input of the methodology is a cloud of points P(X,Y,Z). The colour information shown in Figure 2 is not used in processing. The point cloud is rasterized on the Z-axis to structure the information into an image I(Ix,Iy) with a pixel size l (Soilán et al. 2018). In each pixel the maximum value of Z of the points that correspond to that pixel is saved.
The D8 algorithm (González-Jorge et al. 2015) is then applied to the generated raster I image. The D8 algorithm consists of generating a 3×3 pixel window that runs through the image. The minimum pixel value is selected in the window position. This minimum value indicates where the water drains from the central pixel.

The methodology has been applied to the case study explained, based on a rasterization with a pixel size l = 0.1m. The methodology has been implemented in Matlab the case study has been processed on an Intel Core i7-7700HQ CPU 2.80 GHz with 16 GB RAM in 12 seconds. Figure 3 shows the runoff directions to where the water drains by implementing the D8 algorithm. As can be seen, the directions are related to the slope of the hill and are influenced by the existence of buildings and the paths drawn between them. In the northern part of the case study, water flows southward (including south-east and south-west). In the southern zone, water runs north and northwest. Streets (north to south and east to west) cut the natural runoff direction of the hill. From the directions can also be sensitive areas to accumulate water and areas with greater unevenness in the runoff .

In this work, the D8 algorithm has been applied to a case study of a heritage site with characteristics that make it especially vulnerable to water erosion. The application has made it possible to identify the areas towards which the water flows within the archaeological site. In addition, it is also possible to detect sloping areas and water accumulation correctly.


Balado, J., L. Díaz-Vilariño, P. Arias, and M. Soilán. 2017. “Automatic Building Accessibility Diagnosis from Point Clouds.” Automation in Construction 82 (May). Elsevier: 103–11.

Cassar, M, and R Pender. 2005. “The Impact of Climate Change on Cultural Heritage: Evidence and Response.” In ICOM Committee for Conservation: 14th Triennial Meeting.

Comer, Douglas C. 2012. Tourism and Archaeological Heritage Management at Petra. Springer.

Erkal, Aykut, Dina D’Ayala, and Lourenço Sequeira. 2012. “Assessment of Wind-Driven Rain Impact, Related Surface Erosion and Surface Strength Reduction of Historic Building Materials.” Building and Environment 57: 336–48.

González-Jorge, H, L Díaz-Vilariño, J Martínez-Sánchez, P Arias, and S Lagüela. 2015. “Influence of The Precision of Lidar Data in Surface Water Runoff Estimation for Road Maintenance.” International Archives of the Photogrammetry, Remote Sensing & Spatial Information Sciences, 40 (3/W3), 3–8.

Malavieille, Jacques. 2010. “Impact of Erosion, Sedimentation, and Structural Heritage on the Structure and Kinematics of Orogenic Wedges: Analog Models and Case Studies.” GSA TODAY 20 (1): 4–10.

Robichaud, P R, W J Elliot, F B Pierson, D E Hall, and C A Moffet. 2007. “Predicting Postfire Erosion and Mitigation Effectiveness with a Web-Based Probabilistic Erosion Model.” CATENA 71 (2): 229–41.

Roca, David, Julia Armesto, Susana Lagüela, and Lucia Díaz Vilariño. 2014. “LIDAR-Equipped UAV for Building Information Modelling.” In ISPRS – International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XL-5.

Rubinowicz, P, and K Czyńska. 2015. “Study of City Landscape Heritage Using LiDAR Data and 3D-City Models.” International Archives of the Photogrammetry, Remote Sensing & Spatial Information Sciences.

Sardón de Taboada, María Isabel. 2015. “¿Por Qué Conservar El Patrimonio?” AXA, Una Revista de Arte y Arquitectura 7.

Soilán, Mario, Linh Truong-Hong, Belén Riveiro, and Debra Laefer. 2018. “Automatic Extraction of Road Features in Urban Environments Using Dense ALS Data.” International Journal of Applied Earth Observation and Geoinformation 64: 226–36.

Wischmeier, W H, and D D Smith. 1978. Predicting Rainfall Erosion Losses – a Guide to Conservation Planning.Hyattsville, Maryland: USDA, Science and Education Administration.

Y.F. Tang, MAndy, and Lilian S.C. Pun-Cheng. 2004. “Algorithmic Development of an Optimal Path Computation Model Based on Topographic Map Features.” Isprs 2004.