Abstract:
Visual inspection is a crucial task for the assessment of a dam conservation state. However, these tasks are time consuming and operator dependent, which leads to the aim of developing more objective procedures. Photographs are a way to register the conservation state of a structure at a certain epoch, providing a record that can be later reanalysed and compared to new data. Although they are already used for support of inspection activities, their capture is often limited to accessible areas of the structure, leading to large lacunae in the analysed surface. Photographs obtained from long distances can reach a larger surface area, but the achievable resolution and detail of the identifiable elements is compromised. Furthermore, the manual analysis of the photographs for pathology identification and characterization is also a tedious task.
In this study, a map of pathologies on all visible surfaces of Santa Luzia dam, in Portugal, was performed through photogrammetry and digital image processing. The dam crest, downstream wall and the emerged part of the upstream wall were thoroughly photographed through drone and terrestrial surveys, in order to achieve a digital representation of the dam. The photographs were subjected to photogrammetric procedures, in order to create orthophotographs (distortion-free images of the surfaces). Those images were used to map a set of occurrences observed on the dam surface, such as cracks, calcium carbonate deposits, biologic colonization, leakage and irregularities. The location and extension of these elements were detected in the images using digital image processing algorithms. Both the orthophotographs and the pathology outlines were imported into a spatial database, which can be explored in open-source software. As the orthophotographs and pathology outlines are geometrically corrected, it is possible to measure the dimensions of the affected regions, such as areas, lengths and widths. Furthermore, alphanumeric information can be added to the spatial database, complementing the image-based data with field observations. The initial condition assessment can be updated through new surveys, enabling the analysis of the evolution of the pathologies.
