Combined techniques study for the identification of main drives in coastal 15 cliff processes between portonovo and trave (Ancona, Italy)
- Autores: Nicola Fullin
- Directores de la Tesis: Paolo Ciavola (dir. tes.), Monica Ghirotti (codir. tes.)
- Lectura: En la Universidad de Cádiz ( España ) en 2024
- Idioma: español
- Tribunal Calificador de la Tesis: Enzo Rizzo (presid.), Laura del Río Rodríguez (secret.), Lisa Borgatti (voc.)
- Programa de doctorado: Programa de Doctorado en Ciencias y Tecnologías Marinas por la Universidad de Cádiz
- Materias:
- Texto completo no disponible (Saber más ...)
- Resumen
The stability of slopes is governed by several factors including lithology, geological structures, such as geological arrangement and/or the presence of tectonic disturbances, hydrogeological conditions, and the evolution of relief over time, conditioned in turn by factors external to the slope. Identifying the processes and mechanisms that can lead to instability conditions is a complex operation, also due to the spatial and temporal variability of predisposing and triggering conditions. High coasts, in particular, represent a special case in slope stability due to the interaction between subaerial, marine, and endogenous factors. In particular, the erosive processes affecting high coasts constitute an additional hazard factor for human safety in the case of their use for bathing purposes and for any infrastructure present upstream or at the base of the coast. The development and advancement of remote sensing techniques in slope stability have greatly increased the possibility of having a large amount of geological, hydrogeological, and topographic data over time, even in inaccessible areas, thanks to the repeatability of acquisition campaigns. Several authors have highlighted a gap in research regarding cliff erosion and evolution. During the 20th century, this research was conducted by a few researchers, with limited interest from practitioners in related fields, such as coastal engineering, in integrating data for the reconstruction of a multidisciplinary model. This limitation was exacerbated by the fact that investigations were conducted by small groups of scattered researchers without cooperation in different research areas. Despite this lack, a complete model integrating the multiple factors influencing cliff erosion, such as wave dynamics, beach characteristics, cliff materials, and failure mechanisms, has not yet been developed. The aim of this work is to partially fill this gap. Therefore, the work focused on determining the importance of various factors contributing to cliff erosion, with the aim of better understanding the links between these factors and the erosive process.
The thesis presents three studies conducted in the area between Portonovo and Trave, first examining the geomorphological evolution of the slopes over the time span 1978-2021 and then the annual evolution 2021-2022. Data obtained on decadal erosion were then reused together with geological, topographic, and marine data to determine the importance of each factor with respect to erosion, using machine learning (ML) algorithms.
In the first study, the retreat of the cliff located between Portonovo Beach and Trave, within the Regional Park of Conero (Ancona, Italy), was analyzed and quantified. Morphodynamic changes were evaluated over a total time span of over 40 years and in shorter time intervals (10 years) using aerial orthophoto analysis acquired between 1978 and 2021, data from UAV (Unmanned Aerial Vehicle) digital photogrammetry, and Light Detection and Ranging (LiDAR). The combination of these data, integrated with field surveys, allowed the identification of the main factors causing erosion through ML algorithms. Stability analyses at the 2D limit equilibrium (LEM) were conducted along representative sections of areas showing the highest cliff retreat values, further validating the ML results. Over this forty-year period in the northernmost sector (Trave), erosion of over 40 meters was identified. Furthermore, a Digital Elevation Model (DEM) of Differences (DoD) was calculated to validate the DSAS results, revealing good agreement between the retreat areas identified by DSAS and the coast sections characterized by high DoD values.
In the second study on the Portonovo-Trave Cliffs, two UAV surveys conducted in September 2021 and October 2022 successfully provided dense point clouds, allowing the application of the Multiscale-Model-to-Model Cloud-Comparison (M3C2) point cloud comparison algorithm for change analysis. Volume estimation, performed using the 2.5D Volume tool implemented in the CloudCompare processing software, revealed erosion exceeding 500 m3 in a segment of the Trave sector, interpreted and attributed to both base erosion (beach groove) and landslide processes along the slope. The Mezzavalle sector showed activity, including the formation of the beach groove, not recorded by previous analyses due to different data analysis methods, and differential accretion along a section of Mezzavalle Beach was observed.
In the third study, the UAV survey of 2022 was accompanied by geomechanical data collected during fieldwork. Determining the retreat of the cliff top in a GIS environment involved comparing orthophotos taken in 1978 and a new UAV survey in 2022 using the DSAS tool. Additionally, two machine learning (ML) algorithms, Random Forest (RF) and eXtreme Gradient Boosting (XGB), were employed to analyze the data. The Mean Decrease in Impurity (MDI) methodology assessed the significance of each factor, with both algorithms consistently highlighting slope height as the main influencer of cliff top erosion rates. To validate the machine learning algorithm results, a two-dimensional Limit Equilibrium Method (LEM) was applied. The ten sections showing the most significant retreat of the cliff top, as identified by DSAS, were analyzed through 2D LEM analysis. Factor of Safety (FS) values were compared with the height of each cliff profile, and the results of the 2D LEM analyses were consistent with the results obtained from the machine learning algorithms.
In conclusion, the integration of various remote sensing techniques, as well as machine learning (ML) and limit equilibrium method (LEM) analyses, has proven to be an effective methodology for studying slope and erosion processes in a coastal area. This study underscores the importance of using high-resolution datasets to detect diverse processes and emphasizes the adaptability of UAV devices in complex environments. The results also highlight how continuous monitoring and the use of advanced technologies are crucial for effectively assessing and managing coastal erosion. The findings contribute to a broader understanding of coastal processes, emphasizing the need for a holistic approach that integrates fieldwork, remote sensing, machine learning, and numerical modeling for comprehensive and accurate assessments. A new methodology, accompanied by open-source code, is proposed to study site-specific erosion and obtain the factors that are important in determining the process.
