INVESTIGATING THE INFLUENCE OF FRACTURES AND NEAR-SURFACE TEMPERATURE VARIATIONS ON THE STABILITY OF A SEA ARCH THROUGH AMBIENT VIBRATION MONITORING AND NUMERICAL MODAL ANALYSIS

Structural health monitoring of cultural- or natural-heritage landforms is paramount for their conservation and management. In this context, understanding the role of preparatory and geological predisposing factors in impacting their stability is crucial for defining reliable conservation strategies. In this work, we present results from ambient vibration measurements and 3D finite-element numerical modal analysis to characterize the dynamic behavior of the Wied Il-Mielah coastal arch (Maltese Archipelago). This remarkable landmark represents a natural heritage site of Gozo and is one of the last remaining coastal arches on the island after collapse of the Azure Window in 2017. Persistent joints and continuous environmental stresses make this site an ideal case for investigating the interaction between rock mass fractures, near-surface temperature fluctuations, and their influence on the long-term stability of the arch. Modal properties of the arch were investigated using a combination of geophysical monitoring and numerical modeling techniques. During daily surveys with seasonal recurrence, ambient vibration measurements were performed using a geophone array deployed along the arch length. Spectral and frequency-dependent polarization processing and cross-correlation modal analysis were conducted on the array-based dataset to identify resonance frequencies and polarization attributes, characterize the mode shapes, and study potential temperature-induced frequency shifts through continuous frequency tracking and cross-correlation. 3D finite-element numerical modal analysis was also conducted to predict vibrational modes and compare against field data. Two different models were implemented to test the hypothesis of a fracture-control on the dynamic behavior of the arch. The first model assumed an isotropic and homogeneous material, while the second included a main joint as a zone of reduced modulus. Modeled resonance frequencies and relative modal displacements were extracted for each of the two implemented models to compare numerical results with field data. Spectral analysis of ambient vibrations revealed two resonance frequencies at each array station. The polarization attributes for these two modes allowed us to interpret them as first-order bending modes. However, modal vectors show a complex distribution along the arch length that might depend on the presence of fractures dividing the structure into adjacent compartments. Cross-correlation modal analysis confirmed this evidence describing an identical distribution of relative modal displacements. Frequency tracking highlighted a direct and almost zero-lag correlation between daily temperature changes and frequency variations, suggesting that stress stiffening is the primary mechanism controlling the observed drifts. In contrast, an inverse correlation with a 1-month lag was observed at the seasonal scale. This result was interpreted as caused by rock mass thermal dilation and contraction causing fractures to open and close, thus determining a delayed and periodic variation in rock mass stiffness. A successful match between field data and model results was achieved only for the model containing the simplified fracture. Modeled resonance frequencies match measured values to within 10%, and the associated modal vectors replicated results of cross-correlation modal analysis. The fully isotropic model did not match field data, highlighting how homogeneous models may fail to simulate the dynamic behavior of jointed rock structures. By combining array-based ambient vibration measurements and 3D finite-element numerical modal analysis, this study provides an improved understanding of the role of structurally-relevant joints and near-surface temperature variations on the stability of natural rock landforms.