A Data-Driven Framework for Quantitative NDT Data Fusion in Granite Masonry via Local Window Cross-Analysis: A Case Study Application

The conservation of the coastal built environment faces critical challenges, ranging from capillary moisture rise, particularly in land reclaimed from the sea, and marine aerosol degradation, to cyclic salt crystallization–dissolution processes driven by relative humidity fluctuations. Although Non-Destructive Techniques (NDTs) provide non-intrusive data, existing workflows suffer from a methodological void regarding the quantitative integration of datasets based on distinct physical principles, relying heavily on qualitative visual correlations. To address this gap, this study proposes a data-driven framework that mathematically correlates geometric and thermal alterations, moisture, and salt distributions on granite masonry measured from Terrestrial Laser Scanning (TLS), Infrared Thermography (IRT), pin-mode Electrical Conductivity Technique (pin-mode ECT), and Poultice-based Salt Mapping (P-BSM). The methodology processes multi-platform datasets through sequential rasterisation, resizing, scaling, and smoothing. Normalised horizontal and vertical gradients are subsequently extracted and subjected to Local Window Cross-Analysis (LWCA), enabling a transition from simple anomaly localisation to a rigorous interpretation based on statistical Pearson correlation coefficients. Results demonstrate that directional gradients successfully isolate distinct pathology groups, while Pearson coefficient polarity explicitly unveils the underlying physical mechanisms. Severely degraded regions trigger multiple correlation modalities, enabling cross-detection and robust noise suppression . While the specific empirical correlation results reflect the case-study nature of the investigation across two representative reference stations in a single granite masonry envelope, the overarching scientific contribution resides in the automated, standardised, and transferable multi-NDT fusion framework, eliminating subjective regional assumptions and supporting informed decision-making across diverse structural health monitoring contexts .