Issue 78

C. Vendittozzi et alii, Fracture and Structural Integrity, 78 (2026) 519-540; DOI: 10.3221/IGF-ESIS.78.30

excluding low-quality frames using the integrity and SNR-based masking criteria. The ridge track provides a piecewise descriptor of the evolving dominant spectral content for each sensor and enables compact comparison across sensors and datasets. Figure 12 overlays ridge trajectories for sensors S1–S4 for Dataset A and Dataset B, with the FBG-derived event centered windows marked. In both recordings, changes in the dominant-frequency trajectories cluster around the event centered windows, while smoother evolution is observed during lower-activity intervals. The presence of comparable dominant-frequency plateaus and transitions in the two representative datasets supports the recurrence of sensor-dependent spectral families within the analyzed records. This observation should not be interpreted as statistical repeatability over a population of landings, but as evidence that similar frequency regions can be recovered from independent field recordings acquired with the same sensor installation and processing workflow. Occasional discontinuities in the ridge trajectories are expected when multiple spectral bands compete in amplitude or when the dominant component temporarily falls below the quality threshold. Uncertainty in ridge assignment is also expected to increase around high-energy event transitions, where pre-event and post-event response components may overlap within the same analysis frame. For this reason, the shaded regions in Fig. 12 should be interpreted as event-centered analysis windows rather than as instantaneous modal-identification intervals. Accordingly, the ridge is interpreted here as a compact dominant-frequency descriptor of the evolving response, not as a direct structural-mode track, independently validated touchdown marker, or indicator of damage state. For graphical clarity, consecutive ridge points separated by excessively steep frame-to-frame frequency jumps were not connected in Fig. 12. This operation was used only to avoid visually misleading near-vertical line segments and to emphasize persistent or slowly varying frequency families. It does not modify the extracted ridge frequencies or the quantitative descriptors reported below.

Figure 12: Ridge-tracked dominant-frequency trajectories for sensors S1–S4 extracted from multitaper spectrograms and displayed on a logarithmic frequency axis. (a) Dataset A; (b) Dataset B. At each time frame, the ridge corresponds to the maximum-power frequency bin within the 9–500 Hz analysis band after masking low-quality frames. Light shaded regions denote FBG-derived event-centered analysis windows. The ridge trajectories summarize dominant spectral content and should not be interpreted as independently validated modal tracks or touchdown markers. Quantitative comparison of ridge families and band-limited spectral content To complement the qualitative interpretation of the multitaper spectrograms and ridge trajectories, event-wise quantitative descriptors were computed for each sensor over the three FBG-derived event-centered windows. The analysis used the same 9–500 Hz interval adopted for the time–frequency interpretation and combined ridge-based statistics with band limited spectral metrics. In particular, the ridge median frequency and interquartile range (IQR) were used to quantify the central tendency and temporal dispersion of the maximum-power ridge trajectory within each event window. In parallel, band-limited RMS values were computed over predefined frequency intervals to identify the frequency band carrying the largest integrated spectral contribution within the same event window. The RMS-dominant band and the ridge median frequency therefore describe different aspects of the response. The ridge median is derived from the frame-by-frame maximum-power ridge trajectory, whereas the RMS-dominant band is the interval associated with the largest band-limited RMS contribution. The ridge-persistence value then indicates the percentage of accepted ridge points that fall inside that RMS-dominant band. The predefined RMS bands do not cover the entire 9–500 Hz analysis interval. Consequently, a ridge median within the 9–20 Hz region may coexist with an RMS-dominant band of 20–60 Hz and a low ridge-persistence value within that band. These quantities may differ when broadband energy or competing spectral families are present within the same event-centered window. This distinction is relevant, for example, in Dataset B, sensor S3, Event 1, where the ridge median is near 150 Hz, while the largest band-limited contribution is associated with the 350–500 Hz interval. The main

533