Phenomenological Evaluation of a Nonlinear Ratio-Preserving Stabilizer on Absolute Gravimeter Time-Series Data

This work presents a phenomenological evaluation of a nonlinear ratio-preserving dynamic stabilizer applied to real absolute gravimeter time-series data. Absolute gravimeter measurements are known to exhibit strong non-Gaussian fluctuations, impulsive disturbances, and structured tidal components, posing challenges for standard linear smoothing techniques. Using absolute gravimeter residuals from the HAR station (2010–2011), the nonlinear stabilizer is compared against commonly used linear filters, including moving average and Savitzky–Golay methods. All filters are applied using matched effective smoothing scales and fixed parameters to ensure reproducibility and to avoid adaptive overfitting. Stabilization performance is evaluated using three complementary diagnostics: (i) symbolic entropy to quantify distributional broadening, (ii) kurtosis-based dissonance to measure impulsive, heavy-tailed behavior, and (iii) spectral coherence to assess preservation of dominant tidal structure. Results show that the nonlinear stabilizer suppresses impulsive noise by approximately an order of magnitude relative to the raw signal while maintaining moderate tidal coherence and exhibiting systematically lower entropy inflation than classical linear smoothers. This study does not propose a new physical model of gravitation. Rather, it provides an experimentally grounded characterization of a nonlinear stabilization regime for gravimetric time-series data. The results demonstrate that ratio-preserving nonlinear stabilization produces measurable and reproducible effects on real experimental measurements, motivating further investigation of nonlinear filtering approaches in gravimetry and geophysical signal analysis.