A diagnostic framework for tidal signal recovery under diurnal environmental aliasing: application to a fiber-optic and seismic deployment on Astrolabe Glacier, East Antarctica

Icequake seismicity on coastal glaciers is thought to be controlled by ocean tidal forcing, but short deployments make this difficult to verify: in diurnal tidal environments, wind, temperature, and atmospheric pressure all vary on approximately 24-hour cycles near-indistinguishable in period from the K1 tidal constituent (23.93 hours). This near-collinearity means that naïve tidal analysis on a short record risks measuring the diurnal environmental cycle rather than a physical tidal response, an aliasing problem that has not, to our knowledge, been formally characterised or addressed in the seismological literature. Without a framework to separate the two, short-record analyses cannot determine whether an observed correlation between tidal height and seismicity rate is physical or spurious. A previous deployment on Astrolabe Glacier (Le Bris et al., 2025) identified a tidal phase signal but, suspecting that wind interfered with seismic detection, restricted their analysis to low-wind periods rather than systematically characterising the environmental confounders or quantifying how much of the observed signal could be attributed to aliasing rather than tidal forcing.

The January 2024 SeisAdelice experiment on Astrolabe Glacier, Adélie Land, East Antarctica served as the development and testing site for a deconfounding framework targeting this problem. The network comprised 37 three-component seismic nodes concentrated within a 2 × 0.6 km strip across the glacier's grounding line (~150 m spacing), supplemented by two surface fiber arrays totalling 4 km (linear and z-configuration, 500 Hz, 2.4 m channel spacing). Analysis of the resulting icequake catalog confirms the severity of the aliasing: wind speed, air temperature, and geometric solar elevation together explain the dominant fraction of hourly seismicity variance and are strongly collinear with K1 on the 20-day record, so naïve correlation with tidal height gives a spurious result. We expect to have validated the framework against additional datasets by the time of presentation.

The framework combines three complementary approaches: an environmental Poisson GLM that quantifies the relative contributions of wind, temperature, solar forcing, and tide to icequake rate; tidal phase analysis stratified by tidal regime to isolate the semi-diurnal M2 component; and per-station phase gradient analysis, which is immune to uniform detectability bias and provides spatially coherent corroboration. Preliminary results from Astrolabe confirm that the framework recovers a genuine tidal signal — seismicity preferentially elevated on the falling and low tide — that naïve correlation either mischaracterises or obscures entirely.

The framework makes rigorous tidal analysis tractable for short deployments in tidally forced coastal environments without requiring the multi-year records that would be needed to separate K1 from the solar day spectrally, a practical constraint that makes remote polar fieldwork particularly vulnerable to this aliasing problem.