Detection and characterization of Earth’s interior signals in geodetic data using blind source separation methods

Summary

We assess the potential of blind source separation (BSS) methods to recover signals originating from Earth’s interior, using synthetic tests and satellite gravimetry data. Seven BSS methods are evaluated in terms of their ability to detect and characterize low-amplitude interior signals embedded in dominant noise and climate-induced variability. Our results show that detection is generally feasible even when interior signals have small relative amplitudes; in this context, classical Principal Component Analysis (PCA) provides the most reliable detection performance. However, detailed characterization of the spatial and temporal structure of the interior signal proves more challenging. Independent Component Analysis (ICA) offers improved characterization compared to other methods, but only when the interior signal’s variability is comparable to that of climate signals. We then apply a two-step screening process to GRACE satellite gravity data. First, PCA is applied independently at each grid point across Earth’s surface, successfully retrieving major climate signals previously identified in the literature. Second, we use PCA on an earthquake catalog spanning the GRACE observation period and find that earthquakes with magnitudes below 8.5 can be detected. However, the correlation between the retrieved components and earthquake-induced gravity signals remains too weak to support meaningful geophysical interpretation. These results highlight both the promise and the current limitations of BSS methods for investigating Earth’s internal processes using space-based gravimetry.