Mismatches Between Geodetic Observations and Geophysical Models

21 concepts9 papers1 proposal

Modern geophysics possesses an extraordinary toolkit for monitoring the Earth's dynamic crust. Satellites track minute shifts in the planetary gravity field, Interferometric Synthetic Aperture Radar (InSAR) maps surface strain from orbit, and Global Navigation Satellite System (GNSS/GPS) networks record millimeter-scale position changes over time.

To turn these observations into physical understanding, geophysicists rely on forward models—mathematical formulations of fault friction, elastic dislocation, and environmental mass loading. Ideally, feeding known parameters into a forward model should yield a simulated dataset that matches real-world observations.

In practice, however, a persistent gap remains. Across seismic cycle modeling, earthquake slip inversion, and environmental loading corrections, observational data systematically diverges from model predictions. This discrepancy does not simply stem from noisy data; rather, it reveals that our deterministic models are often too idealized, smooth, or structurally simplified to capture the multi-scale, chaotic, and regionally variable physics of the real Earth.

Why Fault Slip Models Diverge from Reality

When an earthquake strikes, geophysicists invert geodetic and seismic data to map how much a fault slipped and where. However, different observational lenses often yield conflicting pictures of the same event, and forward models built from these inversions frequently fail to replicate independent datasets.

For example, during the 2011 Tohoku-Oki earthquake, gravity gradient changes observed by the GOCE satellite showed stark differences in amplitude and geographical location when compared to the predictions of forward seismic models. As Ding et al. (2019) observed, these multi-pole discrepancies point to deep-seated complexities in fault slip distribution that current models fail to capture, alongside uncharacterized observation errors or limitations in how satellite sensors respond to rapid, large-scale deformation.

[Seismic Waveform Data] ───> Inversion Model A ──┐
                                                 ├──> Discrepancies in Deep Slip & Amplitudes
[Geodetic Data (InSAR/GPS)] ─> Inversion Model B ──┘

This mismatch between data types is a recurring theme. In their analysis of the 2017 Jiuzhaigou earthquake, Hong et al. (2018) demonstrated that while their source model aligned cleanly with surface geodetic data, it diverged significantly from models derived from seismic waveform data, particularly when estimating slip distribution at deeper fault levels. Similarly, when modeling the aftershocks of the 2024 Wushi earthquake, Lv et al. (2025) found that no standard slip model could perfectly explain the observed surface deformation, highlighting fundamental limitations in our assumptions about fault geometry.

The issue compounds when models attempt to capture post-seismic deformation over time. Standard viscoelastic dislocation models treat the Earth's mantle and crust as simplified, homogeneous layers. Yet, tracking the aftermath of massive events like the 2010 Maule earthquake reveals that these models fall short. As Han et al. (2020) argued, deterministic models fail because they omit critical, concurrent real-world processes such as localized afterslip, ongoing aftershocks, fault locking variations, and heterogeneous mantle viscosity.

The Chaotic Nature of Fault Friction

Why do our deterministic models struggle to replicate real-world fault behavior? The answer lies at the microscopic level of rock friction. For decades, the geophysics community has relied on Rate- and State-Dependent Friction (RSF) laws to model how faults transition from locked stability to unstable, catastrophic sliding.

However, laboratory and numerical experiments show that RSF models are incomplete. By analyzing laboratory-generated earthquakes through the lens of dynamical systems theory, Gualandi et al. (2023) demonstrated that fault zones behave as random attractors governed by deterministic and stochastic chaos. Tiny, subtle environmental stress variations can radically alter earthquake recurrence intervals—a chaotic sensitivity that standard, smooth RSF models cannot replicate.

Deterministic Models (RSF) ───> Predicts: Smooth, Average Macroscopic Behavior
Natural Fault Dynamics     ───> Exhibits: Instantaneous Complexity & Chaotic Attractors

This gap becomes highly apparent when comparing natural slow earthquakes to numerical simulations. Current seismic cycle models, even those configured with numerous degrees of freedom, struggle to capture the true complexity of natural events. In a comparative study, Gualandi et al. (2024) showed that while numerical simulations can successfully mimic the average macroscopic complexity of slow slip events, they completely fail to replicate their instantaneous complexity. The real-world system is inherently less predictable, meaning that deterministic models miss the fine-grained, underlying dynamics that dictate how faults actually slip.

The Challenge of Localized Environmental Loading

The mismatch between models and observations extends beyond earthquakes into the study of environmental loading. The Earth's crust flexes elastically under the shifting weight of the atmosphere, oceans, and terrestrial water storage. Geodesists attempt to model these "environmental loads" to clean GNSS time series and isolate pure tectonic signals.

However, global mass-loading models frequently conflict with local geodetic realities due to unmodeled regional factors. For instance, when comparing GPS-derived crustal deformations in the Eastern Tibetan Plateau against hydrologic loading models from the GRACE satellite, Pan et al. (2019) found strong agreement in vertical motions but poor agreement in horizontal motions. The authors noted that this discrepancy likely stems from unmodeled degree-1 spherical harmonic contributions—mass center shifts that global satellite models struggle to resolve but heavily influence local ground stations.

Global Loading Model ───[Applies Standard Correction]───> GNSS Station
                                                              │
                                    Paradoxical Result: Amplitude Increases!
                                                              │
                                    Reason: Model misses local geology & foundation types

Furthermore, the impact of these loads is highly modulated by local geology and climate, creating extreme regional variability. In a comprehensive analysis across mainland China, Zhang et al. (2023) discovered that removing standard surface loading models paradoxically increased the noise amplitude of GNSS vertical time series in southern and southeastern China. This counterintuitive result proves that generalized load models fail to capture complex local environmental conditions and specific geological foundation types.

A similar regional breakdown occurs in Greenland. While GRACE satellite data effectively explains seasonal vertical crustal movements in the eastern and southern regions of the island, Bian et al. (2023) demonstrated that it fails to account for vertical displacements observed by GPS stations in the north. Other unmodeled localized geophysical processes or environmental factors dominate the northern signal, leaving a gap that global models cannot bridge.

Moving Beyond Simplified Determinism

The recurring discrepancies across these nine distinct studies point to a unified conclusion: the Earth is too heterogeneous and dynamically complex for idealized, smooth, deterministic forward models. Whether it is a fault slip model ignoring deep geometric complexities, an RSF model missing the chaotic nature of friction, or a global mass model ignoring local geological foundations, the data-model mismatch highlights where our physics assumptions are oversimplified.

To close this gap, the next generation of geophysical frameworks must move away from homogeneous, isolated models. Progress will require integrating stochastic elements to capture chaos, accounting for multi-scale regional variations, and combining multiple independent geodetic streams (such as combining gravity gradients with surface strain) to force models to adhere to the messy reality of the natural Earth.

History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_discrepancies_between_satellite_gravity__7b6be904,
  title = {discrepancies between satellite gravity and forward models},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/discrepancies-between-satellite-gravity-and-forward-models/rdr_7b6be904c5024968b018501a77bdf97c},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_discrepancies_between_satellite_gravity__7b6be904_v1,
  title = {discrepancies between satellite gravity and forward models},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/discrepancies-between-satellite-gravity-and-forward-models/rdr_7b6be904c5024968b018501a77bdf97c/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake