Researchers have developed a novel unsupervised method for 3D gravity inversion using depth-aware implicit neural representations. This approach represents the subsurface density volume with multiple neural networks optimized directly from gravity measurements, incorporating physics-based priors without needing labeled density models. Experiments demonstrate superior performance over conventional and baseline neural methods, yielding more accurate and spatially coherent density reconstructions, particularly in separating nearby anomalies and preserving vertical extent. The method shows promise for improving subsurface imaging in scenarios where ground-truth density models are unavailable. AI
IMPACT This research introduces a novel AI-driven approach to improve subsurface imaging accuracy in gravimetry, potentially benefiting geological surveys and resource exploration.
RANK_REASON This is a research paper detailing a new method for 3D gravity inversion using neural networks. [lever_c_demoted from research: ic=1 ai=1.0]
- arXiv
- Fourier Features Let Networks Learn High Frequency Functions in Low Dimensional Domains
- neural field
- peak signal-to-noise ratio
- Structural Similarity Index Measure
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
