Recovering missing regions of earth magnetic anomaly grid data (EMAG2) using RePaint based on diffusion model
Big Data and Earth System · 2025-06 · 1 (1) · 100004
DOI: 10.1016/j.bdes.2025.100004 ↗Kurz gefasst
Fachartikel in der Zeitschrift Big Data and Earth System, 2025-06. Verfasst von Fangyuan Mao und 2 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 33 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: National Science and Technology Major Project, Mineral Resources, National Natural Science Foundation of China.
Abstract (Originalsprache)
Earth’s magnetic anomaly fields are essential for understanding the planet’s geophysical structure, yet the current high-resolution global geomagnetic anomaly map, EMAG2, exhibits incomplete coverage due to missing publicly released data, limiting research progress. Traditional interpolation methods often fail to capture complex spatial patterns, producing suboptimal results. To address this, we employ RePaint—a diffusion-based inpainting method—in geophysics, leveraging its iterative optimization and mask-irrelevant sampling strategy to robustly reconstruct missing regions. Building on RePaint, we propose a dual-phase completion framework (global completion and local refinement) to efficiently reconstruct large-scale structures while refining fine details, delivering the first complete high-quality EMAG2 dataset. Experiments show our method significantly outperforms traditional interpolation, offering superior accuracy. We release a complete version of the EMAG2 data, along with the pre-trained model and open-source code.
Förderung
- National Science and Technology Major Project
- Mineral Resources
- National Natural Science Foundation of China
Zitieren
APA
Mao, F., Yang, B., & Jin, S. (2025). Recovering missing regions of earth magnetic anomaly grid data (EMAG2) using RePaint based on diffusion model. Big Data and Earth System, 1(1), 100004. https://doi.org/10.1016/j.bdes.2025.100004
Vancouver
Mao F, Yang B, Jin S. Recovering missing regions of earth magnetic anomaly grid data (EMAG2) using RePaint based on diffusion model. Big Data and Earth System. 2025;1(1):100004. doi:10.1016/j.bdes.2025.100004
BibTeX
@article{Mao2025,
author = {Mao, Fangyuan and Yang, Bo and Jin, Shenyao},
title = {Recovering missing regions of earth magnetic anomaly grid data (EMAG2) using RePaint based on diffusion model},
journal = {Big Data and Earth System},
year = {2025},
volume = {1},
number = {1},
pages = {100004},
doi = {10.1016/j.bdes.2025.100004}
}
EUVIMED