CNN for image super-resolution of airborne magnetic data in Ontario, Canada
Bibliographic record
Abstract
Aeromagnetic surveys have long been a cost-effective tool in mineral exploration, providing data for identifying geological features that are critical for the formation of ore deposits. Many countries have large portions of their territories covered by legacy low-resolution (LR) aeromagnetic surveys, with line spacing varying from 500 m to a few km. In recent years, technological advancements have improved navigation tools, as well as airborne magnetic data acquisition systems, facilitating the acquisition of high-resolution (HR) aeromagnetic data, with line spacing of 200 m or less. These HR maps, however, often cover smaller areas due to the necessity of closer flight paths and associated higher costs. Although older surveys generally offer more extensive coverage, their lower resolution creates difficulties for geological interpretation. To overcome this dilemma, we developed a super-resolution network architecture making use of sub-pixel convolution techniques capable of converting LR to HR aeromagnetic data. Our training and predicting pipeline differ from what is commonly used in aeromagnetic convolutional neural networks applications in two main aspects. First, it samples training data from the full maps, accommodating missing values in the process. Second, being fully convolutional, its capability to generate predictions for data of different sizes than those used during training is only constrained by hardware capacity. We experimented the network using LR (pixel size of 150 m) and HR (50 m) aeromagnetic data acquired over the Ontario province, Canada, evaluating its performance using the Peak Signal to Noise Ratio (PSNR) and R2. We found our architecture to be easy to train, providing robust results across a variety of loss functions. However, it showed weakness in recovering highfrequency components of the HR data. Compared to bicubic interpolation, our approach consistently shows better PSNR (up to +0.52 on the test set) and R2 (up to +0.03) values, as well maps with higher resolution.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.005 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".