IceEB: An ensemble-based method to map river ice type from radar images
Bibliographic record
Abstract
This paper introduces IceEB, i.e., an innovative ensemble-based method that is designed to automate mapping of river ice types using radar imagery. Its goal is the merger of outcomes from three classifiers (IceMAP-R, RIACT, and IceBC) through ensemble-estimation, resulting in a highly performant and fully automated river ice-type map, which is applicable under all meteorological conditions. The first step of our research is the development of a meta-classifier and a confidence estimation index, then we validate our method using ground-truth datasets and finally compare the performance between IceEB and the original classifiers. The anticipated outcome was a map exhibiting superior results compared to individual classifiers. Validation and comparison of IceEB employed six RADARSAT-2 HH-HV C-band images that were selected from historical datasets of Quebec and Alberta rivers (Canada). IceEB integrates RADARSAT-2 satellite imagery, a digital elevation model, and a river mask, undergoing preprocessing tasks before activating the three initial classifiers. The meta-classifier then performs ensemble-based classification, yielding a legend comprised of water, sheet ice and rubble ice. This approach facilitates broad participation in validation data collection, differentiation between ice covers and ice jams, and minimization of assumptions regarding ice formation. We conclude that IceEB successfully combines existing radar remote sensing ice- classification models to create accurate river ice-type maps. IceEB’s ensemble-based approach outperforms individual classifiers, achieving overall accuracy >91 % for each class. Shortcomings of the original classifiers are effectively offset through parallel use, resulting in marked improvements in automation and generalizability across diverse Canadian meteorological conditions.
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 | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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".