Using Remote Sensing, Statistical, and Machine Learning Techniques to Assess Alaskan River Ice Phenology and Thickness
Bibliographic record
Abstract
Climate change is leading to river ice thinning and shorter ice cover durations, posing significant risks to travel safety and ecosystem health. Due to limited in-situ observations in Alaska, models and remote sensing are employed to understand changing river conditions. This study conducts a comparative evaluation of statistical, machine learning, and remote sensing techniques to assess river ice presence and thickness across Alaska and the Yukon River basin. Sentinel-1 synthetic aperture radar data, climate model outputs, and in-situ river ice observations throughout Alaska are used to evaluate the regional applications of these techniques for determining river ice phenology and thickness. Our analysis reveals that ice presence can be accurately identified using Sentinel-1 images and climate data processed through machine learning models, achieving high accuracy across Alaska. Predicting ice break-up and freeze-up with these methods also yields high accuracy, with a root mean square error (RMSE) of 5.3 and 15.0 days, respectively, for machine learning at out-of-sample locations. Statistical, machine learning, and remote sensing techniques each demonstrated similar performance in determining ice thickness, with RMSEs ranging from 18 to 23 cm for out-of-sample years or locations. However, an ensemble of these methods significantly reduced the RMSE to 13 cm. Using the best-performing models, we generated high-resolution estimates of river ice phenology and thickness for major rivers in Alaska. The ensemble river ice thickness methods and the machine learning ice presence model show promise for widespread application in diverse regions, facilitating environmental monitoring and enhancing river ice safety.
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 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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".