Snowpack Stability Prediction with Machine Learning
Bibliographic record
Abstract
Snow avalanches are destructive hazards, impacting life and infrastructure in snow-covered mountain regions around the world. Avalanche warning and management are essential in those regions and involve a diversity of tasks, ranging from numerical modeling of regional avalanche danger, search and rescue of avalanche victims, to artificially triggering avalanches with explosives in a controlled way. Avalanche warning systems rely on in-situ point measurements of snowpack stability in conjunction with numerical models, to predict the avalanche risk at different exposition and height levels. Avalanche teams are performing those in-situ measurements involving various time-intensive tests, such as Rutschblock tests (RB) and manual snow profiles. There are faster measurement approaches, such as using the Snow Micro Pen (SMP), but snowpack stability scores cannot directly be derived from an SMP profile. Machine learning (ML) has supported and improved numerical modeling and forecasting of regional avalanche danger in multiple ways. However, the question if ML can assist in-situ assessments of snowpack stability, remains, to our knowledge, unanswered. In order to retrieve faster and more intuitive assessments, we are proposing to use ML to directly predict snowpack stability scores from SMP profiles. Our ML pipeline 1) is flexible and can be adapted for other measurement devices, 2) allows training to retrieve further physical parameters, and 3) allows the integration of explainability methods. Furthermore, we outline criteria relevant to collecting the data needed for training the ML model. For safe operational use, training on large datasets, uncertainty estimates, and explainability must be addressed. We believe that our approach represents a significant step towards assisting SMP practitioners in the analysis of snowpack stability. We also see the potential for coupling regional avalanche forecasting systems with in-situ snowpack stability estimates further down the line. If done correctly, we believe that ML can be a useful tool to assist in managing in-situ avalanche risk.
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 | 0.005 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 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".