Development and Calibration of Seasonal Probabilistic Forecasts of Ice-Free Dates and Freeze-Up Dates
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Abstract Dynamical forecasting systems are being used to skillfully predict deterministic ice-free and freeze-up date events in the Arctic. This paper extends such forecasts to a probabilistic framework and tests two calibration models to correct systematic biases and improve the statistical reliability of the event dates: trend-adjusted quantile mapping (TAQM) and nonhomogeneous censored Gaussian regression (NCGR). TAQM is a probability distribution mapping method that corrects the forecast for climatological biases, whereas NCGR relates the calibrated parametric forecast distribution to the raw ensemble forecast through a regression model framework. For NCGR, the observed event trend and ensemble-mean event date are used to predict the central tendency of the predictive distribution. For modeling forecast uncertainty, we find that the ensemble-mean event date, which is related to forecast lead time, performs better than the ensemble variance itself. Using a multidecadal hindcast record from the Canadian Seasonal to Interannual Prediction System (CanSIPS), TAQM and NCGR are applied to produce categorical forecasts quantifying the probabilities for early, normal, and late ice retreat and advance. While TAQM performs better than adjusting the raw forecast for mean and linear trend bias, NCGR is shown to outperform TAQM in terms of reliability, skill, and an improved tendency for forecast probabilities to be no worse than climatology. Testing various cross-validation setups, we find that NCGR remains useful when shorter hindcast records (~20 years) are available. By applying NCGR to operational forecasts, stakeholders can be more confident in using seasonal forecasts of sea ice event timing for planning purposes.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| 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 it