How to evaluate sediment fingerprinting source apportionments
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 Purpose Evaluating sediment fingerprinting source apportionments with artificial mixtures is crucial for supporting decision-making and advancing modeling approaches. However, artificial mixtures are rarely incorporated into fingerprinting research and guidelines for model testing are currently lacking. Here, we demonstrate how to test source apportionments using laboratory and virtual mixtures by comparing the results from Bayesian and bootstrapped modeling approaches. Materials and methods Laboratory and virtual mixtures ( n = 79) with known source proportions were created with soil samples from two catchments in Fukushima Prefecture, Japan. Soil samples were sieved at 63 µm and analyzed for colorimetric and geochemical parameters. The MixSIAR Bayesian framework and a bootstrapped mixing model (BMM) were used to estimate source contributions to the artificial mixtures. In addition, we proposed and demonstrated the use of multiple evaluation metrics to report on model uncertainty, residual errors, performance, and contingency criteria. Results and discussion Overall, there were negligible differences between source apportionments for the laboratory and virtual mixtures, for both models. The comparison between MixSIAR and BMM illustrated a trade-off between accuracy and precision in the model results. The more certain MixSIAR solutions encompassed a lesser proportion of known source values, whereas the BMM apportionments were markedly less precise. Although model performance declined for mixtures with a single source contributing greater than 0.75 of the material, both models represented the general trends in the mixtures and identified their major sources. Conclusions Virtual mixtures are as robust as laboratory mixtures for assessing fingerprinting mixing models if analytical errors are negligible. We therefore recommend to always include virtual mixtures as part of the model testing process. Additionally, we highlight the value of using evaluation metrics that consider the accuracy and precision of model results, and the importance of reporting uncertainty when modeling source apportionments.
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 | 0.001 | 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