On the Reliability of Surface Observations and the Pitfalls of Verification Against Own Analyses
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 Near‐surface observations can suffer from significant representativeness errors, especially for Numerical Weather Prediction (NWP) at lower resolution in global applications. Therefore, in Data Assimilation (DA), many operational centers have long been reluctant to assimilate them (e.g., the European Center for Medium‐range Weather Forecast, ECMWF, started assimilating all 6‐h screen‐level temperature reports only in 2024). For forecast verification, some studies advocate that we should not rely on them and use only verification against our own near‐surface analyses. At Environment and Climate Change Canada (ECCC), both temperature and humidity observations from SYNOPs have been assimilated in our global NWP system for more than two decades and, in June 2024, METARs have been added following some positive impacts found only when comparing forecasts against near‐surface observations. To shed light on the impact of the assimilation of screen‐level observations, in this study we present an evaluation of the impact of removing the assimilation of all screen‐level temperature and humidity observations using various verification references: the NWP forecasts were evaluated against radiosondes and surface observations, independent (ECMWF) analysis, our own analysis and surface analysis. Results show that, despite the lack of a proper estimation of representativeness errors in the DA approach, the assimilation of screen‐level temperature and humidity leads to forecast improvements that can be detected from the verification against independent measurement sources, here radiosondes and ECMWF upper‐air analyses. Verification against own analyses, for both upper‐air and screen‐level variables, led instead to opposite and misleading conclusions. In fact, the removal of assimilated screen‐level temperature and humidity measurements renders the NWP forecast more similar to the own analysis, therefore leading to better scores but detachment from the observed world.
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 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| 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