Differences in MOPITT surface level CO retrievals and trends from Level 2 and Level 3 products in coastal grid boxes
Bibliographic record
Abstract
Abstract. Users of MOPITT (Measurement of Pollution in the Troposphere) data are advised to discard retrievals performed over water from analyses. This is because MOPITT retrievals are more sensitive to near-surface CO when performed over land than water, meaning that they have a greater measurement component and are less tied to the a priori CO concentrations (which are taken from a model climatology) that are necessarily used in their retrieval. MOPITT Level 3 (L3) products are a 1∘ × 1∘ gridded average of finer-resolution (∼ 22 × 22 km) Level 2 (L2) retrievals. In the case of coastal L3 grid boxes, L2 retrievals performed over both land and water may be averaged together to create the L3 product, with L2 retrievals over land not contributing to the average at all in certain situations. This conflicts with data usage recommendations. The aim of this paper is to highlight the consequences that this has on surface level retrievals and their temporal trends in “as-downloaded” L3 data (L3O), by comparing them to those obtained if only the L2 retrievals performed over land are averaged to create the L3 product (L3L), for all identified coastal L3 MOPITT grid boxes. First, the difference between surface level retrievals in L3L and the corresponding L2 retrievals performed over water (L3W) is established for days when they are averaged together to create the L3O product for coastal grid boxes (yielding an L3O surface index of “mixed”, L3OM). Mean retrieved volume mixing ratios (VMRs) in L3L differ by over 10 ppbv from those in L3W, and temporal trends detected in L3L are between 0.28 and 0.43 ppbv yr−1 stronger than in L3W, on average. These L3L − L3W differences are clearly linked to retrieval sensitivity differences, with L3W being more heavily tied to the a priori CO profiles used in the retrieval, which are a model-derived monthly mean climatology that, by definition, has no trend year to year. VMRs in the resulting L3OM are significantly different to L3L for 45 % of all coastal grid boxes, corresponding to 75 % of grid boxes where the L3L − L3W difference is also significant. Just under half of the grid boxes that featured a significant L3L − L3W trend difference also see trends differing significantly between L3L and L3OM. Factors that determine whether L3OM and L3L differ significantly include the proportion of the surface covered by land/water and the magnitude of land–water contrast in retrieval sensitivity. Comparing the full L3O dataset to L3L, it is shown that if L3O is filtered so that only retrievals over land (L3OL) are analysed – as recommended – there is a huge loss of days with data for coastal grid boxes. This is because L2 retrievals over land are routinely discarded during the L3O creation process for these grid boxes. There is less data loss if L3OM retrievals are also retained, but the resulting L3O “land or mixed” (L3OLM) subset still has fewer data days than L3L for 61 % of coastal grid boxes. As shown, these additional days with data feature some influence from retrievals made over water, demonstrably affecting mean VMRs and their trends. Coastal L3 grid boxes contain 33 of the 100 largest coastal cities in the world, by population. Focusing on the L3 grid boxes containing these cities, it is shown that mean VMRs in L3OL and L3L differ significantly for 11 of the 27 grid boxes that can be compared (there are no L3OL data for 6 of the grid boxes studied), with 9 of the 18 grid boxes where temporal trend analysis can be performed in L3OL featuring a trend that is significantly different to that in L3L. These differences are a direct result of the data loss in L3OL – data that are available in L2 data (and are incorporated into the L3L product created for this study). The L3L − L3OLM mean VMR difference exceeds 10 (22) ppbv for 11 (3) of these 33 grid boxes, significant in 13 cases, with significant temporal trend differences in 5 cases. It is concluded that an L3 product based only on L2 retrievals over land – the L3L product analysed in this paper, available for public download – could be of benefit to MOPITT data users.
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 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 | 0.001 |
| 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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, 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".