3 November 2026. Set before extracting or inspecting the three additional 2011 channel values in the selected windows. The v7 regional sign test and the 0.065 nT/km near miss of the relaxed 14 km west control are known. An exploratory audit of sensitivity to processing, without changing or optimizing survey windows.
1. Extract MAGCOR, MAGNIV, MAGMIC alongside MAGBRU and MAGCOM, following the same crop then median procedure on native 3087 records. Exclude the IGRF-reduced anomaly channel as a total field. Verify native identities, coverage and relevance of field definitions in the 2011 contractor report, sections 3.1.2.1–5 and Table 22.
2. For all 23 eligible sectors, compare both OLS and median pairwise (Theil–Sen) slopes applied to each complete substituted intensity profile, in both survey years. Keep the existing two 2008 channels/five widths/four bands rule and the 2011 14 paired windows. Test native coordinates and every 1-degree cell center at the original two epochs. Retain each processing outcome and all competitors. For each method repeat the 2008 test with its own estimator, using the native equal-weight row means. For each native channel, count negative targets, positive northern windows and complete pairs and identify failing lines. Report whether common geographic destinations exist if all processing choices must pass. No significance or noise interval inferred from dependent processing stages.
3. Describe paired 2011 per-window slope changes across total-field products on exactly the same flight bins. Compare their scale with the original relaxed margin but avoid treating channel differences as independent measurement errors or as proof of geological cause. If corrections substantially affect any headline results, qualify those results in the manuscript, letter, highlights and portal metadata.

Follow-up after the native fits and first complete 1-degree scan. The full 2011 raw Theil-Sen result at Uruana has one nearly zero slope. Delete each individual occupied 250 m bin from its own window in turn at Uruana, requiring the original minimum sample count and span; report sign failures, all other lines and bins held fixed. Calculate the same native diagnostics for the diurnal, leveled and microleveled products and for compensation. On the 1-degree grid, repeat the prior most-permissive rule (drop the worst complete 2011 line AND the worst complete 2008 width) for each stage and estimator. Finally, evaluate the all-ten geographic intersection at Uruana's native latitude on the previous 0.25-degree longitude grid, and calculate geographic area proportions on the world grid with the original 28 microtesla classes. These are post hoc checks conditional on a known target.
