URUANA DE MINAS — 2011 RAW GRADIENT-INVERSION AUDIT
Date of audit: 2026-08-14

QUESTION
Can the magnetic-gradient inversion be demonstrated directly in measured scalar total-field data, without relying on the 1DV grid, spectral derivatives, or IGRF subtraction?

DATA
Project 3087 original flight-line file Mag-3087.XYZ. Flight lines are N-S, nominally 500 m apart.
Fixed city center: -16.0593, -46.2585. Survey CRS city location used for matched geometry: X=365408.554 m, Y=8224103.341 m.
Frozen target radius: 8 km. Because the 2011 survey begins about 6 km north of the city, the direct raw-line test uses only the actually observed northern cap of r<=8 km.
To limit 10-Hz along-track pseudoreplication, observations were aggregated into 250 m northing bins. A line was retained only if at least 5 bins (>=1 km span) fell inside the observed target cap.

TARGET RESULT
Retained flight lines: 14.
Raw measured total-field northward slope: median -1.649644 nT/km (250 m bin medians).
NOAA/IGRF expected northward slope: median +0.998354 nT/km.
IGRF-reduced residual slope: median -2.672646 nT/km.
Raw-gradient inversion relative to IGRF: 14/14 lines.
Residual opposition to IGRF: 14/14 lines.
Thus the sign inversion is already present in the measured total field before IGRF subtraction and before any 1DV calculation.

MIRRORED NON-OVERLAPPING EDGE CONTROLS
The same 8-km circle geometry was translated in exact 16-km steps along the same southern survey edge, producing non-overlapping mirrored windows with identical boundary geometry.
Target block: raw inversion fraction 1.000; median raw slope -1.649644 nT/km.
Among the 11 non-target mirrored blocks, the largest raw inversion fraction is 0.571, and the most negative median raw slope is -0.107358 nT/km.
Target ranks 1/12 for raw-inversion fraction and 1/12 for most-negative median raw gradient in this mirrored descriptive audit.
This is an exploratory regional rank, not a confirmatory p-value, because this exact statistic was defined after discovery.

INTERPRETATION BOUNDARY
Demonstrated: localized reversal of the N-S spatial gradient of the measured scalar total magnetic field relative to the expected IGRF gradient in the observed part of the frozen 8-km Uruana target.
Not yet demonstrated by this test: direct reversal of the vector Z component. That requires the raw fluxgate fZ channel (ANP file 0032_SAO-FRANCISCO_MED_PROC_00237.ASC) or an independent vector data source.
Do not interpret this result as atmospheric protection or as a gravity anomaly.

ADVERSARIAL 2-D GRADIENT CHECK
A finite-difference 2-D gradient-angle test was also run on the same 250 m x approximately 500 m raw-line lattice. The coordinate-free cosine between the measured raw-field gradient vector and the IGRF gradient vector had median approximately -0.131, with about 52.2% of target cells anti-parallel. A matched control reached about 53.9% anti-parallel. Therefore the full 2-D vector-gradient reversal is NOT discriminating and is not promoted.
The defensible result is directional: the meridional (N-S) spatial derivative of the scalar total field reverses relative to IGRF in the observed target cap. This directional boundary should be explicit in the paper.