URUANA 2026 — WMM vs WMMHR full-vector geometric audit
Frozen center: -16.0593, -46.2585
NOAA grid date in source files: 2026.61644
Altitude: 0.700 km
Gradient convention: derivative per km toward geographic north.
Primary gradient estimate here: 4-latitude linear regression, then interpolation to exact frozen longitude.

At frozen longitude:
WMM gradient vector dB/dN = (3.261565, -1.274607, 3.292682) nT/km
WMMHR gradient vector dB/dN = (3.152388, -1.502944, 2.972344) nT/km
Increment WMMHR-WMM = (-0.109177, -0.228337, -0.320338) nT/km
Increment norm = 0.408257 nT/km
Observed delta F-gradient = 0.165745 nT/km

Angle WMM gradient vs WMMHR gradient = 4.1924 deg
Angle of increment to upward vertical (-Z) = 38.3120 deg
Vertical component / increment norm = 78.4647%
Horizontal component / increment norm = 61.9943%

Projection relative to local field direction:
Radial projection = 0.163493 nT/km
Tangential magnitude = 0.374091 nT/km
Tangential/radial = 2.288
Tangential fraction of squared increment norm = 83.9628%

First-order scalar-F projection by component:
X contribution = -0.084532 nT/km
Y contribution = 0.073456 nT/km
Z contribution = 0.174586 nT/km
X+Y net = -0.011075 nT/km
Z / observed delta-F gradient = 105.334%

Interpretive constraint:
The new X/Y data correct an earlier H-Z-only inference. The WMMHR-WMM gradient increment is NOT
~100% vertical as a 3-D vector. Its vertical component is ~78.5% of its vector magnitude,
and it lies ~38.3 deg from upward vertical. However, in the scalar total-intensity
projection, the X and Y contributions largely cancel, so the Z term alone accounts for ~105.3%
of the observed differential F-gradient. In addition, the increment is predominantly tangential to
the local field direction (tangential amplitude 0.374 vs radial 0.163 nT/km), explaining why
scalar F suppresses much of the directional information.