Data and code for: "Arctic observatories constrain the North Magnetic Pole slowdown to declining geometric sensitivity"

Derived datasets and analysis/figure-generation code supporting the Geophysical Research Letters Letter of the same title. The work formulates the north magnetic (dip) pole as a moving horizontal- field null and uses the local null-velocity operator v = −J⁻¹∂ₜh (evaluated in the local horizontal components (Bθ,Bφ), verified against the analytic degree-1 dipole limit) to factor the speed into a secular-variation forcing Φ = |∂ₜh| and a geometric sensitivity S = v/Φ. It shows (i) that the local operator reproduces the pole velocity (magnitude ratio ≈ 1) across IGRF-14, COV-OBS.x2 and CHAOS-8; (ii) that the record 2010s deceleration is not a weakening of the forcing — the forcing held steady through 2024, confirmed model-independently from polar magnetic observatories — but a decline in the pole's geometric sensitivity; (iii) a quadrupole-dominated, axial-negligible (~1%) per-degree velocity budget; (iv) that the north–south pole-speed asymmetry is a sensitivity asymmetry; and (v) a falsifiable forecast of continued slowing to ~26 km/yr by 2030, consistent with the official IGRF-14 projection (Beggan et al., 2026) while supplying the attribution behind it; rolling-origin hindcasts (declining-S MAE 2.4 km/yr vs fixed-S 14–19 km/yr) calibrate the prediction band. This record contains: the derived polar-observatory annual means (Kyoto WDC definitive to 2020 and INTERMAGNET/BGS provisional 2021–2024), the analysis scripts that compute every reported number, the figure-generation code, the rendered figures, and the derived result tables (per-model series, bootstrap CIs, window stability, stiffness transect, era scan, hindcast, per-degree budget, observatory ratios) as CSV under data/derived/. The global field models (IGRF-14, COV-OBS.x2, CHAOS-8) and the raw observatory data are openly available from their providers (see README) and are not redistributed here.